Identification of a selective G-quadruplex DNA binder using a multistep virtual screening approach

Abstract

To efficiently identify small molecules binding to a G-quadruplex structure while avoiding binding to duplex DNA, we performed a multistep structure-based virtual screening by simultaneously taking into account G-quadruplex DNA and duplex DNA. Among the 13 compounds selected, one outstanding ligand shows significant selectivity for G-quadruplex binding as determined using SPR, FRET-based competition and luciferase activity assay.

---

In addition to the Watson–Crick duplex, DNA and RNA are able to form higher order structures called G-quadruplexes from guanine-rich sequences. It has been shown that G-quadruplex formation in regions of biological significance, such as human telomeres, promoter regions of oncogene and ribosomal DNA, plays important roles in cellular aging and cancer.¹ Recently, an elegant study reported that DNA G-quadruplex structures can be visualized quantitatively in human cells, which confirms the existence of G-quadruplex structures in the human genome and reinforces the idea that DNA G-quadruplexes might be a promising class of molecular targets for anticancer drugs.²

Thus far, a number of small molecules targeting biologically relevant G-quadruplexes have been reported.³ Their anticancer activities and mechanism of actions were carefully studied to direct medicinal chemistry research targeting G-quadruplexes.⁴ However, there is only one small molecule (quarfloxin) that has entered into clinical trials for human cancer and none has made it to the market. The reasons are complex, including inherent difficulties in the drug discovery process, difficulties in translation from in vitro data to in vivo data and the side-toxicity of the small molecules because of their poor selectivity, or nonspecific interactions with the duplex DNA. Typically, fused aromatic G-quadruplex ligands such as BRACO-19,⁵ and SYUIQ-5 bind to duplex DNA via intercalation,⁶ while some unfused aromatic G-quadruplex ligands, such as distamycin A, bind to the groove of duplex DNA.⁷ Importantly, duplex DNAs are more abundant than G-quadruplex DNAs in cells. Therefore, the selectivity for G-quadruplex over duplex DNA is crucial for small molecules to be promising G-quadruplex binders. A number of studies have been devoted to the development of novel G-quadruplex binders or discovery of new chemical platforms for G-quadruplex through virtual screening protocols.⁸ However, to the best of our knowledge, none of them includes the aspect of selectivity screening for G-quadruplex over duplex DNA. Herein, we report on the development of a sequential computational approach to enable the identification of prospective ligands that bind to G-quadruplex selectively without duplex binding. The multistep strategy combines docking-based virtual screening for G-quadruplex binders, duplex DNA intercalators and duplex DNA minor groove binders.

The c-MYC gene promoter G-quadruplex structure was selected as the target in view of its important role linked to human cancers, with the additional advantage that it adopts only one conformation, a propeller-type parallel-stranded G-quadruplex.⁹ A database containing 1.2 million compounds from ChemBridge was used for screening. All the small molecules are drug-like and purchasable. Firstly, only small molecules with more than 3 aromatic rings were selected, as suggested by previously identified G-quadruplex ligands being mostly comprised of more than 3 aromatic rings, which may help the ligands stack over the G-quartet (Fig. 1). Secondly, the remaining 28 530 small molecules were docked to the NMR structure of the c-MYC G-quadruplex (PBD entry: 1XAV)¹⁰ using Surflex-DOCK (Fig. S1, ESI†). Compared to the crystal structure of the c-MYC G-quadruplex, this NMR structure may be more physiologically relevant and has flanking sequences that interact with the G-quartets, thus contributing to the conformational complexity of the structure, which might participate in ligand binding. The docking score was calculated and referred to as Score_G4. Compounds were selected if satisfying the criteria of Score_G4 > 6.4. Thirdly, the resulting 2530 compounds were docked to the duplex DNA structure (PDB entry 1Z3F)¹¹via intercalation interaction. The score was calculated and referred to as Score_intercalator. Compounds were selected if satisfying the criteria of Score_G4/Score_intercalator > 1.0. Fourthly, the resulting 441 compounds were docked to the duplex DNA structure (PDB entry 1K2Z)¹²via groove binding interaction. The score was calculated and referred to as Score_groovebinder. The compounds were selected if satisfying the criteria of Score_G4/Score_groovebinder > 1.1. The resulting 57 compounds were subjected to visual inspection, which has been found to be one of the most critical steps in virtual screening, and undesirable compounds were discarded. The final step yielded 13 candidates that proceeded for experimental validation (Table 1 and Table S1, ESI†).

