Small Molecule Microarray Based Discovery of PARP14 Inhibitors

Article abstract of DOI:10.1002/anie.201609655

Poly(ADP-ribose) polymerases (PARPs) are key enzymes in a variety of cellular processes. Most small-molecule PARP inhibitors developed to date have been against PARP1, and suffer from poor selectivity. PARP14 has recently emerged as a potential therapeutic target, but its inhibitor development has trailed behind. Herein, we describe a small molecule microarray-based strategy for high-throughput synthesis, screening of >1000 potential bidentate inhibitors of PARPs, and the successful discovery of a potent PARP14 inhibitor H10 with >20-fold selectivity over PARP1. Co-crystallization of the PARP14/H10 complex indicated H10 bound to both the nicotinamide and the adenine subsites. Further structure–activity relationship studies identified important binding elements in the adenine subsite. In tumor cells, H10 was able to chemically knockdown endogenous PARP14 activities.

Full text of DOI:10.1002/anie.201609655

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201609655
German Edition: DOI: 10.1002/ange.201609655
High-Throughput Screening
Small Molecule Microarray Based Discovery of PARP14 Inhibitors
Bo Peng, Ann-Gerd Thorsell, Tobias Karlberg, Herwig Schꢀler,* and Shao Q. Yao*
Abstract: Poly(ADP-ribose) polymerases (PARPs) are key
enzymes in a variety of cellular processes. Most small-molecule
PARP inhibitors developed to date have been against PARP1,
and suffer from poor selectivity. PARP14 has recently emerged
as a potential therapeutic target, but its inhibitor development
has trailed behind. Herein, we describe a small molecule
microarray-based strategy for high-throughput synthesis,
screening of > 1000 potential bidentate inhibitors of PARPs,
and the successful discovery of a potent PARP14 inhibitor H10
with > 20-fold selectivity over PARP1. Co-crystallization of
the PARP14/H10 complex indicated H10 bound to both the
nicotinamide and the adenine subsites. Further structure–
activity relationship studies identified important binding ele-
ments in the adenine subsite. In tumor cells, H10 was able to
chemically knockdown endogenous PARP14 activities.
the successful discovery of a cell-permeable inhibitor, H10
(Figure 1B), that showed relatively potent in vitro inhibitory
activity against PARP14 (IC50 = 490 nm) with good selectivity
over other PARPs ( ꢀ 24-fold over PARP1). A crystal
structure of the PARP14/H10 complex revealed that H10 is
the first bidentate inhibitor against an m-ART, binding to
both the nicotinamide and adenine subsites of PARP14.
Subsequent medicinal chemistry efforts provided valuable
structure–activity relationship (SAR) on H10 analogues
interacting with residues in the adenine-binding site of
PARP14. Lastly, in cellulo studies of selected inhibitors with
tumor cells indicated the compounds conferred moderate
cellular PARP14 knockdown activities.[4,5]
Most known PARP inhibitors are structural mimics of the
nicotinamide moiety in NAD+ (Figure 1A), which inevitably
leads to poor selectivity amongst the PARP family members
due to their highly conserved NAD+-binding pockets.[3,8]
Olaparib, for instance, has been shown to display non-
selective binding against PARP1, 2, 3, and 4.[3] Deviating
from our previous strategy to develop PARP1-selective
inhibitors by targeting the BRCT (BRCA1 C-terminus)
domain, which is present only in PARP1 and PARP4,[9] our
new strategy aimed to explore possible secondary binding
sites in PARP14 and other PARPs, that is, the target-binding
site and/or the adenine-binding subsite located next to the
nicotinamide binding pocket (Figure 1A).[3] We reasoned
that, due to different functions (for example, mono- vs. poly-
ADP-ribosylation) as well as various targets modified by
PARPs, such binding sites (or subsites) might be structurally
distinct. A bidentate inhibitor capable of making contacts
