Structural basis and SAR for G007-LK, a lead stage 1,2,4-triazole based specific tankyrase 1/2 inhibitor

Article abstract of DOI:10.1021/jm4000566

Tankyrases 1 and 2 (TNKS1/2) are promising pharmacological biotargets with possible applications for the development of novel anticancer therapeutics. A focused structure-activity relationship study was conducted based on the tankyrase inhibitor JW74 (1). Chemical analoging of 1 improved the 1,2,4-triazole based core and led to 4-{5-[(E)-2-{4-(2-chlorophenyl)-5-[5- (methylsulfonyl)pyridin-2-yl]-4H-1,2,4-triazol-3-yl}ethenyl]-1,3, 4-oxadiazol-2-yl}benzonitrile (G007-LK), a potent, "rule of 5" compliant and a metabolically stable TNKS1/2 inhibitor. G007-LK (66) displayed high selectivity toward tankyrases 1 and 2 with biochemical IC₅₀ values of 46 nM and 25 nM, respectively, and a cellular IC₅₀ value of 50 nM combined with an excellent pharmacokinetic profile in mice. The PARP domain of TNKS2 was cocrystallized with 66, and the X-ray structure was determined at 2.8 ? resolution in the space group P3221. The structure revealed that 66 binds to unique structural features in the extended adenosine binding pocket which forms the structural basis for the compound's high target selectivity and specificity. Our study provides a significantly optimized compound for targeting TNKS1/2 in vitro and in vivo.

Full text of DOI:10.1021/jm4000566

Subscriber access provided by EASTERN KENTUCKY UNIVERSITY
Article
Structural basis and SAR for G007-LK, a lead stage
1,2,4-triazole based specific tankyrase 1/2 inhibitor
Andrey Voronkov, Daniel D Holsworth, Jo Waaler, Steven Ray Wilson, Bie Ekblad, Harmonie
Perdreau-Dahl, Huyen Dinh, Gerard Drewes, Carsten Hopf, Jens Preben Morth, and Stefan Krauss
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/jm4000566 • Publication Date (Web): 11 Mar 2013
Downloaded from http://pubs.acs.org on March 13, 2013
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
Structural basis and SAR for G007-LK, a lead stage 1,2,4-
triazole based specific tankyrase 1/2 inhibitor
7
8
9
1,
1
1,
†,
2
1
†
†
Andrew Voronkov , Daniel D. Holsworth , Jo Waaler *, Steven R. Wilson , Bie Ekblad ,
Harmonie PerdreauꢀDahl3, Huyen Dinh1, Gerard Drewes4, Carsten Hopf4, Jens P. Morth3,5*,
Stefan Krauss1, 3 *.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Authors Affiliations:
1 SFI CAST Biomedical Innovation Center, Unit for Cell Signaling, Oslo University Hospital, Forskningsparken,
Gaustadalleén 21, 0349, Oslo, Norway, 2 Department of Chemistry, University of Oslo, P.O. Box 1033 Blindern, Nꢀ
0315 Oslo, Norway,3 Centre for Molecular Medicine Norway, Nordic EMBL Partnership, University of Oslo and
Oslo University Hospital, Nꢀ0318 Oslo, Norway, 4 Cellzome AG, Meyerhofstrasse 1, 69117 Heidelberg, Germany,
5Institute for Experimental Medical Research, Oslo University Hospital, Nꢀ0424 Oslo, Norway, †These authors
contributed equally.
Financial Support: J.W., A.V. and B.E. and S.K. were supported by the Research Council of Norway, CRI
program and Helse Sør Øst, grant 2010031.
Corresponding Authors: Jo Waaler*, Jens Preben Morth* and Stefan Krauss* (jo.waaler@rrꢀresearch.no;
j.p.morth@ncmm.uio.no; Stefan.krauss@rrꢀresearch.no)
Disclosure of Potential Conflicts of Interest: The described chemical compounds are patented and may have
commercial value.
Running Title: Tankyrase inhibitor, G007ꢀLK
Key Words: Wnt/βꢀcatenin signaling, smallꢀmolecule, inhibitor, cancer, tankyrase (TNKS).
Abbreviations: TRF1, telomeric repeat factor 1; TNKS1/2, tankyrase 1 and tankyrase 2, TRF1ꢀinteracting
ankyrinꢀrelated ADPꢀribose polymerase 1 and 2; PARP, poly(ADPꢀribosyl)ating polymerases; GLUT4, glucose
transporter type 4; IRAP, insulin responsive aminopeptidase; AXIN, Axis Inhibition Protein; KIF3a,
kinesin family member 3A; NuMA, nuclear mitotic apparatus protein; IRAP, interleukin 1 receptor antagonist; Miki,
mitotic kinetics regulator; CPAP, centrosomal P4.1ꢀassociated protein; 3BP2, cꢀAbl Src homology 3 domainꢀ
binding proteinꢀ2; RNF146, ring finger protein 146, SAR, structureꢀactivity relationship; ST, SuperTOP; Luc,
luciferase; SuperTOPFlash plasmid, STꢀLuc; GMDS, GDPꢀmannoseꢀ4,6ꢀdehydratase; TRAIL, TNFꢀrelated
apoptosisꢀinducing ligand; CD95, cluster of differentiation 95; ALDOA, aldolase A; SSSCA1, Sjögren
syndrome/scleroderma autoantigen 1; GPCR, Gꢀprotein coupled receptors; CYP, Cytochrome P450; CYP3A4,
Cytochrome P450 3A4; SOM, site of metabolism; SCS, site of metabolism consensus score; MRCS, metabolism
rate consensus score; HLM, human liver microsomes; NCS, nonꢀcrystallographic symmetry.
