Compounds Designs of CK2α Inhibitors Derived from Virtual Screening Hit Compounds by Computational Chemistry with Crystallography

Abstract

Protein kinase CK2 type α (CK2α) inhibitors are expected to be a new anticancer drug and a treatment for nephritis. Virtual screening for CK2α inhibitors has been conducted and active compounds with various scaffolds have been obtained. Research on compound optimization is currently in progress for some of them with the aim of improving their activity. This process involves the combination of various computational chemistry methods and crystal analyses. In this review, case studies of structure-based compound designs that have efficiently improved the activity of screening hit compounds, including compounds with a thiadiazole ring and a purine scaffold, are introduced.

1. Introduction

Protein kinase CK2 (CK2) is a serine/threonine protein kinase that uses ATP or guanosine 5′-triphosphate (GTP) as coenzymes. These kinases play crucial roles in intracellular signal transduction, including processes such as gene expression, protein synthesis, cell proliferation, and apoptosis.^1,2) CK2 has been related to tumor growth,²⁾ making it a prominent target for anticancer drug development. Research is actively underway to develop novel CK2 inhibitors.³⁾ Additionally, studies have revealed a significant increase in CK2 expression within the glomeruli of patients with nephritis, and known CK2 inhibitors have shown promise in alleviating the symptoms of nephritis in glomerulonephritis model animals.^4,5) Furthermore, recently, a surge in CK2 activity has been observed in cells infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), indicating its involvement in viral proliferation.⁶⁾

As described above, CK2 is a target protein for the treatment of various diseases, and research is ongoing to develop its inhibitors. CK2α serves as the catalytic subunit of CK2,⁷⁾ and the inhibitors^8–14) mentioned in Fig. 1 control its activity by binding to CK2α. Currently, CX-4945 (later named silmitasertib) is the only CK2α inhibitor in clinical development and has received approval as an orphan drug from the U.S. Food and Drug Administration for the treatment of advanced bile duct cancer in January 2017.¹⁵⁾ Given the expectation that developing inhibitors with diverse scaffolds can reduce the side effects, the search for novel inhibitors continues.

Fig. 1. Examples of Reported CK2α Inhibitors and IC₅₀ Values (µM) Prior to the SDO-VS

2. Virtual Screening and Various Hit Compounds

For virtual screening to search for CK2α inhibitors, a novel screening method called SDO-VS has been utilized.¹⁶⁾ This method is based on the observation of a phenomenon known as solvent dipole ordering (SDO) around proteins.¹⁷⁾ SDO refers to the alignment of the dipoles of solvent water molecule, which immobilizes in the presence of protein substituents. SDO is particularly strong within the ligand-binding sites. The three-dimensional shape of the region where SDO occurs, referred to as the ‘SDO region,’ closely resembles the steric shape of the bound ligand of the protein.¹⁸⁾ The SDO-VS method was developed based on this insight.

In this method, the initial step involves the random generation of multiple molecules, called pseudomolecules, inside the SDO region. These pseudomolecules consist solely of sp³ carbon atoms with standard bond lengths and bond angles, and the shape of the SDO region is replicated in the shape of these pseudomolecules. Subsequently, a search is conducted within a database of compounds exhibiting a three-dimensional shape highly similar to that of the generated pseudomolecules. Compounds satisfying this criterion are identified as hits as they closely mimic the shape of the SDO region. Because this screening process does not rely on the structural information of compounds with known activities, hit compounds with structural diversity are acquired. The sp³ carbon atoms in the pseudomolecules are chosen to generate a wide array of shapes. However, in several cases, the shape similarity to existing compounds may not be high. Therefore, methods for generating pseudomolecules that incorporate carbon atoms with various hybrid orbitals and benzene rings are also under development.¹⁹⁾

When the virtual screening process commenced, experimental structures of complexes involving CK2α and several compounds,^20–25) determined through X-ray crystal structure analysis, had already been cataloged in the PDB database.²⁶⁾ To search for ATP-competitive compounds, the complex structure of CK2α with AMP-PNP, an ATP analog of the coenzyme of CK2α (PDB ID: 1JWH),²¹⁾ was utilized. Pseudomolecules were generated using sp³ carbon atoms within the SDO region as analyzed based on this structure, and their three-dimensional shapes were compared with a collection of compounds in the Namiki compound database.²⁷⁾ The use of a database of commercially available compounds for screening offered the advantage of high accessibility and enables the swift verification of results through activity measurements.

