- Email: [email protected]
Identification by Inverse Virtual Screening of magnolol-based scaffold as new tankyrase-2 inhibitors
Bioorganic & Medicinal Chemistry 26 (2018) 3953–3957
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry journal homepage: www.elsevier.com/locate/bmc
Identification by Inverse Virtual Screening of magnolol-based scaffold as new tankyrase-2 inhibitors
T
Simone Di Miccoa, Luana Pulvirentib, Ines Brunoa, Stefania Terraccianoa, Alessandra Russoa, Maria C. Vaccaroa, Dafne Ruggieroa,c, Vera Muccillib, Nunzio Cardullob, Corrado Tringalib, ⁎ Raffaele Riccioa, Giuseppe Bifulcoa, a
Department of Pharmacy, University of Salerno, Via Giovanni Paolo II 132, 84084 Fisciano, Italy Department of Chemical Sciences, University of Catania, Viale Andrea Doria 6, 95125 Catania, Italy c PhD Program in Drug Discovery and Development, University of Salerno, Via Giovanni Paolo II 132, I-84084 Fisciano, SA, Italy b
A R T I C LE I N FO
A B S T R A C T
Dedicated to the memory of our dear colleague and friend Carmela Spatafora.
The natural product magnolol (1) and a selection of its bioinspired derivatives 2–5, were investigated by Inverse Virtual Screening in order to identify putative biological targets from a panel of 308 proteins involved in cancer processes. By this in silico analysis we selected tankyrase-2 (TNKS2), casein kinase 2 (CK2) and bromodomain 9 (Brd9) as potential targets for experimental evaluations. The Surface Plasmon Resonance assay revealed that 3–5 present a good affinity for tankyrase-2, and, in particular, 3 showed an antiproliferative activity on A549 cells higher than the well-known tankyrase-2 inhibitor XAV939 used as reference compound.
Keywords: Tankyrase-2 Inverse Virtual Screening Biomimetic compounds Tankyrase-2 inhibitors Anti-cancer drugs
1. Introduction Inverse Virtual Screening (IVS) aims to identify putative macromolecular targets by computer-aided methodologies.1,2 In this field, an in silico protocol has been developed by our research group,3–5 consisting in a virtual screening of a single compound against a library of proteins by molecular docking and a normalization of the energy values as selection criteria for the macromolecule identification. It represents a rapid and cost efficient methodology for the identification of one or more potential biological targets for a small molecule. This approach, successfully applied in other cases reported in literature,3,6–8 is particularly suitable for natural products, which are well-known to represent a precious source of new chemical skeletons endowed with a wide array of biological activities, whose biological targets are, most of the times, unknown. Recently, we have also demonstrated the applicability of our IVS protocol to the analysis of a synthetic compound library validating our in silico strategy for lead repurposing.9 In this context, the biomimetic syntheses allow to broaden the chemical diversity of the natural scaffold, providing the possibility of increasing the bioactivities, as already reported in literature.10–15 The dimeric neolignan magnolol (1, Fig. 1) has been recently the focus of bioinspired investigations by some of us in order to develop a collection of structurally related derivatives, including compounds 2–5;
⁎
Corresponding author. E-mail address: [email protected] (G. Bifulco).
https://doi.org/10.1016/j.bmc.2018.06.019 Received 5 March 2018; Received in revised form 12 June 2018; Accepted 13 June 2018
Available online 20 June 2018 0968-0896/ © 2018 Elsevier Ltd. All rights reserved.
these were obtained employing both enzymatic dimerization in the presence of horseradish peroxidase (HRP), and regioselective orthohydroxylation with the environmentally benign reagent IBX (2-iodoxybenzoic acid).13 In the present contribution, we investigated by IVS 1 and its derivatives 2–5, succeeding in identifying new inhibitors of tankyrase-2. This macromolecule, along with the isoform 1, is a member of the PARP (poly ADP-ribose polymerases) family of proteins, acting as mono- or poly-ADP-ribosyltransferase and is involved in different biological processes such as telomere elongation and Wnt signal transduction pathway.16,17 Owing to its involvement in different key cellular pathways, tankyrase-2 plays an important role in cancer, resulting as an attractive target for tumor treatment.18–21 2. Results and discussion 2.1. Inverse Virtual Screening Briefly, our IVS protocol consists in molecular docking calculations of a small molecule against a panel of proteins, obtaining a predicted binding energy [V0 (kcal/mol)] for each ligand-protein system. In parallel, we virtually test decoys, namely compounds showing molecular weights and H-bonds donors/acceptors similar to the investigated compound, against the protein panel. For each macromolecule, the
Bioorganic & Medicinal Chemistry 26 (2018) 3953–3957
S. Di Micco et al.
that is used to predict the most promising interacting targets for the case-study small molecule, ranking all the obtained V values from the highest to the lowest one. Such normalization is adopted to prevent the selection of false positives, as emerged by our previous studies.4–9 In the present work, 1–5 were virtually screened on a panel of 308 models of proteins involved in tumor processes, and 162 decoys (“blanks” compounds) were used for the normalization of the docking results.6 In details, for each small molecule (1–5), we obtained the predicted binding energies (see experimental sections) against the 308 protein models from molecular docking calculations. Each predicted value for a specific target was normalized with the corresponding averaged binding energy of the decoys. The normalized values of 1–5 were ranked from the highest (first position in the ranking) to the lowest one (last position in the ranking) obtaining an affinity profile on the whole protein panel (Tables S1–S5) for the small molecules. The next step was the analysis of docked conformations by visual inspection. In particular, we arbitrarily limited the analysis of docked outcomes for the targets ranked in the first 20 positions for each compound (1–5) classification in order to identify macromolecular candidates for the experimental assays.9 From this analysis, putative macromolecular targets were identified for each investigated compound. By comparing the obtained results for 1–5, we observed that each of the small molecules (1–5) targeted the proteins tankyrase-2 (TNKS2), casein kinase 2 (CK2) and bromodomain 9 (Brd9) at the same time. In particular, tankyrase-2 is ranked in the first ten best targets for 1, 2, 4 and 5 (Tables S1, S2, S4, S5). For the compound 3, we found the tankyrase-2 at position twelve of the normalized values (Table S3). Similar considerations were made for casein kinase 2. Indeed, for 1–4 this protein is in the first ten positions (Tables S1–S4), whereas, for 5 it is classified at position fifteen (Table S5). The bromodomain 9 was found at position four and twelve
Fig. 1. Molecular structures of the natural product magnolol (1) and of its synthetic derivatives 2–5.
