Discovery of two aminoglycoside antibiotics as inhibitors targeting the menin–mixed lineage leukaemia interface

Abstract

Menin functions as an oncogenic cofactor of mixed lineage leukaemia (MLL) fusion proteins in leukaemo- genesis. The menin–MLL interface is a potential therapeutic target in acute leukaemia cases. In this study, approximately 900 clinical compounds were evaluated and ranked using pharmacophore-based virtual screening, the top 29 hits were further evaluated by biochemical analysis to discover the inhibitors that target the menin–MLL interface. Two aminoglycoside antibiotics, neomycin and tobramycin, were iden- tified as menin–MLL inhibitors with binding affinities of 18.8 and 59.9 lM, respectively. The results of thermal shift assay validated the direct interactions between the two antibiotics and menin. The results of isothermal titration calorimetry showed that the equilibrium dissociation constant between menin and neomycin was approximately 15.6 lM. We also predicted the binding modes of inhibitors at the menin–MLL interface through molecular docking analysis. The results indicated that neomycin and tobramycin competitively occupy the binding site of MLL. This study has shed light on the development of powerful probes and new therapies for MLL-mediated leukaemogenesis.

Mixed-lineage leukaemia (MLL) protein, the human homolog of trithorax in Drosophila, is required to maintain the expression of homeobox (Hox) family genes which are important for normal hae- matopoiesis.1,2 MLL functions as an H3K4 methyltransferase through the C-terminal SET domain in the form of a large multi- protein complex comprising MLL, RbBP5, Ash2L and WDR5.3,4 Dys- function of MLL by chromosomal translocation up-regulates the expression levels of Hoxa7, Hoxa9 and Meis1, thereby enhancing cell proliferation, blocking haematopoietic differentiation and eventually leading to acute leukaemia in adults and children.5–9 Leukaemia patients harbouring MLL gene abnormality have very poor prognosis under current medical conditions that mainly use conventional chemotherapy and stem cell transplantation.5,10 Therefore, new therapeutic methods are urgently needed.

The leukaemogenic activity of mutant proteins is critically dependent on their interactions with menin,11,12 the product of the multiple endocrine neoplasia type 1 (MEN1) tumour suppres- sor gene.13 MEN1 mutation leads to the formation of tumour in endocrine organs, such as parathyroids, enteropancreatic endocrine tissues and the anterior pituitary.14 Menin has signifi- cant functions in suppressing hyperplasia or tumour in other or- gans, including the lungs15 and the prostate.16 This protein also regulates gene transcription. Recruitment of menin by c-Myc, a transcription factor leading to leukaemia, promotes the expression of several genes, such as Hox, Meis1 and Ezh2, along with MLL and LEDGF.4,17,18 The menin–MLL interface is a candidate therapeutic target for novel drugs for acute leukaemia with MLL rearrange- ments. MLL interacts with menin through two N-terminal motifs, the high-affinity motif menin-binding motif 1 (MBM1) and the low-affinity motif MBM2. MBM1 mutation may significantly dis- turb the interaction between MLL and menin.19 The cocrystal structure of menin complexed with MLL peptide (PDB code 4GQ6) reveals that Pro10MLL occupies the hydrophobic pocket surrounded by Ala242menin, Phe238menin, Tyr276menin and Met278menin; hydrogen bonds are located between Ala11MLL and Tyr323menin, Arg12MLL and Glu359menin or Glu363menin.

One potential therapeutic strategy for MLL-rearranged leukaemia is to block the interaction between menin and MLL. In 2012, the first small-molecule inhibitor21 and the second-generation inhibitor with high affinity20 can reverse the oncogenic activity of MLL fusion proteins in leukaemia by targeting the menin–MLL inter- face. These inhibitors share the same scaffold and function in distinct binding modes compared with MLL. Another macrocyclic peptidomimetic inhibitor was subsequently synthesised and opti- mised.22 However, peptidometics with large molecular weights are less likely to penetrate the cell membranes;23 thus, turning these inhibitors into therapeutically useful compounds is challenging.