with both the primary nicotinamide-binding site and a secon-
dary binding site, if tethered with a suitable linker, should
possess improved binding affinity and specificity for potential
targeting of individual PARP proteins. In fact, bidentate
inhibitors of kinases and phosphatases, which also possess
multiple binding pockets, are known.[10,11] Recently reported
inhibitors of human tankyrases (TNKS1/2, a member of the
PARP subfamily), also showed binding to both the nicotina-
mide and adenine subsites.[12]
Poly(ADP-ribose) polymerases, or PARPs, are ADP-ribo-
sylating enzymes that comprise at least 18 members in
humans.[1] During ribosylation, the enzymatically activated
PARP uses nicotinamide adenine dinucleotide (NAD+) as
a co-substrate to transfer ADP-ribose to its targets while
releasing nicotinamide as the metabolized product (Fig-
ure 1A). Amongst the PARP family members, most research
has thus far focused on PARP1, an enzyme that is critically
involved in DNA repair and participates in a variety of
cellular functions. Consequently, many potent PARP1 inhib-
itors, including some FDA-approved drugs (such as Olaparib
in Figure 1A), have been successfully developed.[2,3] The
other PARP family members, especially the mono(ADP-
ribosyl) transferases (m-ARTs), have remained poorly char-
acterized, in large part due to a lack of cell-permeable small
molecule inhibitors.[3] Emerging evidence suggests important
cellular functions for some of these enzymes. For instance
PARP14, an m-ART, was found to be associated with the
development of inflammatory diseases and various types of
cancer.[4,5] In recognition of the urgent need to develop small
molecule inhibitors against PARP14 and other less-studied
PARPs,[3,6,7] we report herein a small molecule microarray
(SMM)-based strategy suitable for high-throughput synthesis
and screening of potential bidentate inhibitors of PARPs, and
With the above considerations, we devised a strategy to
design, synthesize, and identify PARP bidentate inhibitors.
Previously, we and others used CuI-catalyzed azide–alkyne
cycloaddition (CuAAC) for the rapid synthesis and screening
of bidentate inhibitors against different enzymes.[13,14] While
this method is a step forward compared to traditional
compound synthesis/screening strategies, it is time-consum-
ing, resource-intensive, and often complicated by the risk of
compound impurities. Small molecule microarrays (SMMs)
are miniaturized assemblies of compounds immobilized
across a 2.5 ꢀ 7.5 cm glass slide, on which thousands of
protein–ligand interactions may be simultaneously mea-
sured.[15,16] Over the years, by improving key aspects related
[*] B. Peng, Prof. Dr. S. Q. Yao
Department of Chemistry
National University of Singapore
3 Science Drive, Singapore 117543 (Singapore)
E-mail: chmyaosq@nus.edu.sg
A.-G. Thorsell, Dr. T. Karlberg, Dr. H. Schꢀler
Karolinska Institute
Department of Medical Biochemistry & Biophysics
Scheeles vꢁg 2, 17177 Stockholm (Sweden)
E-mail: herwig.schuler@ki.se
Supporting information for this article can be found under:
http://dx.doi.org/10.1002/anie.201609655.
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

Angewandte
Communications
Chemie
Figure 1. A) Chemical structure of NAD+ and Olaparib (top). Bottom) Scheme showing a bidentate inhibitor upon binding to both the primary
and secondary pockets PARP14 with a suitable linker. B) Scheme showing our microarray-based discovery of PARP14 inhibitors. Trifunctional
compounds (W1–20) were spotted on TCO slides, followed by on-chip CuAAC with azides (A1–56) to generate the corresponding SMM containing
1120 bidentate PARP inhibitors. Subsequent SMM screening led to identification of a potent PARP14 inhibitor H10. C) Overall workflow of the
current study, including SMM screening, solution-phase resynthesis, and in vitro screening, X-ray studies of H10/PARP14 complex, structure-
based medicinal chemistry, and cell-based hit validation.
