Article pubs.acs.org/jmc
Cite This: J. Med. Chem. 2018, 61, 6546−6573
Detailed Exploration around 4‑Aminoquinolines Chemical Space to Navigate the Lysine Methyltransferase G9a and DNA Methyltransferase Biological Spaces
J. Med. Chem. 2018.61:6546-6573. Downloaded from pubs.acs.org by ST FRANCIS XAVIER UNIV on 08/10/18. For personal use only.
Obdulia Rabal,† Juan Antonio Sánchez-Arias,† Edurne San José-Enériz,‡ Xabier Agirre,‡ Irene de Miguel,† Leire Garate,‡ Estibaliz Miranda,‡ Elena Sáez,† Sergio Roa,‡ José Angel Martínez-Climent,‡ Yingying Liu,§ Wei Wu,§ Musheng Xu,§ Felipe Prosper,‡,∥ and Julen Oyarzabal*,† †
Small Molecule Discovery Platform, Molecular Therapeutics Program, ‡Area de Hemato-Oncología, IDISNA, Ciberonc, Center for Applied Medical Research (CIMA), University of Navarra, Avenida Pio XII 55, E-31008 Pamplona, Spain § WuXi Apptec (Tianjin) Co. Ltd., TEDA, No. 111 HuangHai Road, Fourth Avenue, Tianjin 300456, PR China ∥ Departmento de Hematología, Clinica Universidad de Navarra, University of Navarra, Avenida Pio XII 36, E-31008 Pamplona, Spain S Supporting Information *
ABSTRACT: Epigenetic regulators that exhibit aberrant enzymatic activities or expression profiles are potential therapeutic targets for cancers. Specifically, enzymes responsible for methylation at histone-3 lysine-9 (like G9a) and aberrant DNA hypermethylation (DNMTs) have been implicated in a number of cancers. Recently, molecules bearing a 4-aminoquinoline scaffold were reported as dual inhibitors of these targets and showed a significant in vivo efficacy in animal models of hematological malignancies. Here, we report a detailed exploration around three growing vectors born by this chemotype. Exploring this chemical space led to the identification of features to navigate G9a and DNMT1 biological spaces: not only their corresponding exclusive areas, selective compounds, but also common spaces. Thus, we identified from selective G9a and first-in-class DNMT1 inhibitors, >1 log unit between their IC50 values, with IC50 < 25 nM (e.g., 43 and 26, respectively) to equipotent inhibitors with IC50 < 50 nM for both targets (e.g., 13). Their ADME/Tox profiling and antiproliferative efficacies, versus some cancer cell lines, are also reported.
■
in promoting transcriptional silencing of target genes.5 Moreover, pharmacologic or siRNA-mediated inhibition of G9a synergizes with DNMT1 inhibition in reducing cell proliferation.6,7 All these observations strongly suggested that dual inhibitors of both targets could be valuable anticancer agents, and we sought to identify such compounds. A number of selective G9a inhibitors have been reported,8 with substrate-competitive inhibitors 1 (UNC0638)9 and 2 (A-366)10 being extensively used as chemical probes to investigate the biological role of G9a in multiple diseases
INTRODUCTION
Methyltransferases are responsible for the methylation of their corresponding substrates (histones, proteins for protein methyltransferases (PMTs); or DNA, for DNA methyltransferases (DNMTs)) using S-adenosyl-L-methionine (SAM or AdoMet) as cofactor, thus acting as epigenetic writers that control epigenetic gene regulation and carcinogenesis.1,2 G9a (also known as KMT1C or EHMT2) catalyzes mono- or dimethylation of histone H3 lysine 9 (H3K9) and other protein targets such as p53.3 G9a upregulation and overexpression is present in a variety of cancers.4 DNMT1, a protein catalyzing methylation of C5-cytosine of DNA, binds G9a and both proteins cooperate © 2018 American Chemical Society
Received: December 30, 2017 Published: June 11, 2018 6546
DOI: 10.1021/acs.jmedchem.7b01925 J. Med. Chem. 2018, 61, 6546−6573
Journal of Medicinal Chemistry
Article
