Coumarin as attractive Casein Kinase 2 (CK2) inhibitor scaffold: An integrate approach to elucidate the putative binding motif and explain structure-activity relationships

Article abstract of DOI:10.1021/jm070909t

Casein kinase 2 (CK2) is an ubiquitous, essential, and highly pleiotropic protein kinase whose abnormally high constitutive activity is suspected to underlie its pathogenic potential in neoplasia and other diseases. Recently, using different virtual screening approaches, we have identified several novel CK2 inhibitors. In particular, we have discovered that coumarin moiety can be considered an attractive CK2 inhibitor scaffold. In the present work, we have synthetized and tested a small library of coumarins (more than 60), rationalizing the observed structure-activity relationship. Moreover, the most promising inhibitor, 3,8-dibromo-7-hydroxy-4-methylchromen-2-one (DBC), has been also crystallized in complex with CK2, and the experimental binding mode has been used to derive a linear interaction energy (LIE) model.

Full text of DOI:10.1021/jm070909t

752
J. Med. Chem. 2008, 51, 752–759
Coumarin as Attractive Casein Kinase 2 (CK2) Inhibitor Scaffold: An Integrate Approach To
Elucidate the Putative Binding Motif and Explain Structure–Activity Relationships
Adriana Chilin,† Roberto Battistutta,§ Andrea Bortolato,† Giorgio Cozza,†,‡ Samuele Zanatta,† Giorgia Poletto,‡
Marco Mazzorana,§ Giuseppe Zagotto,† Eugenio Uriarte,4 Adriano Guiotto,† Lorenzo A. Pinna,‡ Flavio Meggio,‡ and
Stefano Moro*,†
Dipartimento di Scienze Farmaceutiche, UniVersità di PadoVa, Via Marzolo 5, PadoVa, Italy, Dipartimento di Chimica Biologica, UniVersità di
PadoVa, PadoVa, Italy, Venetian Institute for Molecular Medicine (VIMM) and Dipartimento di Scienze Chimiche, PadoVa, Italy, and
Department of Organic Chemistry Institute of Industrial Pharmacy, Faculty of Pharmacy, UniVersity of Santiago de Compostela, Spain
ReceiVed July 26, 2007
Casein kinase 2 (CK2) is an ubiquitous, essential, and highly pleiotropic protein kinase whose abnormally
high constitutive activity is suspected to underlie its pathogenic potential in neoplasia and other diseases.
Recently, using different virtual screening approaches, we have identified several novel CK2 inhibitors. In
particular, we have discovered that coumarin moiety can be considered an attractive CK2 inhibitor scaffold.
In the present work, we have synthetized and tested a small library of coumarins (more than 60), rationalizing
the observed structure–activity relationship. Moreover, the most promising inhibitor, 3,8-dibromo-7-hydroxy-
4-methylchromen-2-one (DBC), has been also crystallized in complex with CK2, and the experimental binding
mode has been used to derive a linear interaction energy (LIE) model.
Introduction
Casein kinase 2 (CK2)a is probably the most pleiotropic
protein kinase known, with more than 300 protein substrates
already recognized, a feature which might, at least partly,
account for its lack of strict control over catalytic activity.1 Its
catalytic subunits (R or R′) are in fact constitutively active either
with or without the regulatory ꢀ-subunits, which appear to play
a role in targeting and substrate recruiting, rather than controlling
catalytic activity. Although constitutively active CK2 is ubiq-
uitous, essential, and implicated in a wide variety of important
cell functions,2 evidence has been accumulating that its catalytic
subunits may behave as oncogenes,3–6 consistent with the
observation that they display an antiapoptotic effect in prostate
cancer cell lines.7 Actually, they are invariably more abundant
in tumors compared with normal tissues, and their overexpres-
sion causes neoplastic growth in animal and cellular models
presenting alterations in the expression of cellular oncogenes
Figure 1. Chemical structures of known CK2 inhibitors.
or tumor suppressor genes.8 These data, in conjunction with
implemented an in-house molecular database (defined as “MMS-
Database”) in which almost 2500 kinase inhibitor-like com-
pounds are collected for specific virtual screening applications.
Several families of polyphenolic compounds, including a large
class of flavones, flavonols, isoflavones, catechins, anthraquino-
nes, coumarins, and tannic acid derivatives, are represented in
our molecular database. Following some recent successful
reports in discovery of new kinase inhibitors by high-throughput
docking of large collections of compounds,12,13 we have
performed a virtual screening experiment targeting the ATP
binding site of CK2 by browsing the MMS-database. In this
effort, several new derivatives have been reported as potent and
selective CK2 inhibitors such as ellagic acid,14 a naturally
occurring tannic acid derivative, tetrabromocinnamic acid (TB-
CA),15 1,8-dihydroxy-4-nitroxanthen-9-one (MNX), 8-hydroxy-
4-methyl-9-nitrobenzo[g]chromen-2-one (NBC),10 and 3,8-
dibromo-7-hydroxy-4-methylchromen-2-one(DBC),11assummarized
in Figure 1.
the observation that many viruses exploit CK2 as phosphory-
lating agent of proteins essential to their life cycle,1 are raising
interest in CK2 as a potential target for antineoplastic or anti-
infectious drugs.9
In the last few years, we have performed an intensive
screening program, using both conventional and in silico
approaches, with the aim of discovering novel potent and
selective CK2 inhibitors.10,11 In particular, we have recently
* To whom correspondence should be addressed. Mailing address:
Molecular Modeling Section, Department of Pharmaceutical Sciences,
University of Padova, Via Marzolo 5, 35131 Padova, Italy. Tel: +39 049
8275704. Fax: +39 049 827 5366. E-mail: stefano.moro@unipd.it.
†
Dipartimento di Scienze Farmaceutiche, Università di Padova.
Venetian Institute for Molecular Medicine.
Dipartimento di Chimica Biologica, Università di Padova.
University of Santiago de Compostela.