Approximately 2.5 million compounds with molecular weights ranging from 250 to 500 Da were screened, ultimately leading to the purchase of 99 compounds for further assays. After measuring their inhibitory activities, 28 hit compounds were identified, each demonstrating an inhibition rate of 50% or more at a concentration of 30 µM. An additional 36 analogs of these hit compounds featuring different substituents were also procured, resulting in 11 additional hit compounds. Among these 39 compounds, 26 compounds exhibited inhibitory activity values (IC₅₀) stronger than 50 µM and were selected as the final hit compounds (Fig. 2). As previously mentioned, one distinctive aspect of the SDO-VS method is that chemical similarity with known active compounds is not considered during the screening process, leading to the identification of hit compounds with diverse chemical structures.

Fig. 2. Examples of SDO-VS Hit Compounds and IC₅₀ Values (µM)

3. Design from Hit Compounds with Thiadiazole Ring

The 26 hit compounds were optimized with a focus on ligand efficiency.²⁸⁾ Complex crystal structures of hit compound 1 (Fig. 3) and its related compound 2 with CK2α (PDB IDs: 3WIK²⁹⁾ and 3AXW,³⁰⁾ respectively) were successfully analyzed. To introduce the first optimization study,^30,31) a compound design study utilizing these compounds as starting materials was conducted.

Fig. 3. Schematic Diagram of the Design of Thiadiazole Compounds and IC₅₀ Values (µM)

By superimposing both crystal structures, the nitro group was replaced with a more favorable carboxy group considering the interaction with nearby Lys68, and the terminal acetyl group was replaced with a p-methoxybenzoyl group considering the pocket size. Thus, compound 3 exhibited an approximately 10-fold improvement in activity compared with hit compound 1. However, there were concerns regarding the electrostatic repulsion between the central thiadiazole ring of compound 3 and the main-chain carbonyl oxygen atom of Glu114 as a presumed interaction of this compound (Fig. 3). Therefore, thiazole compound 4, intended to form a CH–O hydrogen bond with Glu114, and pyrazole compound 5, designed to establish a hydrogen bond with Glu114, were developed as hetero-ring replacements for thiadiazole compound 3. To validate the design concept, the precise binding energies of the three compounds were estimated using the fragment molecular orbital method.³²⁾

The stability of the designed compounds led to their chemical synthesis and the measurement of their inhibitory activity. The IC₅₀ value was 3.4 µM for thiadiazole compound 3, whereas it was 0.032 and 0.14 µM for thiazole compound 4 and pyrazole compound 5, respectively, resulting in an activity improvement of more than 100 times solely due to the difference in the ring structure.³⁰⁾ Although pyrazole compound 5 was anticipated to have a more stable interaction through a hydrogen bond, it exhibited lower inhibitory activity than thiazole compound 4. This decrease in activity was attributed to unfavorable interactions between the tautomers.

Further compound designs are currently under investigation. By superimposing the CK2α complexes of thiazole compound 4 (PDB ID: 5B0X) with CX-4945 (PDB ID: 3PE1) and IPD (PDB ID: 3AT3), it was anticipated that several CH–π interactions would occur near the benzene ring of the benzoic acid moiety (Fig. 4). Consequently, this moiety was transformed into a naphthalene ring, and compounds featuring the addition of another aromatic ring to the benzene ring were synthesized. These modifications resulted in compounds with various substituents, such as compound 6, which exhibited remarkably strong IC₅₀ values, comparable to those of CX-4945. However, most of these compounds were poorly soluble in water. Because compounds with nitrogen atoms introduced into the benzene ring exhibit improved solubility,³¹⁾ comprehensive N- and F-scan compound designs are currently being developed for thiazole and pyrazole compounds. In recent years, the phenomenon known as an “activity cliff,”³³⁾ in which even slight structural changes, such as those in N- and F-scans, can lead to a significant alteration in activity, has become widely acknowledged. The precise prediction of the activity for such subtle compound changes is a specialty of thermodynamic integration methods in computational chemistry.³⁴⁾ The synthesis and activity of promising compounds are currently advancing based on favorable calculation results. Moreover, compounds with high activity and improved solubility, such as compound 7, which lacks bulky steric substituents, have been obtained.³¹⁾

Fig. 4. Superposition of 5B0X (Compound 4, Magenta), 3PE1 (CX-4945, Cyan) and 3AT3 (IPD, Pink)

The protein residues (Green) were adopted from 5B0X. The significant interactions of compound 4 are highlighted by orange dotted lines.