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 43
1
2
3
4
ABSTRACT
5
6
7
8
Tankyrases 1 and 2 (TNKS1/2) are promising pharmacological biotargets with possible
applications for the development of novel antiꢀcancer therapeutics. A focused structureꢀactivity
relationship study was conducted based on the tankyrase inhibitor JW74 (1). Chemical analoging
of 1 improved the 1,2,4ꢀtriazole based core and led to 4ꢀ{5ꢀ[(E)ꢀ2ꢀ{4ꢀ(2ꢀchlorophenyl)ꢀ5ꢀ[5ꢀ
(methylsulfonyl)pyridinꢀ2ꢀyl]ꢀ4Hꢀ1,2,4ꢀtriazolꢀ3ꢀyl}ethenyl]ꢀ1,3,4ꢀoxadiazolꢀ2ꢀyl}benzonitrile
(G007ꢀLK), a potent, “rule of 5” compliant and a metabolically stable TNKS1/2 inhibitor. G007ꢀ
LK (66) displayed high selectivity toward tankyrases 1 and 2 with biochemical IC50ꢀvalues of 46
nM and 25 nM, respectively and a cellular IC50ꢀvalue of 50 nM combined with an excellent
pharmacokinetic profile in mice. The PARP domain of TNKS2 was coꢀcrystallized with 66 and
the Xꢀray structure was determined at a 2.8 Å resolution in the space group P3221. The structure
revealed that 66 binds to unique structural features in the extended adenosine binding pocket
which forms the structural basis for the compounds high target selectivity and specificity. Our
study provides a significantly optimized compound for targeting TNKS1/2 in vitro and in vivo.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
ACS Paragon Plus Environment

Page 3 of 43
Journal of Medicinal Chemistry
1
2
3
4
INTRODUCTION
5
6
7
8
Adenosine diphosphate (ADP) ribosylation is a widely used catalytic process to modify
proteins whereby nicotinamide adenine dinucleotide (NAD+) is used as a substrate to form postꢀ
translational protein modifications. ADP ribosylating proteins are subdivided into mono(ADPꢀ
ribosyl)ating (ARP) and (oligoꢀ) poly(ADPꢀribosyl)ating (PARP) proteins.1 The telomeric repeat
factor 1 (TRF1)ꢀinteracting ankyrinꢀrelated ADPꢀribose polymerase 1 (tankyrase 1,TNKS1,
PARP5a, ARTD5; 1327 residues) and tankyrase 2 (TNKS2, PARP5b, ARTD6; 1166 residues)
(TNKS1/2) belong to the subgroup of poly(ADPꢀribosyl)ating polymerases. This subgroup is
classified by a sterile alpha motif (SAM) domain and stretches of repetitive ankyrin domains in
the Nꢀterminus that are involved in multimerization. TNKS1/2 have been shown to poly(ADPꢀ
ribosyl)ate a number of substrate proteins by recognizing linear peptide motifs consisting of six
to eight consecutive amino acids with high degeneracy.2ꢀ6 The unique ability of tankyrase to
form multimers while poly(ADPꢀribosyl)ating substrate proteins led to the suggestion that
tankyrases may regulate the assembly and disassembly of large polymerized structures.7
Tankyrases have multiple cellular functions: i) They are involved in telomere
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
maintenance by poly(ADPꢀribosyl)ating TRF1 and releasing TRF1 from telomeres.8, 9 ii)
Tankyrases are implied in vesicle transport modulating the subꢀcellular distribution of glucose
transporter type 4 (GLUT4) vesicles through binding to the insulin responsive aminopeptidase
Interleukin 1 receptor antagonist,(IRAP),10 and it has been shown that tankyrase, AXIN (axis
inhibition protein) and KIF3a (kinesin family member 3A) form a tertiary complex that is crucial
for GLUT4 subꢀcellular localization.11 iii) A role for tankyrase in spindle pole assembly through
interactions with NuMA (nuclear mitotic apparatus protein) 12 and Miki (mitotic kinetics
regulator)13 . In addition, tankyrases are implied in procentriole formation through poly(ADPꢀ
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 43
1
2
ribosyl) ation of CPAP (centrosomal P4.1ꢀassociated protein).14 iv) A connection of TNKS1/2 to
Cherubism has been established through its interaction with the 3BP2 adaptor protein (cꢀAbl Src
homology 3 domainꢀbinding proteinꢀ2).2, 15 v) Recently, an involvement of tankyrase in
attenuating Wnt/βꢀcatenin signaling has been demonstrated.16, 17 In this process, tankyrase
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
poly(ADPꢀribosyl)ates AXIN, the rate limiting structural protein in the βꢀcatenin destruction
complex. Poly(ADPꢀribosyl)ation of AXIN triggers its ubiquitination initiated by the RNF146
(ring finger protein 146) ubiquitin E3 ligase followed by degradation in the proteasome.18, 19
Tankyrases have received attention in chemical biology as promising targets in oncology.
