4-(Phenoxy) and 4-(benzyloxy)benzamides as potent and selective inhibitors of mono-ADP-ribosyltransferase PARP10/ARTD10

Article abstract of DOI:10.1016/j.ejmech.2018.06.047

Human Diphtheria toxin-like ADP-ribosyltranferases (ARTD) 10 is an enzyme carrying out mono-ADP-ribosylation of a range of cellular proteins and affecting their activities. It shuttles between cytoplasm and nucleus and influences signaling events in both compartments, such as nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) signaling and S phase DNA repair. Furthermore, overexpression of ARTD10 induces cell death. We recently reported on the discovery of a hit compound, OUL35 (compound 1), with 330 nM potency and remarkable selectivity towards ARTD10 over other enzymes in the human protein family. Here we aimed at establishing a structure-activity relationship of the OUL35 scaffold, by evaluating an array of 4-phenoxybenzamide derivatives. By exploring modifications on the linker between the aromatic rings, we identified also a 4-(benzyloxy)benzamide derivative, compound 32, which is potent (IC₅₀ = 230 nM) and selective, and like OUL35 was able to rescue HeLa cells from ARTD10-induced cell death. Evaluation of an enlarged series of derivatives produced detailed knowledge on the structural requirements for ARTD10 inhibition and allowed the discovery of further tool compounds with submicromolar cellular potency that will help in understanding the roles of ARTD10 in biological systems.

Full text of DOI:10.1016/j.ejmech.2018.06.047

Accepted Manuscript
4-(Phenoxy) and 4-(benzyloxy)benzamides as potent and selective inhibitors of
mono-ADP-ribosyltransferase PARP10/ARTD10
Sudarshan Murthy, Jenny Desantis, Patricia Verheugd, Mirko M. Maksimainen,
Harikanth Venkannagari, Serena Massari, Yashwanth Ashok, Ezeogo Obaji, Yves
Nkizinkinko, Bernhard Lüscher, Oriana Tabarrini, Lari Lehtiö
PII:
S0223-5234(18)30538-5
10.1016/j.ejmech.2018.06.047
EJMECH 10517
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 3 May 2018
Revised Date: 7 June 2018
Accepted Date: 19 June 2018
Please cite this article as: S. Murthy, J. Desantis, P. Verheugd, M.M. Maksimainen, H. Venkannagari,
S. Massari, Y. Ashok, E. Obaji, Y. Nkizinkinko, B. Lüscher, O. Tabarrini, L. Lehtiö, 4-(Phenoxy) and 4-
(benzyloxy)benzamides as potent and selective inhibitors of mono-ADP-ribosyltransferase PARP10/
ARTD10, European Journal of Medicinal Chemistry (2018), doi: 10.1016/j.ejmech.2018.06.047.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
4-(Phenoxy) and 4-(benzyloxy)benzamides as potent and selective
inhibitors of mono-ADP-ribosyltransferase PARP10/ARTD10
Sudarshan Murthy1, Jenny Desantis2, Patricia Verheugd3, Mirko M. Maksimainen1, Harikanth
Venkannagari1, Serena Massari2, Yashwanth Ashok1, Ezeogo Obaji1, Yves Nkizinkinko1, Bernhard
Lüscher3, Oriana Tabarrini2 & Lari Lehtiö1,*
1Biocenter Oulu and Faculty of Biochemistry and Molecular Medicine, University of Oulu, Oulu,
Finland
2Department of Pharmaceutical Sciences, University of Perugia, 06123 Perugia, Italy
3Institute of Biochemistry and Molecular Biology, RWTH Aachen University, 52074 Aachen,
Germany
*Address correspondence to L.L.: phone: +358 2 9448 1169; E-mail: lari.lehtio@oulu.fi
1

ACCEPTED MANUSCRIPT
ABSTRACT
Human Diphtheria toxin-like ADP-ribosyltranferases (ARTD) 10 is an enzyme carrying out mono-
ADP-ribosylation of a range of cellular proteins and affecting their activities. It shuttles between
cytoplasm and nucleus and influences signaling events in both compartments, such as nuclear factor
kappa-light-chain-enhancer of activated B cells (NF-κB) signaling and S phase DNA repair.
Furthermore, overexpression of ARTD10 induces cell death. We recently reported on the discovery
of a hit compound, OUL35 (compound 1), with 330 nM potency and remarkable selectivity towards
ARTD10 over other enzymes in the human protein family. Here we aimed at establishing a
structure-activity relationship of the OUL35 scaffold, by evaluating an array of 4-
phenoxybenzamide derivatives. By exploring modifications on the linker between the aromatic
rings, we identified also a 4-(benzyloxy)benzamide derivative, compound 32, which is potent (IC50
= 230 nM) and selective, and like OUL35 was able to rescue HeLa cells from ARTD10-induced cell
death. Evaluation of an enlarged series of derivatives produced detailed knowledge on the structural
requirements for ARTD10 inhibition and allowed the discovery of further tool compounds with
submicromolar cellular potency that will help in understanding the roles of ARTD10 in biological
systems.
Key Words:
ADP-ribosylation; PARP; ARTD; Inhibitor; Structure-Activity Relationship
2

ACCEPTED MANUSCRIPT
1. Introduction
ADP-ribosylation is a post translational reversible protein modification catalysed in human cells by
dedicated ADP-ribosyltransferase enzymes and some sirtuins [1]. ADP-ribosyltransferases share
homology to bacterial toxins and they can be divided into membrane anchored cholera toxin-like
(ARTCs) and mainly soluble diphtheria toxin-like (ARTDs or PARPs) enzymes. The ARTD family
includes 17 protein modifying enzymes that share a conserved catalytic domain responsible for
ADP-ribosylation activity [2–4]. Based on the catalytic activity, the members of the human ARTDs
family are further categorised into transferases that are capable of synthesizing polymers of ADP-
ribose (pARTDs) and mono-ADP-ribosyltransferases (mARTDs). The key difference between the
enzymes is in the ‘H-Y-E’ (His-Tyr-Glu) motif where Glu is replaced by a hydrophobic residue in
mARTDs. The lack of the Glu residue limits the activity of mARTD to mono-ADP-ribosylation
(MARylation) of a target residue without poly-APD-ribose formation [5]. mARTDs are involved in
various cellular activities, including transcriptional regulation, immunity and inflammation, stress
response, and DNA damage response [6]. Some of these functions have been linked to diseases such
as cancer and thus the relevant enzymes represent potential therapeutic targets [7,8].
ARTD10 belongs to the mARTD class and it was the first enzyme experimentally demonstrated to
be a true mARTD [5]. ARTD10 was first described as a novel c-Myc interacting partner and it was
shown that it ADP-ribosylates itself and also core histones [9]. The catalytic motif of ARTD10
contains an Ile, but when it was mutated to Glu, analogous to pARTDs, it did not transform
ARTD10 to a polymerase indicating that there are further differences in addition to the simple
amino acid substitution [5,10]. In addition to the C-terminal catalytic domain, ARTD10 contains
several other domains and motifs, including an RNA recognition motif (RRM), two ubiquitin
interacting motifs (UIMs) as well as nuclear localization and export sequences (NLS and NES)
[5,9]. Within the catalytic domain there is a small sequence motif, a PIP box, enabling interaction
with proliferating cell nuclear antigen (PCNA) [11]. The domains and motifs allow ARTD10 to
3

