Colchicine selective interaction with oncogene RET G-quadruplex revealed by NMR

Fei Wang ab, Chunxi Wang ab, Yaping Liu ab, Wenxian Lan ab, Hao Han c, Renxiao Wang ab, Shaohua Huang c and Chunyang Cao *abc
aState Key Laboratory of Bioorganic and Natural Product Chemistry, Center for Excellence in Molecular Synthesis, Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, 345 Lingling Road, Shanghai, 200032, China. E-mail: ccao@mail.sioc.ac.cn; Tel: +86-21-54925491
bUniversity of Chinese Academy of Science, No. 19A, Yuquan Road, Shijingshan District, Beijing, 100049, China
cInstitute of Drug Discovery Technology, Ningbo University, No. 818 Fenghua Road, Ningbo, Zhejiang 315211, China

Received 11th January 2020 , Accepted 24th January 2020

First published on 27th January 2020


G-quadruplexes (G4s) are frequently formed in the promoter regions of oncogenes, considered as promising drug targets for anticancer therapy. Due to high structure similarity of G4s, discovering ligands selectively interacting with only one G4 is extremely difficult. Here, mainly by NMR, we report that colchicine selectively binds to oncogene RET G4-DNA.


Aside from normal double-stranded DNA, the genome is able to fold into a variety of noncanonical three dimensional (3D) structures, for example, G-quadruplexes (G4s) that can interact with small molecules.1 G4s generally occur in G-rich sequences of DNA or RNA, which are characterized by stacks of Hoogsteen-bonded guanine tetrads stabilized by central potassium or sodium ions and flanked by loop regions.1 G4s can arrange as tetramolecular, bimolecular or intramolecular, which are possible by virtue of the changes in strand polarities, the base sequence and the loop topologies. G4 structures can be formed in the human telomere ends and the promoter regions of different oncogenes, such as c-myc,2c-kit,3VEGF4 and Bcl-2.2,5 Small molecules that recognize G4s in the genome with high affinity and selectivity could alter gene expression or regulatory pathways, thus offering potential as therapeutics.6,7 However, challenges to produce these reagents come from the similarity of most G4-DNA or G4-RNA structures, as well as the paucity of atomic level structural information available for G4-small molecule complexes in the RCSB website (http://www.rcsb.org).

RET (RE arranged during Transfection) protein functions as a receptor-type tyrosine kinase and has been found to be important in the initiation and progression of several human cancers.8 It has been reported that RET is overexpressed in breast cancer.9–12 RET expression significantly correlates with larger tumor size, higher tumor stage, decreased metastasis-free survival and lower overall survival.9,10,12 RET inhibition decreases cell growth and metastasis, and sensitizes aromatase inhibitor or tamoxifen treatment in ER positive breast cancer cell.11 This suggests that decreasing RET expression is a potential therapeutic method for the treatment of RET-overexpressing cancers. A highly GC-rich region upstream of the promoter was found to play an important role in the transcriptional regulation of RET.13 A stable G4 structure is formed in this region.14 We previously determined the nuclear magnetic resonance (NMR) solution structure of the major intramolecular G4 formed on the G-rich strand of this region in K+ solution (hereinafter, it was termed as RET G4-DNA for simplicity).15 Our structural studies demonstrate that RET G4-DNA is composed of three stacked G-tetrads, which show distinct features for all parallel-stranded folding topology.15 Its core structure contains one G-tetrad with all syn guanines and two others with all anti-guanines, and three double-chain reversal loops. These characteristics make RET G4-DNA an attractive target amenable to be recognized or stabilized by small molecules. Previously, Lopergolo et al. reported a tri-substituted naphthalene diimide (NDI),16 which significantly impairs RET expression by targeting the RET G4-DNA structure, but this small molecule does not selectively bind to RET G4-DNA. Thus, it's still necessary to discover new small molecules selectively binding to RET G4-DNA as drug design leads for cancer treatment.

