Discovery of potent and selective nonplanar tankyrase inhibiting nicotinamide mimics

Article abstract of DOI:10.1016/j.bmc.2015.06.063

Diphtheria toxin-like ADP-ribosyltransferases catalyse a posttranslational modification, ADP-ribosylation and form a protein family of 17 members in humans. Two of the family members, tankyrases 1 and 2, are involved in several cellular processes including mitosis and Wnt/β-catenin signalling pathway. They are often over-expressed in cancer cells and have been linked with the survival of cancer cells making them potential therapeutic targets. In this study, we identified nine tankyrase inhibitors through virtual and in vitro screening. Crystal structures of tankyrase 2 with the compounds showed that they bind to the nicotinamide binding site of the catalytic domain. Based on the co-crystal structures we designed and synthesized a series of tetrahydroquinazolin-4-one and pyridopyrimidin-4-one analogs and were subsequently able to improve the potency of a hit compound almost 100-fold (from 11 μM to 150 nM). The most potent compounds were selective towards tankyrases over a panel of other human ARTD enzymes. They also inhibited Wnt/β-catenin pathway in a cell-based reporter assay demonstrating the potential usefulness of the identified new scaffolds for further development.

Full text of DOI:10.1016/j.bmc.2015.06.063

Bioorganic & Medicinal Chemistry 23 (2015) 4139–4149
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Discovery of potent and selective nonplanar tankyrase inhibiting
nicotinamide mimics
Yves Nkizinkiko a, B. V. S. Suneel Kumar b, Variam Ullas Jeankumar b, Teemu Haikarainen a,
Jarkko Koivunen a, Chanduri Madhuri b, Perumal Yogeeswari b, Harikanth Venkannagari a, Ezeogo Obaji a,
a,
Taina Pihlajaniemi a, Dharmarajan Sriram b, , Lari Lehtiö
⇑
⇑
a Faculty of Biochemistry and Molecular Medicine & Biocenter Oulu, University of Oulu, PO Box 5400, FIN-90014 Oulu, Finland
b Department of Pharmacy at Birla Institute of Technology and Science-Pilani, Hyderabad campus, Hyderabad 500078, India
a r t i c l e i n f o
a b s t r a c t
Article history:
Diphtheria toxin-like ADP-ribosyltransferases catalyse a posttranslational modification, ADP-ribosylation
and form a protein family of 17 members in humans. Two of the family members, tankyrases 1 and 2, are
involved in several cellular processes including mitosis and Wnt/b-catenin signalling pathway. They are
often over-expressed in cancer cells and have been linked with the survival of cancer cells making them
potential therapeutic targets. In this study, we identified nine tankyrase inhibitors through virtual and
in vitro screening. Crystal structures of tankyrase 2 with the compounds showed that they bind to the
nicotinamide binding site of the catalytic domain. Based on the co-crystal structures we designed and
synthesized a series of tetrahydroquinazolin-4-one and pyridopyrimidin-4-one analogs and were subse-
Received 26 March 2015
Revised 22 June 2015
Accepted 24 June 2015
Available online 2 July 2015
Keywords:
Tankyrase
Poly(ADP-ribose)polymerase
Inhibition
quently able to improve the potency of a hit compound almost 100-fold (from 11 lM to 150 nM). The
most potent compounds were selective towards tankyrases over a panel of other human ARTD enzymes.
They also inhibited Wnt/b-catenin pathway in a cell-based reporter assay demonstrating the potential
usefulness of the identified new scaffolds for further development.
Cancer
Protein crystallography
Virtual screening
Wnt-signalling
Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction
The first indication of the therapeutic potential of tankyrase
inhibition arose from the observation that tankyrases control the
Tankyrases belong to the diphtheria toxin-like ADP-ribosyl-
transferase (ARTD) protein superfamily also known as poly(ADP-
ribosyl)polymerases (PARPs) (EC 2.4.2.30).1 Human tankyrase
1 (TNKS1/ARTD5/PARP5a) and tankyrase 2 (TNKS2/ARTD6/PARP5b)
have a C-terminal catalytic ARTD domain, which is conserved in
the protein family and is responsible for modifying target proteins
by adding one or more ADP-ribose units to specific residues. TNKS1
and TNKS2 are homologous with 82% sequence identity and have
overlapping functions.2 In addition to the ARTD domain, tankyrases
have sterile alpha motifs (SAM) responsible for their oligomeriza-
tion and five ankyrin repeat clusters (ARC) responsible for recog-
nizing and binding target proteins.3
length
(PARsylating)
of
human
telomeres
by
poly-ADP-ribosylating
a
shelterin protein complex component TRF1.
Shelterin protects telomeres by preventing the access of telom-
erase to telomeres.4 PARsylation of TRF1 by tankyrases releases
TRF1 from the telomeres and allows telomerase to extend the
DNA ends. This system is over-activated in cancer cells leading to
an uncontrolled telomere extension.5 Recently many different
functions for tankyrases has been discovered2,6,7 and these have
caused an increasing interest in developing tankyrase inhibitors.
Regarding cancer two functions of tankyrases are of special inter-
est. TNKS1 was found in spindle poles during mitosis and it is
believed to facilitate the formation of normal spindle structure
and function.8 Disrupting this process might be a way to disturb
rapidly dividing cancer cells. Tankyrases also control the Wnt sig-
naling pathway, which is a key survival pathway in many cancer
cells. Tankyrases PARsylate Axin, which is an essential protein for
the formation of the b-catenin destruction complex: a multiprotein
complex controlling b-catenin stability through phosphorylation.9
The PARsylation of axin destabilizes the destruction complex,
Abbreviations: ADE, adenosine subsite; ARC, ankyrin repeat cluster; ARTD, ADP-
ribosyltransferase with diphtheria toxin homology; NI, nicotinamide subsite; PAR,
poly(ADP-ribose); PARP, poly(adp-ribose) polymerase; SAM, sterile alpha motif;
TNKS, tankyrase; TRF1, telomeric repeat binding factor 1.
⇑
Corresponding authors. Tel.: +91 40 66303506 (D.S.), +358 2 9448 1169 (L.L.).
E-mail addresses: dsriram@hyderabad.bits-pilani.ac.in (D. Sriram), lari.lehtio
@oulu.fi (L. Lehtiö).
http://dx.doi.org/10.1016/j.bmc.2015.06.063
0968-0896/Ó 2015 Elsevier Ltd. All rights reserved.

4140
Y. Nkizinkiko et al. / Bioorg. Med. Chem. 23 (2015) 4139–4149
stabilizes b-catenin and leads to the activation of the Wnt signaling
pathway.10 Inhibition of tankyrases, therefore, increases cellular
levels of Axin and decreases the levels of b-catenin, which ulti-
mately decreases the oncogenic expression mediated by b-catenin
and leads to the inhibition of tumorigenesis.9,11
interaction energies. E-pharmacophore is computed by docking
simulation and the pharmacophore features are then ranked based
on their site scores. Here, the E-pharmacophore model was gener-
ated from the co-crystal of 1 complexed with TNKS2 and it was
subjected to re-docking studies (PDB code 3KR8). This structure
was selected as it is a high resolution complex structure of a highly
potent compound (XAV939) with TNKS2, which is also used in the
biochemical testing of the compounds. The RMSD between the
docked pose and crystal conformation is 0.95 Å with glide score
of À9.52. It indicates the docking reliability in terms of reproducing
the experimentally observed binding mode. The detailed method-
ology of E-pharmacophore modelling, docking, and ROCS mod-
elling is discussed in Supporting information.
Glide XP (extra precision) module of Schrödinger 9.2 (Glide, ver-
sion 5.7, Schrödinger, LLC, New York, NY, 81 2011) was utilized for
docking. TNKS2–1 complex structure (PDB: 3KR8) was used for
docking of compounds. The protein was prepared using protein
preparation wizard and glide energy grids were generated for the
prepared protein complex. The binding site was defined by a rect-
angular box surrounding the ligand (1). The ligand was refined
using the ‘Refine’ option in Glide, and the option 70 to output
Glide XP descriptor information was chosen (Glide 71 v5.7,
Schrodinger, LLC, New York, NY). For the refinement and docking
calculations, the default settings as available in the software pack-
age were used. The results from the re-docking studies were used
for E-pharmacophore modelling.
