Molecular and Cellular Biochemistry

, Volume 316, Issue 1, pp 71–85

Expanding the chemical diversity of CK2 inhibitors

Authors

  • Renaud Prudent
    • INSERM, U873
    • CEA, iRTSV/LTS
    • Université Joseph Fourier
  • Virginie Moucadel
    • INSERM, U873
    • CEA, iRTSV/LTS
    • Université Joseph Fourier
  • Miriam López-Ramos
    • Institut CurieCentre de Recherche
    • CNRS, UMR 176
    • Institut Curie, Centre de Recherche, Bâtiment 112Université Paris-Sud
    • Inserm U759, Bâtiment 112Université Paris-Sud
  • Samia Aci
    • CEA, iRTSV/CMBA
    • CEA, iRTSV/LBIM
  • Beatrice Laudet
    • INSERM, U873
    • CEA, iRTSV/LTS
    • Université Joseph Fourier
  • Liliane Mouawad
    • Institut Curie, Centre de Recherche, Bâtiment 112Université Paris-Sud
    • Inserm U759, Bâtiment 112Université Paris-Sud
  • Caroline Barette
    • CEA, iRTSV/CMBA
  • Jacques Einhorn
    • Département de Chimie Moléculaire (SERCO), UMR-5250, ICMG FR-2607CNRS Université Joseph Fourier
  • Cathy Einhorn
    • Département de Chimie Moléculaire (SERCO), UMR-5250, ICMG FR-2607CNRS Université Joseph Fourier
  • Jean-Noel Denis
    • Département de Chimie Moléculaire (SERCO), UMR-5250, ICMG FR-2607CNRS Université Joseph Fourier
  • Gilles Bisson
    • Laboratoire TIMC-IMAG, CNRS/UJF 5525Université de Grenoble Domaine de la Merci
  • Frédéric Schmidt
    • Institut CurieCentre de Recherche
    • CNRS, UMR 176
  • Sylvaine Roy
    • CEA, iRTSV/CMBA
    • CEA, iRTSV/LBIM
  • Laurence Lafanechere
    • CEA, iRTSV/CMBA
  • Jean-Claude Florent
    • Institut CurieCentre de Recherche
    • CNRS, UMR 176
    • INSERM, U873
    • CEA, iRTSV/LTS
    • Université Joseph Fourier
Article

DOI: 10.1007/s11010-008-9828-z

Cite this article as:
Prudent, R., Moucadel, V., López-Ramos, M. et al. Mol Cell Biochem (2008) 316: 71. doi:10.1007/s11010-008-9828-z

Abstract

None of the already described CK2 inhibitors did fulfill the requirements for successful clinical settings. In order to find innovative CK2 inhibitors based on new scaffolds, we have performed a high-throughput screening of diverse chemical libraries. We report here the identification and characterization of several classes of new inhibitors. Whereas some share characteristics of previously known CK2 inhibitors, others are chemically unrelated and may represent new opportunities for the development of better CK2 inhibitors. By combining structure-activity relationships with a docking procedure, we were able to determine the binding mode of these inhibitors. Interestingly, beside the identification of several nanomolar ATP-competitive inhibitors, one class of chemical inhibitors displays a non-ATP competitive mode of inhibition, a feature that suggests that CK2 possess distinct druggable binding sites. For the most promising inhibitors, selectivity profiling was performed. We also provide evidence that some chemical compounds are inhibiting CK2 in living cells. Finally, the collected data allowed us to draw the rules about the chemical requirements for CK2 inhibition both in vitro and in a cellular context.

Keywords

CK2Kinase inhibitorsNon-competitive

Abbreviations

CK2

Casein kinase 2

ATP

Adenosine triphosphate

IQA

IndoloQuinAzolin derivative

TBCA

Tetrabromocinnamic acid

TBB

4,5,6,7-Tetrabromo-1-benzotriazole

DMAT

2-DiMethylAmino-4,5,6,7-tetrabromo-1H-benzimidazole

Introduction

The feasibility of targeting CK2 for cancer therapy has been shown when successful disruption of CK2 activity in vitro with CK2 antisense resulted in induction of cell apoptosis [1, 2]. In addition, it was also found that disruption with the same CK2α antisense oligodeoxynucleotide in xenograft models of prostate cancer induced a dose- and time-dependent tumor cell death [3]. Another way to interfere with CK2 phosphorylation is the use of small molecules to pharmacologically knock-down the kinase activity. Although targeting ATP-binding site of protein kinases had risen some concerns about specificity, it was found to be possible, especially for CK2 where the hydrophobic pocket of the active site of CK2 provide some relatively unique anchoring points [4]. Thus, this pocket has become the principal target of research for new CK2 inhibitors. Various classes of ATP-site-directed inhibitors of CK2 have been reported. Quercetin, Emodin, TBB and IQA are representative of the four main classes of ATP/GTP competitive CK2 inhibitors. Other families were also described like derivatives of hydroxycoumarines (3-carboxy-4(1H)-quinolone) [5], halogenated benzimidazole/triazole [6], pyrazolo[1,5-a][1,3,5]triazine derivatives [7]. Highly potent in vitro CK2 inhibitors were obtained after intensive screening programs, using conventional or in silico approaches, but till today, currently available CK2 inhibitors lack the basic properties required to be suitable for in vivo use such as selectivity [8, 9], cell permeability, metabolic stability, correct pharmacokinetic profile and good inhibitory activity toward CK2 within the cell environment. As CK2 is now considered as a promising target for cancer and viral infection therapies, the need for new classes of CK2 inhibitors satisfying these challenges still remains. Furthermore, development of non ATP-competitive inhibitors is also a promising challenge since such inhibitors do not compete with the millimolar cellular ATP concentration and are thought to be more specific than ATP-competitive inhibitors [10].