It was reported that G4-interactive small molecules are generally composed of three major families.17 The first family has fused aromatic polycyclic systems such as indoloquinoline; the second family mainly contains a macrocycle such as TMPyP4; and the third family members are usually made of non-fused aromatic systems, designated by modular G4-ligands, in which aromatic modules with two or fewer fused rings are combined to create flexible structural motifs. To find ligands specifically binding to RET G4-DNA, firstly, we constructed a database of over 3000 structurally diverse small molecules including some aromatic-ring containing natural products, based on the structural characteristics of G4-interacting small molecules mentioned above. Secondly, an in silico high-throughput docking approach was performed using the reported NMR structure of RET G4-DNA (PDB ID 2L88);15 87 molecules were virtually identified from the library using docking programs GOLD and GLIDE, which potentially interact with RET G4-DNA.

Thirdly, these 87 chemicals were titrated into RET G4-DNA solution at a mole ratio of 1[thin space (1/6-em)]:[thin space (1/6-em)]1, and one-dimensional (1D) 1H-NMR experiments were performed to verify the interaction between DNA and small molecules. Since all chemical compounds were dissolved in isotope labelled d6-DMSO as a stock solution, the titration of solvent d6-DMSO was first performed to eliminate the solvent interference (Fig. S1, ESI). The NMR signals of the imino protons of bound RET G4-DNA were directly assigned based on our previous report of free RET G4-DNA.15Fig. 1A showed that nine compounds were found to produce slight changes in the chemical shifts of imino protons of RET G4-DNA, which have more than one aromatic ring (Fig. S2, ESI). Since almost identical (circular dichroism) CD spectra of RET G4-DNA were obtained (Fig. 1B), their interactions did not change the overall parallel folding or whole secondary structure of RET G4-DNA. Subsequently, we selected c-kit G4-DNA,3,18c-myc G4-DNA2,19 and Tel26wt G4-DNA20 to probe whether these 9 compounds selectively bind to RET G4-DNA. These three G4s have antiparallel and/or parallel topologies, and have some syn bases in G-tetrads, which could represent structural diversity (Fig. S3, ESI). We titrated all these 9 compounds into Tel26wt G4-DNA solution at a mole ratio of 1[thin space (1/6-em)]:[thin space (1/6-em)]1, ran 1D 1H-NMR spectroscopy (Fig. S4, ESI), and found that, except for compound 26, the addition of any of the other 8 compounds results in chemical shift changes of the signals of the imino protons of Tel26wt G4-DNA. So, they were excluded due to non-selectively binding to RET G4-DNA.


image file: d0cc00221f-f1.tif
Fig. 1 Colchicine was screened out to selectively interact with RET G4-DNA. (A) The imino protons region of the 1D 1H NMR spectra of RET G4-DNA titrated with 9 compounds at mole ratio 1[thin space (1/6-em)]:[thin space (1/6-em)]1, respectively. The spectrum of free RET G4-DNA was used as a reference (labelled as free), with the signal assignment of imino protons based on a previous report. (B) CD spectra of free RET G4-DNA and its mixtures with nine chemicals. (C) Assigned signals of imino protons of RET G4-DNA upon being titrated with colchicine at increased mole ratios (RET vs. colchicine). (D) The binding affinity value was fitted using chemical shift changes of protons H1 and H8 against the concentration of colchicine.