Crystal structures of the catalytic domains of both human tan-
kyrases have been solved,12,13 which enables rational design of tan-
kyrase inhibitors. Protein crystallography has also helped to
rationalize the observed selectivity of some of the inhibitors14–16
and it has been utilized in the development of several TNKS inhibi-
tor scaffolds.10,17 The donor NAD+ binding groove of the ARTD
domain has two sub-sites, namely the nicotinamide (NI) and the
adenosine (ADE) sites, which have been targeted by inhibitors.7
The known TNKS inhibitors such as 1–4 (Fig. 1a) bind to the NI sub-
site whereas some inhibitors bind to the ADE subsite.18,19 Also dual
binders interacting with both of the subsites have recently been
developed.20,21 The hit compounds identified in this study bind
to the NI subsite similarly to several other previously characterized
ARTD inhibitors, such as 1 (XAV939). We utilized the available
structural data in structure-based virtual screening approach with
an aim to identify new tankyrase inhibitor scaffolds. The initial hit
compounds were further developed using the existing structural
knowledge as a guide for compound synthesis. Structure–activity
relationship and tankyrase selectivity was rationalized with the
help of protein crystallography. We demonstrate the selectivity
of the new inhibitors and show that the compounds are active in
a cell-based reporter assay.
The generated E-pharmacophore model was further validated
by enrichment factor (EF) studies in screening a database. A small
library, consisting of 250 tankyrase inhibitors were divided into
2. Materials and methods
three bins based on activity range, either highly active (<1
lM),
moderately active (1–10 M) or inactive (>10 M). Database
l
l
2.1. Virtual screening and docking
screening was done by using the pharmacophore model to validate
the predictive power of the model. The results were analyzed using
a set of parameters such as hit list (Ht), number of active percent of
Energy-based pharmacophore modelling (E-pharmacophore) is
a combined effort of pharmacophore perception and protein-ligand
Figure 1. (a) Chemical structure of known tankyrase inhibitors binding to the NI site: 1 XAV939,13 2 G244-LM,38 3 2(tert-butylphenyl)-3,4-dihydroquinazolin-4-one15 and 4
2-4-(propan-2-yl)phenyl-3,4-dihydro-2H-1,3-benzoxazin-4-one.14,30 (b) Schematic representation of 11 and the scaffolds of synthesized pyridopyrimidinone and
dihydroquinazolines analogs. The R1, R2 and R3 represent the ortho-, meta- and para-position, respectively, where substitutions were made.

Y. Nkizinkiko et al. / Bioorg. Med. Chem. 23 (2015) 4139–4149
4141
yields (%Y), percent ratio of actives in the hit list (%A), enrichment
factor (EF), false negatives, false positives, and goodness of hit
score (GH). E-pharmacophore model succeeded in the retrieval of
177 actives (88.5%) of the 200 hits.
2.2.1.2.
2-(o-Tolyl)-2,3-dihydropyrido[2,3-d]pyrimidin-4(1H)-
one (18). The compound was synthesized according to the above
general procedure using 2-aminonicotinonitrile (0.1 g, 0.84 mmol),
2-methylbenzaldehyde (0.1 g, 0.84 mmol), and DBU (0.1 g,
0.84 mmol) to afford 18 (0.13 g, 65%) as off white solid. 1H NMR
(CDCl3) dH 2.34 (s, 3H), 5.83 (s, 1H), 6.71–7.89 (m, 6H), 8.16–8.17
(m, 1H). 13C NMR (DMSO-d6): dc 163.7, 158.4, 153.1, 145.5, 137.6,
135.2, 130.6, 126.9, 126.3, 125.8, 114.8, 110.1, 63.7, 18.7. ESI-MS
m/z 240.1 (M+H)+. Anal Calcd for C14H13N3O; C, 70.28; H, 5.48; N,
17.56. Found: C, 70.19; H, 5.38; N, 17.51.
The validated pharmacophore model was used as 3D query to
screen a commercial database (Asinex) of 4,000,000 compounds
to identify potential hits. Top 5000 hits were selected based on
the pharmacophore feature mapping and visual inspection. The
selected hits were re-ranked using the ROCS model to investigate
how similar the hit compounds are to the top active 1 in terms
of shape and features. The ROCS model was generated using
the redocked conformation of the compound 1 and evaluated fur-
ther based on the model ranking (enrichment factor) to the known
tankyrase inhibitors and decoys. The re-ranked hit list from ROCS
was visually inspected and the top diversified hits were chosen
based on the ROCS Rank and Tanimoto combo score before being
subjected to docking studies. The hit compounds were shortlisted
again based on the docking scores and interaction patterns.
2.2.1.3. 2-(2-Fluorophenyl)-2,3-dihydropyrido[2,3-d]pyrimidin-
4(1H)-one (20). The compound was synthesized according to the
above general procedure using 2-aminonicotinonitrile (0.1 g,
0.84 mmol), 2-fluorobenzaldehyde (0.136 g, 0.84 mmol), and DBU
(0.1 g, 0.84 mmol) to afford 20 (0.12 g, 60%) as off white solid. 1H
NMR (CDCl3) dH 5.86 (s, 1H), 6.74–7.88 (m, 6H), 8.13–8.14 (m,
1H). 13C NMR (DMSO-d6): dc 163.5, 158.7, 156.4, 152.9, 137.9,
130.2, 128.7, 128.4, 123.8, 115.1, 112.1, 109.6, 62.3. ESI-MS m/z
244.1 (M+H)+. Anal Calcd for C13H10FN3O; C, 64.19; H, 4.14; N,
17.28. Found: C, 64.27; H, 4.07; N, 17.21.
2.2. Synthesis and chemicals
All commercially available chemicals and solvents were used
without further purification. TLC experiments were performed on
alumina-backed silica gel 40 F254 plates (Merck, Darmstadt,
Germany). The homogeneity of the compounds was monitored
by thin layer chromatography (TLC) on silica gel 40 F254 coated
on aluminum plates, visualized by UV light and KMnO4 treatment.
Biotage Microwave reactor Initiator was utilized for developing
compounds 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34 and 36. All
1H and 13C NMR spectra were recorded on a Bruker AM-300
(300.12 MHz, 75.12 MHz) NMR spectrometer, Bruker BioSpin
Corp, Germany. Molecular weights of the synthesized compounds
were checked by LCMS 6100B series Agilent Technology.
Chemical shifts are reported in ppm (d) with reference to the inter-
nal standard TMS. The signals are designated as follows: s, singlet;
d, doublet; dd, doublet of doublets; t, triplet; m, multiplet.
Elemental analyses were carried out on an automatic Flash EA
1112 Series, CHN Analyzer (Thermo). The analytical and spectral
data (1H NMR, 13C NMR, mass spectra and elemental analysis) of
all the synthesized compounds were in full agreement with the
proposed structures.
2.2.1.4. 2-(2-(Benzyloxy)phenyl)-2,3-dihydropyrido[2,3-d]pyrim-
idin-4(1H)-one (22). The compound was synthesized according to
the above general procedure using 2-aminonicotinonitrile (0.1 g,
0.84 mmol), 2-benzyloxybenzaldehyde (0.18 g, 0.84 mmol), and
DBU (0.1 g, 0.84 mmol) to afford 22 (0.12 g, 42.9%) as off white solid.
1H NMR (CDCl3) dH 5.19 (s, 2H), 5.85 (s, 1H), 6.71–7.92 (m, 11H),
8.16–8.18 (m, 1H). 13C NMR (DMSO-d6): dc 163.3, 157.6, 154.5,
152.7, 137.8, 136.9, 128.9, 128.2, 127.9, 127.6, 127.3, 123.8, 121,
114.3, 111.8, 109.6, 70.2, 63.1. ESI-MS m/z 332.1 (M+H)+. Anal
Calcd for C20H17N3O2; C, 72.49; H, 5.17; N, 12.68. Found: C, 72.58;
H, 5.24; N, 12.74.