In order to find out new innovative structures for CK2 inhibitors, we performed a high-throughput screening of a collection of 5,160 compounds originating from academic and commercial libraries. We identified and biochemically characterized several classes of new CK2 inhibitors with either classical or innovative chemical structure. Among them, eight compounds were found to be active on CK2 activity in living cells and some were able to trigger tumor cell death. Collectively, our data allowed to draw some rules about the physiochemical requirements for designing relevant cell-potent CK2 inhibitors.

Materials and methods

Human recombinant CK2α subunit was expressed in Escherichia coli and purified to homogeneity as previously described [11].

Chemical libraries

Several libraries of low-molecular-weight compounds were used: NCI diversity set (2,100 compounds) and NCI Mechanistic Set (900 Compounds) from the National Cancer Institute (NCI), Bethesda, Maryland; For more information, see http://dtp.nci.nih.gov); the Chemical Library of DCM (UMR 5250 CNRS-UJF, Grenoble) (960 compounds), the BIO-MOL kinase inhibitor library (80 compounds) and the Prestwick Chemical library (1,120 compounds).

High-throughput screening

High-throughput screening was performed at 15 and 1.5 μM in 96-well plates using a TECAN Genesis robot of the CMBA facilities or manually. CK2 activity was assayed for 65 min at room temperature in a final volume of 30 μl containing 10 μl of compounds or DMSO controls, 10 μl of CK2α (50 ng) and 10 μl of a mixture containing the peptide substrate (RRREDEESDDEE), ATP, MgCl2 at 100 μM, 10 μM and 10 mM, respectively. The final DMSO concentration in the assay was 1%. At the end of the reaction, kinase activity was determined by the luminescence-based Kinase-Glo Assay (Promega) according to the manufacturer recommendations. This assay relies on the use of ATP depletion as a read-out for kinase activity.

Hit validation and CK2 radioactive kinase assay

As a second screen, selected compounds were tested in a radioactive assay performed in a final volume of 18 μl containing 3 μl of compounds at the indicated concentrations, 3 μl of CK2α (36 ng) and a mixture containing 100 μM of peptide substrate (RRREDEESDDEE), 10 mM MgCl2, and 100 μM [γ-32P]-ATP. Assays were performed at room temperature for 5 min before termination by the addition of 60 μl of 4% TCA. 32P incorporation into the peptide substrate was determined as previously described [12].

Kinase profiling

Kinase selectivity was performed by Millipore on a panel of 30 recombinant protein kinases. The assays were performed at 100 μM ATP in the presence of 5-μM inhibitor using the Millipore (formerly Upstate) Kinase profiler panel service. Inhibition was calculated from the residual activity measured in the presence of 5 μM inhibitor and was expressed as the percentage of kinase activity determined in the absence of inhibitor.

Cell culture

Hela (cervical adenocarcinoma) and MDA231 (breast adenocarcinoma) cell lines were routinely cultivated in Dulbecco’s medium (Invitrogen Life Technologies, Inc.) while A549 (lung carcinoma), H1299 (lung carcinoma) and LnCaP (prostate carcinoma) cell lines were cultivated in RPMI (Invitrogen Life Technologies, Inc.). MCF7 (breast adenocarcinoma) cell line was cultivated in Dulbecco’s medium with insulin (10 μg ml−1). MCF10A (breast fibrosis) cell line was cultivated in Dulbecco’s medium/F12 supplemented with horse serum 5%, EGF 25 ng.ml−1, hydrocortisone 0.5 μg.ml−1, cholera toxin 100 ng.ml−1, insulin 10 μg.ml−1. All media were supplemented with 10% (v/v) fetal calf serum (FBS, BioWest) except the MCF10A medium.

Cellular CK2 activity assay

HeLa cells were plated at 106 cells/well in 12 well plates and transfected with the CK2 activity reporter plasmid (pEYFPc1-SβS) using the lipofectamine reagent (Invitrogen). pEYFPc1-SβS was obtained from pEYFPc1-CK2β [11] after two subsequent mutations with the Quickchange-Site Directed mutagenesis kit (Stratagene) using for mutagenesis 1: 5′-GCTCAAGCTTCGGATTCTGAAGACGACGATACCGCGGGCCCG-3′ and 5′-CGGGCCCGCGGTATCGTCGTCTTCAGAATCCGAAGCTTGAGC-3′ and for mutagenesis 2: 5′-GCTCTGAGGAGGTGTCCGAGGTCGACTGGTTCTGAGGGCTCCGT-3′ and 5′-CGGGCCCGCGGTATCGTCGTCTTCAGAATCCGAAGCTTGAGC-3′. One day later, the transfected cells were incubated for 24 or 48 h with fresh medium containing the compounds. Then, cells were collected, washed, and the cell pellets were resuspended in 100 μl of lysis buffer (Tris 50 mM pH = 7.4, 0.15 M NaCl, EDTA 2 mM, Triton X100 1/100, phosphatase inhibitor cocktail 2 (Sigma) 1/100, and leupeptine, aprotinine, AEBSF according to the manufacturer’s recommendations) and incubated for 30 min in ice. Fifty micrograms of proteins were loaded with a mix of glycerol and bromophenol blue and analyzed on 12% native-polyacrylamide gel. After electro-transfer on a nitrocellulose membrane, the blot was blocked with 1% BSA and incubated overnight at 4°C with the mAb anti-GFP (Roche, ref. 1814460) at 1/1,000 dilution. After several washings, the blot was incubated for 1 h with a goat anti-mouse-HRP secondary antibody (Sigma, ref. A4416) at 1/2,000 dilution and GFP was revealed with the ECL plus western blotting detection system (GE Healthcare).