ACCEPTED MANUSCRIPT
interact with various proteins and enable it to function in apoptosis, signalling, DNA repair, and
transcription. ARTD10 shuttles between both cytoplasm and nucleus and it is hypothesized that the
RRM may be involved in control of RNA processing in programmed cell death [12,13].
Protein microarray studies have shown that many of the ARTD10 target proteins are kinases,
receptors and growth factors [14]. ARTD10 has been shown to interact with K63-linked poly-
ubiquitin chains (K63-pUb) through its two UIMs. K63-pUb serves as a scaffold in several
signalling pathways, including in signal transduction by NF-κB. ARTD10 interaction with K63-pUb
facilitates ADP-ribosylation of nuclear factor NF-κB essential modulator (NEMO) and prevents its
poly-ubiquitination. In turn this results in reduced activation of inhibitor of κB kinases, which
prevents NF-κB-dependent gene transcription [15]. ARDT10 is also involved in modulating
glycogen synthase kinase (GSK3β) activity, which has functions in Wnt signalling, metabolism,
neural development, and tumorigenesis [14]. During S phase DNA repair, ARTD10 is recruited by
PCNA to DNA damage sites where MARylation activity is required to promote translesion DNA
synthesis in the event of replication fork stalling [11]. Recently, it has been reported that a
deficiency in ARTD10 correlates with severe developmental delays in a patient and that the
extracted patient cells are sensitive to replication stress induced by hydroxyurea and display
enhanced DNA damage-induced apoptosis [16]. It implies that the DNA repair failure could be due
to the loss of ADP-ribosylation of DNA repair proteins [16].
We described earlier a robust screening method for ARTD10 inhibitors and used it to discover a hit
compound, OUL35 (further referred to as compound 1, Figure 1A), which turned out to be a potent
(IC50 = 330 nM) and selective ARTD10 inhibitor [17,18]. Compound 1 is also able to enter cultured
cells and inhibit ARTD10-dependent cellular processes. Docking-based structural analysis indicated
that compound 1 extends toward the so called “acceptor site” where a potential substrate protein
would bind to be ADP-ribosylated (Figure 1B,C). Compound 1 has two symmetrical aromatic
benzamide moieties designated hereafter as anA-ring binding to the nicotinamide binding site and a
4

ACCEPTED MANUSCRIPT
B-ring extending towards the acceptor site. With the aim to evaluate the structural requirements for
activity and selectivity of compound 1, we analyzed a set of its analogs for ARTD10 inhibition. We
used commercially available compounds and synthesized the missing key compounds in order to
obtain a complete picture of the structure–activity relationship (SAR). We demonstrate that only a
small electron donating ether substituent at the p-position of the A-ring benzamide is needed for
ARTD10 micromolar inhibition, but larger substituents enhanced the potency while maintaining
ARTD10 selectivity. Also the cellular efficacy was improved with the best compounds that showed
submicromolar potency in a colony formation assay.
Figure 1. Binding of compound 1 to the active site. A) Chemical structure of 1 with the aromatic
rings indicated as A- and B-rings. B) Crystal structure 1 bound to the nicotinamide pocket of
ARTD7 (Y598L). Sigma A weighted omit Fo-Fc electron density map is coloured in white and
contoured at 3.0 σ. Amino acid residues and 1 are coloured in magenta and black, respectively. A-
and B-rings of 1 are indicated. Hydrogen bonds between 1 and residues are indicated by black dash
lines. C) Superimposition of the nicotinamide pockets of the ARTD7 (Y598L) complex structure
(colored in magenta) with the modelled binding pose in ARTD10 (colored in light grey). Residues
for ARTD7/ARTD10 are labelled.
2. Chemistry
Among the studied compounds, the synthesis of derivatives 7, 9, 11-17, 19, and 35 was carried out
according to the reactions depicted in Schemes 1-4. 4-Phenoxybenzamides 7, 9, 11, 12, and 35
5

ACCEPTED MANUSCRIPT
(Scheme 1) were obtained through a first coupling reaction between the appropriate phenols and 4-
fluorobenzonitrile furnishing intermediates 36 [19] 37 [20], 38 [19], and 39 [21], which were then
converted into their corresponding benzamides 7 [22], 11 [23], 12 [24], and 35, respectively, by
using H2O2 and K2CO3 in DMSO or H2SO4. O-Demethylation of the methoxy derivative 7,
performed with BBr3 in dry CH2Cl2, led to the corresponding hydroxyl derivative 9. An analogous
procedure was applied for the preparation of compound 13 (Scheme 1), by reacting 4-
hydroxybenzonitrile and 2-bromopyridine yielding the benzonitrile intermediate 40 [25], which was
then hydrated under acid-catalyzed conditions affording benzamide derivative 13.
Scheme 1. Synthesis of compounds 7, 9, 11, 12, and 35. Reagents and conditions: (a) KOtBu, dry
DMSO, 160° C; (b) H2O2 30%, K2CO3, DMSO, rt; (c) H2SO4 conc, 60° C; (d) BBr3, dry CH2Cl2, rt.
Analogously, derivatives 14 and 15, in which a nitrogen atom replaced the oxygen linker, were
synthesized starting with the preparation of intermediate 41 [26] obtained by reacting 4-
aminobenzonitrile and 4-fluorobenzonitrile (Scheme 2). Methylation of intermediate 41 to
compound 42 [27] was performed with MeI and K2CO3. The successive H2O2 mediated conversion
of the nitrile groups of compounds 41 and 42 yielded the corresponding amide derivatives 14 and
15, respectively.
6