Several small molecules have been identified that inhibit tankyrases 1 and 2, and attenuate
Wnt/βꢀcatenin signaling in reporter cell lines and in Wnt dependent cancer cell lines.16, 17, 20, 21
Present tankyrase inhibitors can be classified into two groups: (i) Compounds that bind to the
nicotinamide pocket of the PARP domain, such as 2ꢀ(4ꢀ(trifluoromethyl)phenyl)ꢀ7,8ꢀdihydroꢀ
5Hꢀthiopyrano[4,3ꢀd]pyrimidinꢀ4ꢀol (XAV939)16, 22 and Nꢀ(6ꢀOxoꢀ5,6ꢀdihydrophenanthridinꢀ2ꢀ
yl)ꢀ(N,Nꢀdimethylamino)acetamide (PJ34)22 as well as many generic PARP inhibitors.23, 24 (ii)
Compounds that occupy the adjacent adenosine binding pocket including JW74 (1)25, 26 (Figure
2a) 4ꢀ((3aR,4S,7R,7aS)ꢀ1,3ꢀdioxoꢀ1,3,3a,4,7,7aꢀhexahydroꢀ2Hꢀ4,7ꢀmethanoisoindolꢀ2ꢀyl)ꢀNꢀ
(quinolinꢀ8ꢀyl)benzamide (IWRꢀ1)27, 28, 4ꢀ((3aR,4S,7R,7aS)ꢀ1,3ꢀdioxoꢀ1,3,3a,4,7,7aꢀhexahydroꢀ
2Hꢀ4,7ꢀmethanoisoindolꢀ2ꢀyl)ꢀNꢀ(4ꢀmethylquinolinꢀ8ꢀyl)benzamide (IWRꢀ2)23, 27 and Nꢀ(4ꢀ(((4ꢀ
(4ꢀmethoxyphenyl)tetrahydroꢀ2Hꢀpyranꢀ4ꢀyl)methyl)carbamoyl)phenyl)furanꢀ2ꢀcarboxamide
(JW55).17 For clinical relevance, however, there is a need for tankyrase inhibitors with
significantly improved selectivity and pharmacokinetic properties compared to existing
structures. Here we describe the development of a selective and efficacious compound 1
derivative, compound G007ꢀLK (66)29 (Figure 2a), which has been stabilized against phase I
4
ACS Paragon Plus Environment

Page 5 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
metabolism and shows an excellent pharmacokinetic profile in mice. 66 binds specifically to an
extended flexible adenosine pocket of the PARP domain of TNKS1/2. We describe the
molecular basis for the specific binding of the drug and describe how this interaction is the
premise for the exceptionally high target selectivity observed for 66.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
RESULTS
Affinity-capture identifies TNKS1/2 as the targets of the 1,2,4-triazole based
chemotype
To determine the protein target and specificity of the previously described 1,2,4ꢀtriazole
based chemotype,26 (see structureꢀactivity relationship (SAR) paragraph), a linkable analog of
compound 139 (22)25(Supplementary Figure 1c), was synthesized (compound 161 (24)25, Figure
1a). The activity of 24 was confirmed by a cellꢀbased STꢀLuc reporter activity assay in HEK293
cells (IC50 [STꢀLuc] = 3.92 ꢁM, Figure 1a). Subsequently, 24 was immobilized on Sepharose
beads. The 24 probe matrix was exposed to HEK293 cell extracts that had either been preꢀ
incubated (i) with an active Wnt/βꢀcatenin pathway inhibitor (66) that competes with the
immobilized probes for target protein binding, (ii) with an inactive analog (compound 198, 25)25
or (iii) with a vehicle control (Figure 1a). The reduction in protein capture that resulted from the
inhibitor treatment was quantified by isobaric tagging of tryptic peptides and tandem mass
spectrometry analysis (MS/MS) of the combined peptide pools.30 For each identified protein, the
decrease of the reporter ion signals relative to the vehicle control reflected the competitive
binding of the unbound inhibitor to its target. Unbound compound 66, but not its inactive analog
25, competed with binding of a defined set of proteins to the 24 matrix (Figure 1b). The
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 43
1
2
3
4
5
6
7
8
9
identified proteins were (i) tankyrases 1 and 2 (TNKS1/2), (ii) GMDS (GDPꢀmannoseꢀ4,6ꢀ
dehydratase), a recently identified tankyrase binding protein 31 that renders colon cancer cells
resistant to TRAIL (TNFꢀrelated apoptosisꢀinducing ligand) receptorꢀ and CD95 (cluster of
differentiation 95)ꢀmediated apoptosis,32 (iii) ALDOA (aldolase A), and (iv) the poorly
characterized SSSCA1 (Sjögren syndrome/scleroderma autoantigen 1).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To further test whether 66 directly interfered with the catalytic activity of TNKS1/2 we
used in vitro biochemical assays based on an NAD+ꢀdependent autoꢀpolyꢀ(ADP ribosy)lation of
TNKS1/2 or PARP1 activity. 66 decreased autoꢀpolyꢀ(ADP ribosy)lation of TNKS1 and TNKS2
with IC50ꢀvalues of 46 nM and 25 nM, respectively (Supplementary Figure 4). In contrast to a
benchmark compound XAV939 (2),16 66 did not show biochemical inhibition of PARP1 at doses
up to 20 ꢁM (Supplementary Figure 4). Further proof for high target specificity was established
by testing 66 against a focused panel of kinases, phosphatases and Gꢀprotein coupled receptors
(GPCR). 66 showed no substantial inhibition of the 90 kinases, 16 phosphatases and 73 GPCRs
in the experimental setup conditions used (Supplementary Figure 5 and 6).