4. Design from Hit Compound with Purine Scaffold

Compounds with a purine scaffold are also currently under investigation to explore their structure–activity relationships. A water molecule labeled as W, which is highly conserved in several CK2α complex crystal structures, is located in the proximity of Lys68 deep within the ATP binding site (Fig. 4).

In the prediction of compound binding modes through docking calculations, it is crucial to anticipate the presence or absence of water molecules within the binding site, because they can influence the formation of hydrogen bond donors and acceptors. For instance, the known natural compound CK2α inhibitor apigenin demonstrates different binding modes in CK2α of two distinct species. Although human and maize CK2α exhibit only two differing amino acid residues in the active site, apigenin forms a direct hydrogen bond with CK2α without involving water molecule W in humans (PDB ID: 3AMY). In maize (PDB ID: 4DGM),³⁵⁾ hydrogen bonds are indirectly formed through the crystal water molecule W. This phenomenon has been associated with the stability of water molecule.³⁶⁾ However, precise and quantitative predictions remain challenging and it is necessary to determine whether water molecules should be considered for compound binding.

In the case of purine derivatives, when a benzoic acid moiety is introduced, two potential binding modes are considered: one that retains the crystal water and forms an indirect hydrogen bond, and the other that excludes the crystalline water and establishes a direct hydrogen bond. Consequently, both possibilities were considered during docking calculations to predict a reasonable binding mode.

Based on the binding free energy evaluation using the MM-PB/SA method³⁷⁾ combined with binding entropy correction through the interaction entropy method,³⁸⁾ it was suggested that binding without the crystal water molecule W would yield a more stable complex.³⁹⁾ Additionally, because a CH–π interaction was anticipated for the benzene ring on the opposite side, an electron-donating group and an electron-withdrawing group were introduced to examine the structure–activity relationship. The measured activity exhibited a strong correlation with the Hammet-σ value.⁴⁰⁾ Compound 9 (Fig. 5) resulting from these considerations led to an approximately 5-fold improvement in activity.³⁹⁾ These predictions were subsequently confirmed by experimental results (Fig. 6) involving the crystal structure of the compound without the electron-donating group from compound 9 (PDB ID: 7BU4).

Fig. 5. Schematic Diagram of the Design of Purine Compounds and IC₅₀ Values (µM)

The location of the conserved crystal water molecule W to be avoided is also indicated to represent the design concept of compound 10.

Fig. 6. Superposition of 5B0X (Compound 4, Magenta) and 7BU4 (Compound 9, Gray)

The protein residues (Green) were adopted from 5B0X. The significant of compound 9 are highlighted by orange dotted lines. Notably, the carboxy group of compound 9 displaces the crystal water molecule W, establishing a direct hydrogen bond with Trp176.

By refocusing on the 4-carboxy group within the benzoic acid moiety, it was observed that a direct hydrogen bond with Trp176, without the presence of a stable water molecule W, incurred a significant dehydration penalty. By relocating the 4-carboxy group, we expected to form a hydrogen bond with Lys68 without the need for desolvation. Consequently, compound 10 (Fig. 5), synthesized after exploring the substituents on the opposite benzene ring in anticipation of a shift in the bonding position, achieved an improvement in activity exceeding 100-fold.⁴¹⁾ It was assumed that the presence of W was maintained while binding compounds with a 3-carboxy group, as anticipated. A crystal structure analysis of this compound is currently in progress.

5. Conclusion

The development of CK2α inhibitors with novel scaffolds is anticipated to mitigate potential side effects. Herein, we present case studies of structure-based compound design that facilitated significant enhancement in activity, starting from hit compounds featuring a thiadiazole ring and those with a purine scaffold. In the design of hit compounds with a thiadiazole ring, the strategies employed encompassed heterocyclic conversion, hybrid design by integrating known compounds, and exploring structure–activity relationships through extensive N-scan analysis. In the design based on hit compounds with a purine scaffold, we considered the presence or absence of water molecules during docking to circumvent the dehydration energy and introduced classical structure–activity relationships. Both approaches were successful in substantially improving the activity of the compounds. Further advancements in structure–activity are expected to yield promising new candidate compounds for medicinal applications.