ACCEPTED MANUSCRIPT
Scheme 2. Synthesis of compounds 14 and 15. Reagents and conditions: (a) KOtBu, dry DMSO,
160° C; (b) H2O2 30%, K2CO3, DMSO, rt; (c) MeI, K2CO3, dry DMF, rt.
Derivatives 16 and 17, in which the oxygen linker was replaced by a sulfur atom, were prepared by
treating 4-bromobenzonitrile in the presence of sulfur, CuI and Cs2CO3 in dry DMF, furnishing
derivative 43 [28], which in turn was converted into target compounds by reaction with different
equivalents of H2O2 (Scheme 3).
Scheme 3. Synthesis of compounds 16 and 17. Reagents and conditions: (a) S8, CuI, Cs2CO3, dry
DMF, 100° C; (b) H2O2 30%, K2CO3, DMSO, rt.
Finally, 4-(benzyloxy)benzamide derivative 19 was prepared by applying a Mitsunobu reaction
between 4-hydroxybenzonitrile and 4-(hydroxymethyl)benzonitrile, followed by H2O2 mediated
conversion of the resulting dibenzonitrile intermediate 44 [29] (Scheme 4).
7

ACCEPTED MANUSCRIPT
Scheme 4. Synthesis of compound 19. Reagents and conditions:(a) Ph3P, DIAD, dry THF; (b) H2O2
30%, K2CO3, DMSO, rt.
3. Results
3.1. Structural analysis of the binding site
We previously reported on the discovery of compound 1 (OUL35) as a first-in-class inhibitor of
ARTD10 (Figure 1A) [15]. Based on molecular docking and mutagenesis studies, we proposed that
the compound would bind to the nicotinamide binding pocket of ARTD10 analogously to
previously characterized ARTD/PARP inhibitors, but it would extend to the so called “acceptor
site”. The acceptor site was identified by Ruf and co-workers as the location where the target
protein to be ADP-ribosylated is expected to interact [30]. In order to gain further confidence on its
binding mode, we determined a crystal structure of compound 1 in complex with ARTD7 (Y598L)
at 1.6 Å resolution (Figure 1B). This ARTD7 protein is a gain-of-binding mutant containing the
mutation Y598L that corresponds to the Leu926 in ARTD10. The mutation allows the binding of 1
with similar affinity as ARTD10 itself (IC50 = 300 nM), while we have not been able to get a co-
crystals of ARTD10 with 1 [17]. The crystal structure contains two protein molecules in the
asymmetric unit and compound 1 had well defined electron density in one of the active sites
(Figure 1B). However, in the second molecule compound 1 is only partially visible based on the
electron density (Figure S1). The superimposition of protein chains suggests that the binding of
compound 1 induces a disorder in the D-loop. In molecule binding 1, the residues between Cys578
and Asn582 lack electron density and therefore they are not included in the model. In the second
protein molecule, the D loop is completely visible but its conformation does not allow compound 1
8

ACCEPTED MANUSCRIPT
to bind because the loop would clash with the compound (Figure S1A). Furthermore, inspection of
the crystal packing reveals that a neighboring protein molecule limits the mobility of the D-loop in
molecule not binding 1 (Figure S1B).
The structure revealed that compound 1 is, as predicted, bound to the nicotinamide pocket between
the two tyrosine residues Tyr591 and Tyr604. The A-ring benzamide forms three hydrogen bonds
with Gly560 and Ser599, as typically observed for ARTD inhibitors (Figure 1B). Compound 1
extends from the binding pocket towards the acceptor site and the B-ring benzamide is located
between Leu598 and Leu659. Leu659 is the hydrophobic residue of the H-Y-E motif, which in
pARTD is a Glu that would clash with the compound and explains why 1 does not inhibit pARTDs.
This crystal structure superimposes well with the predicted binding mode of compound 1 to the
active site of ARTD10 (Figure 1C). In addition to the gain-of-binding mutation (Y598L), ARTD7
contains a Thr603, while the corresponding residue in ARTD10 is Arg in postion 931. This residue
could modulate the compound potency and, indeed, was later identified to form a key interaction
when analogs of 1 were tested for ARTD10 inhibition.
3.2. Selection and design of compound 1 analogs
A first important SAR insight was the finding that when amide moieties of compound 1 were mono-
and di-methylated, the compounds completely lost the ARTD10 inhibitory activity [17]. This is in
agreement with the crystal structure revealing that methylation of the A-ring would destroy
hydrogen bonding at the nicotinamide binding site (Figure 1B). In order to understand the
structural requirements for the anti-ARTD10 activity of compound 1, we assayed a panel of
commercially available analogs and complemented the series with designed and synthesized
derivatives. A moiety like benzamide mimicking nicotinamide is found in most of the ARTDs
inhibitors and it is a key feature anchoring the compounds to the active site. We therefore decided to
keep one of the benzamides of compound 1 (A-ring) and modified other parts of the molecule to
9

ACCEPTED MANUSCRIPT
study minimal structural requirements for ARTD10 inhibition (compounds 2-6, Table 1), the role of
the substituent on the B-ring, as well as the role of the linker between the two aromatic rings (7-35,
Tables 2-4).
3.3. Minimum requirements for an ARTD10 inhibitor
To study the importance of the benzamide rings in hit compound 1, the B-ring was deleted in
compound 2 and replaced by various alkyl substituents in compounds 3-6 (Table 1). The small 4-
hydroxybenzamide 2 did not inhibit ARTD10 significantly at 10 µM concentration. This is
analogous to what was previously reported for the structurally related ARTD inhibitor 3-
aminobenzamide [18]. In contrast, the O-methyl group in 3 was enough to achieve an IC50 of 2.8
µM for ARTD10 inhibition. Increasing the length of the alkyl group by one atom, as in compound
4, slightly decreased potency, while isopropyl and cyclobutyl derivatives gave 1.3 µM and 720 nM
inhibitors 5 and 6, respectively. Together, the data indicated that even a small O-linked alkyl group
would be enough to gain ARTD10 inhibitory activity, while a bulkier substituent would lead to
lower IC50 values and thus higher inhibitory potency. Indeed, the region in the catalytic domain
leading to the acceptor site is hydrophobic due to the surrounding amino acid residues Tyr, Leu and
Ile residues (Figure 1C) explaining why an alkyl substituent is suitable at this position.
10

ACCEPTED MANUSCRIPT
Table 1. Exploration of the minimal compound inhibiting ARTD10. Inhibition % at 10 µM
concentration or the measured IC50 (pIC50 ± SEM, n=3) is reported.
Cmpd
R
-p-CONH2Ph
-H
IC50 (pIC50 ± SEM) / inh. %
330 nM
1
2
3
4
5
6
27%
-CH3
2.8 µM (5.56 ±0.19)
3.5 µM (5.47± 0.04)
1.3 µM (5.90 ± 0.05)
720 nM (6.14 ± 0.06)
-CH2CH3
-CH(CH3)2
-c-butyl
3.4. Role of the p-substituent of the B-ring
As we did not observe an increase in potency compared to 1, we maintained the B-ring focusing on
the nature of the p-substituent. Replacement of the amide moiety of compound 1 by methoxy (7)
resulted in an IC50 of 1.4 µM and similar micromolar affinities were observed for amine (8) and
hydroxyl (9) derivatives (Table 2). On the other hand, the carboxylic acid derivative (10) had an
improved IC50 of 180 nM, but the presence of an aldehyde moiety (11) increased the IC50 to 710
nM. These data indicated that while the amide is not strictly required for inhibition, a carbonyl
feature seems important for potency. Moreover, a hydrogen bond donor in the B-ring is not required
as shown by the carboxyl acid derivative 10 that results in even better potency potentially due to
ionic interaction with Arg931. In the docking model of ARTD10 with compound 1 (Figure 1C), we
observed a putative hydrogen bond of Arg931 with the carbonyl. Mutating Arg931 to Ala resulted in
11