Fig. 1 Workflow diagram of the virtual screening cascade protocol.

Table 1 Hit list table from virtual ligand screening

Compd ID	Score_G4	Score_intercalator	Score_groovebinder	A	B
A = Score_G4/Score_intercalator; B = Score_G4/Score_groovebinder.
VS1	9.72	7.29	5.99	1.33	1.62
VS2	9.35	8.51	4.90	1.10	1.91
VS3	9.21	8.11	6.20	1.14	1.49
VS4	8.07	7.85	5.44	1.03	1.48
VS5	8.99	7.62	4.59	1.18	1.96
VS6	8.75	5.71	3.66	1.53	2.39
VS7	7.86	6.37	5.17	1.23	1.52
VS8	9.71	7.55	5.82	1.29	1.67
VS9	10.65	7.33	6.20	1.45	1.72
VS10	9.09	7.94	7.97	1.14	1.14
VS11	8.28	5.36	4.36	1.54	1.90
VS12	6.60	5.67	3.83	1.16	1.72
VS13	6.64	4.86	4.45	1.37	1.49

---

An SPR assay was first performed to rapidly screen candidates by evaluating the G-quadruplex binding affinity as well as selectivity over duplex DNA. Biotinylated c-MYC G-quadruplex and duplex DNA were used. Five compounds were identified to bind to the G-quadruplex DNA with no binding seen to the duplex DNA (Fig. S2, ESI†). The compound VS10 shows strong binding affinity with c-MYC G-quadruplex DNA with a K_D value of 2.0 μM, while the compound has no duplex binding for a concentration of up to 10.0 μM, as indicated by Fig. 2B. K_D values of the other compounds were not determined as they did not bind to the G-quadruplex strongly enough. To verify the effective binding of VS10 towards c-MYC G-quadruplex DNA, the UV titration experiment (Fig. S3, ESI†) and the G-quadruplex fluorescence intercalator displacement (G4-FID) assay (Fig. S4, ESI†) were carried out, and the K_a value of 6.6 × 10⁵ M⁻¹ and DC₅₀ value of 1.8 μM were determined respectively. The result from the G4-FID assay also showed that binding of VS10 towards c-MYC G-quadruplex DNA was comparable to that of the reference compound SYUIQ-5 and bisquinolinium G-quadruplex ligand M2.¹³

Fig. 2 (A) Structure of VS10. (B) SPR sensorgrams overlay for binding of VS10 to G-quadruplex and duplex DNA. (C) Competitive FRET results for VS10 without and with excess of the duplex DNA competitor (CT DNA). The G-quadruplex ligand SYUIQ-5 and M2 were used as reference compounds.

In addition, the stabilization of VS10 to the G-quadruplex was studied using CD spectroscopy through measuring the thermal stability profile of the c-MYC G-quadruplex DNA incubated with the compound. CD studies showed that VS10 increased the melting temperature of the c-MYC G-quadruplex by 5 °C (Fig. S5, ESI†). To further access the G-quadruplex binding selectivity of VS10 under competitive conditions, FRET-based competition assays were performed (Fig. 2C), where the ability of the ligand to retain G-quadruplex stabilizing ability was challenged by the nonfluorescent duplex DNA (CT DNA). In the presence of various amounts of duplex DNA, the thermal stabilization of c-MYC G-quadruplex DNA enhanced by VS10 was only slightly affected until the addition of the 500 μM competitor, while the 20 μM competitor sharply disrupted the binding of SYUIQ-5 to the G-quadruplex. Besides, the selectivity of VS10 for G-quadruplex was also found to be better than that of the bisquinolinium G-quadruplex ligand M2 whose binding to the G-quadruplex was obviously disrupted upon the addition of the 200 μM competitor. These results were in agreement with those found by the SPR assay, suggesting that the multistep virtual screening protocol is effective in identifying selective G-quadruplex ligands.