average value VR (kcal/mol) of binding energies of the decoys is calculated. The resulting ratio between V0 and VR is a dimensionless number V:
V = V0/ VR
(1)
Fig. 2. Three-dimensional model of the interactions given by 1 (a), 2 (b), 3 (c), 4 (d) and 5 (e) with tankyrase-2. Superimposition of 3–5 with co-crystallized XAV939 (f, PDB ID 3KR8). The protein is depicted by tube (colored: C, grey; polar H, white; N, dark-blue; O, red; S, yellow). The small molecules are represented by sticks (pink for 1, cyan for 2, brown for 3, dark green for 4, kaki for 5, aquamarine for XAV939) and balls (colored: C, as for the sticks; polar H, white; N, dark-blue; O, red; F, green; S, yellow). The dashed black lines indicate the hydrogen bonds between ligand and protein. 3954
Bioorganic & Medicinal Chemistry 26 (2018) 3953–3957
S. Di Micco et al.
for 4 (Table S4) and 2 (Table S2), respectively. For the remaining compounds, the normalized value of bromodomain 9 was found at position nineteen (Tables S1, S3, S5). Hence, our investigation suggested that the magnolol-based compounds could potentially fit the binding pockets of the above reported three macromolecules. For sake of simplicity, we describe in details the interactions of 1–5 with tankyrase-2, which gave the best experimental results. The docking outcomes showed that the compounds 1–5 are well accommodated into the binding pocket of tankyrase-2. We observed that the common structural moiety of 1–5, constituted by two aromatic rings and two hydroxyl groups at C-2 and C-2’ (1, Fig. 1) occupied equivalent spatial portions of the binding cavity. In particular, the two phenyl rings presented a normal spatial arrangement respect to each other, and were involved in π-π interactions with His1031, Tyr1050 and Tyr1071 (Fig. 2). The hydroxyl at C-2 established two hydrogen bonds with the backbone amide of Gly1032 (Fig. 2). The OH group at C-2’ was double bonded with the peptide backbone of Tyr1060 (Fig. 2). It is noteworthy that the two H-bonds with Gly1032 and the π-π contacts with both Tyr1050 and Tyr1071 have been already observed (Fig. 2f) for co-crystallized (PDB ID: 3KR8) ligand (XAV939) of tankyrase-222 (Fig. 2f). The two linear chains of 1–5 contribute to the complex stability by van der Waals contacts with Pro1034, Tyr1050, Tyr1060 and Ile1075. The compounds 2–5 structurally differ from the natural product 1 for the substituents at positions C-3, C3’, C-5 and C-5’, which determine further interactions with the protein residues. The compound 2 presented a superimposable docked pose respect to 1, giving the same interactions of the natural product. The derivative 2 differs from 1 for the presence of an additional hydrogen bond with the side chain of Ser1068 established by the OH group at C-3. This H-bond is also found for 3 exerted by its methoxy group at C-3. The hydroxyl group at C3’ of 3 donates an H-bond to the side chain of His1031 and accepts an H-bond from the side chain of Ser1033. The derivative 4 gives the same interactions observed for 3 with the common structural moieties. The methoxy group at C3’ accepts an H-bond from the side chain of Ser1033. Moreover, compound 4 establishes further H-bonds by hydroxyl groups on the linear chains with the side chain of Tyr1071 and NH of Gly1074. The compound 5 is the acetylated derivative of 4. The acetyl group at C-5 made two hydrogen bonds with Met1054 and Tyr1071, whereas, the other acetyl is H-bonded to Ile1075. This docking analysis showed, thus, that the substituents introduced to obtain 2–5 established favorable interactions with protein residues suggesting an improved affinity respect to the natural compound 1, as experimentally confirmed.
Table 1 Thermodynamic constants measured by SPR for the interaction between tested compounds (1–5) and immobilized tankyrase-1, tankyrase-2, casein kinase 2 and PARP1. XAV939 and TBB were used as reference compounds for binding to TNKS2 and CK2. KD (µM) ± SD Compound
Tankyrase-2
Casein kinase 2
Tankyrase-1
PARP1
1 2 3 4 5 XAV939 TBB
no binding no binding 0.021 ± 0.0041 0.0185 ± 0.0007 0.0065 ± 0.0007 0.0042 ± 0.0008 –
no binding no binding no binding no binding no binding – 0.79 ± 0.0201
no binding 5.26 ± 1.46 0.52 ± 0.28 no binding no binding – –
no binding 1.25 ± 0.2 no binding no binding no binding – –
PARP1. For what concerns TNKS1, only compounds 2 and 3 showed a binding for this isoform (Table 1). In particular, compound 3 presented a binding toward TNKS1 and TNKS2, but with higher affinity against isoform 2. Indeed, the measured KD vs. TNKS1 is 25 times larger than the value for TNKS2 (Table 1). Concerning the PARP1 outcomes, compound 2 showed a binding affinity with the tested protein with low micromolar KD value (see Table 1), confirming the good selectivity of the previously disclosed compounds for TANK2 protein. Overall, the experimental outcomes qualitatively confirm the predicted analysis through our IVS protocol. 2.3. Cell-based assays TNKS2 inhibition is well-known to induce a reduction in β-catenindependent signaling in cells with a hyperactive Wnt pathway. As the hyperactivation of this pathway has been documented in samples from aggressive lung cancer and adenocarcinoma,30 we explored the effects of 3–5 and XAV939 on growth of A549 cells. The exposure of A549 cells (1% FBS, Fig. 3), for 72 h, to increasing concentration of compound 3 and XAV939 affects the cell proliferation with comparable IC50 values of 4.7 ± 0.4 and 12.3 ± 1.5, respectively (Table 2 and Fig. 3). In particular, the compound 3 showed a better activity than the reference compound XAV939. Conversely, the compounds 4 and 5 affect the cell viability by 20–30% when used at high concentrations, even though SPR data showed a better affinity of these compounds compared to 3. This behavior could be ascribed to a different cell-permeability. Indeed, we rationalized the experimental results by calculating the permeability
2.2. Surface Plasmon Resonance spectroscopy analyses In order to validate the IVS results, we synthesized compounds 2–5 as previously reported,13 to verify experimentally the binding affinity toward the proposed targets of the natural lead 1 and the synthetic analogues 2–5. Thus, compounds 1–5 were subjected to Surface Plasmon Resonance (SPR)23–26 screening on the immobilized proteins tankyrase-2 and casein kinase 2. The reported TNKS2 and CK2 inhibitors, XAV93927,28 and TBB29 were also tested as positive control. The experimental outcomes showed no binding towards CK2 (Table 1). Based on this assay, the natural compound 1 and 2 were unable to bind the TNKS2; conversely, compounds 3–5 have been identified as high affinity leads for the TNKS2 protein with low micromolar KD values (see Table 1). The compound 5 presented the lowest KD, followed by 4 and 3. It is noteworthy that we found for 5 a KD comparable with XAV939 value. The experimental outcomes confirmed the theoretical predictions, in fact, in the docking analysis, 5 showed the highest number of H-bonds with protein residues and presented the best affinity towards tankyrase-2. In addition, we evaluated the binding toward Brd9 and other related proteins, but negative results emerged (data not shown). Based on the obtained experimental results, we also investigated the putative binding of 1–5 towards two highly structural related isoforms of TNKS2 included in our protein panel: tankyrase-1 (TNKS1) and
Fig. 3. A549 cells were treated for 72 h with the indicated concentrations of 3 and XAV939 in medium with low serum (1.0% FBS) and the cell growth was determined by MTT assay. Data are the means ± SD of two independent experiments. 3955