ACCEPTED MANUSCRIPT
a drop in potency (IC50 600 nM) similarly to the compounds lacking this interaction [17]. If the p-
substituent is removed (12), the potency equals that of the small alkyl substituted compounds
(Table 1) indicating that the aromatic character is not required at the B-ring. Replacement of the
whole benzamide B-ring with a 2-pyridine (13) was very detrimental to the inhibition perhaps due
to unfavourable interaction with Leu929 (Figure 1C). Together the results indicate that a phenyl B-
ring is not mandatory, but when suitably substituted with a carbonyl group at the p-position, such as
carboxylic acid, amide, and to some extent aldehyde, it provides a potent ARTD10 inhibitor.
Docking of 10 and 11 to the active site of ARTD10 suggests similar binding modes to that of
compound 1: A-ring benzamide binds to the nicotinamide pocket, while the B-ring extends towards
the acceptor site (Figure S2A,B).
Table 2. Exploration of the 4-substituent in the B-ring. Inhibition % at 10 µM concentration or
the measured IC50 (pIC50 ± SEM, n=3) is reported.
Cmpd
R
-CONH2
-OCH3
-NH2
-OH
X
IC50 (pIC50 ± SEM)
330 nM
CH
CH
CH
CH
CH
CH
CH
N
1
7
1.4 µM (5.86 ± 0.19)
1.3 µM (5.90 ± 0.11)
1.8 µM (5.76 ± 0.13)
180 nM (6.73 ± 0.05)
710 nM (6.21 ± 0.21)
1.1 µM (5.94±0.07)
49%
8
9
-COOH
-CHO
-H
10
11
12
13
-H
12

ACCEPTED MANUSCRIPT
3.5. Linker modifications
The oxygen linker in 1 restricts the geometry of the symmetric O-linked benzamide rings and the
angle between them to approximately 111°. In order to evaluate the requirement for the particular
geometry of this feature, we studied alternative linking moieties (Table 3). Initially, the oxygen
atom was replaced by other heteroatoms such as nitrogen (14 and 15), sulfur (16 and 17), and
phosphorus (18). All the substitutions were detrimental indicating that the geometry between the
nicotinamide mimicking benzamides was critical for ARTD10 inhibition.
Lower potency (IC50 = 2 µM) was observed also when longer and more flexible -OCH2- was used
as linker (compound 19), while the same linker coupled with the unsubstituted phenyl ring led to
slightly improved potency for compound 20 (IC50 = 1.4 µM). The longer linker would push the B-
ring more towards the acceptor site possibly explaining this observation (Figure 1). Based on this
data, two additional unsubstituted analogs (21 and 22), characterized by an ester or an ethylene
glycol linker, respectively, were also tested, but they were less active than 20. Additional
compounds modified at both the linker and at the C-4 substituent (23, 24 and 25) were almost
completely inactive. Altogether these data indicated that changing the linking moiety decreased the
potency to varying degrees. Thus an ether is required for potent activity revealing the linker
geometry as a critical factor for the inhibitor potency. These data also highlighted that while the
presence of a suitable substituent at the p-position of the 4-phenoxybenzamide B-ring grants potent
ARTD10 inhibitory activity, the 4-(benzyloxy)benzamides, characterized by the longer -OCH2-
linker, are less tolerant to a p-substituent.
13

ACCEPTED MANUSCRIPT
Table 3. Exploration of the linker between the aromatic rings. Inhibition % at 10 µM
concentration or the measured IC50 (pIC50 ± SEM, n=3) is reported.
Cmpd
14
L
-NH-
R
IC50 (pIC50 ± SEM) / inh. %
-CONH2
-CONH2
-CONH2
-CN
0%
-N(CH3)-
-S-
26%
15
30%
16
-S-
16%
0%
17
-PO(CH3)-
-OCH2-
-OCH2-
-OCO-
-OCH2CH2O-
-CO-
-CONH2
-CONH2
-H
18
2.0 µM (5.64 ± 0.11)
1.4 µM (5.87 ± 0.18)
7.3 µM (5.14 ± 0.02)
1.9 µM (5.71 ± 0.13)
24%
19
20
-H
21
-H
22
-NO2
23
-OCH(CH3)- -CH(CH3)2
-CH2O- -OCH2CH3
16%
24
0%
25
Thus, we decided to test if the potency of the 4-(benzyloxy)benzamides could be enhanced by
placing different substituents in the various positions of the B-ring. The presence of a p-methyl
substituent (26) lowered the potency to 9.3 µM, which however slightly increased to 4.5 µM when
two methyl groups were placed at m-positions (27) (Table 4). The acceptor site is located on the
surface of the protein and it contains hydrophilic side chains explaining why lipophilic methyl
14

ACCEPTED MANUSCRIPT
groups would not be preferred on the B-ring. In contrast however, a bulky naphthyl group in 28 was
tolerated resulting in an IC50 of 1.4 µM similar to 20. While the 2-pyridyl ring decreased drastically
potency in the 4-phenoxybenzamide derivative 12 (Table 2) for the longer 4-(benzyloxy)benzamide
scaffold it resulted in a same IC50 as a phenyl B-ring (29, Table 4). Halogen substituents on the B-
ring resulted in varying changes in IC50 values. o- (30) and m-Bromo (31) derivatives were weak
ARTD10 inhibitors with IC50 values of 3.8 µM and 2.2 µM, respectively. Differently, an o-fluoro
substituted derivative (32) was the most potent compound identified in this study with IC50 of 230
nM, which is in strong contrast to IC50 values of m- (33, IC50 = 3.3 µM) and even more of p-fluoro
(34, IC50 = 7.9 µM) derivatives. The data collected analyzing all the 4-(benzyloxy)benzamides
showed that the nature of the B-ring strongly modulates the activity, the p-position was confirmed
as the less suitable to accommodate substituents, and a proper group can confer potent activity as in
the case of o-fluoro derivative 32. The interesting activity of 32 prompted the synthesis of the 4-
(phenoxy)benzamide analog 35, which however was less active (IC50 = 1.7 µM) confirming the
different SAR for 4-(benzyloxy) and 4-(phenoxy)benzamides. The o-fluoro substituent of 35
actually lowered the potency of the corresponding 4-phenoxy(benzamide) 12 (1.7 µM and 1.1 µM,
respectively).
15