Compound 26 is colchicine, a tricyclic, lipid-soluble alkaloid. To further validate its specific interaction with RET G4-DNA, 1D 1H-NMR titration, 1D 1H Carr–Purcell–Meiboom–Gill (i.e., CPMG, which filters out the NMR signals of colchicine in the complex) experiments, saturation transfer difference (STD) experiments and CD spectra were performed. Its addition into c-kit G4-DNA, c-myc G4-DNA and Tel26wt G4-DNA solutions at different mole ratios did not result in any changes in the chemical shift or intensity of the imino proton signals in these G4s, suggesting that colchicine did not interact or interacted with these G4s extremely weakly (Fig. S5, ESI left). In Fig. S6 and S7 (ESI). CPMG spectra showed that the intensity of the signals belonging to colchicine is deduced in a way of DNA dose dependence only when RET G4-DNA is mixed with it. In Fig. 2, only the STD spectrum of colchicine mixed with RET G4-DNA showed some selective enhancement of the proton signals. In contrast, STD spectra of colchicine mixed with c-kit G4-DNA, c-myc G4-DNA and Tel26wt G4-DNA did not display any enhanced signals. Taking all these results together, colchicine potentially prefers to selectively interact with RET G4-DNA in vitro. In Fig. S5 (ESI), right, in the presence of colchicine, RET G4-DNA, Tel26wt G4-DNA, c-kit G4-DNA and c-myc G4-DNA have positive absorption at or close to 264 nm maximal and negative absorption at 245 nm minimal, almost similar to those in the absence of colchicine, indicating unchanged folding of these G4s. Thus, colchicine addition does not lead to obvious changes in the secondary structure of these G4s. Then, we investigated whether colchicine affected cancer cell growth and RET mRNA expression by interacting with RET G4-DNA. Four breast cancer cell lines with different genetic background were used to explore colchicine cytotoxicity. In Fig. 3A, after 3 days treatment, colchicine decreased cell growth in a dose dependent manner, with IC50 values of about 0.001 μM for MCF7, 0.01 μM for T47D, 0.01 μM for MDA-MB-453 and 0.01 μM for BT474, respectively. After 24-hour treatment, RET mRNA expression was decreased by 0.2 to 0.8 folds in all evaluated cells (Fig. 3B), which might be consequent on the fact that colchicine interacts with the G4 structure on the RET promoter and disrupts the mRNA transcription.


image file: d0cc00221f-f2.tif
Fig. 2 Colchicine selective interaction with RET G4-DNA confirmed by saturation transfer difference (STD) experiments. The spectrum of free colchicine (200 μM in 100 mM K+ buffer) is displayed in red lines. The spectra of 200 μM colchicine mixed with 5 μM (A) of RET G4-DNA, (B) of c-kit G4-DNA, (C) of c-myc G4-DNA and (D) of Tel26wt G4-DNA, respectively, are all displayed in blue lines. The irradiation was set at 5.7 ppm, which is in the region of H1’ protons in sugar rings of the bases of RET G4-DNA.

image file: d0cc00221f-f3.tif
Fig. 3 Colchicine displays biological activities probably by selectively interacting with RET G4-DNA. (A) Colchicine decreases breast cancer cell growth. Cancer cells were treated with different concentrations of colchicine for 72 hours. Cell growth was monitored using real-time microscopy and the percentage of surviving cells was normalized by the confluence value of the nontreated cells (control). (B) RET mRNA expression is decreased with colchicine treatment, in which the relative RET mRNA expression was determined by qRT-PCR.

To address how colchicine interacts with RET G4-DNA, we decided to probe the NMR solution structure of colchicine in complex with RET G4-DNA. The binding affinity of colchicine to RET G4-DNA was measured as 1.71 ± 0.13 mM, through NMR chemical shift perturbation (Fig. 1C, D and Fig. S8, ESI). Due to their weak interactions, we made five NMR samples at different mole ratios of 1[thin space (1/6-em)]:[thin space (1/6-em)]1, 1[thin space (1/6-em)]:[thin space (1/6-em)]2, 1[thin space (1/6-em)]:[thin space (1/6-em)]3, 1[thin space (1/6-em)]:[thin space (1/6-em)]4 and 1[thin space (1/6-em)]:[thin space (1/6-em)]8. The total DNA concentration was fixed at 1 mM. Using the KD value, the bound colchicine in each NMR sample was calculated as 0.29 mM, 0.47 mM, 0.59 mM, 0.66 mM and about 0.83 mM, respectively, which was high enough to generate intermolecular NOEs. CPMG experiments (Fig. S6, ESI) demonstrated that 1H NMR signals of bound colchicine just slightly changed. Thus, they could be assigned through two dimensional (2D) 1H–1H TOCSY, COSY and NOESY according to the reported assignments of free colchicine in H2O (ESI, Fig. S9 and Table S1). The NMR signals of bound RET G4-DNA were directly assigned based on our previous report of free RET G4-DNA structure,15 since the chemical shifts of the imino protons were tracked during NMR titration. Through 1H–1H NMR COSY, TOCSY and NOESY spectra of RET G4-DNA in complex with colchicine at different mole ratios in 10% D2O and 100% D2O NMR buffer at 5 °C, 10 °C and 20 °C, respectively, the signals of almost all protons in RET G4-DNA and their intra-molecular NOEs were successfully assigned. 18 inter-molecular NOEs between RET G4-DNA and colchicine were obtained in all 1H–1H NOESY spectra acquired at different molar ratios (Fig. S10, ESI), indicating that colchicine binds to RET G4-DNA at a mole ratio of 1[thin space (1/6-em)]:[thin space (1/6-em)]1. All these intermolecular NOEs are mainly focused on those between colchicine and the G-tetrad at the 3′-end, consistent with the results from NMR titration analysis. No intermolecular NOEs were observed between colchicine and the G-tetrad at the 5′-end, or between colchicine and the central G-tetrad, suggesting that colchicine did not bind to these two G-tetrads.