to immobilization,[16b] on-chip quantitative binding measure-
ments, and dual-color fluorescence screening,[9,16a,d] we have
successfully used this technology for rapid screening and
identification of small molecule inhibitors against a variety of
targets.[16] Inspired by recent work in which on-chip CuAAC
was used to assemble a carbohydrate-based compound
library,[17] our current study made use of a similar strategy
(Figure 1B,C); by using tetrazine (Tz)-trans-cyclooctene
(TCO) ligation to rapidly print 20 alkyne-containing trifunc-
tional compounds (W1 to W20; Table S1) into each well of
a 12-well glass slide, followed by CuAAC coupling with 56
different azides (A1 to A56; Table S2), an 1120-membered
SMM was rapidly generated with minimal compound con-
sumption. The Tz-TCO ligation was previously shown to
allow rapid and highly efficient compound immobilization in
a microarray.[16b] With the nearly quantitative yield of a typical
CuAAC coupling, we expected most if not all of the 1120
bidentate inhibitors to be successfully assembled on the
microarray. An additional advantage of this strategy was that
purification of immobilized compounds could be done
en masse by simple washes of the resulting microarray to
remove contaminating CuAAC reagents/copper catalysts, as
well as excessive azides, all of which are known to cause
severe interference during screening with compounds synthe-
sized from solution-phase CuAAC methods.[13,14]
tached to amine-containing terminal alkynes (L1 to L4).
These linkers were chosen on the basis of their flexibility,
length, and easy chemical access, so as to better exploit the
altered distances between the primary and secondary binding
pockets of different PARPs. The 56 different azides (A1–56)
were synthesized by using a previously published protocol,[18]
with common aliphatic/aromatic amines containing diverse
functional groups. We also included compounds derived from
known kinase inhibitors (Scheme S1); because many kinase
inhibitors are structural mimics of adenine, our chosen azides,
if properly tethered to the primary pocket-binding anchors
(WH-1 to WH-5), might preferentially fit into the secondary
pocket of different PARPs, resulting in an overall increase of
the bidentate inhibitor binding affinity and selectivity.
The successful immobilization of W1–20 to Tz-function-
alized slides followed by subsequent on-chip CuAAC with
azides was confirmed by spotting a tetrazine-containing dye
(TER-Tz) and coupling a control well on the spotted micro-
array with TER-N3 (Figure 2A). To construct the 1120-
membered SMM, a total of five 12-well spotted microarrays
were used, with each well coupled to one azide. The entire on-
chip CuAAC synthesis was completed in one overnight
reaction including post-SMM compound purification by
washing. With the SMM in hand, preliminary microarray
screenings were first carried out with 10 fluorescently labeled
recombinant PARP catalytic domain proteins (Figures 1C,
S1–S9). In total, > 10000 data points were generated in
a single screening experiment. Upon microarray data analysis,
we obtained high-quality bidentate-binding microarray
images against four of the PARPs (TNKS1, TNKS2,
PARP10, and PARP14). For the other six PARPs (PARP1,
As shown in Figure 1B and Scheme S1, W1–20 were
synthesized from five different ligands known to bind to the
nicotinamide-binding pocket of PARP1 (WH-1 to WH-5),
and they were expected to bind well to other PARP proteins.
The Tz moiety was attached to these ligands through four
different linkers derived from aspartic/glutamic acids at-
2
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
These are not the final page numbers!

Angewandte
Communications
Chemie
Figure 2. A) Microarray image of the 440-membered SMM screened with fluorescently labeled PARP14. All of the compounds were spotted in
duplicate. Right) Images of spots showing successful immobilization of TER-Tz on TCO slides and on-chip CuAAC with TER-N3. B) Histogram
showing the relative fluorescence (RFU) of each compound from the SMM in (A). Red: Representative hits. See decoding table (SI_II) for IDs and
structures. C) Dual-color results of representative hits simultaneously screened with PARP14 (green) and another PARP (red; PARP10, TNKS1, or
TNKS2). D) Microarray images of spots from concentration-dependent SMM and a representative Kd curve. E) Data summarizing IC50 values
obtained from in vitro PARP assay against 5 resynthesized hits. F) Structures of two PARP14 inhibitors, H5 and AD3. G) SAR of the adenine-
binding site summarized from X-ray and medicinal chemistry studies.