ACCEPTED MANUSCRIPT
Table 4. Substituents in the B-ring. Inhibition % at 10 µM concentration or the measured IC50
(pIC50 ± SEM, n=3) is reported.
Cmpd
26
R
n
IC50
(pIC50 ± SEM)
Cmpd
27
R
n
IC50
(pIC50 ± SEM)
9.3 µM
(5.03 ± 0.12)
4.5 µM
(5.34 ± 0.03)
1
1
1.4 µM
(5.84 ± 0.04)
1.35 µM
(5.87 ± 0.16)
29
31
1
1
28
3.8 µM
2.2 µM
30
32
1
1
(5.42 ± 0.13)
1
1
(5.66 ± 0.09
)
230 nM
3.3 µM
33
35
(6.63 ± 0.25
)
)
(5.48 ± 0.13)
1.7 µM
7.9 µM
(5.10 ± 0.08
34
(5.77 ± 0.16)
1
0
Based on docking, the B-ring in 4-(benzyloxy)benzamide extends to the acceptor site but is in a
different orientation compared to the more conformationally restricted 4-(phenoxy)benzamides
(Figure S2C). The fluorine atom in 32 could enhance the compound interaction with the aromatic
side chain of Tyr919 with CH-F distance of 2.8 Å in the docking model.
16

ACCEPTED MANUSCRIPT
3.6. Profiling for inhibitor selectivity
The initial hit compound 1 had shown good selectivity for ARTD10 compared to the other tested
ARTD/PARP enzymes [17] and we wanted to test whether the new compounds 10, 11, and 32
retained this property (Table 5). We also included compound 3 in the profiling to assess minimal
requirements for ARTD10 selectivity. Overall, the best ARTD10 inhibitors are also selective
towards it and they are all poor inhibitors of pARTDs ARTD1, 2, 5 and 6. Compound 10 inhibits
also ARTD7/PARP15 (IC50 = 2.6 µM), while 11 was remarkably selective towards ARTD10. It
could be expected that the longer and more flexible -OCH2- linker in 32 would cause poor
selectivity, but we only observed some inhibition of ARTD7/PARP15 and ARTD8/PARP14. We did
not observe much stabilization of ARTD9 and ARTD13 in differential scanning fluorimetry and
only compound 10 had over 1 degree Tm shift for ARTD9, while there was a clear correlation with
the Tm shift and IC50 values for ARTD10 (Table 5, Figure S3A,B). Surprisingly, the minimalistic
compound 3 was also selective towards ARTD10 suggesting that the oxygen in the linker is a key
element for inhibitor selectivity and it can be further modulated by an aromatic B-ring and its
substituents.
17

ACCEPTED MANUSCRIPT
Table 5. IC50 and thermal stabilization by hit compounds against other ARTD family members.
Thermal stabilization was measured at 100 µM.
1#
3
10
11
32
>100 µM
>100 µM
>100 µM
23 µM
57 µM
>100 µM
>100 µM
>100 µM
>100 µM
>100 µM
>100 µM
2.6 µM
>100 µM
>100 µM
86 µM
>100 µM
>100 µM
>100 µM
>100 µM
>100 µM
>100 µM
3.3 µM
ARTD1/PARP1
ARTD2/PARP2
ARTD3/PARP3
ARTD4/PARP4
ARTD5/TNKS1
ARTD6/TNKS2
ARTD7/PARP15
ARTD8/PARP14
ARTD9/PARP9*
ARTD10/PARP10
ARTD10/PARP10*
ARTD12/PARP12
ARTD13/PARP13*
ARTD15/PARP16
77 µM
>100 µM
65 µM
>100 µM
>100 µM
>100 µM
> 10 µM
> 10 µM
0.04 ± 0.44
710 nM
>100 µM
>100 µM
>10 µM
23 µM
>100 µM
>100 µM
>10 µM
>10 µM
0.15 ± 0.38
2.8 µM
> 10 µM
1.2 ± 0.60
180 nM
~ 10 µM
0.76 ± 0.28
230 nM
0.72 ± 0.13
330 nM
3.8 ± 0.10
> 10 µM
0.40 ± 0.08
4.2 µM
2.76 ± 1.0
>10 µM
0.44 ± 0.16
> 10 µM
5.21 ± 0.89
> 10 µM
0.72 ± 0.13
>10 µM
3.60 ±0.80
>10 µM
4.20 ± 0.60
> 10 µM
0.37 ± 0.12
> 10 µM
0.49 ± 0.18
> 10 µM
*∆Tm (DSF), ○C ± SD; #values taken from Venkannagari et al [17].
3.7. Cell activities of the most potent compounds
From the whole set of compounds, 4-(phenoxy)benzamides 10 and 11 and 4-(benzyloxy)benzamide
32 showed highest/best ARTD10 inhibition, and therefore they were selected for further studies. We
tested whether they could rescue HeLa cells from ARTD10 induced cell death (Figure 2A), as
previously shown by compound 1 [17]. This assay demonstrates whether the compounds are able to
enter cells and inhibit overexpressed ARTD10 enzyme, which depends on enzymatic activity
promotes/causes cell death. The carboxylic acid derivative 10 was completely inactive in the colony
18

ACCEPTED MANUSCRIPT
formation assay (Figure 2B), while the aldehyde derivative 11 (IC50 = 1 µM) had similar cellular
activity as 1 (IC50 = 1.35 µM) [15] (Figure 2C, D). The o-fluoro 4-(benzyloxy)benzamide
derivative 32 (IC50 = 2 µM) was slightly less active than compound 1 (Figure 2C,D). At the highest
concentration of 11 and 32 the ARTD10 effect was fully reverted and no toxic effect in control cells
expressing the catalytically inactive ARTD10-G888W was observed (Figure 2C). Most probably,
the lack of activity of compound 10 is due to its inability to enter cells and engage with ARTD10
because of the carboxyl moiety. We also excluded that 10 is simply toxic and thus does not support
cell proliferation because the control cells that express ARTD10-G888W were unaffected by
compound 10 (Figure 2B). This hypothesis was confirmed by a cellular thermal shift assays
(CETSA), which showed that compound 10 did not support enhanced solubility of ARTD10
compared to the DMSO control, while 11 and 32 resulted in increased solubility of ARTD10
between 54°C and 56°C (Fig. 2E).
19

