Rational Design of Coumarin Derivatives as CK2 Inhibitors by Improving the Interaction with the Hinge Region

Article abstract of DOI:10.1002/minf.201500091

Design of novel coumarin derivatives as CK2 inhibitors were attempted by targeting the interaction with the hinge region. A set of substituents capable of forming a hydrogen bond or halogen bond with the hinge region were screened in silico, and trifluoromethyl emerges as a promising motif by forming favorable electrostatic interaction and a presumable halogen bond with the hinge region. As proof of concept, three trifluoromethyl derivatives of coumarin were synthesized and tested in vitro. The results indicated that replacement of methyl by trifluoromethyl leads to a modest 5-fold improvement in potency, with the most active compound being 0.4 μM. The newly designed compounds were further screened on one lung cancer cell line A549, showing low micromolar anti-proliferative activity.

Full text of DOI:10.1002/minf.201500091

Communication
www.molinf.com
DOI: 10.1002/minf.201500091
Rational Design of Coumarin Derivatives as CK2 Inhibitors
by Improving the Interaction with the Hinge Region
Na Zhang,*[a] Wen-juan Chen,[a] Yue Zhou,[a] Hongtao Zhao,[b] and Ru-gang Zhong[a]
Abstract: Design of novel coumarin derivatives as CK2 in-
hibitors were attempted by targeting the interaction with
the hinge region. A set of substituents capable of forming
a hydrogen bond or halogen bond with the hinge region
were screened in silico, and trifluoromethyl emerges as
a promising motif by forming favorable electrostatic inter-
action and a presumable halogen bond with the hinge
region. As proof of concept, three trifluoromethyl deriva-
tives of coumarin were synthesized and tested in vitro. The
results indicated that replacement of methyl by trifluoro-
methyl leads to a modest 5-fold improvement in potency,
with the most active compound being 0.4 mM. The newly
designed compounds were further screened on one lung
cancer cell line A549, showing low micromolar anti-prolifer-
ative activity.
Keywords: Protein kinase CK2 · molecular docking · hinge region · cancer
Casein kinase 2 (CK2) is a multifunctional serine/threonine
kinase, which is involved in a broad range of physiological
events.[1] Aberrant activation of CK2 is a key oncogenic
force underlying human tumorigenesis, and pharmacologi-
cal inhibitors of this attractive therapeutic target have
emerged as promising drug candidates. Several series of
CK2 inhibitors have been reported to date,[2–4] with one
compound CX-4945 advanced into Phase II clinical trial.[5]
Among them, 7-hydroxylcoumarins[6] and its derivatives
such as coumestrol[7] and elliagic acid[8] have shown poten-
tial inhibitory activity against CK2.
The majority of kinase inhibitors form one to three hy-
drogen bonds with the so-called hinge region, which is the
sequence segment that connects the N-lobe to the C-lobe
of the kinase domain. Notably, a potent CK2 inhibitor, 3,8-
dibromo-7-hydroxy-4-methylchromen-2-one (DBC) derived
from coumarin, does not form any hydrogen bond or halo-
gen bond with the hinge region of CK2, as shown in the
complex structure (PDB code 2QC6).[6] Rather than forming
a highly conserved hydrogen bond with the ligand, NH of
Val116 central at the hinge region is unfavorably in close
contact with the methyl group of DBC. It is thus hypothe-
sized that a small polar group rather than methyl might
abolish this unfavorable interaction by forming a hydrogen
bond (H-bond) with NH of Val116. Four small groups capa-
ble of functioning as an H-bond acceptor are chosen to re-
place the methyl group. The four groups are hydroxyl, hy-
droxymethyl, acetyl and carboxamide.
As shown in Figure 1, replacement of the methyl group by
OH retains the overall binding mode, but the distance be-
tween the backbone nitrogen of Val116 and the hydroxyl
oxygen of the ligand is about 4.2 Š, beyond the cutoff of
a hydrogen bond. This short distance does not allow for
a water-bridged interaction either. As a consequence, both
NH of Val116 and OH of the ligand were unfavorably desol-
vated. Indeed, this substitution leads to a decrease in activi-
ty by a factor of 300.[6] Interestingly, instead of forming a hy-
drogen bond with NH of Val116 as an acceptor, the hydrox-
ymethyl group forms a hydrogen bond with the backbone
oxygen of Glu114 as a donor. As such, NH of Val116 remains
in bad contact with hydrophobic groups. In comparison,
both acetyl and carboxamide significantly alter the overall
binding mode, and either of them could form a hydrogen
bond with NH of Val116 as expected. A close examination
reveals that the space between NH of Val116 and DBC is
not big enough to accommodate either the acetyl or car-
boxamide without encountering severe van der Waals
clash.
Since it is rather difficult to rescue NH of Val114 by form-
ing a hydrogen bond, we wonder whether it is possible to
improve the electrostatic interaction, e.g., by introducing
[a] N. Zhang, W.-j. Chen, Y. Zhou, R.-g. Zhong
College of Life Science and Bioengineering,
Beijing University of Technology, Beijing 100124, China
Phone: +86-10-67391667
To shed light on the binding mode of the resulting four
putative compounds to CK2, in particular to validate wheth-
er they can rescue NH of Val116 from the unfavorable con-
tact with the hydrophobic group, molecular docking was
carried out using the program LeDock (http://lephar.com).
*e-mail: nanatonglei@bjut.edu.cn
[b] H. Zhao
Lephar Research, Rindçgatan 21, 11558 Stockholm, Sweden
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/minf.201500091.
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Mol. Inf. 2016, 35, 15 – 18 15