Identifying inhibitors from existing drugs avoids substantial risks because these drugs have well-known safety and pharmaco- kinetic profiles.24 This process also skips the stages of chemical optimisation, toxicology and formulation development. In the present study, we screened a collection of clinical compounds to identify the inhibitors that target the menin–MLL interface. Unlike small-molecule-binding sites in enzymes, protein–protein inter- faces (PPI) are typically flat and lack deep cavities.25 These inter- faces provide large binding surfaces;26 hence, finding small molecules that target PPI is challenging. Structure-based drug de- sign is an effective, inexpensive and time-saving method that facil- itates the identification of lead compounds.27,28 In particular, pharmacophore modelling, which aims to describe the necessary molecular features of a ligand to effectively bind a receptor, is a well-accepted technique in high-throughput virtual screening.29 Along with the increase in high-resolution protein structures, structure-based pharmacophore has become increasingly impor- tant. This method is useful in situations where the number of bio- logically active molecules is limited, which is the requirement for ligand-based pharmacophores. In the present study, we built an in-house library comprising 900 compounds from existing drugs and applied these compounds to screen inhibitors that target the menin–MLL interface.

Several crystal structures of menin are available in the Protein Data Bank,20,22,30,31 which provides a prerequisite for structure- based pharmacophore modelling. We selected menin in the com- plex with the high-affinity motif (MBM1) of MLL with a resolution of 1.55 Å (PDB code 4GQ6)20 to generate pharmacophore models comprising characteristic interaction patterns between these seg- ments. Considering predefined features (e.g., hydrogen bond acceptor, hydrogen bond donor and hydrophobic) and particular shape constraints (e.g., maximum hydrogen bond distance), we generated the 10 best pharmacophore models sorted by an internal scoring function using Discovery Studio 3.0 (Accelrys Inc., San Diego, CA). Excluded volumes were also added based on the locations of atoms in the protein. The same crystal structure was used to predict the hot spots on the binding surface of menin using HSPred methods.32 Hot spots are specific amino acid residues on the PPI having a marked increase in binding free energy greater than 4.0 kcal/mol if mutated to alanine.33 Hot spots are critical for protein–protein interactions. One model with two hydrophobic groups and a hydrogen bond acceptor was selected based on the combination of fitness score and hot spot analysis and then used as queries against the database.

A set of 29 compounds were selected by the pharmacophore model to conduct fluorescence polarisation (FP) competition assay (Supplementary data).21 FITC-MBM1 at 15 nM and menin at 2 lM in FP buffer (50 mM Tris–HCl, pH 7.5, 50 mM NaCl, 1 mM DTT and 0.1 mg/mL BSA) were incubated for 1 h in the dark at room tem- perature. For primary screening, each compound stock with a final concentration of 200 lM was added to the protein–peptide mix- ture and incubated for another hour. Then, 0.1% DMSO and 40 lM MBM1 were used as negative and positive controls, respectively. Fluorescence polarisation signal was monitored on a POLAR- star Omega microplate reader (BMG) and then used to calculate the inhibition percentages. The mixtures with inhibition percentages not less than 30% from the primary screening proceeded to the determination of binding affinity (Ki) with a series of diluted con- centrations. The results of FP indicated that neomycin and tobra- mycin (Fig. 1A and B) inhibited the binding of FITC-MBM1 to menin with Ki values of 18.8 and 59.9 lM, respectively (Fig. 1C and D). Neomycin and tobramycin are aminoglycoside antibiotics that treat bacterial infections by binding to the bacterial ribosome and interfering with bacterial protein synthesis.

Figure 1. Neomycin and tobramycin compete with MBM1 binding to menin. (A and B) Structures of neomycin (A), tobramycin and kanamycin (B). Tobramycin: R1-NH2, R2-H; kanamycin: R1-OH, R2-OH. (C and D) Neomycin and tobramycin inhibit FITC- MBM1 binding to menin with Ki of 18.8 and 59.9 lM, respectively; data represent mean values for triplicates ± SD (E and F) thermal stability of menin is enhanced by the presence of excessive amounts of neomycin or tobramycin, indicating the direct interaction between menin and neomycin or tobramycin. Data represent mean values for triplicates.

Figure 2. Isothermal titration calorimetry experiment confirms the interaction between neomycin and menin with 1:1 stoichiometry.