Bioorganic & Medicinal Chemistry 26 (2018) 3953–3957
S. Di Micco et al.
1.5.634 and added Gasteiger charges and merged non polar hydrogens. The 308 protein 3D models were prepared starting from the deposited structures in the Protein Data Bank database (www.rcsb.org)35 and from homology modelling.36,37 Water molecules were removed by Maestro 11, and .pdb files obtained were then processed with Autodock Tools 1.5.6 and converted in .pdbqt format, merging nonpolar hydrogens and adding Gasteiger charges. Charge deficit was spread over all atoms of related residues. For molecular docking calculations, Autodock Vina software38 was employed, with exhaustiveness value of 64, and saving 10 conformations as the maximum number of binding modes. In docking calculations, 1–5 were treated as flexible and proteins as rigid. Each grid box was centered on binding site of the proteins and the grid points spaced by 1.0 Å. For all the investigated compounds, all bonds were treated as active torsional bonds except double and ring bonds. The analysis of docking outcomes were carried out by Autodock Tools 1.5.6. Predicted binding energies were normalized as previously described6 by using the Eq. (1). Maestro 11 was used to generate the Figures of the 3D models. Predicted apparent permeability was calculated by QuikProp,39 using default parameters and Caco-2 cells as model. The reference calculated values are: < 25 nm/s poor, > 500 nm/s great).39
Table 2 Exposure to TNKS-2 inhibitor compounds affected the A549 cell growth. IC50 (µM) value after 72 h treatment. Compound
1% FBS
3 4 5 XAV939
4.7 ± 0.4 n.i.a n.i.a 12.3 ± 1.5
a
n.i., no inhibition.
properties of 3–5. The predicted data showed a better permeability for 3 (842.518 nm/s) compared to 4 (216.806 nm/s) and 5 (79.995 nm/s), confirming our hypothesis. As expected, compounds 4 and 5 showed comparable cellular activity. Indeed, 5 can be considered a prodrug of 4 because the acetyl groups can be hydrolysed in cell releasing 4. 3. Conclusions In the present work we propose the 1,1′-biphenyl-2,2′-diol scaffold, derived from the natural product magnolol (1, Fig. 1), for the development of new tankyrase-2 inhibitors. By the application of our Inverse Virtual Screening protocol, we filtered a library of 308 putative biological targets for compounds 1–5, proposing for the experimental tests the thankyrase-2, casein kinase 2 and bromodomain 9. The experiments showed a binding towards tankyrase-2 by 3–5, presenting a KD in the low micromolar range. It is noteworthy that 5 showed a comparable KD respect to the reference compound XAV939. The careful molecular docking analysis highlighted the key interactions given by the common 1,1′-biphenyl-2,2′-diol structural moiety for 1–5, mainly represented by π-π interactions with His1031, Tyr1050 and Tyr1071 and H-bonds with Gly1032. Moreover, the analysis of the docked poses, along with the relative biological results pointed that the H-bonds network formed by substituents at C-3, C-3′, C-4 and C-4′ increase the binding affinity towards tankyrase-2, as experimentally confirmed. The anti-proliferative assays, on A549 cells, revealed a better activity of 3 compared to XAV939. Unfortunately, despite the promising KD values obtained by SPR experiments, the cell-based tests showed an absence of antiproliferative activity for 4 and 5. This finding could be ascribed to a lower/scarce membrane permeability as predicted by QikProp calculations. The presented data confirm the Inverse Virtual Screening as a valuable tool to identify the molecular targets of small molecules endowed with interesting biological activities. Moreover, the use of a biosynthetic approach combined with in silico methodology provides a fast and cost efficient lead identification in drug discovery processes. It could give the opportunity of re-evaluating the biological properties of natural products already used in the traditional medicine, as just in the case of magnolol.
4.2. SPR experiments Recombinant Tankyrase-2 (TNKS2) was purchased from MyBioSource (MyBioSource Inc., San Diego, USA). The TNKS2 inhibitor 3,5,7,8-Tetrahydro-2-[4-(trifluoromethyl)phenyl]-4H-thiopyrano-[4,3d]pyrimidin-4-one (XAV939) was purchased from Santa Cruz Biotechnology, Inc. Recombinant Casein Kinase II (CK2) was purchased from New England Biolabs (New England Biolabs Inc., Massachusetts, USA). The CK2 inhibitor 4,5,6,7-Tetrabromo-2-azabenzimidazole (TBB) was purchased from Santa Cruz Biotechnology, Inc. Recombinant TNKS1 (TRF1-interacting ankyrin-related ADP-ribose polymerase 1), also known as PARP5A, TANK1, or tanykyrase-1 was purchased from Sigma Aldrich. Recombinant Poly(ADP-ribose) polymerase 1 (PARP-1) was purchased from Enzo Life Sciences, Inc. SPR analyses were carried out according to our previously published data.23–26 Surface Plasmon Resonance Spectroscopy (SPR) analyses were performed to determine binding of the various molecules to the TNKS1, TNKS2, CK2 and PARP1 proteins using a Biacore 3000 optical biosensor equipped with researchgrade CM5 sensor chips (GE Healthcare). TNKS1, TNKS2, CK2 and PARP1 were coupled to the surface of a CM5 sensor chip using standard amine-coupling protocols, according to the manufacturer’s instructions. The proteins TNKS2 and CK2 (100 µg ml−1 in 10 mM CH3COONa, pH 7.3) were immobilized on individual sensor chip surfaces at a flow rate of 5 µL min−1 to obtain densities of 8–12 kRU. PARP1 (100 µg mL−1 in 10 mM CH3COONa, pH 7.5) was immobilized on individual sensor chip surfaces at a flow rate of 5 µL min−1 to obtain densities of 8–12 kRU. TNKS1 (100 µg mL−1 in 10 mM CH3COONa, pH 4.5) was immobilized on individual sensor chip surfaces at a flow rate of 5 µL min−1 to obtain densities of 8–12 kRU. For the experiments, the recombinant TNKS1, TNKS2 and PARP1 surfaces, a BSA surface and one unmodified reference surface were prepared for simultaneous analyses. For TNKS2 analysis, compounds 1–5 and XAV939 were dissolved to obtain 20 mM solution concentrations, in 100% DMSO and diluted 1:1000 (v/v) in PBS (10 mM NaH2PO4, 150 mM NaCl, pH 7.4) to a final DMSO concentration of 0.5%. For CK2 analysis, compounds 1–5 as well as TBB were dissolved to obtain 20 mM solutions, in 100% DMSO and diluted 1:1000 (v/v) in PBS (10 mM NaH2PO4, 150 mM NaCl, pH 4.5) to a final DMSO concentration of 0.5%. For each molecule a five-point concentration series were set up, including 0 – 0.25 – 1 – 10 – 20 µM, and, for each sample the complete binding study was performed using triplicate aliquots. Changes in mass, due to the binding response, were recorded as resonance units (RU). To obtain the dissociation constant (KD) (Table 1), these responses were fit to a 1:1 Langmuir binding model by nonlinear regression, using the BiaEvaluation software
4. Experimental section Magnolol (1) was purchased from TCI Europe N.V. (Zwijndrecht, Belgium). Compounds 2–5 were synthesized as previously reported.13 4.1. Computational details The three dimensional structures of the investigated compounds (1–5) were built by Build Panel of Maestro (version 11) and minimized by means of OPLS3 force field31 and the Polak-Ribier conjugate gradient algorithm (maximum derivative less than 0.001 kcal/mol), in water by the GB/SA (generalized Born/surface area)32 solvent treatment. Then, the small molecules were processed with LigPrep,33 considering protonation states at a pH of 7.0 ± 1.0. The minimized geometries of 1–5 were converted in the .pdbqt format by Autodock Tools 3956
Bioorganic & Medicinal Chemistry 26 (2018) 3953–3957
S. Di Micco et al.
A. Supplementary data
program provided by GE Healthcare. Simple interactions were suitably fitted to a single-site bimolecular interaction model (A + B = AB), yielding a single KD (Table 1). SPR experiments were performed at 25 °C, using a flow rate of 50 µL min−1, with 60 s monitoring of association and 300 s monitoring of dissociation.