Molecular docking and molecular dynamics simulations were then carried out to investigate the binding mode of VS10 to c-MYC G-quadruplex DNA. The NMR determined propeller-type parallel-stranded G-quadruplex structure used in the above screening process was also employed as the template. Given the possibility that ligand molecules may utilize one or the other side of its aromatic ring system to make stacking interactions with the G-quartet, we selected two docking poses representing the ligand stacking over the 5′ end of the G-quartet and another two docking poses representing the ligand stacking over the 3′ end of the G-quartet. A loop binding mode was also observed, with the ligand binding to the G-quadruplex propeller loop. In total, five docking poses representing all possible binding modes were subjected to molecular dynamics simulations (Fig. S6, ESI†). The binding energy estimated using the MM/PBSA approach shows that the ligand stacking over the 5′ end of the G-quartet has the most negative binding free energy (−14.1 kcal mol⁻¹, Table S2, ESI†), suggesting that VS10 might prefer to bind to the 5′ end of the G-quartet (Fig. 3).

Fig. 3 Modeling structure of VS10 and the c-MYC G-quadruplex complex from MD simulation.

To clarify the binding mode between VS10 and the c-MYC G-quadruplex structure, a NMR spectroscopy study was carried out using G-quadruplex Pu24I derived from c-MYC (PDB entry 2A5P) instead of the 1XAV structure, as Pu24I has clearly assigned and more easily distinguished G-quartet imino proton signals (Fig. S7, ESI†).¹⁴ As shown in Fig. 4, upon addition of VS10 to a solution of Pu24I, the imino proton resonances of the 5′ terminal G-quartet residues (G4, G8 and G13) shift upfield and G17 shifts downfield. In contrast, no shift of the imino proton resonances of the 3′ terminal G-quartet residues (G6, G15, G19 and G24) was observed. The data suggest that VS10 interacts with the G-quadruplex structure by stacking on the 5′ terminal G-quartet. This result is consistent with the molecular modeling study. Particularly, the folding of Pu24I is slightly different in the loop region as compared to that of 1XAV. However, such a difference is not expected to significantly affect VS10 binding to the G-quartet.

Fig. 4 NMR titration of c-MYC G-quadruplex Pu24I with VS10 at various ratios of [VS10]/[Pu24I]. The imino proton resonances of the residues in the G-quartet were assigned based on the data from the literature.¹⁴

In addition to the above studies, it is also important to evaluate the cellular effects of VS10 and see whether it could also bind to the c-MYC G-quadruplex DNA under cellular conditions and accordingly reduce the gene transcriptional and expression levels. Thus, a luciferase activity assay was employed to further explore the cellular effects of VS10 on c-MYC promoter activity. Two luciferase constructs were designed and used. One contained a full-length wild-type promoter of c-MYC with a native G-quadruplex forming sequence, while the other had a mutated sequence that suppressed the G-quadruplex formation as suggested by CD studies (Fig. S8, ESI†). The effects of VS10 on c-MYC promoter activity are shown in Fig. 5A, and a compound SYUIQ-5 with low G-quadruplex selectivity over duplex DNA was used as the reference. Dose-dependent decreased luciferase activities on the wild promoter construct were observed for both compounds (Fig. S9, ESI†). The addition of VS10 at 2.0 μM resulted in 33% reduction of luciferase activity for the wild promoter construct, which is 2.3 times stronger than its reduction of mutant promoter construct (14%). However, at the same concentration, SYUIQ-5 exhibited close inhibitory effects on the luciferase activity of wild and mutant constructs (51% vs. 52%). The results suggest that VS10 can inhibit the activity of c-MYC promoter through interaction with the promoter G-quadruplex structure and VS10 has better selectivity than SYUIQ-5 for the wild promoter construct over the mutant construct.

Fig. 5 (A) Effects of VS10 and SYUIQ-5 at the concentration of 2.0 μM on c-MYC promoter activity. The error bars represent the standard error from a triplicate of three independent experiments. ***: significantly different (P < 0.001). ns: no significant difference (P > 0.05). (B) Effect of VS10 at the concentration of 0.25, 0.5, 1.0 and 2.0 μM on c-MYC transcription.

On the basis of the results from the luciferase activity assay, a study on the effects of VS10 on c-MYC transcription was carried out using two Burkitt's lymphoma cell lines (RAji and CA46). The NHE III1 element of the c-MYC gene containing the G-quadruplex forming sequence is removed together with P1 and P2 promoters in the CA46 cell line, while the RAji cell line still retains this element after translocation.¹⁵ The regulation of the transcription was evaluated by quantitation of mRNA using RT-PCR. As shown in Fig. 5B, upon the addition of VS10, the amount of the c-MYC PCR product did not change much in the CA46 cell line. However, in the RAji cell line with the NHE III1 element deleted, the transcription of c-MYC was inhibited in a dose-dependent fashion. Upon the treatment of VS10 at different concentrations for 24 h, the transcription of c-MYC was reduced by 6%, 18%, 26% and 34%, respectively, related to a control gene β-actin. These results provide additional evidence suggesting that VS10 could bind to the c-MYC G-quadruplex DNA under cellular conditions and accordingly reduce the gene transcriptional level.

