Design and synthesis of N-benzoyl amino acid derivatives as DNA methylation inhibitors

Article abstract of DOI:10.1111/cbdd.12794

The inhibition of human DNA Methyl Transferases (DNMT) is a novel promising approach to address the epigenetic dysregulation of gene expression in different diseases. Inspired by the validated virtual screening hit NSC137546, a series of N-benzoyl amino acid analogues was synthesized and obtained compounds were assessed for their ability to inhibit DNMT-dependent DNA methylation in vitro. The biological screening allowed the definition of a set of preliminary structure–activity relationships and the identification of compounds promising for further development. Among the synthesized compounds, L-glutamic acid derivatives 22, 23, and 24 showed the highest ability to prevent DNA methylation in a total cell lysate. Compound 22 inhibited DNMT1 and DNMT3A activity in a concentration-dependent manner in the micromolar range. In addition, compound 22 proved to be stable in human serum and it was thus selected as a starting point for further biological studies.

Full text of DOI:10.1111/cbdd.12794

Received: 1 February 2016
DOI: 10.1111/cbdd.12794
Revised: 2 May 2016
Accepted: 14 May 2016
R E S E A R C H A R T I C L E
Design and synthesis of N-benzoyl amino acid derivatives as DNA
methylation inhibitors
Davide Garella1 | Sandra Atlante2 | Emily Borretto1 | Mattia Cocco | Marta Giorgis
1
1
|
1
1
1
2
Annalisa Costale | Livio Stevanato | Gianluca Miglio | Chiara Cencioni
Eli Fernández-de Gortari | José L. Medina-Franco | Francesco Spallotta
|
3
3
2
|
2
1
Carlo Gaetano | Massimo Bertinaria
1Dipartimento di Scienza e Tecnologia del
Farmaco, Università degli Studi di Torino, Torino,
Italy
2Division of Cardiovascular Epigenetics
Department of Cardiology, Goethe University,
Frankfurt am Main, Germany
3Facultad de Química Departamento de
Farmacia, Universidad Nacional Autónoma de
México, Mexico City, México
The inhibition of human DNA Methyl Transferases (DNMT) is a novel promising
approach to address the epigenetic dysregulation of gene expression in different
diseases. Inspired by the validated virtual screening hit NSC137546, a series of
N-benzoyl amino acid analogues was synthesized and obtained compounds were
assessed for their ability to inhibit DNMT-dependent DNA methylation in vitro.
The biological screening allowed the definition of a set of preliminary structure–
activity relationships and the identification of compounds promising for further
development. Among the synthesized compounds, L-glutamic acid derivatives 22,
Correspondence
Davide Garella, davide.garella@unito.it
and
2
3, and 24 showed the highest ability to prevent DNA methylation in a total
Carlo Gaetano, gaetano@em.uni-frankfurt.de
cell lysate. Compound 22 inhibited DNMT1 and DNMT3A activity in a
concentration-dependent manner in the micromolar range. In addition, compound
2
2 proved to be stable in human serum and it was thus selected as a starting
point for further biological studies.
K E Y W O R D S
DNA methylation, DNMT inhibitors, docking, epigenetics, structure–activity
relationships
hemimethylated DNA strands.[
6,7]
Another protein, lacking
Epigenetic modifications play an essential role in the estab-
lishment and regulation of cellular differentiation and gene
enzymatic activity, namely DNMT3L, is capable of interact-
ing with DNMT3A and DNMT3B with the consequence of
[1,2]
expression.
DNA methylation is the most stable epigene-
[3]
[8]
tic mark in humans. The DNA methylation occurs at the C5
position of the cytosine ring, particularly in a CpG dinucleo-
tide context, through the action of three active DNA methyl-
transferases (DNMTs): DNMT1, DNMT3A, and DNMT3B.
These enzymes catalyze the transfer of a methyl group from
stimulating their catalytic activity.
In human genome, CpG dinucleotides are typically clus-
tered in regions called CpG islands, which are located in the
[9]
proximal promoter of more than half of all human genes.
When promoter CpG islands are methylated, the correspond-
ing gene is repressed because of poor recognition by transcrip-
tion factors and by other methyl-binding proteins (MBDs)
[
4]
S-adenosyl-L-methionine (SAM) to the C5-cytosine.
DNMT1 is responsible for DNA methylation maintenance
during cell replication by methylation of newly synthesized
DNA strands; however, it was hypothesized that this enzyme
[10]
involved in chromatin remodeling and reorganization.
Aberrant DNA methylation, or the failure to maintain the
appropriate DNA methylation status, results in the expres-
sion of non-optimal level of gene-associated proteins, which
could trigger or exacerbate different pathological responses.
[
5]
can also participate in the de novo methylation process.
DNMT3A and DNMT3B are responsible for de novo DNA
methylation being able to methylate both unmethylated and
Chem Biol Drug Des 2016; 1–14
wileyonlinelibrary.com/cbdd
© 2016 John Wiley & Sons A/S.
|
1

2
Garella et al.
|
[
11]
For instance, in cancer cells,
DNA hypermethylation of
hypermethylation of DNA sequences has been linked to the
onset of cardiac fibrosis.[
16]
CpG islands, joined to a global hypomethylation, gives rise
to genomic instability and inactivation of cancer-suppressor
On these bases, the potential therapeutic application
[12,13]
[17–19]
genes.
Altered DNA methylation has also been found
of DNMT inhibitors is nowadays actively studied.
A
[14]
to regulate synaptic plasticity in postmitotic neurons,
number of DNMT inhibitors has been developed. They can
be classified into two general subsets: nucleoside and non-
nucleoside DNMT inhibitors (Figure 1).
and DNMT1 and DNMT3A have been recently proposed
as new targets for antipsychotic therapy.[
15]
Moreover,
FIGURE 1 Structure of representative DNMT inhibitors