Communication
www.molinf.com
Figure 2. Binding mode of DBC (A) and the predicted binding
mode of its 4-CF3 derivative (B) with CK2a. Hydrogen bonds are
shown in blue dashed lines. The unfavorable contact between CH3
of DBC and NH of Val116 is depicted by red dashed line. In con-
trast, the favorable electrostatic interaction and one presumable
halogen bond between CF3 of the DBC derivative and Val116 are
depicted by green dashed lines.
Figure 1. Predicted binding modes of four putative DBC deriva-
tives: (A) 4-hydroxy (B) 4-hydroxymethyl (C) 4-acetyl and (D) 4-car-
boxamide. Hydrogen bonds are shown in blue dashed lines.
a halogen, which has a negative atomic partial charge. Tri-
fluoromethyl is a privileged motif in drugs such as in kinase
inhibitor drugs Nilotinib and Sorafenib, and it has been
shown that replacement of methyl simply by trifluorometh-
yl group can lead to a dramatic improvement in poten-
cy.[9,10] In addition, trifluoromethyl group can greatly in-
crease the metabolic stability of small molecules.[10] Such
features make it very attractive in drug discovery.[9,10] As
shown in Figure 2, replacement of the methyl group in
DBC by trifluoromethyl retains the overall binding mode,
including the water-bridged interaction. One fluorine forms
favorable electrostatic interaction with NH of Val116, and
moreover, it is in close proximity with the backbone
oxygen of Val116. The angle centered at the fluorine is
close to 180 degrees so that the backbone oxygen directly
faces the region of positive electrostatic potential on the
outer side of fluorine, suggesting a weak halogen bonding
interaction.
Following the hypothesis on the putative binding modes
obtained by docking, trifluoromethyl derivatives of three
DBC analogues were synthesized for in vitro validation. The
choice of these three analogues was largely based on their
synthetic tractability. Besides, their bioactivity against CK2
has not been reported yet. According to Scheme 1, three
compounds 1a-1c were synthesized by a two-step reaction.
The trifluoromethyl-substituted compounds (1a-1c) were
evaluated in an enzymatic CK2 inhibition assay,[5] and three
unsubstituted compounds (2a–2c) were purchased for
comparison. The results were summarized in Table 1, and
five compounds except 2a are in the low-micromolar to
nanomolar range. Among them, compound 1c is the most
Scheme 1. Synthetic route of the target compounds.
Table 1. Inhibitory effects of compounds 1a-2c
Compound
R4
R6
R8
CK2
A549
IC50 (mM)
IC50 (mM)
1a
1b
1c
2a
2b
2c
CF3
CF3
CF3
CH3
CH3
CH3
OH
H
H
OH
H
H
H
OH
Cl
H
OH
Cl
104.7Æ2.4
56.2Æ1.2
0.4Æ0.1
31.7Æ1.0
21.8Æ1.1
20.2Æ1.1
38.5Æ2.7
33.5Æ1.4
29.3Æ2.3
588.8Æ6.5
107.2Æ4.2
2.2Æ0.2
active with an IC50 value of 0.4 mM (Fig. 3). From the pure
perspective of ligand efficiency, it may serve as a starting
point for future lead elaboration. Both compounds 1a and
1c show a modest improvement in potency by a factor of
5, in comparison with 2a and 2c, respectively. Compound
1b shows only a two-fold improvement in potency over
compound 2b. Taken together, it indicates that replace-
ment of CH3 by CF3 indeed improves the interaction with
the hinge region. Analysis on the structure-activity relation-
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Mol. Inf. 2016, 35, 15 – 18 16