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.bmc.2018.06.019. References
4.3. Cell culture and proliferation assay 1. 2. 3. 4. 5.
A549 (human lung carcinoma) cell line was obtained from ECACC European Cell Lines Collection, England. The cells purchased from Invitrogen (Carslbad, CA, USA), were grown in DMEM/high glucose supplemented with heat-inactivated FBS (10%, v/v), 2 mM L−glutamine, penicillin (100 U/mL) and streptomycin (100 mg/mL) at 37 °C in humidified atmosphere with 5% CO2. To ensure logarithmic growth, the cells were subcultured every 2 days. Stock solutions of compounds 3–5 and XAV939 (100 mM in DMSO) were stored at -20 °C in the dark and diluted just before addition to the sterile culture medium. In all the experiments, the final concentration of DMSO was 0.10% (v/v). The proliferation assay was performed with MTT conversion assay, using [3–4,5-dimethyldiazol-2-yl]-2,5-diphenyl tetrazolium bromide (MTT, Sigma-Aldrich). A549 (3000/well) cells were seeded in triplicate in 96 well/plates and incubated with increasing concentrations of compounds 3–5 (concentration between 5 µM and 100 µM) or DMSO 0.10% (v/v) for the 72 h in DMEM medium with 1% FBS. Following the treatment, 20 µL of MTT (5mg/mL in PBS) was added and the cells were incubated for additional 3 h at 37 °C. The formazan crystals, thus formed, were dissolved in 100 μL of buffer containing 50% (v/v) N,Ndimethylformamide, 20% SDS (pH 4.5). The absorbance was measured a 570 nm with a Multiskan™ GO Microplate Spectrophotometer (Thermo Fisher Scientific, USA). IC50 values, defined as the concentration resulting in 50% inhibition of cell survival, were calculated and compared to control cells treated with DMSO.
6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.
Acknowledgements This work was supported by grants from MIUR ITALY PRIN 2015 “Top-down and Bottom-up approach in the development of new bioactive chemical entities inspired on natural products scaffolds” (Project No. 2015MSCKCE_003). We thank Prof. P. Filippakopoulos (Structural Genomics Consortium, Oxford, UK) for tests on bromodomains.
37. 38. 39.
3957
Huang H, Zhang G, Zhou Y, et al. Front Chem. 2018;6:1. Jenkins JL, Bender A, Davies JW. Drug Discov Today Technol. 2006;3:421. Lauro G, Romano A, Riccio R, Bifulco G. J Nat Prod. 2011;74:1401. Scrima M, Lauro G, Grimaldi M, et al. J Med Chem. 2014;57:7798. Proto MC, Fiore D, Piscopo C, et al. Sci Rep. 2017. http://dx.doi.org/10.1038/ s41598-017-11688-x. Cheruku P, Plaza A, Lauro G, et al. J Med Chem. 2012;55:735. Lauro G, Masullo M, Piacente S, Riccio R, Bifulco G. Bioorg Med Chem. 2012;20:3596. Gong J, Sun P, Jiang N, et al. Org Lett. 2014;16:2224. Giordano A, Forte G, Massimo L, Riccio R, Bifulco G, Di Micco S. Eur J Med Chem. 2018;152:253. Di Micco S, Mazué F, Daquino C, et al. Org Biomol Chem. 2011;9:701. Spatafora C, Barresi V, Bhusainahalli VM, et al. Org Biomol Chem. 2014;12:2686. Di Micco S, Spatafora C, Cardullo N, et al. Bioorg Med Chem. 2016;24:820. Pulvirenti L, Muccilli V, Cardullo N, Spatafora C, Tringali C. J Nat Prod. 2017;80:1648. Terracciano S, Di Micco S, Bifulco G, Gallinari P, Riccio R, Bruno I. Bioorg Med Chem. 2010;18:3252. Di Micco S, Terracciano S, Bruno I, et al. Bioorg Med Chem. 2008;16:8635. Amé JC, Spenlehauer C, de Murcia G. BioEssays. 2004;26:882. Hottiger MO, Hassa PO, Lüscher B, Schüler H, Koch-Nolte F. Trends Biochem Sci. 2010;35:208. Kang DH, Lee DJ, Lee S, et al. Nat Commun. 2017;8:40. Lehtiö L, Chi NW, Krauss S. FEBS J. 2013;280:3576. Lakshmi TV, Bale S, Khurana A, Godugu C. Curr Drug Targets. 2017;18:1214. Ferri M, Liscio P, Carotti A, et al. Eur J Med Chem. 2017;142:506. Karlberg T, Markova N, Johansson I, et al. J Med Chem. 2010;53:5352. Strocchia M, Terracciano S, Chini MG, et al. Chem Comm. 2015;51:3850. Terracciano S, Foglia A, Chini MG, et al. RSC Adv. 2016;6:82330. Teracciano S, Chini MG, Vaccaro MC, et al. Chem Comm. 2016;52:12857. Terracciano S, Chini MG, Vaccaro MC, et al. Chem Comm. 2016;52:13515. Chen B, Dodge ME, Tang W, et al. Nat Chem Bio. 2009;5:100. Huang SM, Mishina YM, Liu S, et al. Nature. 2009;461:614. Ruzzene M, Penzo D, Pinna LA. Biochem J. 2002;364:41. Nguyen DX, Chiang AC, Zhang XH, et al. Cell. 2009;138:51. Harder E, Damm W, Maple J, et al. J Chem Theory Comput. 2016;12:281. Still WC, Tempczyk A, Hawley RC, Hendrickson T. J Am Chem Soc. 1990;112:6127. Schrodinger Release 2017–1, LigPrep, Schrodinger, LLC, New York, NY, 2017. Sanner MF. J Mol Graph Model. 1999;17:57. Berman HM, Westbrook J, Feng Z, et al. Nucleic Acids Res. 2000;28:235. Di Micco S, Chini MG, Terracciano S, Bruno I, Riccio R, Bifulco G. Bioorg Med Chem. 2013;21:3795. Di Micco S, Renga B, Carino A, et al. Steroids. 2014;80:51. Trott O, Olson AJ. J Comput Chem. 2010;31:455. Schrodinger Release 2017–1, QikProp, Schrodinger, LLC, New York, NY, 2017.