Garella et al.
3
|
The first subset comprises 5-azacytidine (5-AZA) and
study, the stereochemistry of tested NSC137546 was not
defined. This virtual screening hit showed a moderate selec-
tive inhibition of DNMT1 versus DNMT3B at 100 μm con-
centration. As the virtual screening was conducted at the
substrate-binding site of DNMT1, it was hypothesized that
the active compound binds into this pocket, although this
5
-aza-2′-deoxycytidine (decitabine), which are drugs
currently employed in the treatment of myelodysplastic
syndromes (MDS), chronic myelomonocytic leukemia
CMML), and acute myeloid leukemia (AML). 5-AZA
(
and decitabine are imported into the cell, phosphorylated,
and actively integrated in the DNA structure where they
act through covalent inhibition of DNMTs. This complex
set of events results in an easy onset of resistance and poor
reproducibility of action with more than 1000-fold variabil-
ity in cancer cell lines as demonstrated with decitabine.
A second generation of nucleoside analogues is currently
under study, among them: 5-fluoro-2′-deoxycytidine,
[
32]
was not experimentally checked.
Indeed, currently there
is not reported a crystallographic structure of NSC137546
bound to DNMT that provide evidence for the actual bind-
ing site. Actually, for most of the small-molecule DNMT
inhibitors described to date, the experimental binding site
remains unknown. However, molecular docking has been
helpful to propose binding models that require, of course,
experimental validation. Inspired by the chemical structure
of NSC137546, we decided to explore the chemistry of the
N-benzoyl amino acidic scaffold with the aim of improv-
ing DNMT1 activity and of investigating the selectivity
against DNMT3A and DNMT3B. As the predicted docking
pose of the (S)- and (R)-forms are largely similar, in this
work, we selected the (S) stereochemistry to make use of
natural amino acids. Therefore, the structure of 1(S)-2-
(2,6-dichlorobenzamido) pentanedioic acid (1), used as the
model template, was modulated according to three different
approaches (A–C, Figure 2).
[
20]
5
,6-dihydro-5-azacytidine (DHAC) and zebularine. The
main drawback of these nucleoside DNMT inhibitors is
expected to lie in their mechanism of action, similar to that
of first generation drugs.
Non-nucleoside DNMT inhibitors are represented by
a heterogeneous subset of compounds which can directly
inhibit the enzyme. Different compounds have been identi-
fied either from natural sources, from screening campaigns,
or thanks to synthetic efforts.[
17]
Natural products, such as parthenolide, curcumin, and
nanaomycin A, are proposed to act by covalent binding to
[
21–23]
the catalytic cysteine residue in the enzyme pocket.
A
Synthesis of derivatives bearing natural aminoac-
ids as the terminal acidic portion (A moiety in Figure 2)
was initially considered, to compare a series of sterically
homogenous analogues. A first series of 27 compounds was
designed (Schemes 1–3). Derivatives 1–8 were prepared to
explore the structure–activity relationships (SAR) of the A
moiety. Compound 9, the amino analogue of 1, was syn-
thesized to verify the role of the amide bond (B moiety) in
the activity of this class of compounds. Finally, modulation
of the aromatic substructure (C moiety), as in derivatives
10–27 (Scheme 3), was performed to gain preliminary SAR
information concerning the stereo electronic properties of
the aromatic ring.
main disadvantage of these compounds could be associated
with their lack of selectivity that may reflect in promiscuous
binding to other cellular targets.
Hydralazine, procaine, procainamide, isoxazoline deriva-
tives, and the natural compound (-)-epigallocatechin-3-gallate
are non-covalent inhibitors. They bind into the enzyme
pocket through a complex network of hydrogen bonds in a
[
24,25]
2
′-deoxycytidine-like binding mode.
Moreover, thanks
to the use of computational techniques and ad hoc medicinal
chemistry design, an increasing number of small-molecule
DNMT inhibitors is now emerging.[
26]
A few lead structures have been identified and their ratio-
nal modulation has been pursued. Most studied model com-
poundscompriseRG108,SGI-1027,NSC14778,NSC319745,
and their analogues, which act by competing with SAM or
with DNA strand for binding into the respective enzymatic
[
27–32]
site.
Furthermore, compounds such as NSC14778 are
being used as starting point of computer-assisted drug repur-
posing of novel hypomethylating agents.[
33]
So far, the therapeutic development of non-nucleoside
DNMT inhibitors was hampered by the relatively poor inhib-
itory activity, the lack of isoform selectivity, and a signifi-
cant cytotoxicity. Therefore, new potent and selective DNMT
inhibitors are urgently needed.
Recently, virtual screening of the National Cancer Insti-
tute (NCI) compounds library allowed the identification of
a glutamic acid derivative, namely NSC137546 (Figure 1),
as a potential DNMT1 inhibitor. Of note, in the previous
FIGURE 2 Structure of model compound 1 and chemical modulation
strategies applied

4
Garella et al.
|
The synthesis, the ability of the synthesized compounds
1.1 | Chemistry
to inhibit DNA methylation in vitro, and their preliminary
SAR are described. The ability of selected compounds to
inhibit DNA methylation in cell lysate overexpressing
DNMT1, DNMT3A, and DNMT3B isoforms is reported.
The action of selected compounds on isolated DNMT1
and DNMT3A enzymes is also reported. Finally, the char-
acterization of the putative binding mode of derivative 22
in the substrate-binding site of DNMT1 and DNMT3A is
proposed.
1
.1.1 | General procedure for the synthesis of
compounds 1–6, 10–27
Carboxylic acid (2.6 mmol, 1 eq) was solubilized in SOCl
2
(22 mmol, 8.5 eq) and irradiated under pressure in a
microwave oven (5 bar, N ) at 120 °C for 1.5 h. The
2
solvent was removed under reduced pressure, and the
residual oil was solubilized in 1,4-dioxane (5 mL) and
added dropwise to a stirring solution of appropriate ami-
noacid (2.6 mmol) and Na CO (6.5 mmol) in water
2
3
1
|
METHODS AND MATERIALS
(5 mL). The mixture was stirred overnight at room tem-
perature, poured in 1N HCl (40 mL), and extracted with
EtOAc (3 × 20 mL). The combined organic layers were
washed with brine, dried with anhydrous Na SO , and
Commercially available reagents and solvents were used
without further purification, unless otherwise noted. Reaction
progress was monitored by TLC on precoated silica plates
2
4
concentrated under reduced pressure. The residue was
triturated with CH Cl (about 30 mL), and the obtained
(
Merk 60 F , 250 μm thickness), and spots were stained
254
2
2
by ceric ammonium molibdate, KMnO (0.5 g in 100 mL
precipitate was filtered. The product was purified (when
required) by flash chromatography (CombiFlash, gradient
CH Cl /MeOH 0.1% CF COOH).
4
0
.1 N NaOH), and UV light. Reactions under MW irra-
™
diation were carried out in the Synthwave oven (Milestone,
Shelton, CT, USA). Melting points were measured with
a capillary apparatus (Buchi 540). All the compounds were
2
2
3
1
13
routinely checked by H and C NMR (Bruker Avance
1.1.2 | General procedure for the synthesis of
3
00, Billerica, MA, USA) at 300 and 75 MHz, respec-
compounds 7 and 8
tively. Chemical shifts (δ) are given in ppm relative to
To the compound 3, (500 mg, 1.6 mmol) solubilized in
THF (5 mL) was added DCC (453 mg, 2.2 mmol). The
mixture was stirred overnight at room temperature. The
mixture was filtered, and the filtrate was partially con-
centrated under reduced pressure. The appropriate amine
internal standard TMS (0.00 ppm) or residual solvent peaks
(
2
CDCl = 7.26 ppm, MeOD = 3.31 ppm, DMSO-d =
.50 ppm and D O = 4.79 ppm). H NMR coupling
2
3
6
1
constants (J) are reported in Herz (Hz) and multiplicity
is indicated as follow: s, singlet; bs, broad singlet; d,
doublet; dd, doublet of doublet; t, triplet; m, multiplet;
Im, imidazole ring; Ph, phenyl ring. Low-resolution mass
spectra were recorded on Finnigan-Mat TSQ-700 in chemi-
cal ionization (CI) using isobutane. ESI-MS spectra were
(
isopropyl amine or benzylamine, 1.6 mmol) was added
to the residue, and the mixture was stirred neat (overnight)
at room temperature. The solution was poured in water
and extracted with Et O (1 × 20 mL). The pH was adjusted
2
™
to 1–2 with 1N HCl, and the aqueous phase was extracted
with EtOAc (3 × 20 mL). The organic phases were washed
with brine, dried, filtered, and concentrated under reduced
pressure to obtain the desired compound without any fur-
ther purification.
recorded on a Micromass Quattro micro API (Waters,
Milford, MA, USA). The purification was performed by
®
flash chromatography (CombiFlash Rf Teledyne ISCO)
on appropriate columns (silica gel). Anhydrous sodium
sulfate was used as drying agent for the organic phases.
Organic solvents were removed under vacuum at 30 °C.
Purity of compounds was checked by UHPLC (PerkinElmer,
Waltham, MA, USA) Flexar 15, equipped with UV-Vis
diode array detector using an Acquity UPLC CSH Phenyl-
Hexyl 1.7 μm 2.1 × 50 mm column (Waters, Milford,
MA, USA) and H O/CH CN containing 0.1% CF COOH
1
.1.3 | General procedure for the synthesis of (S)-
Sodium-2-((2,6-dichlorobenzyl)aminopentanedioate) (9)
L-glutamic acid (600 mg, 4.0 mmol) and NaOH (320 mg,
8.0 mmol) were solubilized in H O/MeOH 1:1 (20 mL).
2
3
3
2
and H O/CH OH containing 0.1% CF COOH solvent
2,6-Dichlorobenzaldehyde (700 mg, 4.0 mmol) was added,
and the mixture was stirred at room temperature for 2 h.
2
3
3
systems. Detection was performed at λ = 200, 215, and
2
54 nm. The analytical data confirmed that the purity of
NaBH (227 mg, 6.0 mmol) was added portionwise, and
4
the products was ≥ 95%. Detailed synthetic procedures
are fully described in Supporting Information. Characteri-
the mixture was stirred overnight at room temperature.
The pH of the mixture was adjusted to 5–6 adding acetic
acid. After stirring for further 45 min, the solvent was
completely removed and the sticky mass was treated with
[
34]
[35]
[36]
[37]
zation data of 10–14, 16, 22, 24,
15,
18,
19,
[
38]
2
1,
are in agreement with those reported.