§
‡
4
a Abbreviations: ASA, accessible surface area; CK2, casein kinase 2;
DBC, 3,8-dibromo-7-hydroxy-4-methylchromen-2-one; LIE, linear interac-
tion energy; MMS, molecular modeling section; MNX, 1,8-dihydroxy-4-
nitroxanthen-9-one; NBC, 8-hydroxy-4-methyl-9-nitrobenzo[g]chromen-2-
one; TBCA, tetrabromocinnamic acid; rmsd, root-mean-square deviation;
W, water.
In particular, we focused our attention on the coumarin
derivative DBC because coumarins are natural benzopyrone
derivatives whose members include also flavonoids. Interest-
10.1021/jm070909t CCC: $40.75
2008 American Chemical Society
Published on Web 02/06/2008

Coumarin as AttractiVe CK2 Inhibitor Scaffold
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 4 753
ingly, dietary exposure to benzopyrones is quite significant,
because these compounds are found in vegetables, fruit, seeds,
nuts, coffee, tea, and wine.16 It is estimated that the average
western diet contains approximately 1 g/day of mixed benzopy-
rones.16 It is, therefore, not difficult to see why extensive
research into their pharmacological and therapeutic properties
has been underway over many years.16,17 In particular, coumarin
is a natural substance that has shown antitumor activity in ViVo,
with the effect believed to be due to its metabolites (e.g.,
7-hydroxycoumarin).17 A recent study has shown that 7-hy-
droxycoumarin inhibits the release of cyclin D1, which is
overexpressed in many types of cancer.17 This knowledge may
lead to its use in cancer therapy.
In the present work, we have synthetized and tested a focused
library of coumarin derivatives (more than 60) with the aim of
elucidating the putative binding motif and explaining structure–
activity relationships. In this study, the X-ray diffraction crystal
structure of CK2 in complex with DBC has been determined,
and it was exploited as a starting point for a linear interaction
energy (LIE)18–25 study to rationalize the different free energies
of binding and the key interactions of all coumarin derivatives.
This computational approach is an efficient tool to quantitatively
explore scaffold decorability and to evaluate regions of the
protein active site crucial for ligand binding, as already proven
for a class of CK2 inhibitors such as bromo-benzoimidazole.19
The piperazinesulfonyl-5-bromo derivatives 43 and 44 were
prepared by nucleophilic substitution of the sulfonylchloride
intermediates, obtained by reacting the 5-bromocoumarin with
chlorosulfonic acid at reflux. The piperazinesulfonyl-8-bromo
derivative 45 was obtained in an analogous manner from
8-bromocoumarin.
The remaining compounds (1, 4-7, 9, 10, 13, 15-18, 20,
23-25, 27-30, 32-38, 41, 42, 46-61, 64, 66, and 67) were
already known and were prepared according to literature
methods (see Supporting Information for references). Finally,
compounds 41 and 42 were commercially available.
Crystallography. The use of CK2 R subunit from Zea mays
in cocrystallization experiments has been widely adopted in
recent years due to a better stability of the maize enzyme
compared with the human CK2. This is also supported by its
highly conserved primary structure and by almost identical
catalytic properties toward all CK2 inhibitors tested up to now.
The crystal structure of the complex between the R catalytic
subunit and the ꢀ regulatory one of the human protein has been
determined,26 along with that of the isolated R catalytic subunits
both from human and from maize.27–29 From these structures,
it emerged that the active site of the human protein is practically
identical to that of the maize one (rmCK2). Moreover, it
indicates that the formation of the quaternary complex does not
influence the conformation of the catalytic subunit, so the latter
can be used in design of inhibitors of the holoenzyme. The close
sequence and structural similarities between the Zea mays and
the human R catalytic subunits (which shows 98% sequence
identity with the rat one, with all the variations in the extreme
C-terminal tail) strongly support the choice to work with the
maize enzyme in structural inhibition studies: this enzyme has
a higher tendency to produce good diffracting crystals, and this
is the reason it has been preferred over the human enzyme.
Like the other inhibitors, DBC binds to CK2 inside the ATP-
binding site, in a position similar to those of the anthraquinone
MNA10 and the xanthenone MNX,10 as shown in Figure 2. The
overall structure of the protein is not affected by the binding of
the inhibitor. In the final model, 327 residues out of 332 are
present, from position 2 to 328; the first methionine and the
four C-terminal residues Arg-Thr-Arg-Ala are not visible in the
final electron density. The position of the DBC hydroxyl group
is almost identical to the analogous function in MNX, lying in
an area of positive electrostatic potential near Lys68. This OH
group establishes two hydrogen bonds, one with the amine
function of the Lys68 side chain and another with a water
molecule (W1) that is conserved in all deposited CK2 structures
(Figure 2). This H-bonding network seems to have a crucial
role in the recognition process of all phenol-like CK2 inhibi-
tors.30 DBC does not interact directly with the hinge region;
the closest atom is the bromine ortho to the carbonyl at 3.7 Å
from the Val116 carbonyl.
Results
Focused Coumarin Library. The chemical diversity of our
focused coumarin library is reported in Table 1. Several
derivatives were unknown until now and they have been newly
synthetized as described in the Materials and Methods section.
In particular, bromocoumarin derivatives (19, 21, 26, 31, 62,
63, and 65) were prepared by direct bromination of coumarin
precursors or by appropriate functionalization of bromo-
intermediates. Different bromination schemes were used
according to starting materials and substitution positions.
Bromocoumarins 19 and 21 were synthesized by using N-bro-
mosuccinimide in acetonitrile at reflux, while compounds 26, 31,
62, 63, and 65 were prepared by using bromine in acetic acid at
60 °C.
Iodo or chloro analogs (2, 3, and 22, respectively) were
obtained by direct halogenation at the 8 position of the already
brominated coumarins by using iodine and potassium iodide in
ammonia solution at room temperature or aluminum chloride
at 130 °C.
The acetyl derivative 11 was obtained by esterification of
the phenolic function of DBC (1) with acetic anhydride and
sodium acetate at reflux.