Cell viability

Cells were plated into 96-well plates at 104 cells/well. The following day, the culture medium was replaced with fresh medium containing 100 μM inhibitors or 1% DMSO as control. Two days after, cells were washed twice and cell viability was measured with the CellTiter-Glo® Luminescent Cell Viability Assay (Promega) according to the manufacturer’s recommendations.

Docking

The structure of human CK2α was taken from the PDB (1JWH). Only one CK2α subunit was considered with all the crystal waters in its vicinity. The ligand (AMPPNP) and the other subunits were removed. The protein was prepared using the Protein Preparation Tool of the Biopolymer module of Sybyl 7.1 [13]. All Asp, Glu, Lys and Arg side chains were considered ionized, and His neutral.

Docking of compounds was performed with FlexX [14] as implemented in Sybyl. The active site was defined by the following residues: Arg43, Leu45, Gly46, Arg47, Ser51, Glu52, Val53, Val66, Val67, Lys68, Glu81, Ile84, Leu85, Leu88, Arg89, Ile95, Thr96, Leu97, Leu111, Val112, Phe113, Glu114, Hse115, Val116, Asn117, Asn118, Asp156, Lys158, His160, Asn161, Met163, Ile174, Asp175, Trp176. Default values were kept for all settings. Serial CScore calculations with scoring functions ChemScore, D_Score, G_Score and PMF_Score were requested.

Molecular descriptor calculation

Molecular two-dimensional (2D) structures were provided in SDF files. For standardization purpose, all the molecules have been normalized according to a set of normalization rules that we have built in accordance with usual chemical usage. This task has been achieved thanks to the ChemAxon software Application Programmatic Interface (http://www.chemaxon.com) and standardization rules were formally defined as chemical reactions in an XML configuration file read by the ChemAxon Standardizer object (setup file available upon request). The molecular descriptors of compounds were computed using the ChemAxon software and the Chemistry Development Kit (CDK) Java library [15, 16] (http://almost.cubic.uni-koeln.de/cdk/). The descriptors [17] that we have computed here, belong mainly to three groups, either (1) expressing physico-chemical properties (such as LogP or octanol/water partition coefficient which is a measure of molecular hydrophobicity [18]) or TPSA (Topological Polar Surface Area estimation [19]) or (2) expressing structural properties like for instance the bound count (number of bonds in the molecule including hydrogens) or (3) providing topological analyses for instance like Wiener Index [20], Randic Index [21] or the Eccentric Connectivity Index [22], which enable to estimate branching and volume of the compound. In order to find discriminating descriptors, we used the algorithm PART [23] in the machine learning package Weka (http://www.cs.waikato.ac.nz/ml/weka/).

Results

Discovery of new CK2 inhibitors by high-throughput screening

A collection of 5,160 compounds from academic (National Cancer Institute (NCI), DCM (UMR 5250 CNRS-UJF, Grenoble) and commercial (BIOMOL and Prestwick corporations) libraries of small molecules was screened in a high-throughput luminescence-based in vitro kinase assay against human recombinant CK2 catalytic subunit (CK2α) (Fig. 1). This test is based on the measurement of residual ATP after kinase reaction using a luciferase assay. Screening data were normalized and analyzed thanks to a special home-made software PhenoScreen. As a primary screen, CK2 kinase inhibitory activity was determined at a compound concentration of 15 μM. Hits that showed inhibition above 50% at 15 μM, were further screened at 1.5 μM. Active compounds were then re-ordered and confirmed in a secondary independent screen using a radioactive CK2 kinase assay to exclude any false-positive. Several classes of new CK2 inhibitors harboring classical or innovative chemical structures were discovered and their follow-up is described below.
https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Fig1_HTML.gif
Fig. 1

Flow chart for the high-throughput screening of the libraries and the subsequent analysis of active compounds. For chemically unrelated hits CK2 inhibitors activity, CK2 kinase assays were performed in the absence or presence of 10 μM compound. Results are expressed as a percentage of the control activity without inhibitor

Benzothiazole-based inhibitors

The first new class of CK2 inhibitors is composed of molecules built around a benzothiazole scaffold. Compound 1 of this series is a multimeric benzothiazole derivative (Table 1) that is both positively and negatively charged at physiological pH. To our knowledge, the presence of positively charged groups is an uncommon feature in the CK2 inhibitors area since both the natural ligand of the protein, ATP or GTP and the other previously described inhibitors, are only negatively charged. Compound 2, an analogue of compound 1 that is a negatively charged dimer counterpart (for n = 1), appears to be more active than compound 1 at 2 μM. Therefore, positively charged groups are not required for inhibitory potency. Other compounds with the same phenyl-benzotriazole core were present in the screened libraries (compounds 3–8, Table 1). They exhibit different substitution patterns of the phenyl ring, with either negatively charged or neutral substituents. However, none of these compounds was active. This could be related to their smaller size compared to compounds 1 and 2, or to different arrangements of their charged moieties around the benzothiazole core.
Table 1