2, 3, 4, 12, and 13), the S/N ratios of the fluorescent spots were
on average 5- to 10-fold lower, and were deemed unreliable
for subsequent screening/compound identification (Fig-
ure S3). They were therefore not pursued further. The next
several rounds of microarray screening involved: 1) rescreen-
ing of the 1120-membered SMM with the 4 selected PARPs
(Figures S3 and S4); 2) second-round screening with a 440-
membered focus microarray spotted with compounds deemed
as “positive binders” from the previous round (Figure S5);
and 3) third-round screening of selected compounds with
dual-color and concentration-dependent experiments (Figur-
es S6–8). Dual-color microarray screening, in which spotted
compounds were screened on the same slide with two
different fluorescently labeled protein targets (Cy3-labeled
PARP14 and a Cy5-labeled control), enabled rapid identifi-
cation of compounds having highly selective binding of one
protein over another.[16a] Concentration-dependent micro-
array experiments allowed all spotted compounds to be
simultaneously screened in a single platform with different
concentrations of a protein, thus providing semi-quantitative
binding data (apparent Kd) for rapid and accurate identifica-
tion of high-affinity ligands.[16d] A representative second-
round microarray image obtained from the 440-membered
focus SMM screened against Cy3-labeled PARP14 and its
corresponding histogram showing the relative fluorescence
intensity (RFU) of individual spots are shown in Figure 2A
and B, respectively. In Figure 2B, representative high-fluo-
rescence spots were colored and numbered (according to the
SI II decoding table). Microarray images of the correspond-
ing compounds rescreened in the third round are shown in
Figure 2C and D. Because we were particularly interested in
the discovery of potential PARP14 inhibitors, the dual-color
experiments were carried out with a 1:1 mixture of Cy3-
labeled PARP14 and another Cy5-labeled PARP protein
(PARP10, TNKS1, and TNKS2, respectively). Merged images
of the two resulting fluorescence channels (colored in green
and red, respectively, in Figure 2C) allowed immediate visual
identification of PARP14-selective ligands. For example,
spots derived from compound 587 appeared mostly green
across the various dual-color images, indicating it was
a putative PARP14-selective binder. Similarly, in concentra-
tion-dependent experiments (Figure 2D), the same com-
pound showed PARP14 dose-dependent binding profiles
with an apparent on-chip Kd of 76.96 nm, indicating it was
a high-affinity PARP14 ligand. In total, 20 putative PARP14
ligands were identified from three rounds of microarray
screenings (Table S3). In addition to compound 587, results
from four other compounds (166, 446, 588, 806) are shown in
Figure 2B/C/D. All 20 of the putative hits possessed either
high-binding affinity or selectivity toward PARP14, or both.
Interestingly, all contained the same 3-amino-benzamide
moiety (WH2), which was derived from a previously reported
but relatively weak PARP1 inhibitor.[19]
Our next-step screening efforts involved in vitro PARP
enzymatic assays to evaluate these 20 hits against several
PARP proteins (PARP1, PARP14, and TNKS1; Figures 2E,
S11, and S12). Because the tetrazine immobilization unit was
no longer required, the 20 hits derived from tetrazine-free
W1–W20 (renamed as H1 to H20; Table S3), containing
either a succinic or glutaric acid linker (to minimize molecular
weight and maximize flexibility of the linker), were resynthe-
sized. Structures of two examples, H10 and H5 (correspond-
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

Angewandte
Communications
Chemie
ing to compounds 587 and 166 in the microarray, respectively)
are shown in Figures 1B and 2F. H10 was found to inhibit
PARP14 activity with an IC50 value of 0.49 Æ 0.07 mm in vitro,
which represents the most potent PARP14 inhibitor reported
to date.[6,7,20] Equally important, it also possessed relatively
good selectivity over two other PARPs tested (24- and 18-fold
over PARP1 and TNKS1, respectively). H5, the next-best
PARP14 inhibitor identified (IC50 = 0.76 Æ 0.06 mm), showed
a comparatively poorer selectivity over PARP1 ( ꢀ 6-fold).