We determined the melting temperature (Tm) of menin using differential scanning fluorimetry (Supplementary data) to validate the inhibitors identified by virtual screening. The relative fluorescence intensity of SYPRO orange (Invitrogen) in thermal shift buffer (50 mM Tris–HCl, pH 7.5, 50 mM NaCl and 1 mM DTT) with 2.5 lM menin and different antibiotic concentrations was monitored on a 7500 fast real-time PCR system (ABI). The Tm of menin was fitted using Protein Thermal Shift Software Version 1.1 (ABI). As shown in Figure 1E, the presence of neomycin at 50, 200 and 400 lM shifted the Tm of native menin from 40.46 °C to 42.50 °C, 43.81 °C and 45.19 °C, respectively. For tobramycin, the presence of the compound shifted the Tm of menin from 40.46 °C to 41.43 °C, 42.61 °C and 43.35 °C, respectively (Fig. 1F). Direct interactions were detected between menin and the two antibiotics. Basing on previous FP data, we concluded that neomycin and tobramycin competitively bind to menin against MLL.

To precisely validate our conclusion, we performed isothermal titration calorimetry (Supplementary data) using a Microcal iTC200 isothermal titration calorimeter (GE Healthcare). Freshly purified menin (50 lM) was titrated with 1 mM neomycin and tobramycin in a buffer comprising 50 mM Tris–HCl (pH 7.5) and 150 mM NaCl at 25 °C. The equilibrium dissociation constants (Kd) were fitted using Origin 7 software (OriginLab). As shown in
Figure 2, the Kd between menin and neomycin was 15.6 lM. This result indicates that the binding is specific. The binding interaction is driven cooperatively by entropy and enthalpy (DH= —2.6 kcal/ mol, TDS = 3.95 kcal/mol). Kd cannot be determined for tobramycin because of its weak affinity (data not shown).

Biophysical analysis showed that the inhibitors competitively bound to menin. Molecular docking was performed using GLIDE program to analyse the implicit binding modes and key interac- tions of these potent inhibitors with menin. The initial structures processed by LigPrep and Epik were used to calculate the possible protonation states at pH 7.5. Hydrogen atoms and charges were added using the Protein Preparation module of GLIDE. During the docking process, extra-precision docking was adopted to generate the minimised poses, and the GLIDE scoring function was used to select the final pose for each ligand.35,36 As shown in Figure 3, both tobramycin and neomycin formed a hydrogen bond with Tyr323, and the 30 -deoxyneamine moiety of tobramycin fit in a hydropho- bic pocket shared by Met278, Tyr319, Met322 and Tyr323. These residues are important for the interaction between MLL and menin; Met278 and Tyr323 mutations cause more than 100-fold decrease in activity.20,30,31 Introduction of a hydroxyl group at position 3 of the 2,6-diamino sugar ring in kanamycin generates clashes with neighbouring residues. As a result, kanamycin has lower affinity compared with tobramycin. For neomycin, the 60 -NH2 group of neamine core has a strong propensity to form bifurcated hydrogen bonds with Glu359 and Glu363 when protonated at pH 7.5. In addition, the protonated 30 -NH2 of the 2- deoxystreptamine moiety is more likely to interact with Glu363 via salt bridge, accounting for the better affinity of neomycin than tobramycin. These results suggest that tobramycin and neomycin occupy the binding site by mimicking the binding mode between menin and MLL.

Figure 3. Predicted binding modes of neomycin (lower panel) and tobramycin (upper panel) with menin from docking analysis. The menin surface is colored according to electrostatic potential (left). Neomycin (orange), tobramycin (magenta), as well as surrounding residues (marine) are shown as sticks (middle). Schematic diagrams of predicted interactions (right): antibiotics and menin are colored in purple and orange respectively. Hydrogen bonds are shown in green dashes and hydrophobic contacts are plotted in red arcs.