CK2 inhibitory activity of Benzothiazole derivatives (compounds 19)https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figa_HTML.gif

Compound

R1

R2

R3

R4

R5

CK2 activity (% control)

3

H

H

H

H

H

>95

4

OH

H

OH

H

H

>95

5

OH

H

H

Cl

H

>95

6

COOH

H

H

H

H

>95

7

H

SO3H

NH2

H

H

>95

8

H

SO3H

OMe

H

H

>95

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figb_HTML.gif

CK2 kinase assays were performed in the absence or presence of 2 μM compound. Results are expressed as a percentage of the control activity without inhibitor

Compound 9 is a dimer with a structure very close to that of compound 2, but it contains only two benzothiazole moieties instead of four. Its lack of activity at 2 μM concentration shows the importance of the benzothiazole doublet, which could act either as a spacer to increase the length of the molecule, or as a real pharmacophore that establishes crucial interactions with the protein.

Given the original structure of this class of inhibitors, and in particular the unusual presence of a positive charge, we investigated their mode of CK2 inhibition. Interestingly, preliminary studies revealed that compound 1 did not behave as an ATP competitive inhibitor (Fig. 2). Moreover, experiments performed with the catalytic subunit CK2α alone and with the whole tetramer did not reveal any differential inhibitory potency (data not shown). Taken together, these experiments allow to conclude that for CK2 inhibition, these compounds do not bind either in the ATP binding site nor in a region located in/near the CK2α/CK2β interface. This new mode of inhibition is consistent with the presence of positive charges, which would be detrimental for an ATP-competitive inhibitor that is expected to mimic the negatively charged natural ligand.
https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Fig2_HTML.gif
Fig. 2

Lineweaver–Burk inhibition plots of human recombinant CK2α by compound 1. CK2 kinase activity was determined as described in the experimental section in the absence (♦) or in the presence of 5 μM 1 (▲). The data represent means of experiments run in triplicate with SEM never exceeding 10%

Given that inhibitor selectivity is one of the main problems when targeting kinases, we next assessed the selectivity of compound 1 on a panel of 30 recombinant kinases (Table 2). We chose compound 1 over compound 2 because of its smaller size which made it more prone to hit optimization. From this selectivity panel, it appears that compound 1 is a quite selective inhibitor with only four kinases notably inhibited at 5 μM: CK1, CK2, GSK3β and Rsk2. Thus, we determined IC50 on these kinases: CK2 (500 nM), GSK3β (770 nM), CK1 (4,800 nM), Rsk2 (9,500 nM). Thus, compound 1 appears as a potent dual inhibitor of CK2 and GSK3β and a weak CK1 and Rsk2 inhibitor.
Table 2

Kinase selectivity profile of compound 1

Protein kinase

Residual activity

Protein kinase

Residual activity

Protein kinase

Residual activity

Abl

52

Fes

65

PKA

69

Bmx

86

FGFR3

75

PKBα

83

c-RAF

118

GSK3β

18

PKCα

72

CDK2/cyclinA

98

IKKα

113

PRK2

105

CHK1

125

JNK1α1

91

ROCK-II

109

CK1

50

MAPK2

90

Rsk2

42

CK2

2

MEK1

102

SAPK2α

87

CSK

108

PAK2

100

SGK

86

cSRC

94

PDGFRβ

140

Yes

58

EGFR

92

PDK1

68

ZAP-70

82

Residual kinase activity determined in the presence of 5 μM inhibitor is expressed as a percentage of the control activity without inhibitor. Final concentration of ATP in the experiment was 100 μM

Unfortunately, when tested in a cellular CK2 activity assay, compound 1 did not show any CK2 inhibitory activity in vivo (data not shown).

In conclusion, these thiazole derivatives are uncommon CK2 inhibitors both for their chemical structure and their biochemical mode of inhibition. From a chemical point of view, the presence of positively charged groups is original. Concerning their biochemical properties, these inhibitors exhibit a non ATP-competitive mode of inhibition. This is a non-classical feature for CK2 inhibitors, since only one other compound of this type was already described [24]. However, considering the similar potency on CK2 and GSK3β and the lack of activity in living cells, these inhibitors would require optimization to be useful as in vivo CK2 inhibitor.