To correctly calculate the complex structure, the topology parameters of the natural product colchicine were first generated from the website (http://www.gromacs.org/Downloads/Related_Software/PRODRG), slightly modified, and further integrated into the XPLOR program (NIH version). 366 NOE-derived distance constraints, including 18 inter-molecular NOEs, divided into strong (1.8–2.9 Å), medium (1.8–3.5 Å) and weak (1.8–6.0 Å) groups, respectively, were generated based on the intensity of the NOE cross-peaks (mainly in NOESY spectrum at a mole ratio 1[thin space (1/6-em)]:[thin space (1/6-em)]4, 20 °C). 52 hydrogen-bonds were produced to keep each G-tetrad planar conformation in RET G4-DNA, according to NOE patterns between the protons of the guanines within the G-tetrad observed in NOESY spectra and the reported structure of free RET G4-DNA.15 20 dihedral angle restraints were employed to define anti and syn conformations of all the bases in bound RET G4-DNA, based on the correlation of protons in aromatic rings and sugar rings in the NOESY spectrum, which are similar to those used in the determination of the free RET G4-DNA structure. Using these constraints (Table S2, ESI), the solution structure of RET G4-DNA in complex with colchicine was calculated through XPLOR. In Fig. 4A and B, the ensemble of 20 structures with the lowest energy was displayed with Root Mean Square Deviation (RMSD) values of 0.45 ± 0.08 Å for all residues and 0.63 ± 0.21 Å for all heavy atoms, respectively, indicating that a stable and convergent conformation of the complex has been successfully formed.


image file: d0cc00221f-f4.tif
Fig. 4 Structure of RET G4-DNA in complex with colchicine (in orange). (A) The ensemble of 20 structures with the lowest energy. Base G14 is flexible. (B) One conformer of the complex. An angle 27.5° was formed between the aromatic seven-membered ring and phenyl ring. (C) The structure of free RET G4-DNA. (D) The position of colchicine (in orange) relative to G14 (in green) above G3–G9–G13–G19 tetrad (in cyan) at the 3′-end in the complex structure, compared to the structure of free RET G4-DNA (in line mode, except the base G14 which was termed as G14 free and displayed in blue stick). In (A, B and C), the G-tetrad composed by four syn guanines is shown in magenta line and cartoon modes, respectively. The G-tetrads composed by anti-guanines are shown in cyan lines and cartoon modes, respectively. In all figures, base G14 is in deep-gray lines and cartoon modes. Bases G16 and T20 are in green lines and cartoon modes, respectively. Bases G4, C5, G6 and C10 in loops are in wheat lines or cartoon modes. All nonpolar protons were not displayed.

Compared to the structure of free RET G4-DNA (Fig. 4C), the global folding of RET G4-DNA was not changed upon colchicine binding, with an RMSD value of 0.87 Å when three G-tetrads of free and bound RET G4-DNA were superimposed. Colchicine is located above the outer G3–G9–G13–G19 tetrad at the 3′-end by π–π stacking, supported by 11 intermolecular NOEs between protons in colchicine and bases G9, G13 and G19. The seven-membered aromatic ring (ring A, Fig. S9L, ESI) and phenyl ring (ring C, Fig. S9L, ESI) of colchicine form a small angle of about 27.5° (Fig. 4B). All four methoxyl groups and one acetamino (i.e., CH3CONH–) group, the sidechains of colchicine point away from the G3–G9–G13–G19 tetrad, in favour of colchicine binding to RET G4-DNA. The C-10 atom in the middle aliphatic ring (ring B, Fig. S9L, ESI) slightly folds to the G3–G9–G13–G19 tetrad. Colchicine locates on top of three residues G9–G13–G19 (Fig. 4D), consistent with the chemical shift changes of bases G3, G13 and G19 during NMR titration (Fig. 1C). In addition, the amide proton of colchicine locates the deshielding region of bases G9 and G13, which results in rather significant downfield shifts of this proton, compared to that of free colchicine (Table S1, ESI). The hydrogen-bond between G16 and T20 still exists, as the distance of 2.2 Å observed between the imino nitrogen of base G16 and carbonyl group of base T20 was similar to that observed in the reported structure of free RET G4-DNA.15