In summary, we have identified tobramycin and neomycin as novel inhibitors against menin–MLL interaction through struc- ture-based virtual screening and biochemical evaluation of approximately 900 clinical compounds. The results of biochemical and biophysical experiments combined with molecular docking analysis revealed that these antibiotics inhibit menin by competing with MLL at the binding site. Basing on the results of docking anal- ysis, we assumed that the hydrophobic pocket formed by Met322, Tyr323, Tyr319 and Met278 is critical for interaction. Specifically, neomycin increases affinity by forming bonds with Glu363 and Glu359. The pharmacophore model, which precisely contains two hydrophobic groups and a hydrogen bond acceptor, could be em- ployed to discover other potential inhibitors. This study demon- strates an efficient and cost-effective virtual screening procedure that can be used to identify novel inhibitors that target the me- nin–MLL interface. Moreover, the results of this study could serve as an excellent platform for further development of menin–MLL inhibitors with greater potency and lower toxicity for use as ther- apeutic agents against MLL-rearranged leukaemia.

Acknowledgments

We gratefully acknowledge financial support from the Hi-Tech Research and Development Program of China (2012AA020302), the National Natural Science Foundation of China (91229204, 81230076, and 21210003), the National Science and Technology Major Project ‘‘Key New Drug Creation and Manufacturing Program’’ (2013ZX09507-004, 2013ZX09507001 and 2014ZX09507002-005-012) and the Grant from State Key Laboratory of Drug Research (SIMM1403KF-06).

Supplementary data

Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.bmcl.2014.03. 055.

