Supporting Information

Discovery of Novel DNA Methyltransferase 3A Inhibitors via

Structure-based Virtual Screening and Biological Assays

Zhiyuan Shaoa,†, Pan Xub,†, Wen Xuc,†, Linjuan Lib,d, Shien Liub, Rukang Zhangb,d, Yu-Chih Liue, Chenhua Zhange, Shijie Chenb,*, and Cheng Luob,*

aNano Science and Technology Institute, University of Science and Technology of China,

Suzhou 215123, China.bDrug Discovery and Design Center, State Key Laboratory of Drug Research, Shanghai

Institute of MateriaMedica, Chinese Academy of Sciences, Shanghai 201203, China.cObstetrics & Gynecology Hospital of Fudan University, No.419 Fangxie Road. Shanghai

200011, China.dSchool of Life Science and Technology, Shanghai Tech University, Shanghai 200031, China.

eShanghai ChemPartner Co., LTD, Building 5, 998, Halei Road, Zhangjiang Hi-Tech

Park, Pudong New Area, Shanghai, P.R. China 201203.†These authors contributed equally to this work.

* To whom correspondence should be addressed. E-mail: shijiechen@simm.ac.cn, cluo@simm.ac.cn.

1

Contents:Methods………………..…………………………………………………………p.S3Table S1. The 2D structures of selected 107 compounds for biological evaluation..................................................................................................................p.S7Fig. S1. Pharmacophore models used in pharmacophore-based mapping and the representative mapping results…………………………………………………. ..p.S11Fig. S2. Re-docking results for evaluating different docking programs………….p.S12Fig. S3. In vitro enzymatic activity of 40_3 and 40_8 against DNMT3A determined by radioactive methylation assays………………………………………………...p.S13Fig. S4. Key interactions between DNMT3A and 40 or SAH at SAH binding pocket……………………………………………………………………………..p.S14Fig. S5. Cellular activity of 40 and 40_3 in Hela cell lines……………….....…...p.S15

2

Methods

Virtual screening protocol

Pharmacophore-based mapping. Pharmacophore models were derived from

the crystal structure of DNMT3A (PDB code: 4U7P, 4U7T) using Accelrys

Discovery Studio 3.0.1 By manually editing the generated pharmacophore

features, two pharmacophore models were constructed and used for subsequent

database searching. Both of the models comprised two hydrogen bond

acceptors, a hydrogen bond donor and several excluded volumes (Fig. S1).

Before performing pharmacophore-based mapping, a 3D database was prepared

against SPECS database using the same software with default parameters.

Ligand preparation. The 77,205 molecules obtained from pharmacophore-

based screening were then subjected to ligand preparation in LigPrep program

implemented in Schrödinger Suit.2 All possible ionization states were generated

at pH 7.0± 2.0 using Epik.3 In addition, different tautomers were also generated.

Other parameters were according to the default values. Of note, the second part

of molecules docked with DOCK 6.7 were performed with extra operations

regarding to the speciality,4 including adding AM1BCC charges and assigning

sybyl atom types using antechamber program,5 a useful auxiliary program for

molecular mechanic studies.

Protein preparation. To prepare the receptor, regular protein preparation such

as adding hydrogens, H-bond assignment and restrained minimization was

implemented using Protein Preparation Wizard of Maestro.6

Protein expression and purification

pcDNA3/Myc-DNMT3A and pcDNA3/Myc-DNMT3L were gifts from Arthur

Riggs (Addgene plasmid # 35521 and # 35523).7 These plasmid were used as

templates and DNMT3A sequence (residues 623-908) was cloned into a

modified pET28a vector, which encoded a tag after the N-terminal His6 tag.