An important biological impact of specific inhibition of the TNKS1/2 catalytic PARP
domain is an attenuation of Wnt/βꢀcatenin signaling. By preventing polyꢀ(ADP ribosy)lation of
the tankyrase targets TNKS1/2, RNF146 and AXIN,16, 17 Wnt/βꢀcatenin signaling is, in a context
dependent manner, decreased.16, 17
Structure-activity relationships establishes 66 as a phase I metabolism stabilized
1,2,4-triazole based analogue
The primary hit 126 showed a moderate inhibition of tankyrase and Wnt/βꢀcatenin
signaling (IC50 [TNKS2] = 0.46 ꢁM; IC50 [STꢀLuc] = 1.01 ꢁM) and a low phase I metabolic
6
ACS Paragon Plus Environment

Page 7 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
stability HLM (t½ = 3 min) (Figure 2a). During SAR development, modifications were based on
Hꢀbonding acceptor/donor potential, hydrophilic, hydrophobic and steric interactions, synthesis
feasibility, impact on cLogP (<5), MW (<550), and microsomal stability assessment. In
particular, compound design was guided by cLogPꢀvalues, along with in silico CYP metabolism
consensus scores (Supplementary Figure 4c). “Rule of 5” (RO5) parameters served as a general
guide for analog design. The LogP calculations were performed by using the ALOGP2.1
program, whereby an average value of several methods was applied.33, 34 Sites of metabolism
(SOM) of the proposed compounds were evaluated by consensus scoring on the basis of five
software packages: MetaPrint2D, SmartCYP, MetaDrug, MetaSite and SOME35ꢀ39
(Supplementary Figure 4 and Experimentals Section). The efficacy of new analogs was tested in
a biochemical poly(ADP ribosyl)ation assay that measures the inhibition of TNKS2, and in a
Wnt/βꢀcatenin pathway assay based on STꢀLuc reporter activity in Wnt3aꢀinduced HEK293 cells
(Figure 2a and Supplementary Figure 1ꢀ3). Finally, selected compounds were exposed to human
liver microsomes (HLM), a suitable assay for predicting phase I metabolism (Figure 2a and
Supplementary Figure 1ꢀ3). Figure 2a depicts the progress and evolution of the structure activity
findings starting from molecule 1 (IC50 [STꢀLuc] = 1.01 ꢁM; IC50 [TNKS2] = 0.46 ꢁM),
highlighting selected key compounds during the process of compound improvement and ending
with the lead analog 66 (IC50 [STꢀLuc] = 0.05 ꢁM; IC50 [TNKS1] = 0.046 ꢁM; IC50 [TNKS2] =
0.025 ꢁM).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Five specific regions of JW74 were selected for potential modification (Figure 2a, upper
left panel). Initial SAR studies revealed that the 1,2,4ꢀtriazole core was most likely central for the
activity of this chemotype. For example, when a nitrogen atom in the 1,2,4ꢀtriazole core was
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 43
1
2
replaced with a carbon atom in compound 146 (14),25 activity was lost (IC50 [STꢀLuc] > 10 ꢁM)
(Supplementary Figure 1d).
3
4
5
6
7
8
9
Furthermore, testing of commercially available truncated 1 analogs indicated that in the
context of the central 1,2,4ꢀtriazole, regions AꢀD of 1 were necessary for retaining activity
(Supplementary Figure 1ꢀ3). A replacement of region E by a methyl group resulted in a
substantial reduction of activity (compound 264 (54)29, IC50 [STꢀLuc] = 4.5 ꢁM, Supplementary
Figure 3a). Region E was therefore retained in the further compound development.
Region C was explored to determine the contribution of the methylene thioether linkage
to stability and potency. Replacing sulfur with oxygen (an Hꢀbond acceptor) resulted in a
compound devoid of activity (compound 126 (11)25, IC50 [STꢀLuc] = >10 ꢁM; IC50 [TNKS2] =
>10 ꢁM) (Figure 2a). Inserting an NꢀH group (H bond donor) in place of sulfur also yielded an
inactive molecule (compound 134 (12)25, IC50 [STꢀLuc] = >10 ꢁM; IC50 [TNKS2] = >10 ꢁM)
(Figure 2a). Furthermore, 11 and 12 were unstable when exposed to HLM (Figure 2a).