To determine the structural basis of PARP14 binding to
H10/H5, we next obtained X-ray structures to resolutions of
2.17 ꢁ (for H10) and 2.07 ꢁ (for H5), respectively. As shown
in Figure 3, the PARP14/H10 complex clearly showed the 3-
whether specific interactions might be introduced in the
linker region of H10 to improve its PARP14-binding affinity.
Medicinal chemistry efforts were thus carried out (Figure 2G,
Scheme S3, and Table S4). By first substituting the triazole-
containing linker in H10 with more flexible linkers containing
piperazine, ethylene glycol, or lysine, we found most of the
resulting inhibitors, except the lysine-containing H10 ana-
logue (18 in Scheme S3), had attenuated PARP14 inhibition.
This inhibitor was thus chosen for further SAR studies. We
focused on the adenine-binding subsite in which numerous
interactions between the 3-sulfonamide benzoic acid of H10
and PARP14 were observed (Figure 3). The NH in SO2NH
formed a hydrogen bond with Asn1698 of the protein, while
its O formed another hydrogen bond with His1691. In
addition, the benzoic acid group also formed several favor-
able interactions with the protein.
A
22-membered
arylsulfonamide library based on H10 was thus synthesized
(named AD1 to AD22; Table S4) and tested for PARP14
inhibition (Figure S11C) with results summarized in Fig-
ure 2G. Most of the compounds except AD3 (Figure 2F)
showed attenuated inhibition. Nevertheless, SAR analysis
indicated, as predicted from the crystal structure, that while
large substituents in C1/C3 of the arylsulfonamide were not
tolerated, hydrogen bond-accepting C2 substituents might
improve ligand binding.
Finally, we evaluated whether H10 and AD3 could inhibit
endogenous PARP14 activities from tumor cells. Both com-
pounds showed good cell permeability (Table S6) and hydro-
lytic stability (Figure S15).[16c] Because the majority of poly(-
ADP-ribosyl)ation (PAR) activity in mammalian cells is from
PARP1, we first assessed whether our newly discovered
PARP14 inhibitors cross-inhibited endogenous PARP1 activ-
ity. We treated HepG2 cells (a liver cancer cell line) with H2O2
to induce DNA damage and oxidative stress leading to PAR
polymer formation (Figure 4A).[9] Addition of PARP1 inhib-
itors such as DPQ or Olaparib, but not H10 or AD3 (even at
25 mm), caused apparent inhibition of PAR formation, indi-
cating neither H10 nor AD3 inhibited endogenous PARP1
activities. It was previously reported that PARP14 promotes
tumor survival in multiple myeloma and hepatocellular
carcinoma by blocking the c-Jun N-terminal kinase
1 (JNK1)-mediated cell apoptosis.[4,5] Genetic knockdown of
endogenous PARP14 expression could thus lead to cell death.
As shown in Figure 4B, we first confirmed siRNA knockdown
of PARP14 in HepG2 cells indeed caused activation of JNK1
phosphorylation and apparent formation of cleaved PARP1
(a hallmark of apoptosis). Treatment of the cells by H10 or
AD3 caused similar effects at as low as 10 mm, indicating that
both compounds were able to chemically knockdown endog-
enous PARP14 activities effectively. Subsequent caspase-3/7-
mediated cell apoptosis was confirmed by FACS, in vitro
caspase-3 enzymatic assay, and Western blotting analysis
(Figures 4C and S17). Similar apoptosis was also detected in
compound-treated RPMI-8226 cells (a myeloma cell line).