2.2.1.5. 2-(3-(Benzyloxy)phenyl)-2,3-dihydropyrido[2,3-d]pyrim-
idin-4(1H)-one (24). The compound was synthesized according to
the above general procedure using 2-aminonicotinonitrile (0.1 g,
0.84 mmol), 3-benzyloxybenzaldehyde (0.18 g, 0.84 mmol), and
DBU (0.1 g, 0.84 mmol) to afford 24 (0.13 g, 46.4%) as off white solid.
1H NMR (CDCl3) dH 5.23 (s, 2H), 5.82 (s, 1H), 6.69–7.87 (m, 11H),
8.19–8.20 (m, 1H). 13C NMR (DMSO-d6): dc 162.7, 157.8, 157.4, 153,
148.9, 137.6, 136.9, 129.4, 128.7, 127.7, 126.9, 114.6, 113.9, 111.8,
110.6, 109.7, 69.1, 64.8. ESI-MS m/z 332.1 (M+H)+. Anal Calcd for
Compounds 15, 21, 29, 31, 33, 35, 37,22 17, 19,23 27,24 14, 28, 30,
3225 and 3426 were all known compounds; spectral data obtained
were in agreement with the proposed structures and matched
those reported in literature.
C20H17N3O2; C, 72.49; H, 5.17; N, 12.68. Found: C, 72.41; H, 5.13; N,
12.61.
2.2.1. General procedure for synthesis of 16, 18, 20, 22, 24, 26,
and 36
2.2.1.6. 2-(4-(Benzyloxy)phenyl)-2,3-dihydropyrido[2,3-d]pyrim-
idin-4(1H)-one (26). The compound was synthesized according to
the above general procedure using 2-aminonicotinonitrile (0.1 g,
0.84 mmol), 4-benzyloxybenzaldehyde (0.18 g, 0.84 mmol), and
DBU (0.1 g, 0.84 mmol) to afford 26 (0.15 g, 53.5%) as off white
solid. 1H NMR (CDCl3) dH 5.21 (s, 2H), 5.81 (s, 1H), 6.68–6.74 (m,
3H), 7.32–7.89 (m, 8H), 8.15–8.16 (m, 1H).13C NMR (DMSO-d6):
dc 162.9, 157.5, 154.8, 152.8, 137.6, 136.9, 136.7, 129.2, 128,
127.6, 127.3, 113.9, 112.8, 109.4, 69.9, 65.1. ESI-MS m/z 332.2
(M+H)+. Anal Calcd for C20H17N3O2; C, 72.49; H, 5.17; N, 12.68.
Found: C, 72.57; H, 5.19; N, 12.76.
2-Aminonicotinonitrile (1 equiv), the corresponding aldehyde
(1 equiv), 1,8-diazbicyclo [5.4.0] undec-7-ene (DBU) (1 equiv),
and 1.5 mL water was added in sequential order in a 5 mL micro-
wave vial, sealed and heated to 100 °C for about 5–10 min. The
reaction mixture was then cooled to 28 °C and the precipitate
obtained was filtered and re-crystallized from appropriate solvent
to give the desired compounds in good yields as mentioned below.
2.2.1.1. 2-(2-Chlorophenyl)-2,3-dihydropyrido[2,3-d]pyrimidin-
4(1H)-one (16).
The compound was synthesized according to
the above general procedure using 2-aminonicotinonitrile (0.1 g,
0.84 mmol), 2-chlorobenzaldehyde (0.12 g, 0.84 mmol), and DBU
(0.1 g, 0.84 mmol) to afford 16 (0.14 g, 63.6%) as off white solid.
1H NMR (CDCl3) dH 5.87 (s, 1H), 6.69–7.91 (m, 6H), 8.14–8.16 (m,
1H). 13C NMR (DMSO-d6): dc 163.7, 158.2, 153.6, 143.9, 137.4,
133.6, 128.9, 128.6, 128.2, 126.9, 115.1, 110.3, 62.9. ESI-MS m/z
260.2 (M+H)+. Anal Calcd for C13H10ClN3O; C, 60.12; H, 3.88; N,
16.18. Found: C, 60.01; H, 3.81; N, 16.24.
2.2.1.7. 2-(4-(tert-Butyl)phenyl)-2,3-dihydropyrido[2,3-d]pyrim-
idin-4(1H)-one (36). The compound was synthesized according
to the above general procedure using 2-aminonicotinonitrile
(0.1 g, 0.84 mmol), 4-tert-butylbenzaldehyde (0.136 g, 0.84 mmol),
and DBU (0.128 g, 0.84 mmol) to afford 36 (0.14 g, 58.3%) as off
white solid. 1H NMR (CDCl3) dH 1.29 (s, 9H), 5.79 (s, 1H), 6.69–
7.83 (m, 6H), 8.13–8.15 (m, 1H). 13C NMR (DMSO-d6): dc 163.1,
157.6, 153.6, 152.9, 141.8, 137.7, 125.1, 124.4, 114.3, 109.8, 65.1,

4142
Y. Nkizinkiko et al. / Bioorg. Med. Chem. 23 (2015) 4139–4149
39.6, 36.4. ESI-MS m/z 282.1 (M+H)+. Anal Calcd for C17H19N3O; C,
72.57; H, 6.81; N, 14.94. Found: C, 72.49; H, 6.73; N, 14.98.
Tecan Infinity M1000 plate reader (excitation/emission,
372 nm/444 nm).
2.2.2. General procedure for synthesis of 23 and 25
2.5. Screening of inhibitors and measurement of inhibitor
potencies
The synthesis followed the literature procedure.22 Cyanuric
chloride (10 mol %) was added to a solution of anthranilamide
(1 equiv) and desired aldehyde (1 equiv) in acetonitrile (2 mL).
The reaction mixture was then stirred at 28 °C for about 15 min
(monitored by TLC & LCMS for completion) and allowed to cool.
Solvent was evaporated off and the precipitate formed was filtered
and re-crystallized form ethanol to afford the desired product as
described below.
All compounds were stored at À20 °C in DMSO and diluted in
the TNKS2 assay buffer before use. The preliminary screening
was done at 10 lM and 1 lM concentrations in duplicates.
Furthermore, compound controls were used to exclude the effect
of compound fluorescence and quenching. Inhibitor potencies were
measured for the hit compounds that had IC50 values below
100 lM based on the two point preliminary screening. IC50 values
2.2.2.1. 2-(2-(Benzyloxy)phenyl)-2,3-dihydroquinazolin-4(1H)-
one (23). The compound was synthesized according to the above
general procedure using anthranilamide (0.1 g, 0.73 mmol), 2-ben-
zyloxy benzaldehyde (0.15 g, 0.73 mmol), and cyanuric chloride
(0.013 g, 0.073 mmol) to afford 23 (0.13 g, 54%) as white solid. 1H
NMR (CDCl3) dH 5.21 (s, 2H), 6.01 (s, 1H), 6.88–7.69 (m, 13H) 13C
NMR (DMSO-d6): dc 165.3, 157.1, 144.9, 136.5, 133.2, 129.1,
128.4, 127.8, 127.6, 127.5, 126.9, 123.8, 121, 117.1, 116.7, 113.8,
111.8, 70.9, 64.8. ESI-MS m/z 331.2 (M+H)+. Anal Calcd for
C21H18N2O2; C, 76.34; H, 5.49; N, 8.48. Found: C, 76.31; H, 5.54;
N, 8.42.
were measured using half log dilutions and reactions were carried
in quadruplicates on three separate days. The incubation time was
adjusted so that substrate conversion was 50–60% in the case of
screening and less than 30% in the case of IC50 measurements.
Dose response curves for the compounds were measured using a
half-log dilutions and fitted using 4-parameters with Graphpad
Prism (version 5.0 for windows).