ACCEPTED MANUSCRIPT
Figure 2. Rescue of ARTD10 overexpressing cells by compounds. A-C) HeLa cells (300/well)
expressing doxycycline (Dox)-inducible constructs encoding ARTD10 (WT) or ARTD10-G888W
(GW), which is catalytically inactive, were treated with 500 ng/ml Dox. After 10 days the cells were
stained with methylene blue. Compounds 10, 11 and 32 were added to the final concentrations
indicated and replinished ever second day. Representative wells are shown. D) Colony formation
experiments with compounds 11 and 32 were quantified. The data represent mean values of three
experiments performed in duplicates with standard deviations. E) CETSA assays of U2OS cells
incubated with the indicated compounds (10 µM) showing stabilization of ARTD10 by 11 and 32,
but not by 10.
20

ACCEPTED MANUSCRIPT
4. Discussion and Conclusions
In the current study we used the traditional approach of studying SAR by modifying an already
reported potent ARTD10 inhibitor [17]. Based on the scaffold of compound 1, we purchased,
designed and synthesized new analogs. The strategy was to keep the nicotinamide mimicking
benzamide constant for anchoring the compound to the binding site and to vary other parts of the
molecule in order to better understand the interactions when the compound extends towards the new
inhibitor binding site at the proximity of the so called acceptor site (Figure 1). The first
modifications were to understand the minimum structural features needed for inhibition of ARTD10
and already with an O-linked small alkyl group we could measure micromolar IC50 values. Together
with the observed ARTD10 selectivity of the O-methyl derivative 3 (Table 5), this describes a key
pharmacophore for an ARTD10 inhibitor. We then focused on modifying the linker region between
the aromatic A- and B-rings. The analysis revealed that also an -OCH2- linker in 4-
(benzyloxy)benzamides could be useful for ARTD10 inhibition (Table 3). Potency of the 4-
(benzyloxy)benzamides can be enhanced by the presence of a suitable substituent on the B-ring
such as the o-fluorine that impart 32 an IC50 of 230 nM for ARTD10 inhibition. All analyzed
compounds showed selectivity towards ARTD10 over the other assayed enzymes of the human
ARTD family suggesting that extending the compounds towards the acceptor site would be a way to
create specific mARTD, at least ARTD10, inhibitors. The hydrophobic residue between the NAD+
binding site and the acceptor site is Ile987 in ARTD10 (Figure 1C) and the corresponding residue is
hydrophobic in most mARTDs, in contrast to pARTDs, directly implying that it is possible to
distinguish pARTD and mARTD inhibitors through the interactions at this location. Our study
confirmed that not only 1, but also a handful of analogs extending here are able to inhibit ARTD10
and not pARTDs. The compounds were directed towards ARTD10 and they indeed were selective
towards ARTD10 over other mARTDs (Table 5). Therefore, further modification will be likely
21

ACCEPTED MANUSCRIPT
needed to develop specific inhibitors for other mARTDs from the 4-(phenoxy) and 4-
(benzyloxy)benzamides.
The best compounds described here, 1, 11 and 32, are not only selective towards ARTD10, but are
also nontoxic to cells. They enter cells and inhibit cellular ARTD10 rescuing the cells from
ARTD10 induced cell death (Figure 2). They are therefore useful tools to study ARTD10 biology
and the roles of ARTD10 in DNA repair, cellular signaling, and apoptosis.
5. Experimental Section
5.1. Chemistry
All starting materials were commercially available unless otherwise indicated. All commercially
available starting materials, reagents, and solvents, including anhydrous solvents, were purchased
from commercial sources and used as received. All reactions were routinely checked by thin-layer
chromatography (TLC) using silica gel 60F254 (Merck) and spots were visualized by exposure to
ultraviolet light (UV) or iodine. After extraction, organic solutions were dried over anhydrous
Na2SO4, filtered, and concentrated with a Büchi rotary evaporator at reduced pressure. Flash
column chromatography was performed using Merck silica gel 60 (230-400 mesh). Yields are of
purified product and were not optimized. 1H NMR and 13C NMR spectra were acquired at 200 MHz
(Bruker Avance DPX-200) or at 400 MHz (Bruker Avance DRX-400) using residual solvents signal
such as chloroform (δ = 7.26) or dimethylsulfoxide (δ = 2.48) as an internal standard. Chemical
shifts are reported in ppm (δ) and coupling constant are reported in hertz (Hz). The spectral data are
consistent with the assigned structures. The spin multiplicities are indicated by the symbols s
(singlet), d (doublet), dd (doublet of doublet), t (triplet), q (quartet), m (multiplet), and bs (broad
singlet). Melting points were determined in capillary tubes (Büchi Electrothermal Mod. 9100) and
are uncorrected. High-resolution mass spectra (HRMS) were recorded on Agilent Technologies
6540 UHD Accurate Mass Q-TOF LC/MS, HPLC 1290 Infinity using electrospray ionization (ESI)
22

ACCEPTED MANUSCRIPT
in the positive mode. Purity of the final compounds was determined by LC-MS on Agilent
Technologies 6550 iFUNNEL Q-TOF equipped with HPLC 1290 Infinity with DAD detector and
found to be ≥95%. HPLC conditions to assess the purity of final compounds were as follows:
column, Zorbax Eclipse Plus C12 Rapid Resolution HD 2.1x50mm (1.8 µM); flow rate, 0.5
mL/min; acquisition time, 4 min, stop run at 6 min; DAD 190−650 nm; oven temperature, 50 °C;
Eluent A: water +0.1% Formic Acid, Eluent B: Acetonitrile +0.1% Formic Acid, linear gradient B
from (5%-95% in 4 min, stop run at 6 min).
Compounds 2-6, 8, 10, 20, 23, 24, 25, 27, 28, 32, 34 were purchased form Enamine; 18, 26, 30, 33
from Chembridge; 22 and 29 from Specs; 31 from Princeton and 4 from Combi-Blocks. Their
purity was confirmed to be > 95% based on LC-MS. Compounds 7, 9, 11-17, 19, and 35 were
instead synthesized as described below. Although the synthetic procedures used for the synthesis of
compounds 7 [22] and 11 [23] were already reported, their characterization was reported here for
the first time. For compounds 12 [24] and 41 [29] a different synthetic procedure was applied and
herein described.
5.1.1. General procedure for nitriles hydration to amides
(Method A). To a solution of the appropriate nitrile (1.0 mmol) in DMSO (4 mL) cooled to 0 °C,
K2CO3 (0.15 mmol) and aqueous 30% H2O2 solution (0.7 mL/mmol) were added. The reaction
mixture was then stirred at room temperature until no starting material was detected by TLC. The
mixture was diluted with water, yielding a precipitate that was filtered, washed with water, and
purified as described below.
(Method B). A mixture of the appropriate nitrile (1.0 mmol) and H2SO4 conc (5 mL/mmol) was
stirred at 60 °C until no starting material was detected by TLC. The reaction mixture was then
poured into ice/water and added of saturated solution of NaHCO3, yielding a precipitate that was
filtered and purified as described below.
23