3

The C-terminal residues 160–386 of Dnmt3L (Dnmt3L-C) were cloned into the

pGEX-6p-1 vector. The fusion proteins were expressed in BL21 (DE3) cell

strains. The cells grew at 37°C until OD600 approximately reached 0.6. The

temperature was then shifted to 16°C and the cells were induced by 0.4 mM

isopropyl β-D-1-thiogalactopyranoside (IPTG). After induction, the cells

continued growing overnight. DNMT3A protein was first purified by HisTrap

FF (GE Healthcare). Then, protein was further purified through gel-filtration

chromatography on a Superdex 75 10/300 column (GE Healthcare). The

Dnmt3L protein was sequently purified by GST affinity columns (GE

Healthcare) and Superdex 75 10/300 column (GE Healthcare). The purified

DNMT3A/3L protein were concentrated in buffer containing 20 mM Tris-HCl

(pH 7.4), 200 mM NaCl, 1 mM DTT, and 5% glycerol.

Radioactive methylation assay

DNMT3A/3L inhibition assay were performed in 25 μL reactants containing S-

[methyl-3H]-adenosyl-L-methionine ([3H]-SAM, PerkinElmer),

oligonucleosome (Life technologies) and DNMT3A/3L in modified Tris buffer.

The proteins were preincubated with compounds for 15 min at room

temperature (RT), and then the substrate and [3H]-SAM were added to the assay

plate to start reaction. After 4 hours of incubating at RT, the reaction systems

were stopped by adding SAM and were transferred to Flashplate by Platemate.

Radioactivity was determined by liquid scintillation counting (MicroBeta,

PerkinElmer).

SAR analysis

In order to generate more accurate docking conformation of 40 and its analogs

40_3 and 40_8 for further SAR analysis, induced fit docking against DNMT3A

(PDB: 4U7T) was employed using Induced Fit Docking protocol implemented

in Schrödinger.8, 9 Standard mode was used to generating up to 20 poses, and

4

the grid box was centred on the original ligand SAH with automatically

detected box size. All other parameters were kept default.

Cell culture and cell viability assays

The human acute myeloblastic leukemia lines, Kasumi-1, MV4-11, THP-1 and

KG-1, were purchased from American Type Culture Collection (ATCC) and

cultured in RPMI 1640 medium (Life Technologies). Human cervical cancer

cell line HeLa were obtained from ATCC and cultured in DMEM medium (Life

Technologies). These media were supplemented with 10% fetal bovine serum

and 1% penicillin and streptomycin (Life Technologies). The cells were

cultured at 37°C in a 5% CO2 incubator.

For the cell viability assays, all cells were seeded in 96-well plates and treated

with the tested compounds with no compound as contrast. Then, AlamarBlue

(Thermo Fisher Scientific) reagent was directly added to cells in culture

medium after treating cells with the corresponding compounds after 72 hours.

Cell viability was determined by POLARstar Omega (BMG Labtech).

5

Table S2. The 2D structures of selected 107 compounds for biological evaluation.

6

7

8

9

10

Fig. S4. Pharmacophore models used in pharmacophore-based mapping and the representative mapping results. (A) Pharmacophores derived from the crystal structure of DNMT3A10 (PDB: 4U7P), comprising two hydrogen bond acceptors, one hydrogen bond donor and several excluded volumes. (B) Pharmacophores derived from another crystal structure of DNMT3A with the substrate SAH10 (PDB: 4U7T). (C) The corresponding mapping results of pharmacophores generated from 4U7P against 3D molecule database, which was prepared from SPECS database11. (D) The pharmacophore mapping result of compound 40. The different color schemes represent the following pharmacophore features: HB_ACCEPTOR (green), HB_DONOR (magenta) and excluded volume (grey).

11

Fig. S5. Re-docking results for evaluating different docking programs. (A) The re-docking conformation of SAH superimposed with the original conformation in the crystal structure of DNMT3A (PDB: 4U7T) in its SAH binding pocket. Glide SP mode was used when conducting re-docking study12, and the superimposition result indicated Glide SP mode could reproduce the bioactive conformation of SAH with RMSD value of 0.4507. (B) The re-docking result produced by DOCK 6.74. Similarly, the crystal structure of DNMT3A (PDB: 4U7T) was utilized as receptor and the docking conformation of SAH could superimpose with the conformation in the crystal structure of DNMT3A to a large extent with RMSD value of 0.6132. As depicted in the figure, green sticks represent the original conformation in the crystal structure of DNMT3A, while yellow sticks represent the docking conformation. The white mesh surface displayed the binding pocket of SAH and both figures were produced by PyMOL, version 1.8.0.013. RMSD values were calculated in Superposition tool implemented in Maestro14, with all atom pairs defined except hydrogens.