2.6. Profiling of the inhibitors
Selected TNKS2 inhibitors were profiled against a panel of
human ARTDs using optimized assay conditions (Supporting infor-
mation).27 In order to keep the conditions comparable the concen-
tration of NAD+ was kept low (500 nM) during the screening and
the profiling assays. A dose response curve was measured once
for the best hit compounds using half-log dilutions.
2.2.2.2. 2-(3-(Benzyloxy)phenyl)-2,3-dihydroquinazolin-4(1H)-
one (25). The compound was synthesized according to the above
general procedure using anthranilamide (0.1 g, 0.73 mmol), 3-ben-
zyloxy benzaldehyde (0.15 g, 0.73 mmol), and cyanuric chloride
(0.013 g, 0.073 mmol) to afford 25 (0.11 g,, 45.8%) as off white
solid. 1H NMR (CDCl3) dH 5.24 (s, 2H), 5.99 (s, 1H), 6.84–7.63 (m,
13H) 13C NMR (DMSO-d6): dc 165.6, 160.7, 145.8, 145.3, 136.4,
133.2, 129.8, 128.6, 128.1, 127.6, 126.8, 118.6, 116.9, 116.5,
113.8, 112.5, 110.9, 70.6, 66.7. ESI-MS m/z 331.1 (M+H)+. Anal
Calcd for C21H18N2O2; C, 76.34; H, 5.49; N, 8.48. Found: C, 76.39;
H, 5.42; N, 8.54.
2.7. Crystallization, data collection and refinement
The catalytic domain of TNKS2 (residues 952–1161) was used
for crystallization.14 The crystals were soaked for several days in
a solution containing 22% of PEG3350, 0.2 M Lithium sulfate and
0.1 M Tris (pH 8.5) supplemented with 100 lM of the compound
and with 250 mM NaCl. After soaking the crystals were dipped in
the soaking solution supplemented with 20% glycerol, and flash
frozen in liquid nitrogen.19 The data was collected at ESRF
(Grenoble, France) on beamline ID23-1 and at Diamond Light
source (UK, Didcot) beamlines I02 and I03. The data were pro-
cessed with XDS31 (Supporting information) and the molecular
replacement was done with MOLREP32 using TNKS2 structure com-
plexed with nicotinamide as a template (3U9H). Structures were
refined with REFMAC533,34 and manual building and analysis was
done with Coot.35 The chemical structures of the compounds were
drawn with Marvin (Marvin 5.7.0, 2011, ChemAxon, http://www.
chemaxon.com).
2.3. Expression and purification of the enzymes
The human ARTDs used during this study were expressed and
purified as described shortly below and as reported earlier.15,27
ARTD1-3 were expressed as full length proteins.27 For other pro-
teins, the constructs used consisted of catalytic protein fragments:
ARTD4 (residues 250–565), TNKS1/ARTD5 (1030–1317),
TNKS2/ARTD6 (952–1161 and 873–1161), ARTD7 (460–656) and
ARTD10 (809–1017).27–29 All proteins were purified using a similar
protocol as described before.28,30 Cells were lysed with sonication
in the presence of protease inhibitors and the proteins were puri-
fied using Ni-affinity, TEV-cleavage of the tag and size exclusion
chromatography. A heparin column was additionally used in case
of ARTD1-3 in order to remove protein bound DNA fragments.
Cloning of ARTD3 and detailed protocols for production of recom-
binant ARTD3 and TNKS2 used in the assays are given in the
Supporting information.
2.8. Reporter assay
HEK293,
L Wnt-3a and L cells (ATCC) were cultured in
Dulbecco’s modified Eagle’s medium supplemented with 10% fetal
bovine serum, penicillin and streptomycin at 37 °C in a 5% CO2
atmosphere. Preparation of Wnt3a and control conditioned med-
ium (CM) has been described previously.36 HEK293 cells were pla-
ted on 96-well plate (2.5 Â 104 cells/well) 16 hours before
transfection. Cells were transfected with SuperTopFlash reporter
plasmid37 at the same time with pCMV-ß-galactosidase plasmid
(Clontech) using the FuGENE™ 6 Transfection Reagent (Roche)
according to the manufacturer’s protocol. Following transfection,
cells were incubated 24 hours with 50% of Wnt3a-CM or control-
CM in serum free DMEM containing molecules or DMSO vehicle.
DMSO content was kept below 0.05% in all experiments. Cells were
washed once with PBS, lysed and Wnt3a induced luciferase activity
was measured using the Luciferase assay system (Promega)
2.4. Activity assay
Assays based on quantification of NAD+ were conducted as we
have reported earlier.14,28 Reactions were carried out in 96-well
plates (Greiner bio-one U-shaped) at room temperature. The buffer
used for TNKS2 consisted of 50 mM BisTris propane, pH 7, 0.5 mM
TCEP and 0.01% Triton-X-100 and 500 nM of substrate NAD+ (in
50
none in ethanol and 20
was incubated for 10 min after which 90
added. The plates were recorded after 20 min of incubation using
l
L reaction volume). After the reaction 20
L of 2 M KOH were added and the plate
L of formic acid was
lL of 20% acetophe-
l
l

Y. Nkizinkiko et al. / Bioorg. Med. Chem. 23 (2015) 4139–4149
4143
according to manufacturer’s instructions. b-Galactosidase activity
was measured using the b-galactosidase enzyme assay system
(Promega) to normalize the transfection efficiency. The results
are the mean of three independent experiments SEM. Overall,
the compounds did not affect the luciferase activity when control
medium was used and the control conditioned medium had no sig-
nificant effect on the luciferase activity. Cell viability was also con-
trolled with light microscope.
potent hit compounds, 5 and 6, are both derivates of 2-(piper-
azin-1-yl)-5,6,7,8-tetrahydro-3H-quinazolin-4-one. Interestingly
pyridopyrimidinones 7–13 all had potencies in the
lM range,
although these compounds do not have an aromatic structure typ-
ically reported for ARTDs and tankyrase inhibitors.7,15
3.2. Crystal structures of the hit compounds in complex with
TNKS2
In order to understand and verify the molecular details of the
compound interactions we solved co-crystals structures of the
key compounds with TNKS2 catalytic domain. As expected, 5 and
3. Results
3.1. Virtual screening and design
6 bind to the NI subsite (Fig. 3a and b). They form p–p interaction
with Tyr1071 and hydrogen bonds to the backbone of Gly1032 and
to the side chain hydroxyl of Ser1068. These compounds extend
along the NAD+ binding crevice towards the Phe1035 and the
piperazine moiety forms hydrophobic interactions with Ile1075
of the hydrophobic nook and Tyr1050 of the D-loop. Similarly they
interact with the Phe1035 at the delta of the cleft. Compound 5
extends towards the D-loop with the pyridine moiety and replaces
a water molecule, but does not make direct interactions with the
protein. There is a 15° rotation of Phe1035 with respect to the com-
pound 6 complex structure and the D-loop moves 1 Å in order to
accommodate the bulkier compound 5 (Fig. 3a and b). Both 5
and 6 showed a similar binding mode in the docking studies.
In contrast to 5 and 6, compounds 8, 9, 10 and 11 contain a
nonaromatic B-ring and therefore represent an unconventional
PARP inhibitor scaffold (Table 1, Fig. 1b). We wanted to, despite
the modest potency, to characterize this series further. We solved
We utilized structure and analogue-based approaches to iden-
tify potential tankyrase inhibitors. In the first approach, Energy-
based pharmacophore model (E-pharmacophore) was generated
by utilizing the existing protein-ligand interaction information.
The E-pharmacophore model was validated using TNKS2–1 com-
plex crystal structure (PDB: 3KR8) and a tankyrase inhibitor
library, and the model was then used as query to commercial
(Asinex) database (Fig. 2). Top hits from the pharmacophore
screening were prioritized by shape and feature similarity with
the top active (1) using a ROCS model. The selected compounds
were subjected to docking studies to predict the binding modes
and to reduce the number of false positives. 9 hits were carefully
selected based on the E-pharmacophore, ROCS ranking and dock-
ing analysis and they were screened for tankyrase inhibition.