ACCEPTED MANUSCRIPT
5.1.2. 4-(4-Methoxyphenoxy)benzamide (7) [22]. The title compound was prepared through Method
1
A (2.5 h). White crystals, mp 193-195 °C. H NMR (400 MHz, DMSO-d6): δ 3.65 (s, 3H, OCH3),
6.85 (d, J = 7.9 Hz, 2H, aromatic CH), 6.95 and 7.05 (d, J = 8.1 Hz, each 2H, aromatic CH), 7.25
(bs, 1H, CONH2), 7.75-7.85 (m, 3H, aromatic CH and CONH2); 13C NMR (101 MHz, DMSO-d6): δ
55.7, 115.5, 116.4, 121.7, 128.5, 129.9, 148.7, 156.4, 161.0, 167.5. HRMS (ESI) m/z: [M+H]+ calcd
for C14H13NO3, 244.0974; found 244.0963.
5.1.3. 4-(4-Hydroxyphenoxy)benzamide (9). To a solution of 7 (0.45 g, 1.85 mmol) in dry CH2Cl2 (4
mL), 1M solution of BBr3 in CH2Cl2 (7.40 mL, 7.40 mmol) was added dropwise maintaining the
temperature at 0 °C. The reaction mixture was stirred at room temperature for 4.5 h. Then, the
mixture was quenched with MeOH and water and extracted with CH2Cl2. The organic layers were
evaporated to dryness, yielding a residue that was purified by flash chromatography eluting with
1
CHCl3:MeOH (93:7), to give 9 (0.15 g, 36%) as white solid; mp 202-204 °C. H NMR (400 MHz,
DMSO-d6): δ 6.75 (d, J = 8.7 Hz, 2H, aromatic CH), 7.85 (d, J = 8.5 Hz, 2H, aromatic CH), 7.90 (d,
J = 8.7 Hz, 2H, aromatic CH), 7.20 (bs, 1H, CONH2), 7.75-7.85 (m, 3H, aromatic CH and CONH2),
9.50 (s, 1H, OH); 13C NMR (101 MHz, DMSO-d6): δ 116.1, 116.7, 121.9, 128.3, 129.9, 147.2,
154.7, 161.3, 167.5. HRMS (ESI) m/z: [M+H]+ calcd for C13H11NO3, 230.0818; found 230.0809.
5.1.4. 4-(4-Formylphenoxy)benzamide (11) [23]. The title compound was prepared through Method
1
B. (1 h). White solid, mp 135-136 °C. H NMR (400 MHz, DMSO-d6): δ 7.10-7.20 (m, 4H,
aromatic CH), 7.35 (bs, 1H, CONH2), 7.80-7.90 (m, 4H, aromatic CH), 7.95 (bs, 1H, CONH2), 9.95
13
(s, 1H, CHO); C NMR (101 MHz, DMSO-d6): δ 118.8, 119.7, 130.3, 130.9, 132.2, 132.5, 157.8,
161.9, 167.4, 192.1. HRMS (ESI) m/z: [M+H]+ calcd for C14H11NO3, 242.0818; found 242.0813.
5.1.5. 4-Phenoxybenzamide (12). The title compound was prepared starting from 38 through
24

ACCEPTED MANUSCRIPT
Method A (2 h) and purified by crystallization by EtOH/H2O mixture, in 50% yield as white
crystals; mp 177-178 °C. 1H NMR (400 MHz, DMSO-d6): δ 6.95 (d, J = 8.6 Hz, 2H, aromatic CH),
7.05 (d, J = 7.9 Hz, 2H, aromatic CH), 7.20 (t, J = 7.3 Hz, 1H, aromatic CH), 7.30 (bs, 1H,
CONH2), 7.40 (t, J = 7.8 Hz, 2H, aromatic CH), 7.80 (d, J = 8.6 Hz, 2H, aromatic CH), 7.90 (bs,
13
1H, CONH2); C NMR (101 MHz, DMSO-d6): δ 117.7, 119.9, 124.7, 129.4, 130.1, 130.6, 156.0,
159.9, 167.5. HRMS (ESI) m/z: [M+H]+ calcd for C13H11NO2, 214.0869; found 214.0859.
5.1.6. 4-(Pyridin-2-yloxy)benzamide (13). The title compound was prepared starting from 40 [25]
through Method B (1 h) and purified by crystallization by EtOH, in 37% yield as white crystals; mp
1
165-167 °C. H NMR (400 MHz, DMSO-d6): δ 7.05 (d, J = 8.2 Hz, 1H, aromatic CH), 7.15-7.20
(m, 3H, aromatic CH), 7.30 (bs, 1H, CONH2), 7.80-7.90 (m, 3H, aromatic CH), 7.95 (bs,1H,
CONH2), 8.10-8.20 (m, 1H, aromatic CH); 13C NMR (101 MHz, DMSO-d6): δ 112.4, 119.9, 120.9,
129.6, 130.7, 140.8, 147.9, 156.8, 162.9, 167.7. HRMS (ESI) m/z: [M+H]+ calcd for C12H10N2O2,
215.0821; found 215.0829.
5.1.7. 4,4'-Iminodibenzoamide (14). The title compound was prepared starting from 41 [26]
through Method A (5 min) and purified by crystallization by EtOH, in 40% yield as white crystals;
mp 260-261 °C. 1H NMR (200 MHz, DMSO-d6): δ 7.05 and 7.70 (d, J = 8.6 Hz, each 6H, aromatic
13
CH and CONH2), 8.75 (s, 1H, NH); C NMR (101 MHz, DMSO-d6) 116.3, 126.1, 129.5, 145.6,
167.8. HRMS (ESI) m/z: [M+H]+ calcd for C14H13N3O2, 256.1087; found 256.1082.
5.1.8. 4,4'-(Methylmino)dibenzamide (15). The title compound was prepared starting from 42 [27]
through Method A (5 h) and purified by crystallization by EtOH/DMF mixture, in 55% yield as
1
white crystals; mp 280-281 °C. H NMR (200 MHz, DMSO-d6): δ 3.25 (s, 3H, CH3), 6.95 (d, J =
8.7 Hz, 4H, aromatic CH), 7.20 (bs, 2H, CONH2), 7.70-7.80 (m, 6H, aromatic CH and CONH2);
25