Finally, we carried out combination drug treatment experi-
ments by treating the cells with H10/AD3 and another well-
known drug (doxorubicin (DOX) for HepG2 and dexame-
thasone (DEX) for RPMI-8226 cells, respectively). Both
DOX and DEX are anticancer agents. As shown in Fig-
Figure 3. Close-up view of the H10-binding site in the X-ray structure
of PARP14/H10 complex (PDB entry 5LYH). H10 and PARP14 are
shown in ball and sticks, and ribbons, respectively.
amino-benzamide moiety from H10 resided in the nicotina-
mide (NA)-binding pocket as expected, and the rest of the
molecule extended toward the adenine-binding site in a bind-
ing mode that was not observed previously.[3] The sulfonamide
and benzoic acid moiety at the distal end were found in the
adenine base-binding site of PARP14. Both the linker and
distal end made a number of polar interactions (notably, with
PARP14 side chains His1682, Thr1684, Ser1688, His1691, and
the peptide backbone of Asn1698) that could explain the
slightly improved potency and the significantly better selec-
tivity of the compound. Extending the central linker in H10
by one carbon (giving H15; Table S3) decreased both potency
and selectivity. Surprisingly, the second crystal structure
shows that H5 bound to PARP14 in a different fashion
(Figure S13); its distal phenyl group made hydrophobic
interactions with the Val1773 and Val1784 side chains in
a site around the small hydrophobic residue that characterizes
the mono-ADP-ribosyltransferases of the PARP family,
Leu1782.[3] This was an interesting observation because
binding at the subsite near the A-loop might provide
a means to accomplish truly selective inhibition even within
the mono-ADP-ribosyltransferases.[21] Notwithstanding, our
PARP14/H10 structure clearly showed that, by using the
bidentate strategy, selective PARP14 inhibitors could be
generated by engaging a 3-sulfonamide benzoic acid in the
adenine binding site.
Owing to incomplete density of the flexible D-loop in our
PARP14/H10 structure, this segment could not be completely
modeled. We wondered, by engaging the unresolved D-loop,
4
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
These are not the final page numbers!

Angewandte
Communications
Chemie
Figure 4. A) H2O2-induced PAR formation (1 mm, 1 h) of HepG2 cells were inhibited by PARP1 inhibitors (10 and 1 mm for DPQ and Olaparib, respectively),
but not H10/AD3 (25 mm). B) Effects of PARP14 knockdown (siPARP14, 10–100 mm of H10 or AD3) in HepG2 include promotion of PARP1 cleavage and
JNK1 phosphorylation. C) Caspase-3-activated apoptosis of HepG2 cells was observed as a result of PARP14-knockdown in (B), determined by in vitro
caspase-3 enzymatic assay (black bars) and FACS (orange bars). A positive control was done with cells treated with staurosporine (STS, 200 nm). The
concentration of H10/AD3 was 100 mm. Inset: WB analysis with anti-cleaved caspase-3 Ab. D) XTT cell viability results showing effects of combination drug
treatments with doxorubicin (DOX)-treated HepG2 and dexamethasone (DEX)-treated RPMI-8226 cells in the presence of H10/AD3 (25 mm). The Dox/DEX
concentration was 10 mm.
collection and the beamline staff at DIAMOND (Didcot,
UK) and PETRA (Hamburg, Germany) for assistance.
ure 4D, both of the PARP14 inhibitors showed enhanced cell-
killing properties in the presence of DOX or DEX, respec-
tively, similar to what was observed from the corresponding
siRNA experiments, as well as previous reports.[4,5] All of
these lines of evidence thus confirmed H10 and AD3 are the
first-ever in situ-active PARP14 inhibitors capable of block-
ing endogenous PARP14 activities in both tumor cell lines.