Garella et al.
5
|
EtOH (30 mL) to obtain the precipitation of a white solid
which was filtered and dried to afford the final compound
1.2.3 | Inhibition of DNMT1 and DNMT3A activity
on isolated enzyme
9
as a white solid.
To assess the specific interaction of DNMT inhibitors with
DNMT1 or DNMT3A, the activity assay was performed
on the immunoprecipitated enzymes after their overexpres-
sion. After DNMT transfection, HEK293T cells were freshly
lysed in RIPA buffer as described above (supplemented
with Protease Inhibitors Cocktail), and DNMT1 and
DNMT3A were immunoprecipitated using paramagnetic
beads (Ademtech’s Bioadembeads) as previously
1
.2 | Biology
1
.2.1 | Total DNMT activity assay
All compounds were screened for total DNMT activity
using a DNA methyltransferase (DNMT) activity/inhibitor
assay kit (Epigentek, Farmingdale, NY, USA) according
to manufacturer’s instruction. To measure the effects of
the compounds on DNMT activity, 35 μg of total HaCaT
cellular extract freshly prepared in RIPA buffer (Tris–HCl
pH 7.4 10 mm, NaCl 150 mm, NP-40 1%, sodium deoxy-
cholate (DOC) 1%, SDS 0.1%, glycerol 0.1%, Protease
Inhibitors Cocktail) was incubated with 50 and 100 μm
of the different compounds in DMSO (1% final conc.) or
vehicle alone (1% DMSO) at 37 °C for 2 h. As negative
control, the lysate was denatured at 100 °C for 30 min.
Total lysate was used to avoid the loss of possible coen-
zyme and complex formation. The amount of methylated
DNA, which is proportional to enzyme activity, was col-
[41]
described.
Briefly, 600 μg of transfected cell extract
was incubated at 4 °C for 2 h with DNMT antibodies:
anti-DNMT1 (6 μg, mouse, monoclonal; Abcam ,Cambridge,
UK) and anti-DNMT3A (6 μg, mouse, monoclonal; Abcam).
Normal mouse IgG (Santa Cruz Biotechnology, Inc., Dallas,
TX, USA) was used as immunoprecipitation control. Then,
the immunocomplexes were incubated at 4 °C for 2 h
with 60 μL of paramagnetic beads. All the immunopre-
cipitation steps have been performed on ice to preserve
the enzymatic activity. After elution, the enriched enzymes
were incubated with compounds 1 or 22 at 100 μm (1%
DMSO final conc.) or vehicle alone (1% DMSO) at 37 °C
for 2 h. RG108 (Cayman) was used as reference, while
as negative control, lysates were denatured at 100 °C for
®
orimetrically detected by a plate reader at 450 nm (EnSpire
Multimode Plate Reader – Perkin Elmer).
3
0 min. DNMT activity was detected by DNMT activity/
inhibitor assay kit (Epigentek). Data are presented as
means ± SD; each compound was tested at least three
times. Western blotting analysis was performed, according
to standard procedure, to check the immunoprecipitation.
1
.2.2 | Inhibition of DNA methylation in DNMT1,
DNMT3A, and DNMT3B overexpressing cell lysates
Compounds 1, 22 and 24 were tested to evaluate their
selectivity on the different DNMT isoforms. To work selec-
tively on DNMT1, DNMT3A, and DNMT3B, HEK293T
cells were transfected with the plasmids containing the
three different DNMTs’ sequences and mock control. Cells
were transfected with 2.5 μg of expression plasmid using
Lipofectamine 3000 reagent (Invitrogen, Carlsbad, CA,
USA) according the manufacturer’s instruction. Plasmids
1
.2.4 | Stability of compound 22 in pH 7.4
phosphate-buffered solution
A stirred solution of compound 22 in pH 7.4 phosphate-
buffered solution (final concentration 2 mg/mL) was main-
tained at 37 ± 0.5 °C for 48 h. At different time intervals,
100 μL of this solution was withdrawn, diluted to 1 mL
pcDNA3/Myc-DNMT1 (Addgene plasmid
#
36939),
with methanol containing 1% CF COOH, and 5 μL of
3
pcDNA3/Myc-DNMT3A (Addgene plasmid # 35521), and
pcDNA3/Myc-DNMT3B1 (Addgene plasmid # 35522) were
the resulting solution was analyzed by RP-UHPLC using
a Flexar UHPLC (Perkin Elmer) equipped with a Flexar
Solvent Manager 3-CH-Degasser, a Flexar-FX UHPLC
autosampler, a Flexar-FX PDA UHPLC Detector, a
Flexar-LC Column Oven, and a Flexar-FX-15 UHPLC
[39,40]
a gift from Arthur Riggs.
The presence of exogenous
DNMTs was checked by Western blot (not shown), and
afterward, the transfected cells were freshly lysed in RIPA
buffer as above. Cellular extract of 35 μg were incubated
with selected compounds at different concentration in the
™
Pump. The analytical column was an Acquity CSH
(2.1 × 100 mm, 1.7 μm particle size) (Waters) column.
The samples were analyzed using an isocratic method
employing a mobile phase consisting of methanol/water
(90/10) containing 0.1% trifluoroacetic acid (flow rate
0.6 mL/min). The column effluent was monitored at
λ = 220 nm referenced against a λ = 360 nm wavelength.
Quantification was performed using calibration curves of
compound 22 chromatographed under the same conditions.
1
–150 μm range (1% DMSO final conc.) or with vehicle
alone (1% DMSO) at 37 °C for 2 h. RG108 (Cayman)
was used as positive controls, while as negative control,
lysates were denatured at 100 °C for 30 min. DNMT
activity was detected by DNMT activity/inhibitor assay kit
(
Epigentek). Data are presented as means ± SD; each
compound was tested at least three times.