Six of the 9 hit compounds had an IC50 below 11
lM and three
of the six had potencies lower than 1 M (Table 1). The most
l
Figure 2. Virtual and in vitro screening workflow. The virtual screening was based on the co-crystal structure (3KR8) of compound 1 (XAV939) with TNKS2. The binding
mode of 1 was used to create an E-pharmacophore model, which was used to screen 4 million compounds from vendor databases. Using the ROCS model generated for 1, the
initial hits were re-ranked and 9 final hit compounds were identified for further in vitro studies. Crystallography was used to study co-crystal structures of TNKS2 with the
most potent hit compounds and this lead to the synthesis of a series of analogues which have improved potency against TNKSs.

4144
Y. Nkizinkiko et al. / Bioorg. Med. Chem. 23 (2015) 4139–4149
and selectivity of the inhibitors.14,15 The crystal structures revealed
Table 1
Potencies of the hit compounds against TNKS2
that the rotation of the C-ring of 11 and the presence of para-chlo-
rine caused Tyr1050 and Phe1035 to move away from the com-
pound but the overall structure of the binding pocket did not
change (Fig. 3b and c).
Compound
(PDB code)
Structure
TNKS2 IC50
(pIC50 SEM)
3.3. Derivatives
71 nM
(7.15 0.04), n = 3
5 (5AL5)
3.4.1. Synthesis
The identified compounds are analogous to 1,9 to 2,38 15,39 and
3
to 414,30 (Fig. 1a). We were intrigued by the inhibition of tankyrases
by pyridopyrimidinones containing a chiral center and lacking the
aromatic B-ring. Therefore, using our earlier efforts to create tan-
kyrase selective flavones30 and dihydroquinazolinones15 as guides,
we decided to synthesize analogs of pyridopyrimidinones. N-
methyl group of 11 is restricting the conformation of the phenyl
group and it was removed from the compound at this stage
(Fig. 1b). Derivatives were designed by making systematic substi-
tution either at ortho- or meta- or para-positions at the C-ring
(Fig. 1b).
300 nM
(6.52 0.05), n = 3
6 (5AL4)
720 nM
(6.14 0.05), n = 3
7
The synthesis of 2-(sub)aryl-2,3-dihydroquinazolin-4(1H)-one
and
2-(sub)aryl-2,3-dihydropyrido[2,3-d]pyrimidin-4(1H)-one
derivatives has been well precedent in literature. A series of 12
quinazolinone and 12 pyridopyrimidinone analogues were synthe-
sized by a straightforward methodology as depicted in Scheme 1, as
step towards derivation of a strong structure-activity relationship
and to understand the ideal site for introducing structural diversity.
2-(sub)aryl-2,3-dihydroquinazolin-4(1H)-one analogues com-
pounds 15, 21, 23, 25, 29, 31, 33, 35 and 37 were synthesized using
cyanuric chloride catalyzed protocol as previously reported by
Sharma et al.22 Propylphosphoric anhydride mediated method
was employed for developing compounds 17 and 19.23 Compound
27 was synthesized by condensing anthranilamide and 4-benzyloxy
benzaldehyde in ethanol utilizing catalytic amount of ammonium
chlolride.24 2-(sub)aryl-2,3-dihydropyrido[2,3-d]pyrimidin-4(1H)-
one derivatives (compounds 14, 16, 18, 20, 22, 24, 26, 28, 30, 32,
34 and 36) were generated utilizing a 1,8-diazbicyclo [5.4.0]
undec-7-ene (DBU) catalyzed microwave assisted synthesis as pre-
viously described by Yang et al.25
2300 nM
(5.63 0.05), n = 3
8
6200 nM
(5.2 0.08), n = 3
9
6600 nM
(5.0 0.06), n = 3
10
11,000 nM
(5.0 0.03), n = 3
11 (5AL3)
3.4.2. Evaluation of the analogs
Seven different substituents were used to modify the two start-
ing scaffolds, pyridopyrimidinone 14 and tetrahydroquinazolinone
15. In total, 24 analogs were generated (Table 2). The main differ-
ence between the two series of analogs is the presence of nitrogen
in the pyrimidine-ring (A-ring), whose removal increased the
lipophilicity of the compounds (Table 2). In general, the tetrahy-
droquinazolinones show slightly better potency against TNKS2
(Table 2). Compared to the initial starting compound 11 the most
potent compound of the two series, 37, was a significant improve-
64,000 nM
(4.2 0.38), n = 3
12
13
>100,000 nM
ment in potency from 11 lM to 150 nM (Table 2).
a crystal structure of TNKS2 with 11 containing the chlorine sub-
stituents at the benzene ring. The potency of this nonplanar scaf-
3.4.3. Structure activity relationships
The crystal structures of TNKS2 complexes with selected four
compounds are reported here to illustrate the interaction between
the TNKS2 and the analogs (Fig. 4). The tetrahydroquinazolinone
analogs made same hydrogen bonds as those observed with 5
and 6 whereas the pyridopyrimidinone analogs, in addition to
the typical hydrogen bonds, formed an additional hydrogen bond
between the A-ring nitrogen and a water molecule found in the
vicinity only in these crystal structures (Fig. 4b and c). Similar
hydrogen bond was also formed by 11 (Fig. 3c). The para-substi-
tuted analogs were more potent than the ortho- and meta-substi-
tuted ones and this observation can be explained by the
observation that the para-substitution caused the compound to
extend outwards along the NAD+ binding cavity and form
fold is reduced likely due to the missing capacity to form a
p–p
interaction with Tyr1071. Compound 11 binds also to the NI sub-
site and forms the typical hydrogen bonds (Fig. 3c). The compound
is a racemic mixture, but we clearly identified only the R-enan-
tiomer in the crystal structure. The N-methyl group of the B-ring
interacts with the Tyr1050. Interestingly, the N-methyl group
and the ortho-chlorine substituent of the benzene ring (C-ring)
caused a 90° degree rotation of the C-ring unique for this scaffold
(Fig. 3c and d). Chlorine replaces water molecule hydrogen bonded
to the backbone amide of Tyr1050. Chlorine at the para-position
interacts with the Phe1035. We have previously shown that this
hydrophobic interaction is a potential way to increase affinity

Y. Nkizinkiko et al. / Bioorg. Med. Chem. 23 (2015) 4139–4149
4145
Figure 3. The binding modes of initial hit compounds to TNKS2. Panels show the observed binding modes of (a) 5, (b) 6, and (c) 11. (d) Shows the superposition of compounds
6 and 11 as observed in the crystal structures. Inhibitors are shown in ball and stick models, water molecules as red spheres, surrounding residues as sticks, and hydrogen
bonds as dashed lines.
O
O
CHO
CHO
NC
NH2
NH
a
b
+
+
R
R
NH2
X
N
H
H2N
N
R
X = N, for compounds 14, 16, 18, 20, 22, 24, 26, 28, 20, 32, 34 and 36
X = C, for compounds 15, 17, 19, 21, 23, 25, 27, 29, 31, 33, 35 and 37
a. Cyanuric chloride 10mol%, CH3CN, 27oC (for compounds 15,21,23,25,29,31,33,35 and 37); Propylphosphonic
anhydride (50% in EtOAc), CH3CN, 27oC (for compounds 17,19); NH4Cl (cat:), C2H5OH, 27oC (for compound 27); b.
1,8 - diazbicyclo [5.4.0] undec-7-ene H2O, MW, 100oC (for compounds 14, 16, 18, 20, 22, 24, 26, 28, 30, 32 and 36).