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry journal homepage: www.elsevier.com/locate/bmc
Identification by Inverse Virtual Screening of magnolol-based scaffold as new tankyrase-2 inhibitors
T
Simone Di Miccoa, Luana Pulvirentib, Ines Brunoa, Stefania Terraccianoa, Alessandra Russoa, Maria C. Vaccaroa, Dafne Ruggieroa,c, Vera Muccillib, Nunzio Cardullob, Corrado Tringalib, ⁎ Raffaele Riccioa, Giuseppe Bifulcoa, a
Department of Pharmacy, University of Salerno, Via Giovanni Paolo II 132, 84084 Fisciano, Italy Department of Chemical Sciences, University of Catania, Viale Andrea Doria 6, 95125 Catania, Italy c PhD Program in Drug Discovery and Development, University of Salerno, Via Giovanni Paolo II 132, I-84084 Fisciano, SA, Italy b
A R T I C LE I N FO
A B S T R A C T
Dedicated to the memory of our dear colleague and friend Carmela Spatafora.
The natural product magnolol (1) and a selection of its bioinspired derivatives 2–5, were investigated by Inverse Virtual Screening in order to identify putative biological targets from a panel of 308 proteins involved in cancer processes. By this in silico analysis we selected tankyrase-2 (TNKS2), casein kinase 2 (CK2) and bromodomain 9 (Brd9) as potential targets for experimental evaluations. The Surface Plasmon Resonance assay revealed that 3–5 present a good affinity for tankyrase-2, and, in particular, 3 showed an antiproliferative activity on A549 cells higher than the well-known tankyrase-2 inhibitor XAV939 used as reference compound.
Keywords: Tankyrase-2 Inverse Virtual Screening Biomimetic compounds Tankyrase-2 inhibitors Anti-cancer drugs
1. Introduction Inverse Virtual Screening (IVS) aims to identify putative macromolecular targets by computer-aided methodologies.1,2 In this field, an in silico protocol has been developed by our research group,3–5 consisting in a virtual screening of a single compound against a library of proteins by molecular docking and a normalization of the energy values as selection criteria for the macromolecule identification. It represents a rapid and cost efficient methodology for the identification of one or more potential biological targets for a small molecule. This approach, successfully applied in other cases reported in literature,3,6–8 is particularly suitable for natural products, which are well-known to represent a precious source of new chemical skeletons endowed with a wide array of biological activities, whose biological targets are, most of the times, unknown. Recently, we have also demonstrated the applicability of our IVS protocol to the analysis of a synthetic compound library validating our in silico strategy for lead repurposing.9 In this context, the biomimetic syntheses allow to broaden the chemical diversity of the natural scaffold, providing the possibility of increasing the bioactivities, as already reported in literature.10–15 The dimeric neolignan magnolol (1, Fig. 1) has been recently the focus of bioinspired investigations by some of us in order to develop a collection of structurally related derivatives, including compounds 2–5;
⁎
Corresponding author. E-mail address: [email protected] (G. Bifulco).
https://doi.org/10.1016/j.bmc.2018.06.019 Received 5 March 2018; Received in revised form 12 June 2018; Accepted 13 June 2018
Available online 20 June 2018 0968-0896/ © 2018 Elsevier Ltd. All rights reserved.
these were obtained employing both enzymatic dimerization in the presence of horseradish peroxidase (HRP), and regioselective orthohydroxylation with the environmentally benign reagent IBX (2-iodoxybenzoic acid).13 In the present contribution, we investigated by IVS 1 and its derivatives 2–5, succeeding in identifying new inhibitors of tankyrase-2. This macromolecule, along with the isoform 1, is a member of the PARP (poly ADP-ribose polymerases) family of proteins, acting as mono- or poly-ADP-ribosyltransferase and is involved in different biological processes such as telomere elongation and Wnt signal transduction pathway.16,17 Owing to its involvement in different key cellular pathways, tankyrase-2 plays an important role in cancer, resulting as an attractive target for tumor treatment.18–21 2. Results and discussion 2.1. Inverse Virtual Screening Briefly, our IVS protocol consists in molecular docking calculations of a small molecule against a panel of proteins, obtaining a predicted binding energy [V0 (kcal/mol)] for each ligand-protein system. In parallel, we virtually test decoys, namely compounds showing molecular weights and H-bonds donors/acceptors similar to the investigated compound, against the protein panel. For each macromolecule, the
Bioorganic & Medicinal Chemistry 26 (2018) 3953–3957
S. Di Micco et al.
that is used to predict the most promising interacting targets for the case-study small molecule, ranking all the obtained V values from the highest to the lowest one. Such normalization is adopted to prevent the selection of false positives, as emerged by our previous studies.4–9 In the present work, 1–5 were virtually screened on a panel of 308 models of proteins involved in tumor processes, and 162 decoys (“blanks” compounds) were used for the normalization of the docking results.6 In details, for each small molecule (1–5), we obtained the predicted binding energies (see experimental sections) against the 308 protein models from molecular docking calculations. Each predicted value for a specific target was normalized with the corresponding averaged binding energy of the decoys. The normalized values of 1–5 were ranked from the highest (first position in the ranking) to the lowest one (last position in the ranking) obtaining an affinity profile on the whole protein panel (Tables S1–S5) for the small molecules. The next step was the analysis of docked conformations by visual inspection. In particular, we arbitrarily limited the analysis of docked outcomes for the targets ranked in the first 20 positions for each compound (1–5) classification in order to identify macromolecular candidates for the experimental assays.9 From this analysis, putative macromolecular targets were identified for each investigated compound. By comparing the obtained results for 1–5, we observed that each of the small molecules (1–5) targeted the proteins tankyrase-2 (TNKS2), casein kinase 2 (CK2) and bromodomain 9 (Brd9) at the same time. In particular, tankyrase-2 is ranked in the first ten best targets for 1, 2, 4 and 5 (Tables S1, S2, S4, S5). For the compound 3, we found the tankyrase-2 at position twelve of the normalized values (Table S3). Similar considerations were made for casein kinase 2. Indeed, for 1–4 this protein is in the first ten positions (Tables S1–S4), whereas, for 5 it is classified at position fifteen (Table S5). The bromodomain 9 was found at position four and twelve
Fig. 1. Molecular structures of the natural product magnolol (1) and of its synthetic derivatives 2–5.