In free RET G4-DNA, base G14 stacks with G13 due to hydrogen-bond interaction between its amino group and carbonyl group at the C-6 atom of base G19 (Fig. S11A and B, ESI). Its replacement by I14 results in obvious chemical shift changes of imino protons in the 1D 1H-NMR spectrum (Fig. S11C, ESI). The melting temperature of the RET G14I variant was measured as 69.82 °C (at 10 mM K+), smaller than that (75.84 °C, at 10 mM K+) of RET G4-DNA (Fig. S11D, ESI), implying that G14 plays important roles in stabilizing the conformation of RET G4-DNA. While in the RET G4-colchicine complex, the phenyl ring (ring C) of colchicine clashes with the aromatic ring of G14 (Fig. 4D), making G14 leave away from its original position, and its conformation becomes flexible (Fig. 4A). This result is supported by the disappeared NOEs between the protons of residues G14 and C15 in RET G4-DNA (Fig. S11E and F, ESI) and by the observation of the bigger chemical shift changes of bases G14 and C15 than other residues (Fig. S12, ESI). Thus, colchicine diminishes the roles of G14 in stabilizing the conformation of RET G4-DNA. Interestingly, the stability of free RET G4-DNA was found to be dependent on the K+ concentration (S13A). The Tm value of RET G4-DNA mixed with colchicine was similar to that of free RET G4-DNA, even at a mole ratio of 1[thin space (1/6-em)]:[thin space (1/6-em)]4 (Fig. S13B, ESI). This indicated that colchicine did not strengthen the stability of RET G4-DNA. The real reason for this is that colchicine takes up the original position of the base G14 and impairs its role in stabilizing RET G4-DNA. Thus, colchicine selective interaction with RET G4-DNA might result from this unfavourable interaction. We supposed that, if other G4s could provide the same stacking interactions on a G-tetrad for colchicine, but without any other unfavorable effects on the loop residues, such G4s should display a higher colchicine affinity than RET G4-DNA, unless there are additional specific and favorable interactions when binding RET G4-DNA.

It is known that colchicine interferes with microtubule growth, and affects mitosis and other microtubule-dependent functions. It also has unique anti-inflammatory properties that are used in various cardiovascular conditions and other diseases; thus colchicine has multiple mechanisms of its biological functions. Here, we report that colchicine decreases RET expression through selectively binding RET G4-DNA. The loops at the 3′-end of G4-DNA and hyperfine structure of colchicine are key factors of binding selectivity. Our study links colchicine's new mechanism to its anti-cancer activity, suggesting that it can work as a lead compound. Further modification might be helpful to increase colchicine binding activity without interrupting its selectivity. Therefore, our finding enriches the rational design of small molecules targeting G4 structures.

This work was supported by MOST of China (2017YFE0108200 and 2016YFA0502302), by NSFC (21807105, 91753119, 21977110 and 21778065), and by Strategic Priority Research Program, CAS (XDB 20000000). The authors thank facility team members in NCPSS, and HMFL, SIMM, CAS for their help with NMR spectra acquirement.

Conflicts of interest

There are no conflicts to declare.