Coumarins carrying a formyl, oxime, or cyano in the 8
position, such as compounds 8, 12, and 14, were prepared
starting from the corresponding 3-bromocoumarin. In this way,
the 8-coumarinaldehyde 14 was synthesized by formylating the
starting coumarin with hexamethylenetetramine in acetic acid
at reflux or microwave irradiating at 120 °C, strongly reducing
the reaction time and more than doubling the yield in compari-
son with classical heating. Coumarin 14 was converted to oxime
derivative 12 by methanolic hydroxylamine at room temperature,
and the oxime group was dehydrated to the cyano group in acetic
anhydride at reflux. The 8-cyanocoumarin 8 was also prepared
directly by reacting compound 14 with hydroxylamine in
N-methylpyrrolidone with microwave irradiation at 140 °C.
The 5-bromo derivatives 39 and 40 were synthesized by
condensing 2-bromo-6-hydroxybenzaldehyde with diethylma-
lonate and then hydrolyzing the ethyl ester in alkaline medium.
Linear Interaction Energy Model. A linear interaction energy
(LIE) approach has been used to evaluate the binding free energy
of this class of CK2 inhibitors following a computational ap-
proached recently proposed by our laboratory.19 It was possible to
include a total of 16 coumarins shown in Table 1 in the training
set to generate the quantitative LIE model (Table 2).
Most probably this is linked to the fact that some particular
chemical features, present only in this subset of molecules, are the
key to achieve a DBC-like binding mode, which was used as
starting point for the LIE calculation. First of all, a hydroxyl group
capable of interacting with the conserved water W1 (see Figure 2)
must be present at the 7 position of the coumarin scaffold. This is
the only crucial electrostatic interaction shown by this class of

754 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 4
Table 1. Inhibition of CK2 by Coumarin Analoguesa
Chilin et al.
a The IC50 activity data represent the means of three independent experiments with SEM never exceeding 15%. The in silico prediction of R7 substituent pKa and the
net charge of the most populated ionic state at physiological pH are shown. b The error estimated on the predicted pKa of the phenolic OH at position 7 is (0.02.

Coumarin as AttractiVe CK2 Inhibitor Scaffold
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 4 755
Figure 2. The top panels shows DBC bound to the CK2 ATP binding site. The hydroxyl function is involved in two hydrogen bonds (at a distance
of 2.8 Å) with Lys68 and the conserved water molecule W1. The top panel shows a superimposition of the two inhibitors DBC (yellow carbon
atoms) and MNX (green carbon atoms). Bromine atoms are shown in magenta. Note the identical position of the OH function of DBC and MNX.
As reference, water molecule W1 and the zone of the hinge region are indicated.
Table 2. Ensemble Average LIE Energy Terms for the Inhibitors Used To Build the Energy Modela
b-f
b-f
U
U
HASAb-f
(kcal/mol)
ASAb-f
(kcal/mol)
ASA+b-f
(kcal/mol)
ASA-b-f
(kcal/mol)
CASA+b-f CASA-b-f
VSAb-f
(kcal/mol)
PASAb-f
(kcal/mol)
elect
vdw
name
(kcal/mol)
(kcal/mol)
(kcal/mol)
(kcal/mol)
1 DBC -197.7653
1.5702
4.4171
1.5011
18.9818
2.2074
5.0361
-0.4217
4.3391
3.4665
13.3805
2.0311
3.4807
6.5256
3.7350
3.6139
12.6623
-448.1037 -557.2585 -100.0504 -288.9721
-451.7560 -564.7581 -100.9600 -294.8176
-430.4421 -540.7503 -100.0653 -273.8589
-74.2579
-76.9676
-59.2773
868.0718
-187.3542
2885.0308
2806.5571 -109.1548
2543.4438 -113.0021
1206.3640 -110.3082
2
-193.7874
-215.5250
-220.7834
-167.8966
-169.0103
-219.6383
-231.1221
-182.6900
-229.2684
-212.7549
-192.0742
-197.6469
-218.6269
-191.6230
-202.7459
b-f
3
4
5
6
8
9
10
12
14
17
20
23
25
27
-447.4083 -557.1554
-89.2716 -303.8774 -103.1900
4220.9419 -1929.1779 -109.7472
-397.6865 -512.5418 -113.6143 -233.5788
-398.7990 -516.8976 -113.9363 -237.4862
-374.5649 -553.6124 -126.7920 -260.2887
-414.0914 -570.3610 -109.6718 -292.0350
-379.5748 -494.7814 -113.7104 -217.3335
-381.2005 -574.2347 -160.1861 -248.5138
-388.4405 -543.1951 -145.6445 -230.2745
-373.1952 -560.5029 -101.5250 -291.5494
-319.2815 -513.6910 -112.6414 -237.0944
-320.2022 -503.8472 -139.2490 -199.9986
-64.8164
-66.6977
-60.4464
-89.8760
-50.6239
-88.1841
-61.6509
-66.2913
-56.6301
-52.0343
-46.7456
4264.6226
3336.0249
4194.0859
2417.8528
6509.5933
2366.2527 -114.8553
2459.3225 -118.0987
2781.6345 -179.0476
3626.5566 -156.2696
962.4318 -115.2067
7457.0674 -1039.3051 -193.0342
3919.6196
5628.3013
9666.1660
8096.2407
6553.1392
5357.7446 -154.7546
4908.8560 -187.3077
3728.5420 -194.4096
3481.0146 -183.6451
1366.0836 -114.6586
-356.0089
-70.6675 -122.9562 -235.5503
-332.0560 -528.6724 -173.7345 -192.0848
-78.5291 10961.5870 -1002.8069 -196.6165
a
b-f
U
, U , HASAb-f, ASAb-f, ASA+b-f, ASA-b-f, CASA+b-f, CASA-b-f, VSAb-f, and PASAb-f are, respectively, electrostatic, van der Waals,
elect
vdw
total hydrophobic surface area, water accessible surface area, positive accessible surface area, negative accessible surface area, charge-weighted positive
surface area, charge-weighted negative surface area, van der Waals surface area, and total polar surface area energy terms. All the energy terms are the result
b-f
b-f
of the subtraction from the bound state energy estimation of the free state energy estimation. Together with U
and U , the HASAb-f surface has been
elect
vdw
chosen to build the final energy model.