average value VR (kcal/mol) of binding energies of the decoys is calculated. The resulting ratio between V0 and VR is a dimensionless number V:
V = V0/ VR
(1)
Fig. 2. Three-dimensional model of the interactions given by 1 (a), 2 (b), 3 (c), 4 (d) and 5 (e) with tankyrase-2. Superimposition of 3–5 with co-crystallized XAV939 (f, PDB ID 3KR8). The protein is depicted by tube (colored: C, grey; polar H, white; N, dark-blue; O, red; S, yellow). The small molecules are represented by sticks (pink for 1, cyan for 2, brown for 3, dark green for 4, kaki for 5, aquamarine for XAV939) and balls (colored: C, as for the sticks; polar H, white; N, dark-blue; O, red; F, green; S, yellow). The dashed black lines indicate the hydrogen bonds between ligand and protein. 3954
Bioorganic & Medicinal Chemistry 26 (2018) 3953–3957
S. Di Micco et al.
for 4 (Table S4) and 2 (Table S2), respectively. For the remaining compounds, the normalized value of bromodomain 9 was found at position nineteen (Tables S1, S3, S5). Hence, our investigation suggested that the magnolol-based compounds could potentially fit the binding pockets of the above reported three macromolecules. For sake of simplicity, we describe in details the interactions of 1–5 with tankyrase-2, which gave the best experimental results. The docking outcomes showed that the compounds 1–5 are well accommodated into the binding pocket of tankyrase-2. We observed that the common structural moiety of 1–5, constituted by two aromatic rings and two hydroxyl groups at C-2 and C-2’ (1, Fig. 1) occupied equivalent spatial portions of the binding cavity. In particular, the two phenyl rings presented a normal spatial arrangement respect to each other, and were involved in π-π interactions with His1031, Tyr1050 and Tyr1071 (Fig. 2). The hydroxyl at C-2 established two hydrogen bonds with the backbone amide of Gly1032 (Fig. 2). The OH group at C-2’ was double bonded with the peptide backbone of Tyr1060 (Fig. 2). It is noteworthy that the two H-bonds with Gly1032 and the π-π contacts with both Tyr1050 and Tyr1071 have been already observed (Fig. 2f) for co-crystallized (PDB ID: 3KR8) ligand (XAV939) of tankyrase-222 (Fig. 2f). The two linear chains of 1–5 contribute to the complex stability by van der Waals contacts with Pro1034, Tyr1050, Tyr1060 and Ile1075. The compounds 2–5 structurally differ from the natural product 1 for the substituents at positions C-3, C3’, C-5 and C-5’, which determine further interactions with the protein residues. The compound 2 presented a superimposable docked pose respect to 1, giving the same interactions of the natural product. The derivative 2 differs from 1 for the presence of an additional hydrogen bond with the side chain of Ser1068 established by the OH group at C-3. This H-bond is also found for 3 exerted by its methoxy group at C-3. The hydroxyl group at C3’ of 3 donates an H-bond to the side chain of His1031 and accepts an H-bond from the side chain of Ser1033. The derivative 4 gives the same interactions observed for 3 with the common structural moieties. The methoxy group at C3’ accepts an H-bond from the side chain of Ser1033. Moreover, compound 4 establishes further H-bonds by hydroxyl groups on the linear chains with the side chain of Tyr1071 and NH of Gly1074. The compound 5 is the acetylated derivative of 4. The acetyl group at C-5 made two hydrogen bonds with Met1054 and Tyr1071, whereas, the other acetyl is H-bonded to Ile1075. This docking analysis showed, thus, that the substituents introduced to obtain 2–5 established favorable interactions with protein residues suggesting an improved affinity respect to the natural compound 1, as experimentally confirmed.
Table 1 Thermodynamic constants measured by SPR for the interaction between tested compounds (1–5) and immobilized tankyrase-1, tankyrase-2, casein kinase 2 and PARP1. XAV939 and TBB were used as reference compounds for binding to TNKS2 and CK2. KD (µM) ± SD Compound
Tankyrase-2
Casein kinase 2
Tankyrase-1
PARP1
1 2 3 4 5 XAV939 TBB
no binding no binding 0.021 ± 0.0041 0.0185 ± 0.0007 0.0065 ± 0.0007 0.0042 ± 0.0008 –
no binding no binding no binding no binding no binding – 0.79 ± 0.0201
no binding 5.26 ± 1.46 0.52 ± 0.28 no binding no binding – –
no binding 1.25 ± 0.2 no binding no binding no binding – –
PARP1. For what concerns TNKS1, only compounds 2 and 3 showed a binding for this isoform (Table 1). In particular, compound 3 presented a binding toward TNKS1 and TNKS2, but with higher affinity against isoform 2. Indeed, the measured KD vs. TNKS1 is 25 times larger than the value for TNKS2 (Table 1). Concerning the PARP1 outcomes, compound 2 showed a binding affinity with the tested protein with low micromolar KD value (see Table 1), confirming the good selectivity of the previously disclosed compounds for TANK2 protein. Overall, the experimental outcomes qualitatively confirm the predicted analysis through our IVS protocol. 2.3. Cell-based assays TNKS2 inhibition is well-known to induce a reduction in β-catenindependent signaling in cells with a hyperactive Wnt pathway. As the hyperactivation of this pathway has been documented in samples from aggressive lung cancer and adenocarcinoma,30 we explored the effects of 3–5 and XAV939 on growth of A549 cells. The exposure of A549 cells (1% FBS, Fig. 3), for 72 h, to increasing concentration of compound 3 and XAV939 affects the cell proliferation with comparable IC50 values of 4.7 ± 0.4 and 12.3 ± 1.5, respectively (Table 2 and Fig. 3). In particular, the compound 3 showed a better activity than the reference compound XAV939. Conversely, the compounds 4 and 5 affect the cell viability by 20–30% when used at high concentrations, even though SPR data showed a better affinity of these compounds compared to 3. This behavior could be ascribed to a different cell-permeability. Indeed, we rationalized the experimental results by calculating the permeability
2.2. Surface Plasmon Resonance spectroscopy analyses In order to validate the IVS results, we synthesized compounds 2–5 as previously reported,13 to verify experimentally the binding affinity toward the proposed targets of the natural lead 1 and the synthetic analogues 2–5. Thus, compounds 1–5 were subjected to Surface Plasmon Resonance (SPR)23–26 screening on the immobilized proteins tankyrase-2 and casein kinase 2. The reported TNKS2 and CK2 inhibitors, XAV93927,28 and TBB29 were also tested as positive control. The experimental outcomes showed no binding towards CK2 (Table 1). Based on this assay, the natural compound 1 and 2 were unable to bind the TNKS2; conversely, compounds 3–5 have been identified as high affinity leads for the TNKS2 protein with low micromolar KD values (see Table 1). The compound 5 presented the lowest KD, followed by 4 and 3. It is noteworthy that we found for 5 a KD comparable with XAV939 value. The experimental outcomes confirmed the theoretical predictions, in fact, in the docking analysis, 5 showed the highest number of H-bonds with protein residues and presented the best affinity towards tankyrase-2. In addition, we evaluated the binding toward Brd9 and other related proteins, but negative results emerged (data not shown). Based on the obtained experimental results, we also investigated the putative binding of 1–5 towards two highly structural related isoforms of TNKS2 included in our protein panel: tankyrase-1 (TNKS1) and
Fig. 3. A549 cells were treated for 72 h with the indicated concentrations of 3 and XAV939 in medium with low serum (1.0% FBS) and the cell growth was determined by MTT assay. Data are the means ± SD of two independent experiments. 3955