6
Garella et al.
|
The linearity of the calibration curves was determined in
Differences were judged statistically significant when
p < 0.1.
2
a concentration range of 1–3 mg/mL (r > 0.98). Data
analysis was performed using Chromera Manager (Perkin
Elmer). All experiments were run in triplicate.
2
|
RESULTS AND DISCUSSION
1
.2.5 | Stability of compound 22 in human serum
A solution of compound 22 (20 mg/mL) in methanol was
added to human serum (sterile-filtered from human male
AB plasma; Sigma-Aldrich, Saint Louis, MO, USA) pre-
heated at 37 °C to obtain a final concentration of 2 mg/
mL. The resulting solution was incubated at 37 ± 0.5 °C;
at appropriate time intervals, 100 μL of the reaction mix-
ture was withdrawn and added to 900 μL of methanol
containing 0.1% trifluoroacetic acid in order to deproteinize
the serum. The sample was vortexed, and then centrifuged
for 5 min at 1500 g. The clear supernatant was filtered
by Captiva PES 0.2-μm filters (Agilent, Santa Clara, CA,
USA) and analyzed by RP-UHPLC.
HPLC analyses were performed as indicated above. The
samples (1 μL, injection volume) were analyzed using an iso-
cratic method employing a mobile phase consisting of meth-
anol/water (80/20) containing 0.1% trifluoroacetic acid at a
flow rate of 0.2 mL/min. The column effluent was monitored
at a λ = 220 nm referenced against a λ = 360 nm wavelength.
Quantification was performed as indicated above. All experi-
ments were run in triplicate.
2.1 | Chemistry
Compound 1 and the analogues 2–6, modified in the acidic
moiety, were synthesized according to the procedure
reported in Scheme 1. 2,6-Dichlorobenzoic acid was con-
verted into the corresponding acyl chloride irradiating under
pressure (5 bar, N ) in a microwave oven at 120 °C
2
using SOCl for 1.5 h. This procedure allowed the use
2
of a parallel chemistry approach generating five acyl chlo-
ride derivatives in a single run in almost quantitative
yields. After SOCl evaporation, the crude product was
2
added dropwise to a stirring solution of the appropriate
(S)-amino acid in water using Na CO as the base. The
2 3
mixture was stirred at room temperature to afford com-
pounds 1–6 in 40–72% non-optimized yields (Scheme 1).
The use of water, as the preferred solvent, allowed for
amino acids dissolution, but surely affected the reaction
yield. Nevertheless, it avoided the use of polar aprotic
solvents (e.g. DMF) which could complicate isolation and
purification of the final compounds. Compounds 7 and 8
were synthesized by coupling of 1 with 2-propylamine or
benzylamine using dicyclohexylcarbodiimide (DCC) in THF
in 50% and 55% yield, respectively (Scheme 1). The
structural identity of compounds 7 and 8 was supported
by 2D-NMR spectra. For both compounds, the 2D-COSY
experiments allowed to discriminate the proton in position
1
.2.6 | Molecular docking
All flexible ligand docking and scoring calculations were
performed with ICM-Pro, version 3.8–4. ICM is based
on Monte Carlo optimization of the ligand internal
co-ordinates in the space of pocket grid potential maps.
The crystallographic structures of DNMT1 (PDB ID: 3PTA)
and DNMT3A (PDB ID: 2QRV) were employed. Before
docking, the structures of the proteins were prepared using
ICM using default settings. To ensure the convergence of
the Monte Carlo algorithms, three different cycles were
performed for each docked ligand. For the best ten poses,
a manual clustering was performed and the docked poses
were checked visually. Visualizations and analysis of the
protein–ligand complexes were conducted with ICM. The
2
from the chiral proton in position 4. The 2D-HMBC
experiments showed the correlation of the chiral proton
with both the amide carbon atoms, confirming that the
new amide bond was formed at the alpha acid (see
Supporting Information).
The amino derivative 9 was obtained in 70% yield via reduc-
tive amination of 2,6-dichlorobenzaldehyde with L-glutamic
acid and NaBH (Scheme 2).
4
Derivatives 10–27 (Scheme 3) were synthesized by reac-
tion of L-glutamic acid or 4-aminobutanoic acid with the
appropriately substituted benzoyl chloride using the expe-
ditious protocol previously developed for the synthesis of
compounds 1–6.
2
D interaction diagram was generated with the molecular
[42]
operatinG environment (moe) software, version 2014.09.
1
.3 | Statistics
2
2
.2 | Biological activity
Data were collected and analyzed blindly by an observer.
Results were expressed as mean ± SD of at least three
experiments performed in triplicate. Statistical significance
was evaluated by anova and Bonferroni post hoc test
.2.1 | Inhibition of total DNA methylation
The ability of the newly synthesized compounds to inhibit
DNA methylation was evaluated on HaCaT total cell lysate
(
Prism 5, GraphPad Software, La Jolla, CA, USA).