inhibitors, and it seems to be strongly modulated in its intensity
by the presence of an electron-withdrawing substituent at the 6
position of the coumarin scaffold. According to the Hammett
theory, the presence of an electron-withdrawing substituent in the
ortho position to phenol OH increases the acidic behavior of the
phenolic group. To simulate this highly polarized hydrogen
bonding, we created two possible working hypothesis: (I) in the
first case, the hydroxyl group in 7 becomes anionic before forming
the complex with CK2, and it is considered with a formal net charge
equal to -1; (II) in the second assumption, phenol OH participates
at the active site in its neutral form, and the hydrogen bonding
with water W1 is so highly polarized that can be more suitably
modeled as an ionic couple between a hydronium ion (water W1
is transformed into H3O+) and the anionic form of the phenol OH
(Figure 2). Both these possibilities are in accordance with the fact
that only ligands with a low pKa, estimated using the ACDLabs31
package, can be studied using the LIE approach considering a
DBC-like binding mode. Both hypotheses resulted in high-quality
and roughly comparable energy models, the best one presented
here is derived by in situ ionic couple theory. Finally, almost all
possess hydrophobic groups at the 8 position of the coumarin
moiety, a hydrogen atom in position 5, and a nonhydrophobic group
at the 6 position.
The calculated energy terms (Table 2) result in a final model
with R ) 0.029, ꢀ ) -0.003, and γ ) 0.016. The obtained
correlation factor is r2 ) 0.771 with a rmsd of 0.303 kcal/mol,
and the cross-validated q2 is slightly lower (q2 ) 0.633, cross-
validated rmsd ) 0.491 kcal/mol). Free energies of binding,
19
calculated from experimental IC50 and predicted ∆Gbind are
shown in Table 3 and finally plotted in Chart 1.
Discussion
Qualitative Structure–Activity Relationship (SAR). Several
important structural features of coumarin derivatives can be
identified from a first qualitative analysis of their activity on
CK2 (Table 1). Primarily, the hydroxyl group at the 7 position
of the coumarin moiety can be considered an essential feature
even if not sufficient to achieve an IC50 in a submicromolar
range. Indeed, to increase the inhibitory potency, an electron-
withdrawing substituent should be simultaneously present at
position 8. For instance, in derivative 5 (IC50 ) 0.74 µM), a
simple substitution of the bromide atom with a methoxy group
(derivative 50) results in a complete loss of activity. This
important effect can be rationalized according to the increase
of the acidity of the 7-hydroxyl group as confirmed by our in
silico predictions, where to achieve a submicromolar activity,

756 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 4
Chilin et al.
Table 3. Comparison of Experimental (∆Gexp, kcal/mol), Predicted
(∆Gpred, kcal/mol) and Cross-Validated (∆Gcv, kcal/mol) Free Energies
of Binding for the 16 Inhibitors Included in the Model
relationship of coumarin derivatives from a more quantitative point
of view. Using the linear interaction energy method, it has been
possible to build up a quantitative model using 16 coumarin
derivatives and including in this training set all submicromolar
inhibitors (Figure 3). The final model is statistically acceptable.
This computational technique allows us to evaluate the
principal contributions for inhibitor activity. The strongest effect
results from van der Waals interactions (R ) 0.029), in
agreement with the steric complementarily between the CK2
active pocket and almost all coumarin derivatives studied. This
is followed in absolute importance by the hydrophobic effect
(γ ) 0.016), constituting the most variable part among the test
set. Indeed this aspect is the most important to discriminate the
relative activity of coumarins studied, as shown by the γ
normalized value of 0.854. In contrast, the very low absolute
weight of the electrostatic contribution (ꢀ ) -0.003) is linked
to the fact that the strong interaction with water molecule W1
is common in all inhibitors. For this reason, even if this
interaction is crucial for the inhibitor orientation in the active
pocket, it is not important to discriminate the activity difference
inside the series studied.
In particular, the nonpolar contribution to the free energy of
solvation in a LIE study is represented by the difference between
the solvent-accessible surface area of the ligand in complex and
that of the ligand free in solution (ASAb-f). Since this
contribution seems crucial for the binding of CK2 inhibitors,2,28
we tried to include it in our LIE model also using alternative
types of solvent-accessible surface, such as the positive and
negative accessible surface area (respectively, ASA+b-f and
ASA-b-f), the charge-weighted positive and negative surface
area (respectively, CASA+b-f and CASA-b-f), the van der
Waals surface area (VSAb-f), the total polar surface area
(PASAb-f), and the total hydrophobic surface area (HASAb-f) as
collected in Table 2. Interestingly, using only the total hydrophobic
surface area (HASAb-f), it is possible to obtain an acceptable LIE
model, pointing out the important role of the hydrophobic contribu-
tion in the final free energy of binding. This is a clear indication
of the major role played by the several nonpolar amino acid side
chains that characterize the CK2 active pocket.
name
∆Gexp (kcal/mol)
∆Gpred (kcal/mol)
∆Gcv (kcal/mol)
1 DBC
2
3
4
5
6
8
9
10
12
14
17
20
23
25
27
-9.61
-9.00
-8.92
-8.48
-8.42
-8.37
-7.85
-7.82
-7.77
-7.69
-7.52
-7.41
-7.41
-7.38
-6.82
-6.45
-9.04
-9.03
-8.69
-8.44
-8.30
-8.23
-7.82
-8.29
-7.92
-7.49
-8.00
-7.78
-6.79
-6.81
-7.50
-6.80
-8.87
-9.04
-8.63
-8.37
-8.25
-8.18
-7.81
-8.45
-7.94
-7.41
-8.07
-7.82
-6.59
-6.57
-7.59
-6.97
it is necessary that the pKa of the 7-hydroxyl group is lower
than 7. This hypothesis is also supported by the experimental
evidence that the substitution on DBC of the 7-hydroxyl group
with the 7-amine group results again in a complete loss of
activity (see derivative 62).