Bioorganic & Medicinal Chemistry 26 (2018) 3953–3957
S. Di Micco et al.
1.5.634 and added Gasteiger charges and merged non polar hydrogens. The 308 protein 3D models were prepared starting from the deposited structures in the Protein Data Bank database (www.rcsb.org)35 and from homology modelling.36,37 Water molecules were removed by Maestro 11, and .pdb files obtained were then processed with Autodock Tools 1.5.6 and converted in .pdbqt format, merging nonpolar hydrogens and adding Gasteiger charges. Charge deficit was spread over all atoms of related residues. For molecular docking calculations, Autodock Vina software38 was employed, with exhaustiveness value of 64, and saving 10 conformations as the maximum number of binding modes. In docking calculations, 1–5 were treated as flexible and proteins as rigid. Each grid box was centered on binding site of the proteins and the grid points spaced by 1.0 Å. For all the investigated compounds, all bonds were treated as active torsional bonds except double and ring bonds. The analysis of docking outcomes were carried out by Autodock Tools 1.5.6. Predicted binding energies were normalized as previously described6 by using the Eq. (1). Maestro 11 was used to generate the Figures of the 3D models. Predicted apparent permeability was calculated by QuikProp,39 using default parameters and Caco-2 cells as model. The reference calculated values are: < 25 nm/s poor, > 500 nm/s great).39
Table 2 Exposure to TNKS-2 inhibitor compounds affected the A549 cell growth. IC50 (µM) value after 72 h treatment. Compound
1% FBS
3 4 5 XAV939
4.7 ± 0.4 n.i.a n.i.a 12.3 ± 1.5
a
n.i., no inhibition.
properties of 3–5. The predicted data showed a better permeability for 3 (842.518 nm/s) compared to 4 (216.806 nm/s) and 5 (79.995 nm/s), confirming our hypothesis. As expected, compounds 4 and 5 showed comparable cellular activity. Indeed, 5 can be considered a prodrug of 4 because the acetyl groups can be hydrolysed in cell releasing 4. 3. Conclusions In the present work we propose the 1,1′-biphenyl-2,2′-diol scaffold, derived from the natural product magnolol (1, Fig. 1), for the development of new tankyrase-2 inhibitors. By the application of our Inverse Virtual Screening protocol, we filtered a library of 308 putative biological targets for compounds 1–5, proposing for the experimental tests the thankyrase-2, casein kinase 2 and bromodomain 9. The experiments showed a binding towards tankyrase-2 by 3–5, presenting a KD in the low micromolar range. It is noteworthy that 5 showed a comparable KD respect to the reference compound XAV939. The careful molecular docking analysis highlighted the key interactions given by the common 1,1′-biphenyl-2,2′-diol structural moiety for 1–5, mainly represented by π-π interactions with His1031, Tyr1050 and Tyr1071 and H-bonds with Gly1032. Moreover, the analysis of the docked poses, along with the relative biological results pointed that the H-bonds network formed by substituents at C-3, C-3′, C-4 and C-4′ increase the binding affinity towards tankyrase-2, as experimentally confirmed. The anti-proliferative assays, on A549 cells, revealed a better activity of 3 compared to XAV939. Unfortunately, despite the promising KD values obtained by SPR experiments, the cell-based tests showed an absence of antiproliferative activity for 4 and 5. This finding could be ascribed to a lower/scarce membrane permeability as predicted by QikProp calculations. The presented data confirm the Inverse Virtual Screening as a valuable tool to identify the molecular targets of small molecules endowed with interesting biological activities. Moreover, the use of a biosynthetic approach combined with in silico methodology provides a fast and cost efficient lead identification in drug discovery processes. It could give the opportunity of re-evaluating the biological properties of natural products already used in the traditional medicine, as just in the case of magnolol.
4.2. SPR experiments Recombinant Tankyrase-2 (TNKS2) was purchased from MyBioSource (MyBioSource Inc., San Diego, USA). The TNKS2 inhibitor 3,5,7,8-Tetrahydro-2-[4-(trifluoromethyl)phenyl]-4H-thiopyrano-[4,3d]pyrimidin-4-one (XAV939) was purchased from Santa Cruz Biotechnology, Inc. Recombinant Casein Kinase II (CK2) was purchased from New England Biolabs (New England Biolabs Inc., Massachusetts, USA). The CK2 inhibitor 4,5,6,7-Tetrabromo-2-azabenzimidazole (TBB) was purchased from Santa Cruz Biotechnology, Inc. Recombinant TNKS1 (TRF1-interacting ankyrin-related ADP-ribose polymerase 1), also known as PARP5A, TANK1, or tanykyrase-1 was purchased from Sigma Aldrich. Recombinant Poly(ADP-ribose) polymerase 1 (PARP-1) was purchased from Enzo Life Sciences, Inc. SPR analyses were carried out according to our previously published data.23–26 Surface Plasmon Resonance Spectroscopy (SPR) analyses were performed to determine binding of the various molecules to the TNKS1, TNKS2, CK2 and PARP1 proteins using a Biacore 3000 optical biosensor equipped with researchgrade CM5 sensor chips (GE Healthcare). TNKS1, TNKS2, CK2 and PARP1 were coupled to the surface of a CM5 sensor chip using standard amine-coupling protocols, according to the manufacturer’s instructions. The proteins TNKS2 and CK2 (100 µg ml−1 in 10 mM CH3COONa, pH 7.3) were immobilized on individual sensor chip surfaces at a flow rate of 5 µL min−1 to obtain densities of 8–12 kRU. PARP1 (100 µg mL−1 in 10 mM CH3COONa, pH 7.5) was immobilized on individual sensor chip surfaces at a flow rate of 5 µL min−1 to obtain densities of 8–12 kRU. TNKS1 (100 µg mL−1 in 10 mM CH3COONa, pH 4.5) was immobilized on individual sensor chip surfaces at a flow rate of 5 µL min−1 to obtain densities of 8–12 kRU. For the experiments, the recombinant TNKS1, TNKS2 and PARP1 surfaces, a BSA surface and one unmodified reference surface were prepared for simultaneous analyses. For TNKS2 analysis, compounds 1–5 and XAV939 were dissolved to obtain 20 mM solution concentrations, in 100% DMSO and diluted 1:1000 (v/v) in PBS (10 mM NaH2PO4, 150 mM NaCl, pH 7.4) to a final DMSO concentration of 0.5%. For CK2 analysis, compounds 1–5 as well as TBB were dissolved to obtain 20 mM solutions, in 100% DMSO and diluted 1:1000 (v/v) in PBS (10 mM NaH2PO4, 150 mM NaCl, pH 4.5) to a final DMSO concentration of 0.5%. For each molecule a five-point concentration series were set up, including 0 – 0.25 – 1 – 10 – 20 µM, and, for each sample the complete binding study was performed using triplicate aliquots. Changes in mass, due to the binding response, were recorded as resonance units (RU). To obtain the dissociation constant (KD) (Table 1), these responses were fit to a 1:1 Langmuir binding model by nonlinear regression, using the BiaEvaluation software
4. Experimental section Magnolol (1) was purchased from TCI Europe N.V. (Zwijndrecht, Belgium). Compounds 2–5 were synthesized as previously reported.13 4.1. Computational details The three dimensional structures of the investigated compounds (1–5) were built by Build Panel of Maestro (version 11) and minimized by means of OPLS3 force field31 and the Polak-Ribier conjugate gradient algorithm (maximum derivative less than 0.001 kcal/mol), in water by the GB/SA (generalized Born/surface area)32 solvent treatment. Then, the small molecules were processed with LigPrep,33 considering protonation states at a pH of 7.0 ± 1.0. The minimized geometries of 1–5 were converted in the .pdbqt format by Autodock Tools 3956
Bioorganic & Medicinal Chemistry 26 (2018) 3953–3957
S. Di Micco et al.
A. Supplementary data
program provided by GE Healthcare. Simple interactions were suitably fitted to a single-site bimolecular interaction model (A + B = AB), yielding a single KD (Table 1). SPR experiments were performed at 25 °C, using a flow rate of 50 µL min−1, with 60 s monitoring of association and 300 s monitoring of dissociation.