12

Fig. S3. In vitro enzymatic activity of 40_3 and 40_8 against DNMT3A determined by radioactive methylation assays.

13

Fig. S4. Key interactions between DNMT3A and 40 or SAH at SAH binding pocket. (A) The 3D interaction diagram between 40 and neighboring residues of DNMT3A in the binding pocket. As it depicted, the carboxyl group formed two hydrogen bonds with Arg887 and a hydrogen bond with the side chain of Asn711. Besides, the guanidine group of Arg887 formed another hydrogen bond with the carbonyl moiety of 40. Significantly, the terminal aromatic ring of 40 interacted with the side chain of Phe640 by forming edge-to-face aromatic stacking as indicated in the figure. (B) The 3D interaction diagram of SAH with surrounding residues in the binding site. SAH formed several hydrogen bonds with other residues, including Thr645, Ile643, Glu664 and Val687. (C) The corresponding 2D interaction diagram of 40 with surrounding residues of DNMT3A. (D) The 2D interaction diagram between SAH and residues in the binding pocket. The first two figures were produced in PyMOL, version 1.8.0.013, in which cartoons represent protein, the gray sticks are residues interacting with 40 or SAH, the green sticks and yellow ones are represent 40 and SAH respectively. Hydrogen bonds are displayed as red dashed lines while the aromatic stacking is indicated by red cylinders. The 2D interaction diagrams were generated with LigPlot+, version v.1.4.515. Green dashed lines are indicated as hydrogen bonds and the corresponding distances are displayed. The red dashed lines represent non-ligand residues involved in hydrophobic contacts.

14

Fig. S5. cellular activity of 40 and 40_3 in Hela cell lines. Cellular activity of 40 and 40_3 in Hela cell lines for 4 days.

References1. Accelrys Discovery Studio (version 3.0), Accelrys, San Diego, CA, 2010.2. LigPrep (version 2.3), Schrödinger, LLC, New York, NY, 2009.3. Shelley, J.C.; Cholleti, A.; Frye, L.L.; Greenwood, J.R.; Timlin, M.R.;Uchimaya, M. J Comput Aided Mol Des 2007, 21, 681.4. Allen, W.J.; Balius, T.E.; Mukherjee, S.; Brozell, S.R.; Moustakas, D.T.; Lang, P.T.; Case, D.A.; Kuntz, I.D.;Rizzo, R.C. J Comput Chem 2015, 36, 1132.5. Wang, J.; Wang, W.; Kollmann, P.;Case, D. J. Comput. Chem. 2005, 25, 1157.6. Maestro (version 9.0), Schrödinger, LLC, New York, 2009.7. Chen, Z.X.; Mann, J.R.; Hsieh, C.L.; Riggs, A.D.;Chedin, F. J Cell Biochem 2005, 95, 902.8. Glide (version 6.7), Schrödinger, LLC, New York, NY, 2015.9. Prime (version 4.0), Schrödinger, LLC, New York, NY, 2015.10. Guo, X.; Wang, L.; Li, J.; Ding, Z.; Xiao, J.; Yin, X.; He, S.; Shi, P.; Dong, L.; Li, G.; Tian, C.; Wang, J.; Cong, Y.;Xu, Y. Nature 2015, 517, 640.11. Specs, http://www.specs.net/, (accessed April, 2015).12. Glide (version 5.5), Schrödinger, LLC, New York, 2009.13. PyMOL (version 1.8.0.0), Schrödinger, LLC, New York, 2015.14. Maestro (version 10.2), Schrödinger, LLC, New York, NY, 2015.15. Laskowski, R.A.;Swindells, M.B. J Chem Inf Model 2011, 51, 2778.

15