Hydrophobic groups at positions 3 and 4, such as bromide
or methyl, improve the activity, while substitutions at the 5
position result in a loss of inhibition efficiency. Halogen or nitro
substituents at position 6 can be compatible with an activity
around 1–4 µM according to the other decorations of coumarin
moiety.
SAR Considering DBC-CK2 Complex Structural Infor-
mation. The crystal structure of DBC in complex with CK2
nicely explains the pharmacophore derived by SAR consider-
ations, as summarized in Figure 3.
In fact, hydrophobic groups in positions 3, 4, and 8 can
interact through van der Waals interactions with the CK2 active
site, while the importance of the 7-hydroxyl group is explained
by its key interactions with Lys68 and water W1, a water
molecule very well conserved in all CK2-inhibitor complexes
available. These represent the only electrostatic interactions
exploited by DBC in the binding to CK2, and most probably
all other coumarin derivatives must possess this ionizable
hydroxyl group to achieve a similar binding mode.
However, considering the strong hydrophobic nature of the
CK2 ATP-binding pocket, the crucial role of both hydrophobic
interactions and desolvation effects in ligand binding is doubt-
less. It is most likely that when the hydrophobic contribution
to the binding becomes sufficiently strong, it is possible to have
active compounds that could present a binding mode different
from DBC as seen for the dibromobenzo imidazole and triazole
derivatives.19 In fact, crystal structure analysis and molecular
modeling studies have demonstrated that these scaffolds can
bind the CK2 active pocket into two different orientations. When
the inhibitor can interact through a polarized hydrogen bonding
with the water molecule, like tetrabromobenzotriazole (TBB)
derivatives, the position of the planar aromatic scaffold is similar
to the one presented by the DBC-CK2 crystal structure; in
contrast, the ligand can be shifted and rotated nearer the hinge
region, like already observed in K44, K37, K25, or K17 crystal
structures.32 In other words, even if DBC and TBB present
different chemical structures, their ATP-binding pocket recogni-
tions are comparable, and water molecules seem to adapt the
protein to the ligand peculiarities.
The final energy model derived was able to predict almost
all inhibitors with a hydrophobic group at the 8 position and
with the pKa value of the 7-hydroxyl group lower than 7 units.
A possible reason can be related to the peculiarity of CK2 ATP
binding site to allow different binding modes, even among similar
inhibitors, as demonstrated by several structural data available.
Indeed these chemical-physical requisites of coumarins seem to
discriminate between a DBC-like binding mode, used as starting
point for the LIE study, and other possible ligand orientations.
Conclusion
A CK2 inhibitor design strategy that started more than 15 years
ago with the first ribofuranosyl-benzimidazole derivatives33 has
been continuously evolving with a progressive development of
many submicromolar inhibitors, constantly improving the structural
knowledge of their activity. Results presented here can be
considered a new important tessera in the always more complete
CK2 mosaic. The multidisciplinary approach used in this work
connects in a virtual circle chemical synthesis, biological assays,
structural information, and different molecular modeling techniques
to rationally design new CK2 inhibitors. More than 60 coumarin
derivatives synthesized and tested represent a strong base to
analyze, propose, and corroborate hypotheses about essential
inhibitor features for CK2 binding. In this particular effort, the
crystallization of the DBC-CK2 complex represents a very
Quantitative Free Energy of Binding Model. Starting from
these qualitative considerations, we analyzed the structure–activity

Coumarin as AttractiVe CK2 Inhibitor Scaffold
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 4 757
Chart 1. Free Energy of Binding Estimated by the LIE Model of Each Inhibitor Used versus the Experimentally Measured Dataa
a On the bottom, the correlation coefficient (r2) and cross-validated coefficient (q2) are shown. For the whole set, rmsd and cross-validated rmsd (rmsdcv)
were calculated. R, ꢀ, and γ are the resulting coefficients of the LIE equation for CK2 score.
Figure 3. The 16 inhibitors used to build the LIE energy model using the DBC binding mode as starting position for the computational study are
shown in the upper left. The more the scaffold satisfies the base pharmacophore shown on the upper right (the green spheres represent hydrophobic
molecular regions, the small red region stands for a strongly polarized hydrogen bonding donor), the higher is the probably that it will possess the
chemical features to achieve a DBC similar binding mode in the CK2 ATP binding pocket.
important improvement in mapping the most crucial chemical
features into CK2 recognition processes. Full understanding of
principal binding features is the first step in the inhibitor design
process based on biostructural information. Finally, starting from
this new crystallographic information, a linear interaction energy
method has been used as an efficient computational approach to
understand the importance of different energy contributions to the
final free energy of binding.
Materials and Methods
Chemistry. All the details about synthetic methods and analytical
and spectral data for the newly synthesized compounds are

758 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 4
Chilin et al.
Table 4
Phosphorylation Assay. Native CK2 was purified from rat liver,
as previously described.39 CK2 phosphorylation assays was carried
out at 37 °C in the presence of increasing amounts of each inhibitor
in a final volume of 25 µL containing 50 mM, Tris-HCl (pH 7.5),
100 mM NaCl, 12 mM MgCl2, and 0.02 mM [33P]-ATP (500–1000
cpm/pmol), unless otherwise indicated. The phosphorylatable
substrates were the synthetic peptide substrate RRRADDSDDDDD
(100 µM). Reaction started with the addition of the kinase and was
stopped after 10 min by addition of 5 µL of 0.5 M orthophosphoric
acid before spotting aliquots onto phosphocellulose filters. Filters
were washed in 75 mM phosphoric acid (5–10 mL each) four times
and then once in methanol and dried before counting.40 Each
determination was repeated in three independent experiments, and
the means of the resulting values were calculated.