Scheme 1. Synthetic strategy employed for generating compounds 14–37.
enhanced interactions with the protein similarly to 5. The removal
of the N-methyl and the ortho- and the para-chlorine found in 11
was not favorable and resulted in the basic scaffold 14, which
importance of the hydrophobic substituent at this position. The
hydrophobic interactions made by chlorine in the para-position
with Phe1035 observed in the crystal structure (Fig. 4a) were
clearly favorable compared to the chlorine substitutions at ortho-
had lower potency of 12 lM. However, the removal of the A-ring
nitrogen from 14 yielded 15 that is two fold more potent than
14. The ortho-chlorine substitution at the C-ring in 16 and 17
caused lower potencies and other small substituents in this posi-
tion 18–21 did not improve the potency of the starting compounds
14 and 15. Interestingly, the potency was improved by the use of a
larger ortho-substituent, benzyloxy, in 22 and 23. Benzyloxy sub-
stituent decreased the potency when attached to the meta-position
24 and 25, but it is tolerated at the para-position (Table 2).
The initial hit compound 11 contained two chlorine sub-
stituents and the increased potency of 28 (IC50 760 nM) and 29
(IC50 410 nM) where para-chlorine was preserved highlights the
position in 16 and 17, which have decreased potencies of 31
lM
and 140 M, respectively. Small methyl (30 and 31) or methoxy
l
(32 and 33) substituents were not beneficial over chlorine
(Table 2). Replacement of the para-chlorine with a larger aliphatic
substituent, isopropyl, increased the hydrophobic interaction of 34
and 35 with Phe1035 (Fig. 3b). This substitution improved the
potency 2-fold when comparing the compounds in the same series
28 and 34 as well as 29 and 35. Consistently the removal of the A-
ring nitrogen in tetrahydroquinazolinones improved the potency
against TNKS2 (Table 2). Increasing the size of the aliphatic sub-
stituent to tert-butyl yielded the most potent analog in the series

4146
Y. Nkizinkiko et al. / Bioorg. Med. Chem. 23 (2015) 4139–4149
Table 2
Potencies of the synthesized analogs against TNKS2. Dose response curves were measured three times for the most active compounds
Compound (PDB Structure
code)
TNKS2 IC50
(pIC50 SEM)
Compound Structure
TNKS2 IC50
(pIC50 SEM)
12,000 nM
(4.94 0.03)
5500 nM
(5.26 0.06)
14
15
16
18
140,000 nM
93,000 nM
17
19
31,200 nM
92,000 nM
15,000 nM
(4.82 0.03)
20
63,000 nM
21
4500 nM
(5.35 0.04)
580 nM
(6.24 0.05)
22
23
7300 nM
(5.14 0.06)
24
230,000 nM
25
1900 nM
(5.71 0.09)
2700 nM
(5.57 0.15)
26
28
30
32
27
760 nM
(6.12 0.10)
29
(5AKW)
410 nM
(6.38 0.28)
5400 nM
(5.27 0.02)
2700 nM
(5.57 0.229)
31
33
2100 nM
(5.67 0.08)
990 nM
(6.00 0.147)

Y. Nkizinkiko et al. / Bioorg. Med. Chem. 23 (2015) 4139–4149
4147
Table 2 (continued)
Compound (PDB Structure
code)
TNKS2 IC50
(pIC50 SEM)
Compound Structure
TNKS2 IC50
(pIC50 SEM)
360 nM
(6.45 0.07)
230 nM
(6.63 0.06)
34 (5AL2)
36 (5AL1)
35
53,000 nM
(4.27 0.14)
150 nM
(6.81 0.18)
37 (5AKU)
Figure 4. Crystal structures of TNKS2 in complex with the synthesized inhibitors (a) 29, (b) 34, (c) 36 and (d) 37. Inhibitors are shown in ball and stick models, water
molecules as red spheres, surrounding residues as sticks, and hydrogen bonds as dashed lines.
(37) with IC50 of 150 nM, but interestingly 36 showed a remarkably
low potency despite the presence of tert-butyl substituent
(Table 2). Both 36 and 37 bind to the TNKS2 catalytic domain in
compounds have high potency even though they lack the aromatic
B-ring found in other potent TNKS inhibitors such as in 1–4
(Fig. 1a).
a
similar manner as demonstrated by the crystal structure
(Fig. 4c and d). The only clear difference in the structures is the
additional hydrogen bond to a water molecule made by 36.
Taken together, we noticed that the increase in size and
hydrophobicity of the para-substituent from para-methyl, to
para-isopropyl and to para-tert-butyl enhanced the potency of
the compound. Furthermore, the removal of the A-ring nitrogen
produced compounds, which had even better potency. This led to
the removal of the water molecule, increased hydrophobicity and
enhanced potency (Figs. 3 and 4). Notably, the most potent
3.4.4. Inhibition of Wnt signaling
The best compounds (5, 6, 35 and 37) were tested in a
SuperTopFlash (STF) based Wnt responsive reporter assay. When
the canonical Wnt signaling is activated, b-catenin is stabilized
and translocated to the nucleus. In the nucleus, beta-catenin asso-
ciates with TCF/LEF transcription factors and activates the tran-
scription of Wnt target genes. The STF reporter is Wnt responsive
luciferase construct with TCF/LEF binding sites. Thus, the STF assay
measures the accumulation of b-catenin, which is controlled by

4148
Y. Nkizinkiko et al. / Bioorg. Med. Chem. 23 (2015) 4139–4149
and in most cases we did not observe any inhibition of other tested
ARTDs. The potency of 35 and 37 is slightly lower for TNKS1
explaining why the potency in the cell based assay is at the micro-
molar range. Overall 37 emerged as the most potent and selective
compound against tankyrases.
4. Discussion and conclusion
We conducted the virtual screening in search of new TNKS inhi-
bitor scaffolds based on the known TNKS inhibitors. The virtual
screening was successful in that three out of nine initial hits had
potencies, which were less than 1 lM. While the most potent com-
pound 5 is analogous to the compound 4 reported earlier, some of
the initial hit compounds had a unique scaffold lacking an aromatic
B-ring. We wanted to explore this feature by testing pyridopyrim-
idinones and tetrahydroquinazolinones for tankyrase inhibition.
The results showed that the aromaticity of the B-ring is not an
absolute requirement for achieving high potency. The synthesized
24 analogs were distinguishable by their substituents in the C-ring
and by the presence of the A-ring nitrogen, whose removal
increased the hydrophobicity and potency of the compounds.
This observation, which is a trend among the 24 analogs, was in
compliance with our previously published results regarding the
hydrophobicity of potent tankyrase inhibitors.2,7,19 Furthermore,
the obtained structure–activity relationship was in agreement
with our previous studies done with other tankyrase inhibitors
such as flavones and para-substituted 2-phenyl-3,4-dihydroquina-
zolin-4-ones, which bind to nicotinamide subsite.15,30 The removal
of the A-ring nitrogen and the ortho-chlorine from 11 produced
nearly parallel conformation of the aromatic rings upon binding
to TNKS2. The para-substitution of the benzene ring was favorable
over the ortho- and the meta-substitution because of the conforma-
tional restraint caused by the narrow binding site (Figs. 3 and 4).
The substitution at either ortho- or meta-position would clash with
the side chains of Tyr1060, Tyr1050 or His1031, depending on the
size of the substituent. However, the para-substitution extends the
molecule in an opening towards hydrophobic residues Pro1034
and Phe1035 (Fig. 4). In addition, increasing the size of the para-
substituent from methyl to tert-butyl increases the potency of
the compounds due to tighter interaction with the Phe1035 and
Pro1034 lining the pocket.
Figure 5. Inhibition of Wnt/b-catenin signaling pathway in STF cell-based reporter
assay. Compound 1 also known as XAV939 was used as a control. Dose response
with three concentrations is shown for 1, 5, 35 and 37. Compound 6 did not have
any response at the tested concentrations and was omitted from the figure.