Stibonic acid-based inhibitors

One of the striking features of this cluster (Table 3) is the presence of compounds bearing an antimony atom (Sb). Even if antimony is a well known toxic agent, it is used in several therapies, especially against some parasites [25]. Given the role of CK2 in parasite proliferation, we decided to characterize more thoroughly these compounds.
Table 3

CK2 inhibitory activity of stibonic acid derivatives (compounds 1020) https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figc_HTML.gif

Compound

R

CK2 activity (% control)

10

SbO3H2

7

11

H

>95

12

F

>95

13

Cl

>95

14

Br

>95

15

OH

>95

16

NH2

>95

Compound

 

CK2 activity (%)

10

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figd_HTML.gif

7

17

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Fige_HTML.gif

64

18

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figf_HTML.gif

29

19

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figg_HTML.gif

96

20

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figh_HTML.gif

100

CK2 kinase assays were performed in the absence or presence of 2 μM compound. Results are expressed as a percentage of the control activity without inhibitor

The most active compound of this series is compound 10, a benzene substituted by a stibonic (SbO3H2) and a vinylic acid. In order to test whether the rather toxic antimony moiety is necessary for the inhibitory potency, analogues bearing other groups at the same position were tested (Table 3). Thus far, it appears that none is active albeit no analogues with acid groups at this position were available. Therefore, these data do not exclude that other acid groups may favorably replace this stibonic acid. In order to analyze the influence of the vinylic acid, which is also present in the known CK2 inhibitor tetrabromocinnamic acid TBCA [26], we tested derivatives 17–20 (Table 3). From these results, it can be noticed that vinylic acid in meta position of vinylic acid as in compound 10 (IC50 = 0.15 μM) is the optimal substituent. It can be replaced by smaller acidic groups (compounds 17, 18) even in different positions (compound 18), albeit with a loss of inhibitory potency. Short and acidic substituents seem to be important for activity in this series of compounds, since bulkier substituents (compounds 19, 20) lead to inactive derivatives. This is also the case for tetrabromocinnamic acid derivatives [26]. However, due to the lack of other related compounds in the chemical library, it was impossible to draw definitive conclusions.

As compound 10 is the most potent compound of this class, we investigated its selectivity (Table 4). It appears that compound 10 is a quite selective inhibitor with only two kinases notably inhibited at 5 μM concentration: CK2 (IC50 = 220 nM), GSK3β (IC50 = 180 nM).
Table 4

Kinase selectivity profile of compound 10

Protein kinase

Residual activity

Protein kinase

Residual activity

Protein kinase

Residual activity

Abl

107

Fes

55

PKA

86

Bmx

91

FGFR3

87

PKBα

98

c-RAF

149

GSK3β

12

PKCα

99

CDK2/cyclinA

101

IKKα

92

PRK2

108

CHK1

105

JNK1α1

85

ROCK-II

116

CK1

76

MAPK2

157

Rsk2

82

CK2

2

MEK1

103

SAPK2α

125

CSK

89

PAK2

100

SGK

63

cSRC

98

PDGFRβ

119

Yes

39

EGFR

80

PDK1

58

ZAP-70

87

Residual kinase activity determined in the presence of 5 μM inhibitor is expressed as a percentage of the control activity without inhibitor. Final concentration of ATP in the experiment was 100 μM

The active compounds of this cluster are closely related to the previously described tetrabromocinnamic acid derivatives. It is worth noting that in this class, stibonic acid plays an important role in inhibitory potency. Compound 10 the most potent compound in this series is a dual CK2 and GSK3β inhibitor in the high nanomolar range. Given their lack of inhibitory activity in a cellular context that could probably be related to their highly charged groups (data not shown) together with their similarity to previously described tetrabromocinnamic derivatives, this class of compounds was not investigated further.

N-hydroxyphthalimide-based inhibitors

This N-hydroxyphthalimide cluster is similar to the 4,5,6,7-tetrahalogeno-1H-isoindole-1,3(2H)-diones [27] but in our case we find an hydroxyl as N-substituent (Table 5). Among the tested compounds, only compound 21, in which all the substituents are phenyl groups, displays an inhibitory activity (IC50 = 1.75 μM). This highlights the requirement for highly hydrophobic substituents on the aromatic core, since other less hydrophobic substituents like chlorine lead to inactive compounds (compound 28). This is consistent with previously reported structure–activity relationships performed on similar compounds [27]. We also investigated other substitutions on the aromatic core (Table 5). The main conclusion that can be drawn from these experiments is that the bare aromatic core of compound 29 is not sufficient to achieve inhibitory activity, whereas active compounds are obtained with supplementary substitutions on positions 4 and 9 of the 2-hydroxybenzo[f]isoindole-1,3-dione core. Methoxy substituent confers the greatest potency (compound 32). This class of compounds displays an ATP-competitive mode of inhibition in line with previously described observations on similar compounds (Fig. 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Fig3_HTML.gif
Fig. 3

Lineweaver–Burk inhibition plots of human recombinant CK2α by compound 21. CK2 kinase activity was determined as described in the experimental section in the absence (♦) or in the presence of 2.5 (■) and 5 (▲) μM 21. The data represent means of experiments run in triplicate with SEM never exceeding 10%

Table 5

CK2 inhibitory activity of N-hydroxyphthalimide derivatives (compounds 2134) https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figi_HTML.gif

Compound

R1

R2

R3

R4

CK2 activity (% control)

21

Ph

Ph

Ph

Ph

72

22

H

H

OMe

H

99

23

H

H

NHAc

H

100

24

H

H

NO2

H

100

25

H

H

H

NO2

100

26

H

H

Cl

H

99

27

H

Cl

Cl

H

99

28

Cl

Cl

Cl

Cl

99

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figj_HTML.gif

Compound

R

CK2 activity (% control)

   

29

 

100

   

30

H

59

   

31

Me

36

   

32

OMe

25

   

33

CF3

51

   

34

Ph

30

   

CK2 kinase assays were performed in the absence or presence of 5 μM compound. Results are expressed as a percentage of the control activity without inhibitor