Data Collection Statisticsa
data collection wavelength (Å)
space group
a, b, c (Å)
0.979 709
C2
141.68, 60.25, 44.90
90.0, 103.0, 90.0
49%
1.84 (1.84-1.94)
27 834 (2671)
3.2 (1.5)
9.8 (1.5)
5.7 (32.9)
86.5 (57.4)
R, ꢀ, γ (deg)
solvent content
max resolution (Å)
independent reflns
multiplicity
1/σ
Rmerge (%)
completeness (%)
Final Model Statistics
Protein and Inhibitors Preparation. Using the software pack-
age Molecular Operating Environment (MOE 2006.08),41 all the
ligands and the X-ray diffraction crystal structure of CK2 in
complex with the ATP competitive inhibitor DBC were prepared
for the linear interaction energy analysis. The water W1 was
included in the protein active site for the molecular modeling study.
This molecule indeed is well-conserved in almost all CK2-inhibitor
complex crystal structures and seems crucial for their binding.
Hydrogens were added and energy minimized using the AMBER99
force field42 until the energy gradient reached 0.05 kcal/mol. This
protocol was exploited for both the neutral and anionic states of
DBC to achieve two possible orientations of the water as donor or
acceptor of hydrogen bonding. Furthermore, to fully understand
the importance of acidity of the 7-hydroxy coumarin derivatives,
the hypothesis of the ionic couple formation in the active site created
by the anionic DBC form and the hydronium ion has been
investigated. All inhibitors were built with MOE9 and energy
minimized using the MMFF94x force field43 until a 0.01 energy
gradient was attained. Their ionic states were evaluated using
ACDLabs suite,32 and as starting pose of the LIE calculation, the
molecules were superimposed to DBC, since all these ligands have
a similar scaffold and it is likely that they share a common binding
mode.
protein atoms
protein residues
water molecules
R/Rfree
2729
327
218
21.4/26.8
28.0
overall mean B value (Å2)
Ramachandran plot statistics (MolProbity validation)37
residues in favored regions
residues in allowed regions
rms on distances (Å)
rms on angles (deg)
98%
100%
0.011
1.27
a Numbers in parentheses refer to the highest resolution bin (1.84–1.94 Å).
described in Supporting Information. For the already known
compounds whose 1H NMR and HRMS data were not yet available
in the literature, these data are reported in Supporting Information,
together with elemental analyses.
Crystal Preparation of Maize CK2r in Complex with
DBC and Data Collection. The recombinant CK2 R-subunit from
Zea mays was expressed in Escherichia coli, isolated, and purified
according to a previously described method.34 Crystals of the CK2
complex with the inhibitor DBC were obtained by cocrystallization
with the sitting drop vapor diffusion technique. A 25 mM inhibitor
solution (100% DMSO) was added to the 8 mg/mL protein stock
solution in the proper amount to have an inhibitor-protein molar
ratio of 3 to 1 and not to exceed a 5% DMSO concentration in the
final protein solution. Each solution was diluted with 2 volumes of
water and allowed to stand for 2 h in ice. Crystallization drops
wereobtainedbymixinga3µLdropofpreincubatedprotein-inhibitor
solution with 1 µL of precipitant solution (10–20% PEG 4000, 0.2
M sodium acetate, 0.1 M Tris, pH 8). Each drop was equilibrated
against 500 µL of the same precipitant solution (20% PEG 4000).
Crystals grew in a few days at 293 K.
Linear Interaction Energy Method Calculation. The LIE
method is based on two different simulations, one describing the
molecule free in solution (Uf) and the other considering the
ligand-protein complex (Ub). In this approach, the inhibitor free
energy of binding to a macromolecule is considered to be linearly
correlated to several energy terms calculated using a molecular
mechanics force field1 as shown in the following formula:
∆Gbind ) R Ub
- Ufvdw + ꢀ Ub
- Ufelec
+
(
)
(
)
X-ray diffraction data were collected at ESRF beamline BM30A
in Grenoble at a temperature of 100 K. Before mounting, crystals
were cryoprotected by a very rapid soaking in type B immersion
oil (Hampton Research, Aliso Viejo, CA). Data were indexed with
MOSFLM35 and then scaled with SCALA from the CCP4 software
package.36
vdw
elec
γ Ub
- Ufcav
(1)
(
)
cav
Brackets indicate that the ensemble average of the energy values
of van der Waals interactions (Uvdw), electrostatic contributions
(Uelec), and the cavity parameter (Ucav)
44 are considered during the
Structure Determination and Refinement. The CK2-DBC
complex crystallizes in space group C2, with one molecule in the
asymmetric unit. To find the best position inside the unit cell, an
initial rigid body transformation on the model of the apoenzyme
was adequate, using the rigid body refinement option of REFMAC
5 (CCP4 suite).36 The refinement was carried out using REFMAC5.
Data collection and final model statistics are reported in Table 4.
The presence of the inhibitors in the active site was clear since the
beginning of the refinement in both Fo - Fc and 2Fo - Fc maps.
The definition files for the inhibitor were created using the monomer
library sketcher routine of CCP4. Refinement was carried out by
alternating automated cycles and manual inspection steps using the
graphic program Coot v.0.2.38 The stereochemistry and geometric
properties of the final model were checked using Coot validate
analysis and Sfcheck and Procheck in CCP4 suite and were adequate
for the resolution.
simulation. Using a training set of molecules with known activity,
we built a semiempirical energy model by fitting the three different
parameter coefficients (R, ꢀ, and γ) to the free energy of binding.19
The LIE energy model was created using the default options of
the MOE-LIE suite.41 Briefly, using the MMFF94x force field and
the surface generalized Born (SGB) implicit solvent,45 we per-
formed a Truncated Newton minimization for ligands in complex
with CK2 and free in solution with a residue-based cutoff distance
of 15 Å, a 0.5 root-mean-square (rms) gradient for convergence,
and a maximum of 500 steps. The energy terms calculated with
this protocol were used to build the binding free energy model.
Acknowledgment. The molecular modeling work coordi-
nated by S.M. has been carried out with financial support from
the University of Padova, Italy. S.M. is also very grateful to
Chemical Computing Group for the scientific and technical
partnership. In particular, we specially thank Dr. Niall English
for LIE implementation into MOE suite.
Data Deposition. The coordinates for the model of CK2 in
complex with DBC has been deposited at the RCSB Protein Data
Bank with ID code 2QC6.