In conclusion, by using on-chip CuAAC, we have
developed a small molecule microarray capable of rapid
synthesis and screening of potential PARP inhibitors. The
successful discovery of H10, a potent and selective PARP14
inhibitor that showed moderate inhibitory properties against
endogenous PARP14 in different tumor cells, indicates this
microarray-based strategy might be amenable for the future
discovery of selective inhibitors against other PARP proteins.
The X-ray structure of PARP14/H10 also unequivocally
established the bidentate binding mode of H10 for the first
time.
Conflict of interest
The authors declare no conflict of interest.
Keywords: click chemistry · high-throughput screening ·
microarrays · PARP inhibitors · X-ray crystallography
[1] B. A. Gibson, W. L. Kraus, Nat. Rev. Mol. Cell Biol. 2012, 13,
411 – 424.
[2] L. Virag, C. Szabo, Pharmacol. Rev. 2002, 54, 375 – 429.
[3] E. Wahlberg, T. Karlberg, E. Kouznetsova, N. Markova, A.
Macchiarulo, A. G. Thorsell, E. Pol, A. Frostell, T. Ekblad, D.
Oncu, B. Kull, G. M. Robertson, R. Pellicciari, H. Schuler, J.
Weigelt, Nat. Biotechnol. 2012, 30, 283 – 288.
[4] A. Barbarulo, V. Iansante, A. Chaidos, K. Naresh, A. Rahem-
tulla, G. Franzoso, A. Karadimitris, D. O. Haskard, S. Papa, C.
Bubici, Oncogene 2013, 32, 4231 – 4242.
Coordinates and structure factors have been deposited in
the Protein Data Bank (PARP14-H5, PDB entry 5LXP;
PARP14-H10, PDB entry 5LYH).
[5] V. Iansante, P. M. Choy, S. W. Fung, Y. Liu, J. G. Chai, J. Dyson,
A. Del Rio, C. DꢂSantos, R. Williams, S. Chokshi, R. A. Anders,
C. Bubici, S. Papa, Nat. Commun. 2015, 6, 7882.
[6] C. D. Andersson, T. Karlberg, T. Ekblad, A. E. Lindgren, A. G.
Thorsell, S. Spjut, U. Uciechowska, M. S. Niemiec, P. Wittung-
Stafshede, J. Weigelt, M. Elofsson, H. Schuler, A. Linusson, J.
Med. Chem. 2012, 55, 7706 – 7718.
[7] T. Ekblad, A. E. Lindgren, C. D. Andersson, R. Caraballo, A. G.
Thorsell, T. Karlberg, S. Spjut, A. Linusson, H. Schuler, M.
Elofsson, Eur. J. Med. Chem. 2015, 95, 546 – 551.
[8] D. V. Ferraris, J. Med. Chem. 2010, 53, 4561 – 4584.
[9] Z. Na, B. Peng, S. Ng, S. Pan, J. S. Lee, H. M. Shen, S. Q. Yao,
Angew. Chem. Int. Ed. 2015, 54, 2515 – 2519; Angew. Chem.
2015, 127, 2545 – 2549.
Acknowledgments
Funding was provided by the National Medical Research
Council (CBRG/0038/2013), Ministry of Education
(MOE2013-T2-1–048) of Singapore, the IngaBritt & Arne
Lundberg Foundation, the Swedish Cancer Society, the
Swedish Research Council, and the European Communityꢂs
Seventh Framework Programme (FP7/2007–2013) under
BioStruct-X (grant agreement No. 283570). We thank the
Protein Science Facility at Karolinska Institute for technical
support, Joseph Brock for help with synchrotron data
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!

Angewandte
Communications
Chemie
[10] C. M. Gower, M. E. K. Chang, D. J. Maly, Crit. Rev. Biochem.
Mol. Biol. 2014, 49, 102 – 115.
W. L. Lee, J. Wang, S. Q. Yao, J. Am. Chem. Soc. 2007, 129,
13110 – 13117.
[11] R. Srinivasan, M. Uttamchandani, S. Q. Yao, Org. Lett. 2006, 8,
713 – 716.