A Topliss tree analysis40 indicated an enhanced potency when there was a substitution at
the 2ꢀposition of the phenyl ring (region D). The comparative analog analysis (Supplementary
Figure 1c) showed also that this 2ꢀchlorophenyl in region D yielded the best potency
improvement. Together with the 1,3,4ꢀoxadiazole ring in region B in compound 154 (26),25 the
potency improved 7.2 fold over 1 in a cellular assay, and 9 fold in the TNKS2 biochemical assay
(26, IC50 [STꢀLuc] = 0.14 ꢁM; IC50 [TNKS2] = 0.051 ꢁM) (Figure 2a). The properties of 26
indicated that metabolism may be caused by the methylene thioether moiety and/or the pyridyl
group in region E (Figure 2b and Supplementary Figure 4a).41
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
8
ACS Paragon Plus Environment

Page 9 of 43
Journal of Medicinal Chemistry
1
2
Activity of compound 199 (28)25(Figure 2a) (IC50 [STꢀLuc] = 4.53 ꢁM; IC50 [TNKS2] =
1.8 ꢁM) demonstrated that a ring fusion of regions A and B can be tolerated, although a 4.5 fold
loss of activity relative to 1 was seen in a cellular assay in this particular configuration.
In compound 197 (29)25 (Figure 2a), a replacement of the methylene thioether moiety
with an ethylene chain was tolerated with a 3.5ꢀfold improvement in potency relatively to 1
(cellular assay), but no stability improvement against HLM was seen (t½ = 7 min) (29, (IC50
[STꢀLuc] = 0.42 ꢁM; IC50 [TNKS2] = 0.29 ꢁM). Pyridine rings with a 4ꢀsubstitution (region E)
can metabolize via Nꢀoxide formation, therefore it was not surprising that stability remained an
issue.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compound 251 (43)25 (Figure 2a) was designed in an effort to install a stable moiety in
region E while maintaining optimal moieties in regions A, B and D. Since it was envisioned that
hydrogen bond acceptors and steric substitutes in the paraꢀposition of region E were favorable
for potency (Supplementary Figure 2b), a methyl sulfone moiety was installed.
Furthermore, a nitrogen atom in the 2ꢀposition of the ring was installed to lower the
cLogP. Compound 43 was potent (IC50 [STꢀLuc] = 0.05 ꢁM; IC50 [TNKS2] = 0.043 ꢁM), but
unstable to HLM (t½ = 4 min) (Figure 2a). This finding indicated that the methylene thio unit
(region C) of 43 was the culprit of instability in HLM. However, during the course of the work, it
was shown that potent tankyrase inhibitors could be derived from the 1,2,4ꢀtriazole based (1)
chemotype.42
Conformational statistical analysis using Frog2 for 3D conformations, and DIVCF for
clustering, was performed on compounds 1, 26, 29 and 66.43,44 This analysis indicated that a
transꢀlike conformation provides the most biologically active conformation. In addition,
calculated consensus scores (MetaPrint2D, SmartCYP, MetaDrug, MetaSite) and SOM
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 43
1
2
3
4
5
6
7
8
9
prioritization (Figure 2b and Supplementary Figure 4), indicated that regions C and/or E were the
main sources of HLM instability for compounds studied up to this point. Installation of a transꢀ
alkene (mimicking the geometry of the methylene thio group) in region C (compound 244, 55)29
solved the stability issue (HLM t½ = 128min), but potency (IC50 [STꢀLuc] = 6.3 ꢁM; IC50
[TNKS2] = 12.3 ꢁM) was an issue since nonꢀoptimized groups in region A and E were present
(Figure 2a).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Although region E was shown not to be required for potency (54, IC50 [STꢀLuc] = 4.5
ꢁM) (Supplementary Figure 3a), when installing the 4ꢀsulfoxideꢀ2ꢀpyridyl ring in region E and
methoxy group in region A of 55, the resultant compound 268 (65)29 (Figure 2a) showed
improved potency (IC50 [STꢀLuc] = 0.07 ꢁM; IC50 [TNKS2] = 0.054 ꢁM) and a favorable cLogP
(3.5) profile, but exhibited reduced stability in HLM (t½ = 37 min).
Since the 3ꢀmethoxy group in region A of 65 is susceptible to demethylation, and after
taking a Craig Plot replacement scheme into account, 45, 46 a 4ꢀcyanophenyl group was installed.
The resultant compound, 66, was a potent inhibitor of TNKS1, TNKS2 and canonical Wnt/βꢀ
catenin signaling (IC50 [TNKS1] = 0.046 ꢁM; IC50 [TNKS2] = 0.025 ꢁM; IC50 [STꢀLuc] = 0.05
ꢁM) (Figure 1c and 2a). 66 had also a favorable cLogP (3.90), a high CYP3A4 inhibition IC50ꢀ
value (>25 ꢁM, Supplementary Figure 4c) and was stable in HLM (t½ = 101 min, Figure 2a).
Compared to the starting compound 1, compound 66 showed 18.7 fold increased tankyrase
inhibition in a biochemical assay (TNKS2 biochemical assay), and a 20.2 fold increased
inhibition of STꢀLuc in a cellular Wnt assay (in HEK 293 cells) while the HLM stability
increased 33.7 fold. 66 has a molecular weight of 529.96 daltons and a calculated polar surface
area of 140.5.
10
ACS Paragon Plus Environment

Page 11 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
66 exhibited an excellent pharmacokinetic profile in ICR mice (Figure 2cꢀd).