Communication
www.molinf.com
which is similar to the H-bond between OH of elliagic
acid[11] or Urolithin derivatives[12] with the hinge region. In
contrast, the small CH3 group of compound 2c could not
interact with the hinge region, which is consistent with the
X-ray structure of CK2a in complex with DBC. This may ex-
plain the modest inhibitory activity of compound 1c over
2c.
As shown in Table 1, all of the compounds inhibited cell
growth when tested against CK2 over-expressed A549
cancer cell lines, with the modest IC50 values ranging from
20.2 to 38.5 mM. The CF3-subtituted serials are slightly more
potent than their corresponding unsubstituted ones. Nota-
bly, there is a big discrepancy between the enzymatic po-
tency and cell growth inhibition, presumably due to their
low aqueous solubility and poor cell membrane permeabili-
ty. Further optimization of this series of compounds is on-
going to resolve these problems.
In summary, we screened in silico a small set of substitu-
ents capable of forming hydrogen bond or halogen bond
with the hinge region, with the aim to improve the binding
potency of coumarin derivatives. As proof of concept, three
trifluoromethyl derivatives suggested by computations
were synthesized and tested against CK2. They all showed
a modest improvement in potency.
Supporting Information
Experimental procedures and structural data of synthesized
compound 1a-1c were supplied in the supporting informa-
tion.
Figure 3. Dose response measurement of compound 1c against
CK2 enzymatic activity (A) and A549 cell proliferation (B).
Acknowledgements
This work was financially supported by National Natural Sci-
ence Foundation of China (Grant No. 81402852), Beijing
Natural Science Foundation (Grant No.7152013), and Train-
ing Program for Creative Graduate Students of Beijing Uni-
versity of Technology (Grant No. 015000541115008).
ship reveals that compounds 1c and 2c, bearing a chlorine
atom at the 8-position, exhibited a dramatic improvement
by a factor of 100 over compounds 1b and 2b. This trend
is consistent with the previous findings.[6,11] An electron-
withdrawing substituent at position 8 decreases pKa of the
phenol at position 7, thus strengthening the crucial water-
mediated H-bond with the positive electrostatic region.
The 6-hydroxyl is detrimental to the activity, as it interacts
with the hydrophobic side chain of the gatekeeper residue
Phe. As such, both compounds 1a and 2a are least active.
Since compound 2c is an analogue of DBC, to have
a better understanding on the overall interaction modes
compounds 1c and 2c were docked into CK2, respectively
(Fig. S2). Both compounds were located at the active site of
CK2a and binds primarily through hydrophobic and polar
interactions. The aromatic rings of compound 1c and 2c
embedded into a hydrophobic cavity consisting of Leu45,
Val53, Val66, Ile95, Phe119, Met163, and Ile174. The polar in-
teractions differ. For compound 1c, the CF3 group may
form favorable electrostatic interaction with NH of Val116,
Conflict of Interest
None declared.
References
[1] F. Piazza, S. Manni, M. Ruzzene, L. A. Pinna, C. Gurrieri, G. Se-
menzato, Leukemia 2012, 26, 1174–1179.
[2] G. Cozza, L. A. Pinna, S. Moro, Curr. Med. Chem. 2013, 20, 671–
693.
[3] G. Cozza, A. Bortolato, S. Moro, Med. Res. Rev. 2010, 30, 419–
462.
[4] H. P. Sun, J. Zhu, F. H. Chen, Q. D. You, Mol. Inf. 2011, 30, 579–
592.
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Mol. Inf. 2016, 35, 15 – 18 17

Communication
[5] A. Siddiqui-Jain, D. Drygin, N. Streiner, P. Chua, F. Pierre, S. E.
O’Brien, J. Bliesath, M. Omori, N. Huser, C. Ho, C. Proffitt, M. K.
Schwaebe, D. M. Ryckman, W. G. Rice, K. Anderes, Cancer Res.
2010 ,70, 10288–10298.
[6] A. Chilin, R. Battistutta, A. Bortolato, G. Cozza, S. Zanatta, G.
Poletto, M. Mazzorana, G. Zagotto, E. Uriarte, A. Guiotto, L. A.
Pinna, F. Meggio, S. Moro, J. Med. Chem. 2008, 51, 752–759
[7] S. Liu, D. Hsieh, Y. L. Yang, Z. Xu, C. Peto, D. M. Jablons, L. You,
BMC Pharmacol. Toxicol. 2013, 14, 36.
[8] G. Cozza, P. Bonvini, E. Zorzi, G. Poletto, M. A. Pagano, S.
Sarno, A. Donella-Deana, G. Zagotto, A. Rosolen, L. A. Pinna, F.
Meggio, S. Moro, J. Med. Chem. 2006, 49, 2363–2366.
[9] L. A. Hardegger, B. Kuhn, B. Spinnler, L. Anselm, R. Ecabert, M.
Stihle, B. Gsell, R. Thoma, J. Diez, J. Benz, J. M. Plancher, G.
www.molinf.com
Hartmann, Y. Isshiki, K. Morikami, N. Shimma, W. Haap, D. W.
Banner, F. Diederich, ChemMedChem. 2011, 6, 2048–2054.
[10] H. J. Bohm, D. Banner, S. Bendels, M. Kansy, B. Kuhn, K. Muller,
U. Obst-Sander, M. Stahl, Chembiochem 2004, 5, 637–643.
[11] Y. Sekiguchi, T. Nakaniwa, T. Kinoshita, I. Nakanishi, K. Kitaura,
A. Hirasawa, G. Tsujimoto, T. Tada, Bioorg. Med. Chem. Lett.
2009, 19, 2920–2923.
[12] G. Cozza, A. Gianoncelli, P. Bonvini, E. Zorzi, R. Pasquale, A. Ro-
solen, L.A Pinna, F. Meggio, G. Zagotto, S. Moro, Chem. Med.
Chem. 2011, 6, 2273–2286.
Received: July 22, 2015
Accepted: August 6, 2015
Published online: September 14, 2015
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Mol. Inf. 2016, 35, 15 – 18 18