Coumarin as AttractiVe CK2 Inhibitor Scaffold
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 4 759
reductase from Pneumocystis carinii. J. Med. Chem. 2001, 44, 2391–
2402.
(22) Huang, D.; Caflisch, A. Efficient evaluation of binding free energy
using continuum electrostatics solvation. J. Med. Chem. 2004, 47,
5791–5797.
Supporting Information Available: Experimental procedures
and spectral data of all newly synthesized compounds and 1H NMR
and HRMS data of some already known compounds. This material
is available free of charge via the Internet at http://pubs.acs.org.
(23) Zhou, R.; Friesner, R. A.; Ghoshs, A.; Rizzo, R. C.; Jorgensen, W. J.;
Levy, R. M. New linear interaction method for binding affinity
calculations using a continuum solvent model. J. Phys. Chem. B 2001,
105, 10388–10397.
(24) Tounge, B. A.; Reynolds, C. H. Calculation of the binding affinity of
ꢀ-secretase inhibitors using the linear interaction energy method.
J. Med. Chem. 2003, 46, 2074–2082.
(25) Singh, P.; Mhaka, A. M.; Christensen, S. B.; Gray, J. J.; Denmeade,
S. R.; Isaacs, J. T. Applying linear interaction energy method for
rational design of noncompetitive allosteric inhibitors of the sarco-
and endoplasmic reticulum calcium-ATPase. J. Med. Chem. 2005, 48,
3005–3014.
(26) Ermakova, I.; Boldyreff, B.; Issinger, O. G.; Niefind, K. Crystal
structure of a C-terminal deletion mutant of human protein kinase CK2
catalytic subunit. J. Mol. Biol. 2003, 330, 925–934.
References
(1) Meggio, F.; Pinna, L. A. One-thousand-and-one substrates of protein
kinase CK2? FASEB J. 2003, 17, 349–368.
(2) Litchfield, D. W. Protein kinase CK2: Structure, regulation and role
in cellular decisions of life and death. Biochem. J. 2003, 369, 1–15.
(3) Seldin, D. C.; Leder, P. Casein kinase II alpha transgene-induced
murine lymphoma: relation to theileriosis in cattle. Science 1995, 267,
894–897.
(4) Kelliher, M. A.; Seldin, D. C.; Leder, P. Tal-1 induces T cell acute
lymphoblastic leukemia accelerated by casein kinase IIR. EMBO J.
1996, 15, 5160–5166.
(5) Landesman-Bollag, E.; Channavajhala, P. L.; Cardiff, R. D.; Seldin,
D. C. p53 deficiency and misexpression of protein kinase CK2R
collaborate in the development of thymic lymphomas in mice.
Oncogene 1998, 16, 2965–2974.
(6) Orlandini, M.; Semplici, F.; Ferruzzi, R.; Meggio, F.; Pinna, L. A.;
Oliviero, S. Protein kinase CK2R is induced by serum as a delayed
early gene and cooperates with Ha-ras in fibroblast transformation.
J. Biol. Chem. 1998, 273, 21291–21297.
(7) Guo, C.; Yu, S.; Wang, H.; Davis, A. T.; Green, J. E.; Ahmed, K. A
potential role of nuclear matrix-associated protein kinase CK2 in
protection against drug-induced apoptosis in cancer cells. J. Biol.
Chem. 2001, 276, 5992–5999.
(8) Tawfic, S.; Yu, S.; Wang, H.; Faust, R.; Davis, A.; Ahmed, K. Protein
kinase CK2 signal in neoplasia. Histol. Histopathol. 2001, 16, 573–
582.
(9) Unger, G. M.; Davis, A. T.; Slaton, J. W.; Ahmed, K. Protein kinase
CK2 as regulator of cell survival: implications for cancer therapy. Curr.
Cancer Drug. Targets 2004, 4, 77–84.
(27) Niefind, K.; Guerra, B.; Ermakowa, I.; Issinger, O. G. Crystal structure
of human protein kinase CK2: insights into basic properties of the
CK2 holoenzyme. EMBO J. 2001, 20, 5320–5331.
(28) Niefind, K.; Guerra, B.; Pinna, L. A.; Issinger, O. G.; Schomburg, D.
Crystal structure of the catalytic subunit of protein kinase CK2 from
Zea mays at 2.1 A resolution. EMBO J. 1998, 17, 2451–2462.
(29) Pechkova, E.; Zanotti, G.; Nicolini, C. Three-dimensional atomic
structure of a catalytic subunit mutant of human protein kinase CK2.
Acta Crystallogr. Sect. D: Biol. Crystallogr. 2003, 59, 2133–2139.
(30) Battistutta, R.; Mazzorana, M.; Cendron, L.; Bortolato, A.; Sarno, S.;
Kazimierczuk, Z.; Canotti, G.; Moro, S.; Pinna, L. A. Role of water
and a positive electrostatic potential in the binding of protein kinase
CK2 inhibitors. ChemBioChem 2007, 8, 1804–1809.
(31) ACD/pKa, version 10.0, Advanced Chemistry Development, Inc.,
Toronto ON, Canada, www.acdlabs.com, 2006.
(32) Battistutta, R.; De Molier, E.; Sarno, S.; Zanotti, G.; Pinna, L. A.
Structural features underlying selective inhibition of protein kinase
CK2 by ATP site-directed tetrabromo-2-benzotriazole. Protein Sci.
2001, 10, 2200–2206.
(33) Meggio, F.; Shugar, D.; Pinna, L. A. Ribofuranosyl-benzimidazole
derivatives as inhibitors of casein kinase-2 and casein kinase-1. Eur.
J. Biochem. 1990, 187, 89–94.
(10) De Moliner, E.; Moro, S.; Sarno, S.; Zagotto, G.; Zanotti, G.; Pinna,
L. A.; Battistutta, R. Inhibition of protein kinase CK2 by an-
thraquinone-related compounds. A structural insight. J. Biol. Chem.
2003, 278, 1831–1836.