Bioavailability (F) was on average 94.3% for per oral (p.o.) administration (male: 116%, female:
72.7%) and >100% for intraperitoneal (i.p.) administration (male: 157%, female: 152%). The
clearance rate (CL) was on average 520 mL/h/Kg (male: 600 mL/h/Kg, female: 430 mL/h/Kg),
the volume of distribution was on average 2.42 L/Kg (male: 2.7 L/Kg, female: 2.2 L/Kg), and in
vivo compound halfꢀlives (t½) were on average 3.29 hours for an i.v. administration (male: 3.12,
female: 3.46), 2.61 hours for a p.o. administration (male: 2.60, female: 2.63), and 2.81 hours for
an i.p. administration (male: 2.89, female: 2.73)(Figure 2cꢀd).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Co-crystallization of 66 and TNKS2 establishes an extended adenosine binding
pocket as the molecular target of 66
The drug candidate 66 was coꢀcrystallized with the TNKS2ꢀPARP domain residues 946ꢀ
1162 including an Nꢀterminal hexaꢀhistidine tag and a TEV cleavage site (TNKS2946ꢀ1162). This is
identical to the constructs that were used to characterize the binding of XAV939 (2).47 Soaking
attempts of 66 with TNKS2946ꢀ1162 had been unsuccessful.48 Only prior addition of 66 to the
TNKS2946ꢀ1162 domain would form crystals and resulted in a new crystal condition with new and
higher symmetry space group (P3221) (Supplementary Table 1) when compared to previous
reports.24, 28, 42, 47 Despite the relatively low pH value (pH 4.5) in the applied crystal condition,
we did not expect that the acidic environment would affect the interpretation of the 66 binding
properties. TNKS2946ꢀ1162 crystallizes with three molecules in an asymmetric unit and a solvent
content of ~80%. The refinement was initially performed with nonꢀcrystallographic symmetry
(NCS) restrains. However, the restrains were released in the later stages of the refinement. This
significantly improved the model, which implies that minor atomic differences are present
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 43
1
2
3
4
5
6
7
8
9
between the three molecules. The polypeptide chain of all three molecules was traced completely
in the electron density map, including a fragment of the TEV cleavage peptide. There are no
significant differences between the NCS related molecules as assessed by the Rapido server.49
The structural basis for selective binding of 66 is rationalized as a combination of
hydrophobic interactions, πꢀπ stacking, and hydrogen bond interaction formed within the
adenosine binding pocket of the TNKS2ꢀPARP domain. The drug binding pocket of 66 is
primarily stabilized by hydrophobic interactions and anchored by two hydrogen bonds. One
hydrogen bond is formed between the nitrogen atom (N10 in PDB ID 4HYF) of the 1,2,4ꢀ
triazole core of 66 (Figure 3b) and the backbone amide of Tyr1060. The second hydrogen bond
is formed between the free electron pair of the nitrogen (N16) in the oxadiazole ring (region B,
Figure 3b) and the backbone amide of Asp1045. There are three key πꢀπ stacking interactions
between 66 and TNKS2: (i) The nitrile substituted aromatic ring system in region A of 66 forms
a parallel πꢀπ stacking interaction with His1048 (boxed in the alignment, Figure 3c), an
equivalent interaction was previously found for the aromatic system substituted by an ether in 3ꢀ
(4ꢀmethoxyphenyl)ꢀ5ꢀ((4ꢀ(4ꢀmethoxyphenyl)ꢀ5ꢀmethylꢀ4Hꢀ1,2,4ꢀtriazolꢀ3ꢀyl)thio)ꢀ1,2,4ꢀ
oxadiazole (compound 24 (c24), Figure 3a).42 ii) The chlorinated benzene ring (region D, Figure
3a and b) fits into the hydrophobic pocket that is formed by the side chains of amino acids
Ser1033, Pro1034, and Tyr1071 to Ile1075 of the Gꢀloop. We suspect that the benzene ring
(region D) is supported through Tꢀshaped πꢀπ stacking interactions with the aromatic moiety of
Phe1035 (Figure 3a and highlighted with a black box in the alignment of Figure 3c). (iii) The
side chains of Tyr1060 and Tyr1071 participate in stacking interactions with the aromatic
systems of region E of 66.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
12
ACS Paragon Plus Environment

Page 13 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
In summary, we believe that the parallel πꢀπ stacking between region A and His1048, the
Tꢀshaped πꢀπ stacking between region D and Phe1035, the extended aromatic group in region E
and the two described hydrogen bounds are core features that rationalize the efficacy and
selectivity of 66.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
DISCUSSION AND CONCLUSIONS
The drug candidate 66 shows a favorable combination of high target specificity to
TNKS1/2, high potency and in vitro as well as in vivo stability, leading to a good
pharmacokinetic profile in mice. Earlier iTRAQ studies using the tankyrase inhibitor 2,16
identified TNKS1 and TNKS2, several members of the PARP family and a number of further
proteins that do not contain a PARP domain. Compared to 2, 66 shows a greater selectivity
towards TNKS proteins. Furthermore, no other PARP proteins were found displaced by 66.
Interestingly, 66 displacement also enriched GMDS and SSSCA1 proteins in the eluate. Both
proteins may be residing in a complex with the TNKS1/2 targets. In this context it is notable that
GMDS and SSSCA1 (but not ALDOA) had also been identified by iTRAQ experiments using 2.
In contrast, known PARP1 interacting proteins such as XRCC5/6, that were found displaced by
2,16 were not found using the 66ꢀbased matrix. Recently, GMDS was identified as a TNKS1
binding partner that can inhibit TNKS1 in vitro.31 It is therefore a possibility that also SSSCA1
associates with TNKS proteins.