Measurements were done in triplicates and the figure shows the mean value of
three independent experiments with SEM.
tankyrases though the stability of the b-catenin destruction com-
plex. STF reporter assay is a widely used method to measure the
interference with the Wnt/b-catenin signaling pathway and for
the effectiveness of tankyrase inhibitors.11,14–17 5, 35 and 37 were
found to be effective Wnt/b-catenin signaling inhibitors at concen-
tration of 10 lM (63–82% inhibition) (Fig. 5). The inhibition is in all
cases dose dependent, but does not reach full inhibition of the
reporter assay as in the case of highly potent control compound
1. Interestingly 6 was not effective in this assay at all (data not
shown) even though it is a potent TNKS2 inhibitor in vitro. The
molecules did not have an effect on b-galactosidase activity and
consequently, the compounds do not alter HEK293 cell viability
at tested concentrations.
3.4.5. Profiling of the best inhibitors
In order to evaluate the specificity of the best compounds we
profiled them against a panel of other ARTD family members. All
the tested compounds showed different degrees of selectivity
(Table 3). Compound 5 and 6 inhibited also other ARTDs with
micromolar IC50 values. The potencies of the compounds against
TNKS1 and TNKS2 were almost equal except 6, which has 21-fold
selectivity towards TNKS2. This indicates that inhibition of
TNKS2 alone would not be enough to inhibit Wnt signaling in
the reporter assay (Fig. 5). As 6 is a selective inhibitor of TNKS2
it could be a useful tool to probe differences between tankyrase
isoforms. Compounds 35 and 37 were selective towards tankyrases
The STF cell-based reporter assay revealed that three out of four
tested compounds were active in cells with compound 35 having
the highest activity. Four compounds were profiled against other
human ARTDs and compound 35 and 37 had higher selectivity
towards tankyrase than other compounds (Table 3). The hydropho-
bic para-substituent is a source for selectivity as it interacts with a
poorly conserved region where, in other ARTDs, they are replaced
with hydrophilic residues explaining the observed specificity for
tankyrases.7 In case of, for example, ARTD1 the para-substituent
would clash with a negatively charged environment of regulatory
domain, not present in tankyrases.15
Using existing structural information we have been able to
identify a new inhibitor scaffold and through synthesis of analogs
we were able to improve the potency of the compounds. We also
demonstrated the selectivity of the inhibitors towards tankyrases
and showed that they are active in cells. The selectivity results
from enhanced interactions with the hydrophobic residues
Phe1035 and Pro1034 found in tankyrases (Figs. 3 and 4). In other
ARTDs this region is more hydrophilic and in case of ARTD1-3 a
regulatory domain interacting with the catalytic domain provides
Table 3
Summary of selectivity profiling of the best compounds with IC50 against different
human ARTDs
5
6
35
37
TNKS1
TNKS2
370 nM
71 nM
5.9
4.2
13
33
>10
5.4
6.3
300 nM
3.6
3.6
91
17
>10
>10
l
M
1.1
230 nM
>100
>100
50
>100
>10
>10
l
M
620 nM
150 nM
>100
>100
34
>100
ARTD1/PARP1
ARTD2/PARP2
ARTD3/PARP3
ARTD4/PARP4
ARTD7/PARP15
ARTD10/PARP10
l
l
M
M
l
l
M
M
l
l
M
M
l
l
M
M
M
lM
l
l
M
M
l
M
l
lM
l
M
l
l
l
M
further polar environment and causes
a selectivity of the
l
M
l
l
M
M
l
l
M*
M*
>10
>10
M*
M*
hydrophobic para-substituted analogs towards tankyrases. We
have reported this observation also for other scaffolds.14,15,40 37
was identified as the most potent inhibitor in the series with IC50
lM
*
No inhibition at the highest concentration used (10
lM).

Y. Nkizinkiko et al. / Bioorg. Med. Chem. 23 (2015) 4139–4149
4149
9. Huang, S.-M. A.; Mishina, Y. M.; Liu, S.; Cheung, A.; Stegmeier, F.; Michaud, G.
A.; Charlat, O.; Wiellette, E.; Zhang, Y.; Wiessner, S.; Hild, M.; Shi, X.; Wilson, C.
J.; Mickanin, C.; Myer, V.; Fazal, A.; Tomlinson, R.; Serluca, F.; Shao, W.; Cheng,
H.; Shultz, M.; Rau, C.; Schirle, M.; Schlegl, J.; Ghidelli, S.; Fawell, S.; Lu, C.;
Curtis, D.; Kirschner, M. W.; Lengauer, C.; Finan, P. M.; Tallarico, J. A.;
Bouwmeester, T.; Porter, J. A.; Bauer, A.; Cong, F. Nature 2009, 461, 614.
10. Bregman, H.; Chakka, N.; Guzman-Perez, A.; Gunaydin, H.; Gu, Y.; Huang, X.;
Berry, V.; Liu, J.; Teffera, Y.; Huang, L.; Egge, B.; Mullady, E. L.; Schneider, S.;
Andrews, P. S.; Mishra, A.; Newcomb, J.; Serafino, R.; Strathdee, C. A.; Turci, S.
M.; Wilson, C.; DiMauro, E. F. J. Med. Chem. 2013, 56, 4320.
11. Waaler, J.; Machon, O.; Tumova, L.; Dinh, H.; Korinek, V.; Wilson, S. R.; Paulsen,
J. E.; Pedersen, N. M.; Eide, T. J.; Machonova, O.; Gradl, D.; Voronkov, A.; von
Kries, J. P.; Krauss, S. Cancer Res. 2012, 72, 2822.
12. Lehtiö, L.; Collins, R.; van den Berg, S.; Johansson, A.; Dahlgren, L.-G.;
Hammarström, M.; Helleday, T.; Holmberg-Schiavone, L.; Karlberg, T.;
Weigelt, J. J. Mol. Biol. 2008, 379, 136.
13. Karlberg, T.; Hammarström, M.; Schütz, P.; Svensson, L.; Schüler, H.
Biochemistry 2010, 49, 1056.
14. Narwal, M.; Koivunen, J.; Haikarainen, T.; Obaji, E.; Legala, O. E.; Venkannagari,
H.; Joensuu, P.; Pihlajaniemi, T.; Lehtiö, L. J. Med. Chem. 2013, 56, 7880.
15. Haikarainen, T.; Koivunen, J.; Narwal, M.; Venkannagari, H.; Obaji, E.; Joensuu,
P.; Pihlajaniemi, T.; Lehtiö, L. ChemMedChem 2013, 8, 1978.
16. Voronkov, A.; Holsworth, D. D.; Waaler, J.; Wilson, S. R.; Ekblad, B.; Perdreau-
Dahl, H.; Dinh, H.; Drewes, G.; Hopf, C.; Morth, J. P.; Krauss, S. J. Med. Chem.
2013, 56, 3012.
17. Shultz, M. D.; Majumdar, D.; Chin, D. N.; Fortin, P. D.; Feng, Y.; Gould, T.; Kirby,
C. A.; Stams, T.; Waters, N. J.; Shao, W. J. Med. Chem. 2013, 56, 7049.
18. Karlberg, T.; Markova, N.; Johansson, I.; Hammarström, M.; Schütz, P.; Weigelt,
J.; Schüler, H. J. Med. Chem. 2010, 53, 5352.
value of 150 nM for TNKS2 and 6 was demonstrated to have a high
selectivity for TNKS2 over TNKS1. Catalytic domains of tankyrases
are highly similar and the residues lining the active site are identi-
cal. Despite the similarity it is usual for tankyrase inhibitors to dis-
play some degree of selectivity for one or the other isoform.41,42 It
is not clear where the selectivity comes from as the crystal struc-
tures do not elucidate features that could be responsible for the
selectivity. The D-loop adopts different conformations in the crys-
tal structures and also the overall dynamics of the catalytic
domains may provide reasons of the selectivity. The crystal struc-
tures consist of only the catalytic domain and the SAM domain
responsible for the oligomerization could contribute on the selec-
tivity. However, 6 is the most isoform selective compound reported
so far and might allow the use of 6 as a chemical probe to decipher
the roles of the tankyrase isoforms in various cellular contexts.
PDB codes
Coordinates and structure factors are deposited to the protein
data bank with codes 5AL1, 5AL2, 5AL3, 5AL4, 5AL5, 5AKU and
5AKW.