In summary, a multistep structure-based virtual screening was carried out, incorporating docking with G-quadruplex and duplex DNA in order to identify selective G-quadruplex binders. A new ligand, VS10, was identified, which exhibited high selectivity for c-MYC G-quadruplex versus duplex DNA as determined using SPR, FRET-based competition and luciferase activity assays. The results demonstrate the feasibility of applying a multiple step structure-based virtual screening approach for selective targeting of G-quadruplex. This work may shed light on the search for selective binders for G-quadruplexes against duplex DNA. Furthermore, it represents an important first step that can be used in the process of discovery of selective binders that target a specific G-quadruplex structure.

The interaction between VS10 and c-MYC G-quadruplex DNA was further studied by UV titration, G-quadruplex fluorescent intercalator displacement, molecular modeling and NMR spectroscopy methods. Notably, there is still room for improvement of the binding of VS10 to the G-quadruplex. Structural modification of VS10 to further improve the binding affinity is currently underway. It is also worth mentioning that an identical compound emerged recently as a multi-target antimalarial agent,¹⁶ which provides added evidence that this compound is a potential drug lead that requires further comprehensive structural modification and understanding of its mechanism of action.

This work was financially supported by the National Science Foundation of China (No. 91213302, 81330077 and 21272291) and the Program for Changjiang Scholars and Innovative Research Team in the University of China (No. IRT1298). We also thank the Shared Hierarchical Academic Research Computing Network (SHARCNET, http://www.sharcnet.ca) for a generous allocation of computer resources.