A few key binding elements in the adenosine binding pocket of TNKS2 explains the
apparent high target specificity of 66. His1048 is specific to TNKS1/2 and not conserved in other
PARP domains (Figure 3c). It likely forms a parallel π−π interaction with the aromatic moiety of
66 (Figure 3a) ꢀ an interaction that has also been discussed by Shultz and coꢀworkers for
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 43
1
2
compound 5.42 In addition, the molecular groove formed by the amino acids His1031 to Phe1035
(Fꢀloop, Figure 3c) and Tyr1071 to Ile1075 (Gꢀloop), adjacent to the nicotinamide binding
pocket, have low sequence conservation throughout PARP domains in humans (Figure 3c). The
structural feature of this groove is highly variable across members of the PARP family, and is in
particular a distinguishing feature between tankyrases and PARP1.50 However, the groove
structure is similar in both TNKS1 and TNKS2. The molecular groove present between the Gꢀ
loop and the Fꢀloop is occupied by Tyr1203 in the apo form of TNKS1 and possibly involved in
substrate release following the ADPꢀribosyltransferase reaction and NAD+ exchange binding by
Tyr1203 (TNKS2 numbering Tyr1050),50 whereby Tyr1203forms a hydrogen bond with the
backbone carbonyl of Tyr1224 (equivalent toTyr1071). Thus both 66 and 5 are predicted to
interfere sterically with a natural binding site of Tyr1050 from the Dꢀloop. It is widely accepted
that antagonistic drugs ensure specificity by hitting key functional elements in the binding
pocket. Examples of this have been described for the quinazolinone and pyridolꢀpyrimidine
classes of p38 mitogenꢀactivated protein (MAP) kinase inhibitors, a class of kinase inhibitors
with a high degree of specificity due to a p38 MAP kinase flexible region (Met109ꢀGly110) that
allows a peptide flip to accommodate the inhibitor binding.51
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Interestingly, the conserved loop motif between beta3 and alpha3 [HQ]ꢀ[GQ]ꢀ[STV],
found in all human PARPs (Figure 3c), extends into Pro1034ꢀPhe1035 in the TNKSꢀPARP
domain. This forms a unique structural motif that is only found in tankyrases. It may be
speculated that the positioning of both 66 and 542 in the binding pocket of tankyrase might be
guided in combination by His1048 and Phe1035. Furthermore, we interpret the clear density of
the chloride position in the anomalous difference density map as an indicator for possible
significant hydrophobic interactions that guide the compound placement in the pocket.
14
ACS Paragon Plus Environment

Page 15 of 43
Journal of Medicinal Chemistry
1
2
Although there are clear similarities between the binding of IWRꢀ1 (3), PJ34 (4)22, c24
(5) and 66, there are also significant differences. In particular the Dꢀloop (Phe1044ꢀIle1059) with
residue Tyr1050 seems to be forced away by 66. This Dꢀloop feature is not present for 5 and 3,
when comparing to the drug pocket of TNKS1 derived from coꢀcrystallography with 5 (PDB ID
3UDD)42 and in the drug pocket of TNKS2 derived from coꢀcrystallography with IWRꢀ1 (PDB
ID 3UA9)28. We hypothesize that the benzene/sulfone moiety of 66 forces the Dꢀloop to allow an
induced fit between Tyr1060 and Tyr1071. This may also contribute significantly to the
specificity of 66.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The drug pocket itself, in the presence of 66 and 5, is dry, no water molecules have been
modelled in between 66 or 5 and TNKS2 to form intercalating hydrogen bonds. Indeed there are
no water molecules that seem to directly support the binding affinity. Thus the dynamics of how
NAD+ enter the pocket in the catalytic process forming poly(APDꢀribosyl) chains and the ability
of 66 to interfere with the process at the molecular level should be features to study in future
scientific interrogation.
The novel tankyrase inhibitor 66 extends the possibilities to selectively inhibit TNKS1/2
in vitro and in vivo. As such, 66 provides an excellent smallꢀmolecule for analyzing the functions
of TNKS1/2, Wnt/βꢀcatenin signaling in cell biology, cancer and disease models.
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 43
1
2
3
4
EXPERIMENTALS SECTION
5
6
7
8
9
Chemistry. The following catalog compounds were purchased: 1 (Ambinter) and 26
(ChemRoyal). The following compounds were synthesized and purchased: 10, 11, 28, 29,
(Medicilon) and 43, 55, 65 and 66 (TC scientific Inc).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
All solvents were purchased from commercial sources. 1H spectra were recorded using 300 or
400 MHz instruments (Varian), or a 600 MHz instrument (Bruker). Compounds were dissolved
in either CDCl3 or DMSOꢀd6. Chemical shifts were reported in δ values (parts per million, ppm)
and calibrated against solvent peaks δ 2.50 (DMSOꢀd6) or δ 7.24 (CDCl3). HRMS was
performed by direct injection of compounds dissolved in acetonitrile/water/formic acid
(70/29.9/0.1, v/v/v) in to a QꢀExactive mass spectrometer (Thermo) with an electrospray
interface, set to positive mode with mass m/z = 445.120025 used for internal calibration. All
compounds tested were of ≥95 % purity, confirmed by HPLCꢀDAD with the following
instrumentation and conditions: Agilent 1100 series pump and photodiode array detection (180ꢀ
400 nm), WP Octadocyl column (250 mm length, 4.6 inner diameter, 5 mm particle size) at 22
°C, MeOH/H20 (75/25, v/v) mobile phase, 1 mL/minute flow rate.