(11) Meggio, F.; Pagano, M. A.; Moro, S.; Zagotto, G.; Ruzzene, M.; Sarno,
S.; Cozza, G.; Bain, J.; Elliott, M.; Deana, A. D.; Brunati, A. M.;
Pinna, L. A. Inhibition of protein kinase CK2 by condensed polyphe-
nolic derivatives. An in vitro and in vivo study. Biochemistry 2004,
43, 12931–12936.
(34) Battistutta, R.; De Moliner, E.; Sarno, S.; Zanotti, G.; and Pinna, L. A.
Protein Sci. 2001, 10, 2200–2206.
(12) Vangrevelinghe, E.; Zimmermann, K.; Schoepfer, J.; Portmann, R.;
Fabbro, D.; Furet, P. Discovery of a potent and selective protein kinase
CK2 inhibitor by high-throughput docking. J. Med. Chem. 2003, 46,
2656–2662.
(13) Toledo-Sherman, L.; Deretey, E.; Slon-Usakiewicz, J. J.; Ng, W.; Dai,
J. R.; Foster, J. E.; Redden, P. R.; Uger, M. D.; Liao, L. C.; Pasternak,
A.; Reid, N. Frontal affinity chromatography with MS detection of
EphB2 tyrosine kinase receptor. 2. Identification of small-molecule
inhibitors via coupling with virtual screening. J. Med. Chem. 2005,
48, 3221–3230.
(14) Cozza, G.; Bovini, P.; Zorzi, E.; Paletto, G.; Pagano, M. A.; Sarno,
S.; Donella-Deana, A.; Zagotto, G.; Rosolen, A.; Pinna, L. A.; Meggio,
F.; Moro, S. Identification of ellagic acid as potent inhibitor of protein
kinase CK2: a successful example of a virtual screening application.
J. Med. Chem. 2006, 49, 2363–2366.
(35) Leslie, A. G. W. In Crystallographic Computing V; Moras, D.,
Podjarny, A. D., Thierry, J. P., Eds.; Oxford University Press, Oxford,
U.K., 1991; pp 27–38.
(36) Collaborative Computational Project, Number 4, The CCP4 suite:
Programs for protein crystallography. Acta Crystallogr. 1994, D50,
760–763.
(37) Lovell, S. C.; Davis, I. W.; Arendall, W. B., 3rd; de Bakker, P. I.;
Word, J. M.; Prisant, M. G.; Richardson, J. S.; Richardson, D. C.
Structure validation by C-alpha Geometry: phi, psi and C-beta
deviation. Proteins: Struct., Funct., Genet. 2003, 50, 437–450.
(38) Emsley, P.; Cowtan, K. Coot: Model-building tools for molecular
graphics. Acta Crystallogr. 2004, D60, 2126–2132.
(39) Meggio, F.; Donella Deana, A.; Pinna, L. Endogenous phosphate
acceptor proteins for rat liver cytosolic casein kinases. J. Biol. Chem.
1981, 256, 11958–11961.
(15) Pagano, M.A.; Paletto, G.; Di Maira, G.; Cozza, G.; Ruzzane, M.;
Sarno, S.; Bain, J.; Elliott, M.; Moro, S.; Zagotto, G.; Meggio, F.;
Pinna, L. A. Tetrabromocinnamic acid (TBCA) and related compounds
represent a new class of specific protein kinase CK2 inhibitors.
ChemBioChem 2007, 8, 129–139.
(16) Borges, F.; Roleira, F.; Milhazes, N.; Santana, L.; Uriarte, E. Simple
coumarins and analogues in medicinal chemistry: Occurrence, synthesis
and biological activity. Curr. Med. Chem. 2005, 12, 887–916.
(17) Lacy, A.; O’Kennedy, R. Studies on coumarins and coumarin-related
compounds to determine their therapeutic role in the treatment of
cancer. Curr. Pharm. Des. 2004, 10, 3797–3811.
(18) Åquist, J.; Medina, C.; Samuelsson, J.-E. A new method for predicting
binding affinity in computer-aided drug design. Protein Eng. 1994, 7,
385–391.
(19) Bortolato, A.; Moro, S. In Silico binding free energy predictability
by using linear interaction energy (LIE) method: Bromo-benzimidazole
CK2 inhibitors as case study. J. Chem. Inf. Model. 2007, 47, 572–
582.
(40) Sarno, S.; Vaglio, P.; Meggio, F.; Issinger, O.-G.; Pinna, L. A. Protein
kinase CK2 mutants defective in substrate recognition. Purification
and kinetic analysis. J. Biol. Chem. 1996, 271, 10595–10601.
(41) MOE (The Molecular Operating Environment), version 2006.08,
software available from Chemical Computing Group Inc., 1010
Sherbrooke Street West, Suite 910, Montreal, Canada H3A 2R7; http://
www.chemcomp.com.
(42) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M.;
Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman,
P. A. A second generation force field for the simulation of proteins,
nucleic acids and organic molecules. J. Am. Chem. Soc. 1995, 117,
5179–5196.
(43) Halgren, T. Merck molecular force field. I. Basis, form, scope,
parameterization, and performance of MMFF94. J. Comput. Chem.
1996, 17, 490–519.
(44) Carlson, A.; Jorgensen, W, L. An extended linear response method
for determining free energies of hydration heather. J. Phys. Chem.
1995, 99, 10667–10673.
(20) Hansson, T.; Åquist, J. Estimation of binding-free energies for HIV
proteinase inhibitors by molecular dynamics simulation. Protein Eng.
1995, 8, 1137–1144.
(21) Graffner-Nordberg, M.; Kolmodin, K.; Aqvist, J.; Queener, S. F.;
Hallberg, A. Design, synthesis, computational prediction, and biologi-
cal evaluation of ester soft drugs as inhibitors of dihydrofolate
(45) Gallicchio, E.; Zhang, L. Y.; Levy, R. M. The SGB/NP hydration
free energy model based on the surface generalized born solvent
reaction field and novel nonpolar hydration free energy estimators.
J. Comput. Chem. 2002, 23, 517–529.
JM070909T