[12] a) M. D. Shultz, et al., J. Med. Chem. 2012, 55, 1127 – 1136; b) T.
Haikarainen, M. Narwal, P. Joensuu, L. Lehtio, ACS. Med.
Chem. Lett. 2014, 5, 18 – 22; c) J. W. Johannes, et al., ACS Med.
Chem. Lett. 2015, 6, 254 – 259.
[17] C. D. Rillahan, E. Schwartz, C. Rademacher, R. McBride, J.
Rangarajan, V. V. Fokin, J. C. Paulson, ACS Chem. Biol. 2013, 8,
1417 – 1422.
[18] a) R. Srinivasan, L. P. Tan, H. Wu, P.-Y. Yang, K. A. Kalesh, S. Q.
Yao, Org. Biol. Chem. 2009, 7, 1821 – 1828; b) Z. Li, P. Hao, L. Li,
C. Y. Tan, X. Cheng, G. Y. Chen, S. K. Sze, H. M. Shen, S. Q.
Yao, Angew. Chem. Int. Ed. 2013, 52, 8551 – 8556; Angew. Chem.
2013, 125, 8713 – 8718.
[13] R. Srinivasan, J. Li, S. L. Ng, K. A. Kalesh, S. Q. Yao, Nat.
Protoc. 2007, 2, 2655 – 2664.
[14] X. S. Wang, B. S. Huang, X. Y. Liu, P. Zhan, Drug Discovery
Today 2016, 21, 118 – 132.
[15] J. A. Hong, D. V. Neel, D. Wassaf, F. Caballero, A. N. Koehler,
Curr. Opin. Chem. Biol. 2014, 18, 21 – 28.
[16] a) Z. Na, S. Pan, M. Uttamchandani, S. Q. Yao, Angew. Chem.
Int. Ed. 2014, 53, 8421 – 8426; Angew. Chem. 2014, 126, 8561 –
8566; b) C. J. Zhang, C. Y. Tan, J. Ge, Z. Na, G. Y. Chen, M.
Uttamchandani, H. Sun, S. Q. Yao, Angew. Chem. Int. Ed. 2013,
52, 14060 – 14064; Angew. Chem. 2013, 125, 14310 – 14314; c) H.
Wu, J. Ge, P.-Y. Yang, J. Wang, M. Uttamchandani, S. Q. Yao, J.
Am. Chem. Soc. 2011, 133, 1946 – 1954; d) M. Uttamchandani,
[19] M. R. Purnell, W. J. Whish, Biochem. J. 1980, 185, 775 – 777.
[20] P. Y. Wang, J. Li, X. Jiang, Z. Q. Liu, N. Ye, Y. J. Xu, G. F. Yang,
Y. C. Xu, A. Zhang, Tetrahedron 2014, 70, 5666 – 5673.
[21] H. Venkannagari, P. Verheugd, J. Koivunen, T. Haikarainen, E.
Obaji, Y. Ashok, M. Narwal, T. Pihlajaniemi, B. Lꢃscher, L.
Lehtiç, Cell Chem. Biol. 2016, 23, 1251 – 1260.
Manuscript received: October 2, 2016
Revised: November 22, 2016
Final Article published: && &&, &&&&
6
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
These are not the final page numbers!

Angewandte
Communications
Chemie
Communications
High-Throughput Screening
B. Peng, A.-G. Thorsell, T. Karlberg,
H. Schꢁler,* S. Q. Yao*
&&&& — &&&&
Small Molecule Microarray Based
Discovery of PARP14 Inhibitors
“Spotting” the Hit: A small molecule
microarray was developed for miniatur-
ized on-chip synthesis and screening of
>1000 potential bidentate inhibitors
against poly(ADP-ribose) polymerases
(PARPs), which led to the successful
discovery of a potent PARP14 bidentate
inhibitor (IC50 =0.49 mm) with >20-fold
selectivity over PARP1.
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7
These are not the final page numbers!