Acknowledgments

The authors would like to express their gratitude to the students involved in the compound synthesis. This study used the supercomputer system of the Academic Center for Computing and Media Studies at Kyoto University.

Conflict of Interest

The authors declare no conflict of interest.

References

• 1) Cozza G., Gianoncelli A., Bonvini P., Zorzi E., Pasquale R., Rosolen A., Pinna L. A., Meggio F., Zagotto G., Stefano M., ChemMedChem, 6, 2273–2286 (2011).
• 2) Litchfield D. W., Biochem. J., 369, 1–15 (2003).
• 3) Jain A. S., Drygin D., Streiner N., Chua P., Pierre F., O’Brien S. E., Bliesath J., Omori M., Huser N., Ho C., Proffitt C., Schwaebe M. K., Ryckman D. M., Rice W. G., Anderes K., Cancer Res., 70, 10288–10298 (2010).
• 4) Tawfic S., Yu S., Wang H., Faust R., Davis A., Ahmed K., Histol. Histopathol., 16, 573–582 (2001).
• 5) Guerra B., Issinger O. G., Electrophoresis, 20, 391–408 (1999).
• 6) Miranda J., Bringas R., Fernandez-de-Cossio J., Perera-Negrin Y., Mol. Med., 27, 161 (2021).
• 7) Pechkova E., Zanotti G., Nicolini C., Acta Crystallogr. D Biol. Crystallogr., 59, 2133–2139 (2003).
• 8) Sarno S., Moro S., Meggio F., Zagotto G., Dal Ben D., Ghisellini P., Battistutta R., Zanotti G., Pinna L. A., Pharmacol. Ther., 93, 159–168 (2002).
• 9) Pierre F., Chua P. C., O’Brien S. E., et al., J. Med. Chem., 54, 635–654 (2011).
• 10) Wang Z. Q., Johnson C. L., Kumar A., Molkentine D. P., Molkentine J. M., Rabin T., Mason K. A., Milas L., Raju U., Anticancer Res., 34, 6981–6989 (2014).
• 11) Cozza G., Bonvini P., Zorzi E., Poletto G., Pagano M. A., Sarno S., Donella-Deana A., Zagotto G., Rosolen A., Pinna L. A., Meggio F., Moro S., J. Med. Chem., 49, 2363–2366 (2006).
• 12) Suzuki Y., Cluzeau J., Hara T., Hirasawa A., Tsujimoto G., Oishi S., Ohno H., Fujii N., Arch. Pharm. (Weinheim), 341, 554–561 (2008).
• 13) Fuchi N., Iura Y., Kaneko H., Nitta A., Suyama K., Ueda H., Yamaguchi S., Nishimura K., Fujii S., Sekiya Y., Yamada M., Takahashi T., Bioorg. Med. Chem. Lett., 22, 4358–4361 (2012).
• 14) Qiao Y., Chen T., Yang H., Chen Y., Lin H., Qu W., Feng F., Liu W., Guo Q., Liu Z., Sun H., Eur. J. Med. Chem., 181, 111581 (2019).
• 15) FDA Updates, Oncol. Times, 39, 23 (2017).
• 16) Murata K., Nagata N., Nakanishi I., Kitaura K., J. Comput. Chem., 31, 2714–2722 (2010).
• 17) Higo J., Nakasako M., J. Comput. Chem., 23, 1323–1336 (2002).
• 18) Murata K., Nagata N., Nakanishi I., Kitaura K., J. Comput. Chem., 31, 791–796 (2010).
• 19) Nakamura S., Kitayoshi H., Nakanishi I., J. Comput. Aided Chem., 18, 149–158 (2017).
• 20) Niefind K., Guerra B., Ermakowa I., Issinger O. G., EMBO J., 20, 5320–5331 (2001).
• 21) Kinoshita T., Sekiguchi Y., Fukada H., Nakaniwa T., Tada T., Nakamura S., Kitaura K., Ohno H., Suzuki Y., Hirasawa A., Nakanishi I., Tsujimoto G., Mol. Cell. Biochem., 356, 97–105 (2011).