Chart 1. Known G9a Inhibitors (1, 2), Nucleoside DNMT1 Inhibitors (3, 4), and Inhibitor (5)a
a
G9a IC50 values of 1, 3, 4, and 5 and DNMT1 IC50 values for 1 and 5 as determined internally (see footnote in Table 1). Alternatively, the reported IC50 value for compound 1 (UNC-0638) was 1 log unit
■
RESULTS SAR in Biochemical G9a and DNMT1 Assays. SAR examination of the 2-position of the quinoline scaffold revealed 5-alkyl-2-furyl groups (5-methyl in 5 and 5-ethyl in 6, Table 1) as the optimal groups for yielding functional cellular activity.13 Thus, we decided to alternatively use these substituents as initial reference points for subsequent SAR exploration of the 4-, 6-, and 7-positions of the quinoline scaffold, using a sequential approach (Chart 2). The 4-amino moiety was first explored (Table 1). According to the predicted binding mode of 56 and 6 (Figure 1A,B) into both methyltransferases and X-ray structure of 1 and related quinazoline analogues into G9a,9 this position extends out toward the so-called solvent accessible area, suggesting that chemically diverse groups, exploring the biological space,16 could be 6547
DOI: 10.1021/acs.jmedchem.7b01925 J. Med. Chem. 2018, 61, 6546−6573
Journal of Medicinal Chemistry
Article
Table 1. Exploration of the 4-Position
the piperidine ring by limiting molecular flexibility17 with spiropiperidines (17, 18, 20, 21), spiropyrrolidines (19, 22), and bridged piperazines (23−26). In general, these conformationally constrained analogues share a biochemical profile similar to that of 5, especially in terms of selectivity toward G9a over DNMT1. Exceptionally, racemic analogues 20−22, whose basic nitrogen predictably positions more distant from that of the piperidine (in green in Figure 1D compared to the other docked spiro derivatives in orange and 5 in violet, that overlap well), were less potent against G9a (IC50 > 50 nM). DNMT1 SAR was flat, with no general improvements in activity (200− 300 nM, pIC50 ranging between 6.5 and 6.7, with less than 0.3 log units compared to 5 and 6) except for some remarkable analogues such as the homologated 3-azabicyclo[3.2.1]octane of 25 (IC50 = 80 nM). As expected,18 loss of the hydrogen bond donor of the 4-amino group of 5 by methylation (27) or replacement by an oxygen atom (28)6 was detrimental for G9a
well-tolerated. Removal of the basic nitrogen of 5 (compounds 7−11, 16) caused a significant drop in activity against both targets, and only the piperidone derivative 11 retained potent DNMT1 activity (IC50 < 500 nM). This result is consistent with the predicted binding mode of 6 with G9a, as the positively charged nitrogen of the piperidine ring interacts with Asp1078 (Figure 1A). With respect to DNMT1, no explicit contacts were predicted by docking between the piperidine ring and the protein (Figure 1B), although a negative electrostatic potential is observed in this area (Figure 1C). Homologation of the piperidine ring of 5 with a methylene linker (compounds 126−13) greatly improved DNMT1 activity (IC50 < 50 nM) while causing minor potency loss against G9a (IC50 ∼ 15 nM). Given the poor pharmacokinetic profile of 12,6 the gem-dimethyl analogue 15 was synthesized as the methylene linker was viewed as a potential metabolic liability. As 15 was significantly less potent against both targets, we concentrated on alternatives to 6548
DOI: 10.1021/acs.jmedchem.7b01925 J. Med. Chem. 2018, 61, 6546−6573
Journal of Medicinal Chemistry
Article
selectivity profile of lead compound 5 (e.g., with constrained rings) or equally potent compounds were obtained at the cost of losing G9a activity (e.g., 11). Only the homologated compounds 12 and 13 exhibit both G9a and DNMT1 activities in the low nanomolar range (