19. Narwal, M.; Venkannagari, H.; Lehtiö, L. J. Med. Chem. 2012, 55, 1360.
20. Bregman, H.; Gunaydin, H.; Gu, Y.; Schneider, S.; Wilson, C.; DiMauro, E. F.;
Huang, X. J. Med. Chem. 2013, 56, 1341.
21. Shultz, M. D.; Cheung, A. K.; Kirby, C. A.; Firestone, B.; Fan, J.; Chen, C. H.-T.;
Chen, Z.; Chin, D. N.; Dipietro, L.; Fazal, A.; Feng, Y.; Fortin, P. D.; Gould, T.;
Lagu, B.; Lei, H.; Lenoir, F.; Majumdar, D.; Ochala, E.; Palermo, M. G.; Pham, L.;
Pu, M.; Smith, T.; Stams, T.; Tomlinson, R. C.; Touré, B. B.; Visser, M.; Wang, R.
M.; Waters, N. J.; Shao, W. J. Med. Chem. 2013, 56, 6495.
22. Sharma, M.; Pandey, S.; Chauhan, K.; Sharma, D.; Kumar, B.; Chauhan, P. M. S. J.
Org. Chem. 2012, 77, 929.
23. Desroses, M.; Scobie, M.; Helleday, A. New J. Chem. 2013, 37, 3595.
24. Shaabani, A.; Maleki, A.; Mofakham, H. Synth. Commun. 2008, 38, 3751.
25. Yang, L.; Shi, D.; Chen, S.; Chai, H.; Huang, D.; Zhang, Q.; Li, J. Green Chem. 2012,
14, 945.
26. Parish, H. A.; Gilliom, R. D.; Purcell, W. P.; Browne, R. K.; Spirk, R. F.; White, H.
D. J. Med. Chem. 1982, 25, 98.
27. Haikarainen, T.; Venkannagari, H.; Narwal, M.; Obaji, E.; Lee, H.-W.; Nkizinkiko,
Y.; Lehtiö, L. PLoS One 2013, 8, e65404.
28. Narwal, M.; Fallarero, A.; Vuorela, P.; Lehtiö, L. J. Biomol. Screen. 2012, 17, 593.
29. Venkannagari, H.; Fallarero, A.; Feijs, K. L. H.; Lüscher, B.; Lehtiö, L. Eur. J. Pharm.
Sci. 2013, 49, 148.
Acknowledgements
The work was funded by Biocenter Oulu, Sigrid Jusélius
Foundation, Academy of Finland (Grant No. 266922 for Teemu
Haikarainen and 284605 for the Centre of Excellence). This work
was carried out with the support of the Diamond Light Source
(Didcot, UK) and the European Synchrotron Radiation Facility
(ESRF, Grenoble, France). We are grateful to Local Contacts at
ESRF and at Diamond for providing assistance in using beamlines
ID23-1 and I02, I03, I04-1. The research leading to these results
has received funding from the European Community’s Seventh
Framework Programme (FP7/2007-2013) under BioStruct-X
(Grant agreement N° 283570). The use of the facilities and exper-
tise of the Biocenter Oulu core facility, a member of Biocenter
Finland and Instruct-FI, is gratefully acknowledged.
30. Narwal, M.; Haikarainen, T.; Fallarero, A.; Vuorela, P. M.; Lehtiö, L. J. Med. Chem.
2013, 56, 3507.
Supplementary data
31. Kabsch, W. Acta Crystallogr. D Biol. Crystallogr. 2010, 66, 125.
32. Vagin, A.; Teplyakov, A. Acta Crystallogr. D Biol. Crystallogr. 2010, 66, 22.
33. Dodson, E. J.; Winn, M.; Ralph, A. Meth. Enzymol. 1997, 277, 620.
34. Murshudov, G. N.; Skubák, P.; Lebedev, A. A.; Pannu, N. S.; Steiner, R. A.;
Nicholls, R. A.; Winn, M. D.; Long, F.; Vagin, A. A. Acta Crystallogr. D Biol.
Crystallogr. 2011, 67, 355.
35. Emsley, P.; Cowtan, K. Acta Crystallogr. D Biol. Crystallogr. 2004, 60, 2126.
36. Willert, K.; Brown, J. D.; Danenberg, E.; Duncan, A. W.; Weissman, I. L.; Reya, T.;
Yates, J. R.; Nusse, R. Nature 2003, 423, 448.
37. Veeman, M. T.; Slusarski, D. C.; Kaykas, A.; Louie, S. H.; Moon, R. T. Curr. Biol.
2003, 13, 680.
38. Lau, T.; Chan, E.; Callow, M.; Waaler, J.; Boggs, J.; Blake, R. A.; Magnuson, S.;
Sambrone, A.; Schutten, M.; Firestein, R.; Machon, O.; Korinek, V.; Choo, E.;
Diaz, D.; Merchant, M.; Polakis, P.; Holsworth, D. D.; Krauss, S.; Costa, M. Cancer
Res. 2013, 73, 3132.
Supplementary Table 1 contains the information regarding data
collection and refinement statistics for the crystal structures.
Supplementary Table 2 describes the conditions used for each
ARTD isoenzyme. Supplementary Table 3 lists the catalogue num-
bers for the purchased compounds. Supplementary methods
describe expression and purification of ARTD3 and TNKS2 and
details of the virtual screening protocol. Supplementary data asso-
ciated with this article can be found, in the online version, at
http://dx.doi.org/10.1016/j.bmc.2015.06.063.
References and notes
39. Nathubhai, A.; Wood, P. J.; Lloyd, M. D.; Thompson, A. S.; Threadgill, M. D. ACS
Med. Chem. Lett. 2013, 4, 1173.
1. Hottiger, M. O.; Hassa, P. O.; Lüscher, B.; Schüler, H.; Koch-Nolte, F. Trends
Biochem. Sci. 2010, 35, 208.
2. Lehtiö, L.; Chi, N.-W.; Krauss, S. FEBS J. 2013, 280, 3576.
3. Guettler, S.; LaRose, J.; Petsalaki, E.; Gish, G.; Scotter, A.; Pawson, T.; Rottapel,
R.; Sicheri, F. Cell 2011, 147, 1340.
4. Smith, S.; Giriat, I.; Schmitt, A.; de Lange, T. Science 1998, 282, 1484.
5. Smith, S.; de Lange, T. Curr. Biol. 2000, 10, 1299.
6. Riffell, J. L.; Lord, C. J.; Ashworth, A. Nat. Rev. Drug Discov. 2012, 11, 923.
7. Haikarainen, T.; Krauss, S.; Lehtio, L. Curr. Pharm. Des. 2014.
8. Chang, P.; Coughlin, M.; Mitchison, T. J. Nat. Cell Biol. 2005, 7, 1133.
40. Kumpan, K.; Nathubhai, A.; Zhang, C.; Wood, P. J.; Lloyd, M. D.; Thompson, A. S.;
Haikarainen, T.; Lehtiö, L.; Threadgill, M. D. Bioorg. Med. Chem. 2015, 23, 3013.
41. Shultz, M. D.; Kirby, C. A.; Stams, T.; Chin, D. N.; Blank, J.; Charlat, O.; Cheng, H.;
Cheung, A.; Cong, F.; Feng, Y.; Fortin, P. D.; Hood, T.; Tyagi, V.; Xu, M.; Zhang, B.;
Shao, W. J. Med. Chem. 2012, 55, 1127.
42. Larsson, E. A.; Jansson, A.; Ng, F. M.; Then, S. W.; Panicker, R.; Liu, B.;
Sangthongpitag, K.; Pendharkar, V.; Tai, S. J.; Hill, J.; Dan, C.; Ho, S. Y.; Cheong,
W. W.; Poulsen, A.; Blanchard, S.; Lin, G. R.; Alam, J.; Keller, T. H.; Nordlund, P. J.
Med. Chem. 2013, 56, 4497.