Garella et al.
7
|
SCHEME 1 Synthesis of 2,6-dichlorobenzoyl aminoacid derivatives 1–8. Reagents and conditions. (a) (i) SOCl , MW irradiation 120 °C, N (5 bar),
2
2
1
.5 h; (ii) L-amino acid, Na CO , H O, RT, 12 h. (b) (i) DCC, THF, RT, 12 h; (ii) R-NH , RT, 12 h
2
3
2
2
had no effect on DNA methylation in our test system. We
would like to point out that many studies highlighted that,
depending on the experimental conditions of the enzymatic
assay, the inhibitory potency can vary greatly for the same
compound; critical factors are the DNA methyltransferase
employed, the method of detection, the concentration of
the cofactor, and the nature of the DNA duplex used as the
SCHEME 2 Synthesis of compound 9. Reagents and conditions. (a)
i) L-glutamic acid, NaOH, H O/MeOH 1/1, RT, 2 h; (ii) NaBH , RT, 12 h
[43]
substrate.
(
2
4
The ability of synthesized compounds to inhibit total
DNA methylation, although initially studied only at two fixed
concentrations (100 and 50 μm), was helpful to draw prelim-
inary SAR indications and to address future chemical modu-
lation of this class of compounds.
Compound 1 (100 μm) significantly decreased DNA
methylation by 39 ± 12% (p < 0.05). Removal of the α-car-
boxylic group (2) afforded an inhibition of DNA methylation
(40 ± 2%; p < 0.05) comparable to that showed by 1. Both
1 and 2 were ineffective when tested at 50 μm (Table 1).
Conversion of α-COOH into a substituted amide gave com-
pounds 7 and 8, showing no significant ability to prevent
DNA methylation under our assay conditions. The obtained
in order to avoid the loss of possible coenzyme(s) and complex
formation. Compounds were incubated at either 100 or 50 μm
for 2 h at 37 °C in a freshly obtained cell lysate. The use
of two different concentrations of test compounds was chosen
to overcome cellular variability and to afford reliable results,
thus avoiding the identification of false-positive hits.
The amount of methylated DNA was quantified using
the EpiQuik DNA methyl transferases activity/inhibition
assay in comparison with vehicle-treated (DMSO 1%) cell
lysate. Results, expressed as the percentage of the residual
DNA methylation, are reported in Table 1. Vehicle alone

8
Garella et al.
|
SCHEME 3 Synthesis of compounds 10–27. Reagents and conditions. (a) (i) SOCl , MW irradiation 120 °C, N (5 bar), 1.5 h; (ii) L-glutamic acid,
2
2
Na CO , H O, RT, 12 h. (b) (i) SOCl , MW irradiation 120 °C, N (5 bar), 1.5 h; (ii) 4-aminobutanoic acid, Na CO , H O, RT, 12 h
2
3
2
2
2
2
3
2
TABLE 1 Ability of compounds 1–27 to inhibit total DNA methylation expressed as residual relative DNA methylation
a
b
a
b
Residual DNA methylation (%) (mean ± SD )
Residual DNA methylation (%) (mean ± SD )
Compound
1
100 μm
61 ± 12c
50 μm
Compound
15
100 μm
50 μm
96 ± 6
72 ± 7
8
2 ± 6
2
3
4
5
6
7
8
9
60 ± 2c
99 ± 11
126 ± 29c
90 ± 1
58 ± 7c
82 ± 16
94 ± 39
63 ± 1d
116 ± 9
85 ± 20
72 ± 23
57 ± 4c
70 ± 6c
87 ± 15
122 ± 26
96 ± 2
16
17
18
19
20
21
22
23
24
25
26
27
75 ± 2
69 ± 3
58 ± 3c
64 ± 11
81 ± 2
60 ± 5d
45 ± 1c
51 ± 9c
68 ± 21d
74 ± 4
89 ± 5
97 ± 2
92 ± 4
93 ± 6
85 ± 5
86 ± 18
63 ± 16d
68 ± 3d
65 ± 1d
95 ± 10
103 ± 11
93 ± 5
98 ± 22
104 ± 15
118 ± 1c
90 ± 17
111 ± 2
108 ± 36
122 ± 16
100 ± 1
106 ± 13
100 ± 16
1
1
1
1
1
0
1
2
3
4
107 ± 10
99 ± 14
anova and Bonferroni post hoc test.
a
Determined in fresh HaCaT cell lysate using Epiquik DNA methyltransferase activity/inhibition assay.
b
Data are expressed as percentage of residual-methylated DNA relative to vehicle (DMSO 1%)-treated cell lysate. Results are the mean of at least three independent ex-
periments run in triplicate.
c
p < 0.05 versus vehicle.
d
p < 0.1 versus vehicle.
results indicated that substitution of α-carboxylic group in
ring in γ-position, respectively. Results showed that the imid-
azole derivative 6 could conserve the inhibitory activity
(42 ± 7% inhibition; p < 0.05) while the strongly basic 5
reflected in a complete loss of activity. This result was not
unexpected taking into account that γ-COOH in 1 is nega-
tively charged at pH 7.4, while 5 is positively charged. The
compound 1 with uncharged and bulky groups (compounds
7
and 8) is not a promising strategy to improve DNMT inhi-
bition in this series of L-glutamic acid derivatives, while
the role of free α-COOH needs further studies to be fully
elucidated.
The modulation of the γ-carboxylic functionality was then
considered. The replacement of this carboxylic group with
a carboxyamido group (3), or its removal (4), resulted in a
complete loss of activity (Table 1).
The replacement of the γ-COOH with basic groups fur-
nished compounds 5 and 6 bearing a guanidine and imidazole
imidazole ring in 6 (calculated pK 7.08) is also present in
a
a positively charged form for about 32% at pH 7.4. Conse-
quently, its residual ability to decrease DNA methylation
could be attributed to other properties of this moiety (e.g.
metal-complexing ability or 1,3-prototropic tautomerism).
Unfortunately, similarly to derivatives 1 and 2, compound 6