The active compounds of this cluster (compounds 21 and 30–34) were docked into the ATP binding site of CK2α. The aromatic core of these compounds is well buried into the hydrophobic region of this active site and the N-hydroxyphthalimide moiety interacts through a hydrogen bond with the backbone of Leu45. A supplementary interaction is established between the oxygen of the methoxy group of compound 32 and Lys68 (Fig. 4).
https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Fig4_HTML.jpg
Fig. 4

Docking of compound 32 in CK2α. (a) Oxygen atoms are represented in red, nitrogen in blue, carbon in cyan and hydrogen in white; hydrogen bonds are represented as dotted lines. (b) The cavity is represented as a solid surface in which hydrophobic residues are figured in white, negatively charged in red, positively charged in blue, and polar residues in green. AMPPNP is represented in green. VMD [42] software was used for molecular visualization

Interestingly, when tested in the cellular CK2 activity assay, compounds 21 and 30–34 were able to inhibit CK2 in living cells. When tested at 100 μM during 24 h compound 21 was as active as TBB (Fig. 5a). In order to test whether compound 21, like TBB, could promote tumor cell death, we compared their effect on seven different tumor cell lines. At 50 μM (Fig. 5b) and 100 μM (data not shown), compound 21 and TBB can both reduce cell viability to various extents depending on cell types, compound 21 being more efficient than TBB, except in HeLa cells.
https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Fig5_HTML.gif
Fig. 5

(a) Inhibition of CK2 activity by compounds 21 and 3034 in living cells. HeLa cells transfected with a plasmid expressing a chimeric CK2 activity YFP-based reporter were incubated with inhibitory compounds or DMSO for 24–48 h. Cell extracts were then analyzed by native electrophoresis and YFP was revealed by immunoblotting. (b) Effect of compound 21 on cell viability. Different cancer cell lines were incubated with 100 μM compound 21 for 48 h and the cell viability was assayed by the CellTiter-Glo® Luminescent Cell Viability Assay (Promega)

The compounds of this series share the common feature of ATP-competitive CK2 inhibitors, an aromatic core and a polar moiety (the N-hydroxyphthalimide group). Despite their poor aqueous solubility, some (compounds 21 and 30–34) are cell active.

Xanthene-based inhibitors

The compounds of this cluster (Table 6), that are xanthene derivatives have in common negatively charged groups (carboxylate or sulfonate). Compound 35 is one of the most potent CK2 inhibitors described to date. However, it lacks specificity since it has been previously reported to be a cell-permeable inhibitor of initiation of translation [28] and of the interaction between Bax and Bcl-X(L) [29]. What is remarkable with this inhibitor is the presence of a tetrabromobenzoic acid moiety, which may play the same role in this compound as the tetrabrominated aromatic core in TBB derivatives. However, it can be noticed that the presence of bromine is not a requisite since other compounds sharing a similar scaffold but without bromines are also potent CK2 inhibitors (compounds 36–43). These compounds are built around a xanthenone moiety, which is dihydroxylated compared to compound 35. However, given the potency of compound 35, these hydroxyl groups do not seem necessary for the inhibitory potency of the molecule, although they might help to increase aqueous solubility. Indeed, the variations in inhibitory activity are driven by the substitution pattern at position 9 (R). Too bulky groups are unfavorable for inhibitory potency (compounds 41–43), despite the presence of halogens as in compound 35. The most active compounds in this series are compounds 37 and 38, which bear a vinylic (37) or propanoic (38) acid moiety. Vinylic acid is also present in compound 10 and in the previously described tetrabromocinnamic acid derivatives (TBCA) [26].
Table 6

CK2 inhibitory activity of Xanthene derivatives (compounds 3543) https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figk_HTML.gif

Compound

R

IC50 (μM)

 

35

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figm_HTML.gif

0.08

 

36

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figl_HTML.gif

0.4

 

37

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Fign_HTML.gif

0.12

 

38

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figo_HTML.gif

0.13

 

39

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figp_HTML.gif

0.18

 

40

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figq_HTML.gif

0.8

 

41

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figr_HTML.gif

>2

 

42

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figs_HTML.gif

1.5

 

43

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Figt_HTML.gif

1.3

 

CK2 kinase assays were performed in the absence or increasing concentrations of compounds. Results are expressed as IC50 values

As compound 37 is one of the most potent compounds, we investigated its selectivity. It was observed that this compound lacks specificity since it was active on 11 kinases out of 30 (Table 7).
Table 7

Kinase selectivity profile of compound 37

Protein kinase

Residual activity

Protein kinase

Residual activity

Protein kinase

Residual activity

Abl

39

Fes

29

PKA

81

Bmx

29

FGFR3

31

PKBα

52

c-RAF

123

GSK3β

7

PKCα

107

CDK2/cyclinA

80

IKKα

66

PRK2

83

CHK1

92

JNK1α1

70

ROCK-II

127

CK1

61

MAPK2

110

Rsk2

31

CK2

0

MEK1

94

SAPK2α

103

CSK

87

PAK2

91

SGK

19

cSRC

38

PDGFRβ

79

Yes

2

EGFR

48

PDK1

34

ZAP-70

56

CK2 residual activity determined in the presence of 5 μM inhibitor is expressed as a percentage of the control activity without inhibitor. Final concentration of ATP in the experiment was 100 μM

Finally, despite of their known cell-permeability, none of the compounds of this cluster was able to inhibit CK2 activity in a cellular context. However, since they share some structural similarities with fluorescein, these compounds might serve as fluorescent CK2 markers, providing that their affinity and selectivity for CK2 are optimized.