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.bmc.2018.06.019. References
4.3. Cell culture and proliferation assay 1. 2. 3. 4. 5.
A549 (human lung carcinoma) cell line was obtained from ECACC European Cell Lines Collection, England. The cells purchased from Invitrogen (Carslbad, CA, USA), were grown in DMEM/high glucose supplemented with heat-inactivated FBS (10%, v/v), 2 mM L−glutamine, penicillin (100 U/mL) and streptomycin (100 mg/mL) at 37 °C in humidified atmosphere with 5% CO2. To ensure logarithmic growth, the cells were subcultured every 2 days. Stock solutions of compounds 3–5 and XAV939 (100 mM in DMSO) were stored at -20 °C in the dark and diluted just before addition to the sterile culture medium. In all the experiments, the final concentration of DMSO was 0.10% (v/v). The proliferation assay was performed with MTT conversion assay, using [3–4,5-dimethyldiazol-2-yl]-2,5-diphenyl tetrazolium bromide (MTT, Sigma-Aldrich). A549 (3000/well) cells were seeded in triplicate in 96 well/plates and incubated with increasing concentrations of compounds 3–5 (concentration between 5 µM and 100 µM) or DMSO 0.10% (v/v) for the 72 h in DMEM medium with 1% FBS. Following the treatment, 20 µL of MTT (5mg/mL in PBS) was added and the cells were incubated for additional 3 h at 37 °C. The formazan crystals, thus formed, were dissolved in 100 μL of buffer containing 50% (v/v) N,Ndimethylformamide, 20% SDS (pH 4.5). The absorbance was measured a 570 nm with a Multiskan™ GO Microplate Spectrophotometer (Thermo Fisher Scientific, USA). IC50 values, defined as the concentration resulting in 50% inhibition of cell survival, were calculated and compared to control cells treated with DMSO.
6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.
Acknowledgements This work was supported by grants from MIUR ITALY PRIN 2015 “Top-down and Bottom-up approach in the development of new bioactive chemical entities inspired on natural products scaffolds” (Project No. 2015MSCKCE_003). We thank Prof. P. Filippakopoulos (Structural Genomics Consortium, Oxford, UK) for tests on bromodomains.
37. 38. 39.
3957
Huang H, Zhang G, Zhou Y, et al. Front Chem. 2018;6:1. Jenkins JL, Bender A, Davies JW. Drug Discov Today Technol. 2006;3:421. Lauro G, Romano A, Riccio R, Bifulco G. J Nat Prod. 2011;74:1401. Scrima M, Lauro G, Grimaldi M, et al. J Med Chem. 2014;57:7798. Proto MC, Fiore D, Piscopo C, et al. Sci Rep. 2017. http://dx.doi.org/10.1038/ s41598-017-11688-x. Cheruku P, Plaza A, Lauro G, et al. J Med Chem. 2012;55:735. Lauro G, Masullo M, Piacente S, Riccio R, Bifulco G. Bioorg Med Chem. 2012;20:3596. Gong J, Sun P, Jiang N, et al. Org Lett. 2014;16:2224. Giordano A, Forte G, Massimo L, Riccio R, Bifulco G, Di Micco S. Eur J Med Chem. 2018;152:253. Di Micco S, Mazué F, Daquino C, et al. Org Biomol Chem. 2011;9:701. Spatafora C, Barresi V, Bhusainahalli VM, et al. Org Biomol Chem. 2014;12:2686. Di Micco S, Spatafora C, Cardullo N, et al. Bioorg Med Chem. 2016;24:820. Pulvirenti L, Muccilli V, Cardullo N, Spatafora C, Tringali C. J Nat Prod. 2017;80:1648. Terracciano S, Di Micco S, Bifulco G, Gallinari P, Riccio R, Bruno I. Bioorg Med Chem. 2010;18:3252. Di Micco S, Terracciano S, Bruno I, et al. Bioorg Med Chem. 2008;16:8635. Amé JC, Spenlehauer C, de Murcia G. BioEssays. 2004;26:882. Hottiger MO, Hassa PO, Lüscher B, Schüler H, Koch-Nolte F. Trends Biochem Sci. 2010;35:208. Kang DH, Lee DJ, Lee S, et al. Nat Commun. 2017;8:40. Lehtiö L, Chi NW, Krauss S. FEBS J. 2013;280:3576. Lakshmi TV, Bale S, Khurana A, Godugu C. Curr Drug Targets. 2017;18:1214. Ferri M, Liscio P, Carotti A, et al. Eur J Med Chem. 2017;142:506. Karlberg T, Markova N, Johansson I, et al. J Med Chem. 2010;53:5352. Strocchia M, Terracciano S, Chini MG, et al. Chem Comm. 2015;51:3850. Terracciano S, Foglia A, Chini MG, et al. RSC Adv. 2016;6:82330. Teracciano S, Chini MG, Vaccaro MC, et al. Chem Comm. 2016;52:12857. Terracciano S, Chini MG, Vaccaro MC, et al. Chem Comm. 2016;52:13515. Chen B, Dodge ME, Tang W, et al. Nat Chem Bio. 2009;5:100. Huang SM, Mishina YM, Liu S, et al. Nature. 2009;461:614. Ruzzene M, Penzo D, Pinna LA. Biochem J. 2002;364:41. Nguyen DX, Chiang AC, Zhang XH, et al. Cell. 2009;138:51. Harder E, Damm W, Maple J, et al. J Chem Theory Comput. 2016;12:281. Still WC, Tempczyk A, Hawley RC, Hendrickson T. J Am Chem Soc. 1990;112:6127. Schrodinger Release 2017–1, LigPrep, Schrodinger, LLC, New York, NY, 2017. Sanner MF. J Mol Graph Model. 1999;17:57. Berman HM, Westbrook J, Feng Z, et al. Nucleic Acids Res. 2000;28:235. Di Micco S, Chini MG, Terracciano S, Bruno I, Riccio R, Bifulco G. Bioorg Med Chem. 2013;21:3795. Di Micco S, Renga B, Carino A, et al. Steroids. 2014;80:51. Trott O, Olson AJ. J Comput Chem. 2010;31:455. Schrodinger Release 2017–1, QikProp, Schrodinger, LLC, New York, NY, 2017.