Garella et al.
9
|
did not exert any detectable inhibitory effect at lower con-
centration (50 μm; Table 1). These preliminary SARs suggest
that a putative electrostatic interaction, rather than just hydro-
gen bonding, could be involved in the binding of the γ-COOH
group of 1 with its target (vide infra).
assay, indicating that α-COOH group can play a relevant role
in the activity of the o-chloro-substituted compound 24.
2
.2.2 | Inhibition of DNA methylation in cell lysates
selectively over expressing DNMT1, DNMT3A, and
DNMT3B
Interestingly, when the amide bond of 1 was reduced to
afford the amino derivative 9 (B moiety in Figure 2), a behav-
ior similar to that of the parent compound was found, with
inhibition of methylation at 100 μm (37 ± 1%; p < 0.1) and
no significant inhibition at 50 μm.
Based on the results of the inhibition of DNA methylation
discussed in the previous section, we selected two of the
newly designed molecules, 22 and 24, to further investigate
their DNMT inhibition properties. The choice of o-Cl-
substituted compound 24, which is slightly less active than
o-F-substituted 23 (at 100 μm) was performed to directly
compare 24 versus 1 over a larger range of concentrations,
in order to test the hypothesis that mono-halo-substituted
benzoyl amino acids could be better DNMT inhibitors than
disubstituted analogues. The new compounds, along with
model compound 1, and RG108 were studied at seven
different concentrations (range 1–150 μm) for their ability
to prevent DNA methylation in a HEK293T cell lysate
selectively overexpressing DNMT1, DNMT3A, and
DNMT3B. All the tested compounds were able to inhibit
DNMT1- and DNMT3A-dependent DNA methylation in a
concentration-dependent manner with a residual methylation
in the 28–49% range in cell lysate overexpressing DNMT1
(Figure 3) and 39–41% in cell lysate overexpressing
DNMT3A (Figure 4) at the maximal concentration tested.
Interestingly, in this test model, compound 22 proved as
active as the reference DNMT inhibitor RG108 maintaining
Based on the preliminary SAR data discussed above with
the modification of the acidic and amide moieties (A and B,
respectively, in Figure 2), we decided to explore the influ-
ence of benzene ring substitution (C moiety) keeping the L-
glutamic acid side chain present in 1 as the preferred moiety
A and the amide group.
Moving one chlorine atom from ortho-to-para position of
benzene ring, to obtain the 2,4-dichloro-substituted compound
1
0, afforded an inactive compound. Derivatives 24, 16, and 14
bearing one chlorine atom in ortho, meta, and para positions,
proved able to slightly reduce DNA methylation at 100 μm
showing a residual methylation of 68 ± 21%, 75 ± 2%, and
70 ± 6%, respectively. Interestingly, the o-chloro-substituted
compound 24 maintained an inhibitory activity (35 ± 1% inhi-
bition) at the lower concentration, demonstrating an improve-
ment with respect to the reference 1.
These preliminary SAR data, at two compound concentra-
tions, suggest that substitution of benzene ring with a halogen
atom in ortho position could afford derivatives with improved
efficacy as compared to double halogen substitution.
Use of an electron-donating substituent (i.e. a methoxy
group) in either position of benzene ring (compounds 15, 17,
2
0) did not show an increase in the activity when compared
with o-Clsubstituted 24, when compounds where tested at
5
0 μm concentration.
In order to obtain more hints about the benzene ring sub-
stitution preference, and to verify whether we could further
improve the activity of 24, we synthesized and evaluated a
series of compounds (11–13, 18, 19, 21–23, and 26) bear-
ing substituents endowed with different steric and electronic
properties. Among the synthesized compounds, 13 (p-F
substituted), 22 (o-Br substituted), and 23 (o-F substituted)
showed the most attractive activity, being able to inhibit DNA
methylation by 43 ± 4% (p < 0.05), 55 ± 1% (p < 0.05), and
FIGURE 3 Concentration–response experiments for compounds
1
, 22, 24, and reference RG108 showing inhibition of DNA methylation
4
9 ± 9%, (p < 0.05), respectively, at 100 μm. Compounds
in cell lysate selectively overexpressing DNMT1. HEK293T cells were
transfected with plasmids containing DNMT1 sequence and mock
control. Fresh cellular extracts were incubated either in the presence of
test compounds or with vehicle (1% DMSO) alone for 2 h. The amount of
methylated DNA was determined using Epiquik DNA methyltransferase
activity/inhibition assay. Data are expressed as percentage of residual-
methylated DNA ± SD relative to vehicle (DMSO 1%)-treated cell lysate.
Results are the mean of at least three independent experiments run in
triplicate. *p < 0.05, ∇p < 0.01, and p < 0.001 versus vehicle-treated
lysates; anova and Bonferroni post hoc test
2
2 and 23 were also able to prevent DNA methylation up
to a significant extent when tested at 50 μm (37 ± 16% and
3
2 ± 3% inhibition). Unsubstituted compound 25 as well as
p-phenyl-substituted derivative 19 retained a modest activity
only at 100 μm.
Finally, to explore the role of α-COOH group, we combined
o-chloro substitution (as in 24) with α-COOH group removal
#
(as in 2). The obtained derivative 27 proved inactive in our

1
0
Garella et al.
|
TABLE 2 Ability of compounds 1, 22, and reference compound RG108
to inhibit DNA methylation mediated by isolated DNMT1 and DNMT3A
enzymes
Compound
Residual relative enzymatic activity (%)a
DNMT1 (mean ± SDb)
DNMT3A (mean ± SDb)
1 ± 17c
1
66 ± 15c
6
2
2
58 ± 11 d
65 ± 8c
51 ± 13c
RG108
79 ± 3
anova and Bonferroni post hoc test.
a
HEK293T cells were transfected with plasmids containing the different DNMTs’
sequences and mock control. Cells were freshly lysed in RIPA buffer, and
DNMT1 and DNMT3A were immunoprecipitated using Paramagnetic beads and
anti-DNMT1, anti-DNMT3A as antibodies. Normal mouse IgG was used as
immunoprecipitation control. All the immunoprecipitation steps have been
performed on ice to preserve the enzymatic activity. The immunoprecipitated
enzymes were incubated with compounds 1 and 22 at 100 μm (1% DMSO final
conc.) or vehicle alone (1% DMSO) at 37 °C for 2 h. RG108 was used as
reference at 100 μm. DNMT activity was detected by DNMT activity/inhibitor
assay kit.
FIGURE 4 Concentration–response experiments for compounds 1,
2, 24, and reference RG108 showing inhibition of DNA methylation in
2
cell lysate selectively overexpressing DNMT3A. HEK293T cells were
transfected with plasmids containing DNMT3A sequence and mock
control. Fresh cellular extracts were incubated either in the presence of
test compounds or with vehicle (1% DMSO) alone for 2 h. The amount of
methylated DNA was determined using Epiquik DNA methyltransferase
activity/inhibition assay. Data are expressed as percentage of residual-
methylated DNA ± SD relative to vehicle (DMSO 1%)-treated cell lysate.
Results are the mean of at least three independent experiments run in
b
Data are expressed as percentage of residual enzymatic activity relative to vehi-
cle (DMSO 1%)-treated cell lysate ± SD. Results are represented as mean of, at
least, three independent experiments run in triplicate.
∇
#
triplicate. *p < 0.05, p < 0.01, and p < 0.001 versus vehicle-treated
c
p < 0.1.
lysates; anova and Bonferroni post hoc test
d
p < 0.05 versus vehicle.
a significant and encouraging inhibition of DNA methyla-
tion up to 25 μm (22 = 41 ± 18% versus RG108 = 47 ± 10%
inhibition in DNMT1-expressing lysate; 22 = 38 ± 6%
versus RG108 = 32 ± 17% inhibition in DNMT3A-expressing
lysate). Compounds 1 and 24 proved somewhat less active
on both DNMT1 and DNMT3A. All the compounds showed
poor or non-significant inhibitory activity in DNMT3B
overexpressing cell lysate up to 100 μm (Table S1).
The obtained data suggest that these compounds are non-
selective DNMT1/DNMT3A inhibitors.
2
.2.4 | Docking studies of 22 with DNMT1 and
DNMT3A
Molecular docking and other computational techniques have
shown to be useful to elucidate the binding mode of
experimentally known DNMT1 and DNMT3A inhibitors.
[
26]
To explore the putative protein–ligand interactions of com-
pound 22, we conducted flexible docking of its structure
within the substrate-binding site of the catalytic domain
of human DNMT1 and DNMT3A, respectively.
Crystallographic structures of DNMT1 (Protein Data Bank,
2
.2.3 | Inhibition of DNMT1 and DNMT3A in
[44]
PDB ID: 3PTA)
and DNMT3A (PDB ID: 2QRV) were
enzymatic-based assays
[8]
employed. Docking was conducted with the program
[45]
To further evaluate the ability of compounds 1, 22, and
RG108 to act directly on DNMT1 and DNMT3A, the
enzymes were overexpressed in HEK293T cells and enriched
by immunoprecipitation using the specific antibodies.
Enriched enzymes were treated with 100 μm of compounds
Internal Coordinates Mechanics (ICM) software.
The
docking protocol is presented in the Experimental part
section. Figure 5 shows a three- and bidimensional (3D
and 2D) binding model of 22 in complex with DNMT1.
According to this model, the γ-carboxylate group of 22
makes two important hydrogen bonds with the side chains
of Arg1312 and Arg 1310. An additional hydrogen–arene
bond is predicted between the phenyl moiety of 22 with
Cys1226. The overall position of the compound in the
binding pocket and, in particular, the interactions with
the catalytic Cys1226, suggests that 22 could inhibit DNMT1
by blocking the substrate-binding site.
1
, 22, and RG108; DNA methylation inhibition was meas-
ured as above. Results are reported in Table 2. All tested
compounds are able to inhibit DNMT1- and DNMT3A-
mediated DNA methylation. Compound 22 inhibited both
DNMT1 and DNMT3A activity by 42% and 49%, respec-
tively. In this assay, the activity of 1 was very close to
that previously reported for NSC137546 (stereochemistry
[32]
not defined) on recombinant DNMT1.
Collectively, these
Figure 6 depicts the docking model of 22 with DNMT3A.
Similar to Figure 5, a 3D model is shown along with a 2D-in-
teraction diagram. In this docking model, the α-carboxylate
group makes a hydrogen bond with the side chains of Arg887,
observations suggest that the chemistry of N-benzoyl-L-
glutamic acid derivatives could be further explored to
generate mixed DNMT1/DNMT3A inhibitors.