References

  1. S. Burge, G. N. Parkinson, P. Hazel, A. K. Todd and S. Neidle, Nucleic Acids Res., 2006, 34(19), 5402–5415 CrossRef CAS PubMed .
  2. A. Siddiqui-Jain, C. L. Grand, D. J. Bearss and L. H. Hurley, Proc. Natl. Acad. Sci. U. S. A., 2002, 99(18), 11593–11598 CrossRef CAS PubMed .
  3. V. Kuryavyi, A. T. Phan and D. J. Patel, Nucleic Acids Res., 2010, 38(19), 6757–6773 CrossRef CAS PubMed .
  4. Y. Liu, W. Lan, C. Wang and C. Cao, J. Biol. Chem., 2018, 293(23), 8947–8955 CrossRef CAS PubMed .
  5. M. Nambiar, G. Goldsmith, B. T. Moorthy, M. R. Lieber, M. V. Joshi, B. Choudhary, R. V. Hosur and S. C. Raghavan, Nucleic Acids Res., 2011, 39(3), 936–948 CrossRef CAS PubMed .
  6. S. Neidle, J. Med. Chem., 2016, 59(13), 5987–6011 CrossRef CAS PubMed .
  7. M. Duchler, J. Drug Targeting, 2012, 20(5), 389–400 CrossRef CAS PubMed .
  8. M. Takahashi, Y. Buma, T. Iwamoto, Y. Inaguma, H. Ikeda and H. Hiai, Oncogene, 1988, 3(5), 571–578 CAS .
  9. A. Boulay, M. Breuleux, C. Stephan, C. Fux, C. Brisken, M. Fiche, M. Wartmann, M. Stumm, H. A. Lane and N. E. Hynes, Cancer Res., 2008, 68(10), 3743–3751 CrossRef CAS PubMed .
  10. S. Esseghir, S. K. Todd, T. Hunt, R. Poulsom, I. Plaza-Menacho, J. S. Reis-Filho and C. M. Isacke, Cancer Res., 2007, 67(24), 11732–11741 CrossRef CAS PubMed .
  11. A. Morandi, L. A. Martin, Q. Gao, S. Pancholi, A. Mackay, D. Robertson, M. Zvelebil, M. Dowsett, I. Plaza-Menacho and C. M. Isacke, Cancer Res., 2013, 73(12), 3783–3795 CrossRef CAS PubMed .
  12. A. Gattelli, I. Nalvarte, A. Boulay, T. C. Roloff, M. Schreiber, N. Carragher, K. K. Macleod, M. Schlederer, S. Lienhard, L. Kenner, M. I. Torres-Arzayus and N. E. Hynes, EMBO Mol. Med., 2013, 5(9), 1335–1350 CrossRef CAS .
  13. S. D. Andrew, P. J. Delhanty, L. M. Mulligan and B. G. Robinson, Gene, 2000, 256(1-2), 283–291 CrossRef CAS .
  14. K. Guo, A. Pourpak, K. Beetz-Rogers, V. Gokhale, D. Sun and L. H. Hurley, J. Am. Chem. Soc., 2007, 129(33), 10220–10228 CrossRef CAS PubMed .
  15. X. Tong, W. Lan, X. Zhang, H. Wu, M. Liu and C. Cao, Nucleic Acids Res., 2011, 39(15), 6753–6763 CrossRef CAS PubMed .
  16. A. Lopergolo, R. Perrone, M. Tortoreto, F. Doria, G. L. Beretta, V. Zuco, M. Freccero, M. G. Borrello, C. Lanzi, S. N. Richter, N. Zaffaroni and M. Folini, Oncotarget, 2016, 7(31), 49649–49663 CrossRef PubMed .
  17. A. R. Duarte, E. Cadoni, A. S. Ressurreicao, R. Moreira and A. Paulo, ChemMedChem, 2018, 13(9), 869–893 CrossRef CAS .
  18. D. Wei, J. Husby and S. Neidle, Nucleic Acids Res., 2015, 43(1), 629–644 CrossRef CAS PubMed .
  19. J. Dai, M. Carver, L. H. Hurley and D. Yang, J. Am. Chem. Soc., 2011, 133(44), 17673–17680 CrossRef CAS PubMed .
  20. W. Liu, Y. F. Zhong, L. Y. Liu, C. T. Shen, W. Zeng, F. Wang, D. Yang and Z. W. Mao, Nat. Commun., 2018, 9(1), 3496 CrossRef PubMed .

Footnotes

Electronic supplementary information (ESI) available: Methods, partial results, figures and tables. See DOI: 10.1039/d0cc00221f
Equal contributions.

This journal is © The Royal Society of Chemistry 2020