Unrelated unique hits

Screening of diverse chemical libraries allowed us to isolate chemically unrelated unique hits (Fig. 1). Compound 44 (GW5074 [30]) is a well-known ATP-competitive c-RAF inhibitor. It was also active in our CK2 cellular assay and at 100 μM during 24 h compound 44 was as efficient as 100 μM TBB to block the cellular CK2 activity (Fig. 6a). Furthermore, when tested on seven different tumor cell lines, compound 44 (50 μM) showed a comparable or a better efficiency to trigger cell death than TBB (Fig. 6b). Molecular modeling shows that the indolinone moiety of compound 44 interacts with Lys68, while the hydroxyl establishes a hydrogen bond with the backbone of Leu45 (Fig. 7a). This dual polar anchoring is completed by the hydrophobic interactions established by the aromatic carbons and the halogen atoms, like in all the reported brominated CK2 inhibitors. Compound 45 (Indirubine 3′-monoxime [3134]), which is an ATP-competitive inhibitor of several kinases, displays a binding mode close to that of compound 44, but there is no interaction with Leu45 because of the absence of hydroxyl (Fig. 7b). This compound was inactive in our cellular assay (data not shown). The binding mode of compound 46 (SP600125 [35]), which has been described as a JNK inhibitor, recalls that of TBB, with the pyrazole moiety interacting with Lys68 and the aromatic cycles establishing hydrophobic interactions (Fig. 7c). It was also inactive in our cellular assay (data not shown). Compound 47 (Mitoxanthone [36]) is a substituted anthraquinone with a core structure very close to that of the known CK2 inhibitor emodin. Although it was a weak in vitro CK2 inhibitor, this compound displayed a CK2 inhibitory activity in living cells (Fig. 6a). Molecular modeling shows that, while the aromatic core occupies the hydrophobic cavity, as in the case of emodin, one of the flexible aliphatic chains establishes hydrogen bonds with Glu114 and the other occupies the same region as the phosphates of ATP (Fig. 7d).
https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Fig6_HTML.gif
Fig. 6

(a) Inhibition of CK2 activity by compounds 44 and 47 in living cells. HeLa cells transfected with a plasmid expressing a chimeric CK2 activity YFP-based reporter were incubated with inhibitory compounds or DMSO for 24–48 h. Cell extracts were then analyzed by native electrophoresis and YFP was revealed by immunoblotting. (b) Effect of compound 44 on cell viability. Different cancer cell lines were incubated with 50 μM compound 44 for 48 h and the cell viability was assayed by the CellTiter-Glo® Luminescent Cell Viability Assay (Promega)

https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Fig7_HTML.jpg
Fig. 7

Docking of unique hits in CK2α. (a) Compound 44; (b) Compound 45; (c) Compound 46; (d) Compound 47. Molecular modeling of the interactions established between inhibitors and CK2α. AMPPNP is represented in green. Oxygen atoms are represented in red, nitrogen in blue, carbon in cyan, bromine in green and hydrogen in white. Hydrogen bonds are represented as dotted lines

Interestingly, docking of these inhibitors reveals some variation in their binding mode (Fig. 7): in the case of compound 44, the brominated moiety occupies a position close to the one occupied by the ribose in ATP and docking of compound 47 reveals that flexible side chains are positioned in the phosphate binding site.

We did not carry on the study of these inhibitors because of the lack of available analogues to build an SAR that may have helped us to design more potent inhibitors. Moreover, compounds 46 and 47 have scaffolds similar to the well-characterized CK2 inhibitor emodin.

Physico-chemical properties of CK2 inhibitors

We next summed up data generated from this screening in an attempt to identify molecular properties that confer in cellulo activity to CK2 inhibitors (Fig. 8). We added TBB, DMAT and others known cell-potent CK2 inhibitors for comparison purpose. Clinical kinase inhibitors (Imatinib [37], Dasatinib [38], Nilotinib [39], Bosutinib [40]) were also added in order to compare their properties to CK2 inhibitors. We tested available molecular descriptors using the Chemaxon, the CDK and the WEKA softwares to find out if any of them could discriminate between cell active and inactive compounds.
https://static-content.springer.com/image/art%3A10.1007%2Fs11010-008-9828-z/MediaObjects/11010_2008_9828_Fig8_HTML.gif
Fig. 8

Display of the ratio of the Topological Surface Polar Area (TPSA) on the Eccentric Connectivity Index ξc versus LogP. Values for 54 tested compounds; 23 are inactive (black squares), 15 are in vitro active but in vivo inactive (red squares), 12 are both in vitro and in vivo active (purple squares) and 4 are clinically used kinase (not against CK2) inhibitors (blue circles). Green lines delimit an area where LogP and TPSA/ξc are respectively superior to 2 and inferior to 0.14. Compound 1 was removed from the analysis as its polymeric structure does not allow any descriptor determination

Active compounds can be clustered mainly thanks to three descriptors: LogP value, TPSA (i.e. the Polar Surface Area calculated by a fast method) and, interestingly, a topological descriptor, namely the Eccentric Connectivity Index (ξc), which reflects the overall shape (compactness) of the compounds. We found that the ratio of TPSA/ξc could be really helpful to select compounds with an enrichment of cell-potent inhibitors: 100% of the cell-active compounds have a value inferior to 0.35 (See Fig. 8); more over, using only one rule with this descriptor (TPSA/ξc < 0,145) allows to predict the in cell-activity or the in cell-inactivity of the compounds with a successful rate of 89% (see Supplementary data).