Garella et al.
11
|
FIGURE 6 Docking model of compound 22 with the catalytic domain of
human DNMT3A. The 3D and 2D interaction map show selected amino acid
residues of the binding site. Nonpolar hydrogen atoms are omitted for clarity. In
the 2D interaction map, the color coding and symbols are as in Figure 5
FIGURE 5 Docking model of compound 22 with the catalytic domain
of human DNMT1. The 3D and 2D interaction map show selected amino
acid residues of the binding site. Nonpolar hydrogen atoms are omitted for
clarity. In the 2D interaction map, green and blue arrows indicate hydrogen
bonding to side chain and backbone atoms, respectively. Blue ‘clouds’ on
ligand atoms indicate the solvent-exposed surface area of ligand atoms.
Light blue ‘halos’ around residues indicate the degree of interaction
with ligand atoms. The dotted contour reflects steric room for methyl
substitution
incubated at 37 °C for 48 h in pH 7.4 phosphate-buffered
solution and in human serum (Figure S1) at the concen-
tration of 2 mg/mL. The stability was monitored for 48 h
by measuring the compound concentration at different time
intervals via RP-UHPLC. Compound 22 was completely
stable in both the conditions tested.
but it does not make interactions with the catalytic Cys662.
Similar to the binding model with DNMT1, the docking
results suggest that 22 could inhibit DNMT3A by blocking
the substrate-binding site. In both models, with DNMT1 and
DNMT3A, the two carboxylate groups are involved in the
formation of hydrogen bonds with residues in the binding
site. This result is in overall agreement the SAR discussed
above.
3
|
CONCLUSIONS
Inspired by the structure of the validated hit compound
NSC137546, in this work, we synthesized a series of
N-benzoyl amino acids and explored their ability to inhibit
DNMT-dependent total DNA methylation. The SAR derived
from this study indicated that the N-benzoyl-substituted
L-glutamic acid scaffold is the minimal requirement for
the activity of this class of benzamide derivatives. Ortho-
halogeno-substituted compounds 22, 23, and 24 emerged
as the most interesting hits. In particular, compound 22
inhibited both DNMT1- and DNMT3A-mediated DNA
2
.2.5 | Stability of compound 22 under physiological
conditions and in human serum
In view of future in vivo studies, the chemical stability
of 22 under physiological conditions and in human serum
was checked. In separate experiments, compound 22 was