Notes and references

1. (a) T. A. Brooks and L. H. Hurley, Nat. Rev. Cancer, 2009, 9, 849; (b) S. Balasubramanian, L. H. Hurley and S. Neidle, Nat. Rev. Drug Discovery, 2011, 10, 261; (c) M. L. Bochman, K. Paeschke and V. A. Zakian, Nat. Rev. Genet., 2012, 13, 770.
2. G. Biffi, D. Tannahill, J. McCafferty and S. Balasubramanian, Nat. Chem., 2013, 5, 182.
3. (a) D. Monchaud and M.-P. Teulade-Fichou, Org. Biomol. Chem., 2008, 6, 627; (b) C. Shan, J.-H. Tan, T.-M. Ou and Z.-S. Huang, Sci. China: Chem., 2013, 56, 1351; (c) B. Maji and S. Bhattacharya, Chem. Commun., 2014, 50, 6422.
4. (a) S. Neidle, FEBS J., 2010, 277, 1118; (b) S. A. Ohnmacht and S. Neidle, Bioorg. Med. Chem. Lett., 2014, 24, 2602.
5. R. J. Harrison, J. Cuesta, G. Chessari, M. A. Read, S. K. Basra, A. P. Reszka, J. Morrell, S. M. Gowan, C. M. Incles, F. A. Tanious, W. D. Wilson, L. R. Kelland and S. Neidle, J. Med. Chem., 2003, 46, 4463.
6. X.-D. Wang, T.-M. Ou, Y.-J. Lu, Z. Li, Z. Xu, C. Xi, J.-H. Tan, S.-L. Huang, L.-K. An, D. Li, L.-Q. Gu and Z.-S. Huang, J. Med. Chem., 2010, 53, 4390.
7. L. Martino, A. Virno, B. Pagano, A. Virgilio, S. Di Micco, A. Galeone, C. Giancola, G. Bifulco, L. Mayol and A. Randazzo, J. Am. Chem. Soc., 2007, 129, 16048.
8. (a) D.-L. Ma, T.-S. Lai, F.-Y. Chan, W.-H. Chung, R. Abagyan, Y.-C. Leung and K.-Y. Wong, ChemMedChem, 2008, 3, 881; (b) Q. Li, J. Xiang, X. Li, L. Chen, X. Xu, Y. Tang, Q. Zhou, L. Li, H. Zhang, H. Sun, A. Guan, Q. Yang, S. Yang and G. Xu, Biochimie, 2009, 91, 811; (c) S. Cosconati, L. Marinelli, R. Trotta, A. Virno, L. Mayol, E. Novellino, A. J. Olson and A. Randazzo, J. Am. Chem. Soc., 2009, 131, 16336; (d) H.-M. Lee, D. S.-H. Chan, F. Yang, H.-Y. Lam, S.-C. Yan, C.-M. Che, D.-L. Ma and C.-H. Leung, Chem. Commun., 2010, 46, 4680; (e) S.-B. Chen, J.-H. Tan, T.-M. Ou, S.-L. Huang, L.-K. An, H.-B. Luo, D. Li, L.-Q. Gu and Z.-S. Huang, Bioorg. Med. Chem. Lett., 2011, 21, 1004; (f) D. S.-H. Chan, H. Yang, M. H.-T. Kwan, Z. Cheng, P. Lee, L.-P. Bai, Z.-H. Jiang, C.-Y. Wong, W.-F. Fong, C.-H. Leung and D.-L. Ma, Biochimie, 2011, 93, 1055; (g) D.-L. Ma, D. S.-H. Chan, W.-C. Fu, H.-Z. He, H. Yang, S.-C. Yan and C.-H. Leung, PLoS One, 2012, 7, e43278; (h) S. Alcaro, C. Musetti, S. Distinto, M. Casatti, G. Zagotto, A. Artese, L. Parrotta, F. Moraca, G. Costa, F. Ortuso, E. Maccioni and C. Sissi, J. Med. Chem., 2013, 56, 843; (i) H. R. Nasiri, N. M. Bell, K. I. McLuckie, J. Husby, C. Abell, S. Neidle and S. Balasubramanian, Chem. Commun., 2014, 50, 1704.
9. (a) C. V. Dang, L. M. S. Resar, E. Emison, S. Kim, Q. Li, J. E. Prescott, D. Wonsey and K. Zeller, Exp. Cell Res., 1999, 253, 63; (b) T.-M. Ou, Y.-J. Lu, C. Zhang, Z.-S. Huang, X.-D. Wang, J.-H. Tan, Y. Chen, D.-L. Ma, K.-Y. Wong, J. C.-O. Tang, A. S.-C. Chan and L.-Q. Gu, J. Med. Chem., 2007, 50, 1465; (c) V. Gonzalez and L. H. Hurley, Annu. Rev. Pharmacol. Toxicol., 2010, 50, 111.
10. A. Ambrus, D. Chen, J. Dai, R. A. Jones and D. Yang, Biochemistry, 2005, 44, 2048.
11. A. Canals, M. Purciolas, J. Aymami and M. Coll, Acta Crystallogr., Sect. D: Biol. Crystallogr., 2005, 61, 1009.
12. K. Uytterhoeven, J. Sponer and L. Van Meervelt, Eur. J. Biochem., 2002, 269, 2868.
13. (a) D. Peng, J.-H. Tan, S.-B. Chen, T.-M. Ou, L.-Q. Gu and Z.-S. Huang, Bioorg. Med. Chem., 2010, 18, 8235; (b) S.-B. Chen, Q.-X. Shi, D. Peng, S.-Y. Huang, T.-M. Ou, D. Li, J.-H. Tan, L.-Q. Gu and Z.-S. Huang, Biochim. Biophys. Acta, Gen. Subj., 2013, 1830, 5006.
14. A. T. Phan, V. Kuryavyi, H. Y. Gaw and D. J. Patel, Nat. Chem. Biol., 2005, 1, 167.
15. (a) A. Siddiqui-Jain, C. L. Grand, D. J. Bearss and L. H. Hurley, Proc. Natl. Acad. Sci. U. S. A., 2002, 99, 11593; (b) T.-M. Ou, Y.-J. Lu, C. Zhang, Z.-S. Huang, X.-D. Wang, J.-H. Tan, Y. Chen, D.-L. Ma, K.-Y. Wong, J. C.-O. Tang, A. S.-C. Chan and L.-Q. Gu, J. Med. Chem., 2007, 50, 1465.
16. D. Reker, M. Seet, M. Pillong, C. P. Koch, P. Schneider, M. C. Witschel, M. Rottmann, C. Freymond, R. Brun, B. Schweizer, B. Illarionov, A. Bacher, M. Fischer, F. Diederich and G. Schneider, Angew. Chem., Int. Ed., 2014, 53, 7079.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental procedures, and supplemental tables, spectra and graphs. See DOI: 10.1039/c4cc06951j

---