• 22) Battistutta R., Mazzorana M., Sarno S., Kazimierczuk Z., Zanotti G., Pinna L. A., Chem. Biol., 12, 1211–1219 (2005).
• 23) Raaf J., Brunstein E., Issinger O.-G., Niefind K., Chem. Biol., 15, 111–117 (2008).
• 24) Sekiguchi Y., Nakaniwa T., Kinoshita T., Nakanishi I., Kitaura K., Hirasawa A., Tsujimoto G., Tada T., Bioorg. Med. Chem. Lett., 19, 2920–2923 (2009).
• 25) Berman H. M., Westbrook J., Feng Z., Gilliland G., Bhat T. N., Weissig H., Shindyalov I. N., Bourne P. E., Nucleic Acids Res., 28, 235–242 (2000).
• 26) Namiki Shoji Co., LTD., Kensei Shinjuku Bldg., 5–5–3 Shinjuku, Shinjuku-ku, Tokyo 160–0022, Japan.
• 27) Nakanishi I., Murata K., Nagata N., Kurono M., Kinoshita T., Yasue M., Miyazaki T., Takei Y., Nakamura S., Sakurai A., Iwamoto N., Nishiwaki K., Nakaniwa T., Sekiguchi Y., Hirasawa A., Tsujimoto G., Kitaura K., Eur. J. Med. Chem., 96, 396–404 (2015).
• 28) Hopkins A. L., Groom C. R., Alex A., Drug Discov. Today, 9, 430–431 (2004).
• 29) Nakanishi I., Murata K., Nagata N., Kurono M., Kinoshita T. Yasue M., Miyazaki T., Hirasawa A., Tsujimoto G., Kitaura K. To be published. (PDB ID: 3WIK)
• 30) Hou Z., Nakanishi I., Kinoshita T., Takei Y., Yasue M., Misu R., Suzuki Y., Nakamura S., Kure T., Ohno H., Murata K., Kitaura K., Hirasawa A., Tsujimoto G., Oishi S., Fujii N., J. Med. Chem., 55, 2899–2903 (2012).
• 31) Ohno H., Minamiguchi D., Nakamura S., Shu K., Okazaki S., Honda M., Misu R., Moriwaki H., Nakanishi S., Oishi S., Kinoshita T., Nakanishi I., Fujii N., Bioorg. Med. Chem., 24, 1136–1141 (2016).
• 32) Kitaura K., Ikeo E., Asada T., Nakano T., Uebayasi M., Chem. Phys. Lett., 313, 701–706 (1999).
• 33) Stumpfe D., Bajorath J., J. Med. Chem., 55, 2932–2942 (2012).
• 34) Kirkwood J. G., J. Chem. Phys., 3, 300–313 (1935).
• 35) Lolli G., Cozza G., Mazzorana M., Tibaldi E., Cesaro L., Donella-Deana A., Meggio F., Venerando A., Franchin C., Sarno S., Battistutta R., Pinna L. A., Biochemistry, 51, 6097–6107 (2012).
• 36) Nakamura S., CICSJ Bulletin, 34, 17–20 (2016).
• 37) Kollman P. A., Massova I., Reyes C., Kuhn B., Huo S., Chong L., Lee M., Lee T., Duan Y., Wang W., Donini O., Cieplak P., Srinivasan J., Case D. A., Cheatham T. E. 3rd, Acc. Chem. Res., 33, 889–897 (2000).
• 38) Duan L., Liu X., Zhang J. Z. H., J. Am. Chem. Soc., 138, 5722–5728 (2016).
• 39) Nishiwaki K., Nakamura S., Yoshioka K., Nakagawa E., Nakatani S., Tsuyuguchi M., Kinoshita T., Nakanishi I., Chem. Pharm. Bull., 71, 558–565 (2023).
• 40) Hammett L. P., J. Am. Chem. Soc., 59, 96–103 (1937).
• 41) Nishiwaki K., Nakatani S., Nakamura S., Yoshioka K., Nakagawa E., Tsuyuguchi M., Kinoshita T., Nakanishi I., RCS Med. Chem., 15, 1274–1282 (2024).