References and notes

1. Tkachuk, D. C.; Kohler, S.; Cleary, M. L. Cell 1992, 71, 691.
2. Hess, J. L. Crit. Rev. Eukaryot. Gene Expr. 2004, 14, 235.
3. Yokoyama, A.; Wang, Z.; Wysocka, J.; Sanyal, M.; Aufiero, D. J.; Kitabayashi, I.; Herr, W.; Cleary, M. L. Mol. Cell. Biol. 2004, 24, 5639.
4. Dou, Y.; Milne, T. A.; Ruthenburg, A. J.; Lee, S.; Lee, J. W.; Verdine, G. L.; Allis, C. D.; Roeder, R. G. Nat. Struct. Mol. Biol. 2006, 13, 713.
5. Sorensen, P. H.; Chen, C. S.; Smith, F. O.; Arthur, D. C.; Domer, P. H.; Bernstein, I. D.; Korsmeyer, S. J.; Hammond, G. D.; Kersey, J. H. J. Clin. Invest. 1994, 93, 429.
6. Rozovskaia, T.; Feinstein, E.; Mor, O.; Foa, R.; Blechman, J.; Nakamura, T.; Croce,
C. M.; Cimino, G.; Canaani, E. Oncogene 2001, 20, 874.
7. Imamura, T.; Morimoto, A.; Takanashi, M.; Hibi, S.; Sugimoto, T.; Ishii, E.; Imashuku, S. Br. J. Haematol. 2002, 119, 119.
8. Ayton, P. M.; Cleary, M. L. Genes Dev. 2003, 17, 2298.
9. Cox, M. C.; Panetta, P.; Lo-Coco, F.; Del Poeta, G.; Venditti, A.; Maurillo, L.; Del Principe, M. I.; Mauriello, A.; Anemona, L.; Bruno, A.; Mazzone, C.; Palombo, P.; Amadori, S. Am. J. Clin. Pathol. 2004, 122, 298.
10. Eguchi, M.; Eguchi-Ishimae, M.; Greaves, M. Int. J. Hematol. 2003, 78, 390.
11. Yokoyama, A.; Somervaille, T. C.; Smith, K. S.; Rozenblatt-Rosen, O.; Meyerson, M.; Cleary, M. L. Cell 2005, 123, 207.
12. Caslini, C.; Yang, Z.; El-Osta, M.; Milne, T. A.; Slany, R. K.; Hess, J. L. Cancer Res.
2007, 67, 7275.
13. Chandrasekharappa, S. C.; Guru, S. C.; Manickam, P.; Olufemi, S. E.; Collins, F. S.; Emmert-Buck, M. R.; Debelenko, L. V.; Zhuang, Z.; Lubensky, I. A.; Liotta, L. A.; Crabtree, J. S.; Wang, Y.; Roe, B. A.; Weisemann, J.; Boguski, M. S.; Agarwal, S. K.; Kester, M. B.; Kim, Y. S.; Heppner, C.; Dong, Q.; Spiegel, A. M.; Burns, A. L.; Marx, S. J. Science 1997, 276, 404.
14. Chandrasekharappa, S. C.; Teh, B. T. J. Intern. Med. 2003, 253, 606.
15. Gao, S. B.; Feng, Z. J.; Xu, B.; Wu, Y.; Yin, P.; Yang, Y.; Hua, X.; Jin, G. H. Oncogene
2009, 28, 4095.
16. Seigne, C.; Fontaniere, S.; Carreira, C.; Lu, J.; Tong, W. M.; Fontaniere, B.; Wang,
Z. Q.; Zhang, C. X.; Frappart, L. Bmc Cancer 2010, 10, 395.
17. Jin, S.; Zhao, H.; Yi, Y.; Nakata, Y.; Kalota, A.; Gewirtz, A. M. J. Clin. Invest. 2010,
120, 593.
18. Yokoyama, A.; Cleary, M. L. Cancer Cell 2008, 14, 36.
19. Grembecka, J.; Belcher, A. M.; Hartley, T.; Cierpicki, T. J. Biol. Chem. 2010, 285, 40690.
20. Shi, A.; Murai, M. J.; He, S.; Lund, G.; Hartley, T.; Purohit, T.; Reddy, G.; Chruszcz, M.; Grembecka, J.; Cierpicki, T. Blood 2012, 120, 4461.
21. Grembecka, J.; He, S.; Shi, A.; Purohit, T.; Muntean, A. G.; Sorenson, R. J.; Showalter, H. D.; Murai, M. J.; Belcher, A. M.; Hartley, T.; Hess, J. L.; Cierpicki, T. Nat. Chem. Biol. 2012, 8, 277.
22. Zhou, H.; Liu, L.; Huang, J.; Bernard, D.; Karatas, H.; Navarro, A.; Lei, M.; Wang, S. J. Med. Chem. 2013, 56, 1113.
23. Wells, J. A.; McClendon, C. L. Nature 2007, 450, 1001.
24. Ashburn, T. T.; Thor, K. B. Nat. Rev. Drug Disc. 2004, 3, 673.
25. Pettit, F. K.; Bowie, J. U. J. Mol. Biol. 1999, 285, 1377.
26. Horton, N.; Lewis, M. Protein Sci. 1992, 1, 169.
27. Lyne, P. D. Drug Discovery Today 2002, 2, 1047. 28. Hou, T.; Xu, X. Curr. Pharm. Des. 2004, 10, 1011.
29. Yang, S. Y. Drug Discovery Today 2010, 15, 444.
30. Huang, J.; Gurung, B.; Wan, B.; Matkar, S.; Veniaminova, N. A.; Wan, K.; Merchant, J. L.; Hua, X.; Lei, M. Nature 2012, 482, 542.
31. Murai, M. J.; Chruszcz, M.; Reddy, G.; Grembecka, J.; Cierpicki, T. J. Biol. Chem.
2011, 286, 31742.
32. Lise, S.; Buchan, D.; Pontil, M.; Jones, D. T. PLoS ONE 2011, 6.
33. Pons, J.; Rajpal, A.; Kirsch, J. F. Protein Sci. 1999, 8, 958.
34. Le Goffic, F.; Capmau, M. L.; Tangy, F.; Baillarge, M. Eur. J. Biochem. 1979, 102, 73.
35. Friesner, R. A.; Banks, J. L.; Murphy, R. B.; Halgren, T. A.; Klicic, J. J.; Mainz, D. T.; Repasky, M. P.; Knoll, E. H.; Shelley, M.; Perry, J. K.; Shaw, D. E.; Francis, P.; Shenkin, P. S. J. Med. Chem. 2004, 47, 1739.
36. Friesner, R. A.; Murphy, R. B.; Repasky, M. P.; Frye, L. L.; Greenwood,Menin-MLL Inhibitor J. R.; Halgren, T. A.; Sanschagrin, P. C.; Mainz, D. T. J. Med. Chem. 2006, 49, 6177.