If this Index is displayed versus LogP value for all the tested compounds (Fig. 8) two areas can be drawn: one area composed of compounds with a LogP and a TPSA/ξc, respectively, superior to 2 and inferior to 0.14; 100% of the compounds which are in this area are cell-active molecules whereas the rest of the chemical space according to these descriptors presents 88% of cell-inactive compounds. This kind of analysis could help chemists to design CK2 inhibitors with higher probability to be cell-potent.

From a biological point of view, these finding are coherent. LogP restriction may be explained by constraint of membrane permeability excluding highly polar or too hydrophobic compounds. TPSA/ξc descriptor may reflect a restraint in the overall shape and polarity of the molecule. Molecules with appropriate polar surface area together with a compact shape seems to be more suitable for, at least, in cellulo activity. However, the precise explanation remains unclear. Moreover, a few compounds do not follow this general trend. Notably, mitoxantrone does not fit in this area despite being cell-potent, this may be due to its specific uptake in cell, which does not rely on a classical passive diffusion [41]. At the opposite, it can be observed that some compounds within this defined area are not cell potent CK2 inhibitors, this is probably due to lack of specificity or intra-cellular instability. It is worth noting that kinase inhibitors currently in clinical use are more narrowly clustered. This is probably due to additional constraints such as metabolic stability or intestinal permeability (if oral route is the mode of administration). Those differences between current CK2 inhibitors and clinically used kinase inhibitors indicate that CK2 inhibitors need to be further optimized to confer them acceptable pharmacokinetic profiles, and thus, to become good candidates for cancer and viral infection treatments.

Discussion

We performed a high-throughput screening on diverse chemical libraries to isolate new CK2 inhibitors. Whereas some compounds belong to previously known scaffolds, we found new CK2 inhibitors that exhibit innovative structures. These new structures (benzothiazoles, stibonic acid) open the way to more chemically diverse scaffolds, which expands the opportunity for the development of more clinically suitable CK2 inhibitors. Structure-activity relationships analysis combined with docking allowed us to propose a coherent binding mode for most of these inhibitors, thus, rationalizing their activity and the structure-based optimization of their properties.

Another interesting point is the identification and characterization of non-competitive CK2 inhibitors, represented by some benzothiazole derivatives. This mode of CK2 inhibition implies that CK2 possesses other druggable site in addition to the canonical ATP-binding site. ATP non-competitive inhibitors are an emerging class of kinase inhibitors with promising characteristics [8]. Indeed, they are generally specific and do not have to compete with millimolar cellular ATP concentration.

In order to assess the cell potency of the discovered inhibitors, we set up a cellular CK2 assay performed on living cells. Several of the identified CK2 inhibitors showed a good cellular activity. This cellular assay is rather stringent and weak CK2 inhibitors may not be classified as active in this test, which does not preclude that in other cellular contexts, these weak inhibitors may be found to be active. Despite this high stringency, we found that commercially available protein kinase inhibitors or therapeutics were also cell potent CK2 inhibitors, an observation that may be of particular relevance given the broad implication of CK2 in many cellular processes.

Finally, using this cellular assay we are able to draw some guidelines about the molecular requirements for a CK2 inhibitor to be active in living cells. Whereas in vitro inhibitors may be highly diverse in term of charge, mass and polarity, the properties of cell-potent compounds are much more restricted. We can then conclude that these restrictions are cell-dependent. Taken together, these preliminary results should be confirmed with more data but they are very encouraging. Extending this study about cell-potency further with different ligand-based SAR methods is warranted. Such analysis will provide guidelines to optimize compounds only active in vitro to render them also efficient in a cellular context. Indeed, comparison with clinically used kinase inhibitors reveals that the available CK2 inhibitors need further optimizations.

Taken together, our results show that CK2 inhibitors remain an open field with new promising opportunities for the development of better inhibitors.

Acknowledgments

This work was supported by the Institut National de la Santé et de la Recherche Médicale (INSERM), the Centre National pour la Recherche Scientifique (CNRS), the Commissariat à l’Energie Atomique (CEA), the Institut Curie, the Ligue Nationale Contre le Cancer (équipe labellisée 2007), the Institut National du Cancer (Grant Number 57). The authors thanks the Drug Synthesis & Chemistry Branch, Developmental Therapeutics Program, Division of Cancer Treatment and Diagnosis, NCI, for the library of low-molecular-weight compounds, the ChemAxon company (http://www.chemaxon.com) for allowing academics to freely use their software, especially here the Standardizer and the Calculator Plugins to compute descriptors and the MarvinView package, for viewing structure files in our home-made software, PhenoScreen.

Supplementary material

11010_2008_9828_MOESM1_ESM.xls (21 kb)
Supporting Information Available: Detailed list of compounds with their trivial names and molecular descriptors, machine learning resulting rules used to determine TPSA/ξc threshold are available upon request. MOESM1 (XLS 21 kb)

Copyright information

© Springer Science+Business Media, LLC. 2008