ACCEPTED MANUSCRIPT
13C NMR (101 MHz, DMSO-d6) 40.1, 119.7, 127.0, 129.3, 150.5, 168.1. HRMS (ESI) m/z:
[M+H]+ calcd for C15H15N3O2, 270.1243; found 270.1234.
5.1.9. 4,4'-Thiodibenzamide (16). The title compound was prepared starting from 43 [28] through
Method A (5 h) and purified by crystallization by EtOH/DMF mixture, in 17% yield as white
1
crystals; mp 299 °C. H NMR (400 MHz, DMSO-d6): δ 7.30 (d, J = 8.2 Hz, 4H, aromatic CH),
13
7.40 (bs, 1H, CONH2), 7.85 (d, J = 8.2 Hz, 4H, aromatic CH), 7.95 (bs, 2H, CONH2); C NMR
(101 MHz, DMSO-d6) 129.1, 130.6, 133.7, 138.2, 167.5. HRMS (ESI) m/z: [M+H]+ calcd for
C14H12N2O2S, 273.0698; found 273.0691.
5.1.10. 4-[(4-cyanophenyl)thio)benzamide (17). The title compound was prepared starting from 43
[28] through Method A (5 min) but using 0.1 mL/mmol of aqueous 30% H2O2 solution and purified
by flash chromatography eluting with CHCl3:MeOH (98:2), in 39% yield as white solid; mp
1
207 °C. H-NMR (200 MHz, DMSO-d6): δ 7.25 (d, J = 8.3, 2H, aromatic CH), 7.40 (bs, 1H,
CONH2), 7.50 (d, J = 8.2, 2H, aromatic CH), 7.70 (d, J = 8.3 Hz, 2H, aromatic CH), 7.85 (d, J = 8.2
Hz, 2H, aromatic CH), 8.00 (bs, 1H, CONH2); 13C NMR (101 MHz, DMSO-d6): 109.2, 119.0,
129.0, 129.4, 133.0, 133.5, 135.0, 135.1, 143.5, 167.4. HRMS: m/z calcd for C14H10N2OS 255.0593
(M+H)+, found 255.0598.
5.1.11. 4-{[4-(Aminocarbonyl)benzyl]oxy}benzamide (19). The title compound was prepared
starting from 44 through Method A (2.5 h) and purified by crystallization by DMF, in 36% yield as
1
white crystals; mp 305-306 °C. H NMR (400 MHz, DMSO-d6): δ 5.20 (s, 2H, CH2), 7.00 (d, J =
8.8 Hz, 2H, aromatic CH), 7.20 and 7.35 (bs, each 1H, CONH2), 7.50 (d, J = 8.2 Hz, 2H, aromatic
CH), 7.80 (d, J = 8.7 Hz, 3H, aromatic CH and CONH2), 7.85 (d, J = 8.2 Hz, 2H, aromatic CH); 13C
NMR (101 MHz, DMSO-d6): δ 69.1, 114.6, 127.1, 127.6, 128.0, 129.7, 134.1, 140.3, 160.8, 167.7,
26

ACCEPTED MANUSCRIPT
167.9. HRMS (ESI) m/z: [M+H]+ calcd for C15H14N2O3, 271.1083; found 271.1077.
5.1.12. 4-(2-Fluorophenoxy)benzamide (35). The title compound was prepared starting from 39
through Method A (3 h) and purified by crystallization by cyclohexane/EtOAc mixture, in 68%
yield as white crystals; mp 152-153 °C. 1H NMR (400 MHz, DMSO-d6): δ 6.90 (d, J = 8.7 Hz, 2H,
aromatic CH), 7.20-7.30 (m, 4H, aromatic CH and CONH2), 7.35-7.45 (m, 1H, aromatic CH), 7.85
(d, J = 8.8 Hz, 2H, aromatic CH), 7.90 (bs, 1H, CONH2); 13C NMR (101 MHz, DMSO-d6): δ
116.1, 117.8 (JH-F = 17.9 Hz), 123.4, 126.1 (JH-F = 3.7 Hz), 126.8 (JH-F = 7.2 Hz), 129.4, 130.1,
142.2 (JH-F = 11.6 Hz), 154.2 (JH-F = 247.9 Hz), 159.8, 167.5. HRMS (ESI) m/z: [M+H]+ calcd for
C13H10FNO2, 232.0775; found 232.0764.
5.1.13. 4-[(Cyanobenzyl)oxy]benzonitrile (44) [29]. To a solution of diisopropyl azodicarboxylate
(DIAD) (0.35 mL, 1.76 mmol) in dry THF (10 mL), Ph3P (1.3 g, 5.20 mmol) was added at 0 °C
under N2 atmosphere. The mixture was stirred until there was a white precipitate. Then, 4-
hydroxybenzonitrile (0.2 g, 1.67 mmol) and 4-(hydroxymethyl)benzonitrile (0.23 g, 1.76 mmol)
were added and the reaction mixture was maintained to room temperature overnight. The mixture
was then poured into ice/water, yielding a precipitate that was filtered and treated with Et2O, to give
44 (0.17 g, 44%) as white solid. 1H NMR (400 MHz, DMSO-d6): δ 5.25 (s, 2H, CH2), 7.15 and 7.75
(d, J = 8.6 Hz, each 2H, aromatic CH), 7.65 and 7.85 (d, J = 8.0 Hz, each 2H, aromatic CH).
5.2. PAINS and aggregators filters.
The most active compounds (1, 10, 11, 20, 32,) were examined for known classes of pan-assay
interference compounds (PAINS) [31] and aggregators by using the ZINC 15 remover filter at
http://zinc15.docking.org [32]. None of the compounds were found as potential PAINS or
aggregators.
27

ACCEPTED MANUSCRIPT
5.3. Protein expression and purification
All proteins used in the current study were expressed in E. coli, and purified in using Ni-affinity and
size exclusion chromatography according to our previously reported protocols [17]. The details of
the constructs used for each ARTD enzyme are listed in Table S1.
5.4. Activity assay
Dose response experiments were carried out using our previously reported activity assay for
mARTD enzymes [18]. Half log dilutions of inhibitors were used and reactions were performed in
quadruplicates. IC50 curves were fitted using sigmoidal dose response curve (four variables) in
GraphPad Prism version 5.04 (GraphPad Software). Details of the assay conditions of different
ARTD enzymes are available in Table S1.
5.5. Differential Scanning Fluorimetry
DSF experiments were carried out for ARTD9, ARTD10 and ARTD13. Protein used was diluted to
0.25 mg/ml in PBS. SyproOrange (Life Technologies) at a concentration of 5X was used as the
reporter dye. The compound concentration used for the experiment was 100 µM. The experiments
were repeated three times with appropriate control. It was performed on a real-time PCR machine
(Applied Biosystems) with the temperature increasing from 21°C up to 90°C (70 cycles) with 1°C
increment per minute.
5.6. Colony Formation and Cellular Thermal Stability Assay
The colony formation assays and the cellular thermal stability assays were performed precisely as
described earlier [17]. The compounds were dissolved in DMSO and added to the final
concentrations as indicated in the figure. New medium and compounds were added every other day.
28