1
2
Garella et al.
|
[
[
18] J. M. Foulks, K. M. Parnell, R. N. Nix, S. Chau, K. Swierczek, M. Saunders,
K. Wright, T. F. Hendrickson, K. K. Ho, M. V. McCullar, S. B. Kanner, J.
Biomol. Screen 2012, 17, 2.
19] N. Martinet, B. Y. Michel, P. Bertrand, R. Benhida, MedChemComm 2012,
3, 263.
[20] T. Qin, J. Jelinek, J. Si, J. Shu, J. P. Issa, Blood 2009, 113, 659.
21] Z. Liu, S. Liu, Z. Xie, R. E. Pavlovicz, J. Wu, P. Chen, J. Aimiuwu, J. Pang,
D. Bhasin, P. Neviani, J. R. Fuchs, C. Plass, P.-K. Li, C. Li, T. H.-M. Huang,
L.-C. Wu, L. Rush, H. Wang, D. Perrotti, G. Marcucci, K. Chan, J. Pharma-
col. Exp. Ther. 2009, 29, 505.
methylation in a concentration-dependent manner and
proved active on both DNMT1 and DNMT3A isolated
by immunoprecipitation. Docking studies suggest putative
binding to the substrate site of both DNMT isoforms
studied. Compound 22 proved to be stable under physi-
ological conditions in human serum, and it is currently
being tested in different models of cardiac fibrosis. Results
of these studies will be reported in due course.
[
[
[
22] Z. Liu, Z. Xie, W. Jones, R. E. Pavlovicz, S. Liu, J. Yu, P.-K. Li, J. Lin, J. R.
Fuchs, G. Marcucci, C. Li, K. K. Chan, Bioorg. Med. Chem. Lett. 2009, 19,
706.
ACKNOWLEDGMENTS
23] D. Kuck, C. Caulfield, F. Lyko, J. L. Medina-Franco, Mol. Cancer Ther.
This research was supported by funding from the University
of Turin, Ricerca locale 2013 ‘quota B’ to D. G., and
Ricerca locale 2014 and 2015. C.G. and F.S. are partially
supported by a grant from LOEWE Cell & Gene Therapy
Center (LOEWE-CGT), Goethe University Frankfurt. F.S.
is recipient of the LOEWE-CGT grant # III L 5 - 518/17.004
2
010, 9, 3015.
[24] N. Singh, A. Duecas-Gonzalez, F. Lyko, J. L. Medina-Franco, ChemMed-
Chem 2009, 4, 792.
25] W. J. Lee, J.-Y. Shim, B. T. Zhu, Mol. Pharmacol. 2005, 68, 1018.
26] J. L. Medina-Franco, O. Mendez-Lucio, A. Dueñas-González, J. Yoo, Drug
Discov. Today 2015, 20, 569.
[27] S. Valente, Y. Liu, M. Schnekenburger, C. Zwergel, S. Cosconati, C.
Gros, M. Tardugno, D. Labella, C. Florean, S. Minden, H. Hashimoto,
Y. Chang, X. Zhang, G. Kirsch, E. Novellino, P. B. Arimondo, E. Miele,
E. Ferretti, A. Gulino, M. Diederich, X. Cheng, A. Mai, J. Med. Chem.
[
[
(
2013) and funded by the DFG (German Research
Foundation), Excellence Cluster Cardio Pulmonary System
ECCPS). E.F.-deG. is grateful to CONACyT for the Ph.D.
(
2
014, 57, 701.
Fellowship granted # 348291/240072. We also thank the
National Autonomous University of Mexico (UNAM), grant
PAPIIT IA204016 to JL.M-F and the program ‘Nuevas
Alternativas de Tratamiento para Enfermedades Infecciosas’
[28] S. Asgatay, C. Champion, G. Marloie, T. Drujon, C. Senamaud-Beaufort,
A. Ceccaldi, A. Erdmann, A. Rajavelu, P. Schambel, A. Jeltsch, O.
Lequin, P. Karoyan, P. B. Arimondo, D. Guianvarch, J. Med. Chem.
2
014, 57, 421.
[
29] S. Castellano, D. Kuck, M. Viviano, J. Yoo, F. Lopez-Vallejo, P. Conti, L.
Tamborini, A. Pinto, J. L. Medina-Franco, G. Sbardella, J. Med. Chem.
2011, 54, 7663.
(
NUATEI-IIB-UNAM) for the acquisition of the software
MOE. Authors wish to thank Prof. Giancarlo Cravotto,
DSTF, University of Turin, Italy, and Dr. Marco Lucio
Lolli, DSTF, University of Turin, Italy, for access to
[
30] A. Kabro, H. Lachance, I. Marcoux-Archambault, V. Perrier, V. Dore, C.
Gros, V. Masson, J.-M. Gregoire, F. Ausseil, D. Cheishvili, N. Bibens Lau-
lan, Y. St-Pierre, M. Szyf, P. B. Arimondo, A. Gagnon, MedChemComm
™
Synthwave oven and UHPLC instrumentations. Authors
2
013, 4, 1562.
wish to thank Prof. Loretta Lazzarato for mass spectra.
[31] J. Yoo, S. Choi, J. L. Medina-Franco, PLoS ONE 2013, 8(4), e62152.
[
[
32] D. Kuck, N. Singh, F. Lyko, J. L. Medina-Franco, Bioorg. Med. Chem.
2
010, 18, 822.
33] O. Méndez-Lucio, J. Tran, J. L. Medina-Franco, N. Meurice, M. Muller,
REFERENCES
ChemMedChem 2014, 9, 560.
[
1] A. Bird, Nature 2007, 447, 396.
[34] L. Rovati, F. Makovec, R. Chiste, P. Senin, Patent, BE 902726 A1 19851016,
1985.
[
2] S. L. Berger, T. Kouzarides, R. Shiekhattar, A. Shilatifard, Genes Dev.
2
009, 23, 781.
[35] T. Shiokari, S. Ueda, M. Iwata, K. Yukinori, Patent, EP 226304 A1
19870624, 1987.
[36] A. Rosowsky, H. Baader, R. A. Forsch, R. G. Moran, J. H. Freisheim, J.
Med. Chem. 1988, 31, 763.
[37] G. Groth, Patent, WO 2013093007 A1 20130627, 2013.
[38] Y.-J. Xiao, J.-G. Wang, X.-H. Liu, Y.-H. Li, Z.-M. Li, Gaodeng Xuexiao
Huaxue Xuebao 2007, 28, 1280.
[39] Z. X. Chen, J. R. Mann, C. L. Hsieh, A. D. Riggs, F. Chedin, J. Cell. Bio-
chem. 2005, 95, 902.
[40] H. Li, T. Rauch, Z. X. Chen, P. E. Szabo, A. D. Riggs, G. P. Pfeifer, J. Biol.
Chem. 2006, 281, 19489.
[41] F. Spallotta, C. Cencioni, S. Straino, S. Nanni, J. Rosati, S. Artuso, I. Manni,
C. Colussi, G. Piaggio, F. Martelli, S. Valente, A. Mai, M. C. Capogrossi, A.
Farsetti, C. Gaetano, J. Biol. Chem. 2013, 288, 11004.
[42] Molecular Operating Environment (MOE), version 2014.09, Chemical
Computing Group Inc., Montreal, Quebec, Canada. Available at http://
www.chemcomp.com
[43] D. Guianvarc’h, P. B. Arimondo, Fut. Med. Chem. 2014, 6, 1237.
[44] J. Song, O. Rechkoblit, T. H. Bestor, D. J. Patel, Science 2011, 331,
1036.
[
[
[
3] W. Reik, Nature 2007, 447, 425.
4] M. G. Goll, T. H. Bestor, Annu. Rev. Biochem. 2005, 74, 481.
5] G. D. Kim, J. Ni, N. Kelesoglu, R. J. Roberts, S. Pradhan, EMBO J. 2002,
2
1, 4183.
[
6] C. Gros, J. Fahy, L. Halby, I. Dufau, A. Erdmann, J. M. Gregoire, F. Ausseil,
S. Vispé, P. B. Arimondo, Biochimie 2012, 94, 2280.
[
[
7] R. Z. Jurkowska, T. P. Jurkowski, A. Jeltsch, ChemBioChem 2011, 12, 206.
8] D. Jia, R. Z. Jurkowska, X. Zhang, A. Jeltsch, X. Cheng, Nature 2007, 449,
2
48.
[
9] R. S. Illingworth, A. P. Bird, FEBS Lett. 2009, 583, 1713.
[
10] A. Fournier, N. Sasai, M. Nakao, P. A. Defossez, Briefings Funct. Genomics
2
012, 11, 251.
[
[
[
[
11] J. Fahy, A. Jeltsch, P. B. Arimondo, Exp. Opin. Ther. Pat. 2012, 22, 1427.
12] M. Esteller, N. Engl. J. Med. 2008, 358, 1148.
13] S. Sharma, T. K. Kelly, P. A. Jones, Carcinogenesis 2010, 31, 27.
14] B. Bie, J. Wu, H. Yang, J. J. Xu, D. L. Brown, M. Naguib, Nat. Neurosci.
2
014, 17, 223.
[
[
[
15] A. Guidotti, D. R. Grayson, Dialogues Clin. Neurosci. 2014, 16, 419.
16] H. Tao, J.-J. Yang, K.-H. Shi, Z.-Y. Deng, J. Li, Toxicology 2014, 323, 125.
17] A. Erdmann, L. Halby, J. Fahy, P. B. Arimondo, J. Med. Chem. 2015, 58,
[45] M. A. C. Neves, M. Totrov, R. Abagyan, J. Comp. Aided Mol. Des. 2012,
26, 675.
2
569.

Garella et al.
13
|
1
13
SUPPORTING INFORMATION
Data S2. H and C-NMR spectra of compounds 1–9, 17, 20, 23, 25-27.
Table S1. Ability of compounds 1, 22, 24, and reference compound RG108
to inhibit DNA methylation in cell lysate selectively overexpressing DN-
MT3B.
Additional Supporting Information may be found online in
the supporting information tab for this article.
Data S1. Characterization of compounds 1–9, 17, 20, 23, 25-27.
Figure S1. Stability of compound 22 in human serum.