Compounds
Compound 66
To a 100 mL round bottle flask with 5ꢀbromopicolinic acid (7.0 g, 35 mmol) in methanol (80
mL) thionyl chloride (3.0 mL) was added dropwise at room temperature. After addition, the
reaction mixture was heated to reflux for 3 h. Methanol was removed and ethyl acetate (100 mL)
16
ACS Paragon Plus Environment

Page 17 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
was added to the residue and was adjusted pH to 7.0 by addition of sodium bicarbonate solution.
The organic phase was separated and dried over sodium sulfate. The organic solvent was
removed and methyl 5ꢀbromopicolinate was obtained as white solid (6.57 g, 86.9%), which was
used for next step reaction without further purification. A solution of methyl 5ꢀbromopicolinate
(6.20 g, 28.7 mmol) in THF anhydrous (150 mL) was added sodium methythionide (2.51 g, 35.9
mmol) and the reaction was heated to reflux for overnight. The solvent was removed and the
residue was purified by column chromatography (eluting with hexane and ethyl acetate 1:1).
Methyl 5ꢀ(methylthio)picolinate as a white solid was isolated (3.0 g, 57.1%): 1HꢀNMR (300 Hz,
CDCl3) δ ppm 8.58 (d, 1H), 8.02 (d, 1H), 7.60 (dd, 1H), 3.99 (s, 3H), 2.58 (s, 3H). To a 100 mL
round bottle flask with methyl 5ꢀ(methylthio)picolinate (1.50 g, 8.2 mmol) in 30 mL of DCM,
mCPBA (5.51 g, 77%, 25.0 mmol) was added. The reaction was kept at room temperature
overnight. After solvent was removed, the residue was purified by column chromatography
(eluting with hexane and ethyl acetate 1:1) to give methyl 5ꢀ(methylsulfonyl)picolinate as a
white solid (1.35g, 76.5%): 1HꢀNMR (300 Hz, CDCl3) δ ppm 9.22 (d, 1H), 8.40 (dd, 1H), 8.36
(d, 1H), 4.02 (s, 3H), 3.10 (s, 3H). 2ꢀChloroaniline (1.20 g, 0.0093 mol) in toluene (20 mL) was
added trimethylaluminum (2.0 M 4.6 mL) then methyl 5ꢀ(methylsulfonyl)picolinate (1.0 g, 4.65
mol) was added and the mixture was heated to 80ꢀ90 °C for 2 h. The reaction was cooled down
and 1N HCl solution (10 mL) was added for acidification, and dichloromethane (100 mL) was
then added. The organic phase was further washed with water (100 mL) and dried over sodium
sulfate. The solvent was removed and the residue was mixed with ether (50 mL) and stirred for
0.5 h. The solid was filtered and dried to give Nꢀ(2ꢀchlorophenyl)ꢀ5ꢀ
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(methylsulfonyl)picolinamide as light yellow solid ( 1.26 g, 87.2%): 1HꢀNMR (300 Hz, CDCl3) δ
ppm 10.6 (s, 1H), 9.21 (m, 1H), 8.62 (dd, 1H), 8.51 (dd, 1H), 8.46 (dd, 1H), 7.45 (dd, 1H),7.34
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 43
1
2
3
4
5
6
7
8
9
(td, 1H), 7.12 (td, 1H), 3.20 (s, 3H). Nꢀ(2ꢀchlorophenyl)ꢀ5ꢀ(methylsulfonyl)picolinamide (1.26 g,
0.0040 mol) in benzene anhydrous (20 mL) was added PCl5 (1.26 g, 6.0 mmol) and the mixture
was heated to reflux for overnight. The solvent was removed and the residue was further dried
under high vacuum, and crude Nꢀ(2ꢀchlorophenyl)ꢀ5ꢀ(methylsulfonyl)picolinimidoyl chloride
was obtained as a yellow solid (1.60 g). The Nꢀ(2ꢀchlorophenyl)ꢀ5ꢀ
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(methylsulfonyl)picolinimidoyl chloride was dissolved into THF anhydrous (30 mL) and the
reaction was cooled down to 0 °C and hydrazine monohydrate (9.0 mL) was added. The reaction
was kept at 0 °C for 10 min and warmed to room temperature in 0.5 h. The solvent was removed
and the residue was purified by column chromatography (eluting with hexane and ethyl acetate
1:1) to give N'ꢀ(2ꢀchlorophenyl)ꢀ5ꢀ(methylsulfonyl) picolinimidohydrazide as yellow solid (1.22
g, 92.0 %): 1HꢀNMR (300 Hz, CDCl3) δ ppm 9.02 (dd, 1H), 8.25 (dd, 1H), 8.17 (dd, 1H) 7.44 (s,
1H), 7.37 (dd, 1H), 7.16 (td, 1H), 6.89 (td, 1H), 6.45 (dd, 1H), 5.62 (s, 2H), 3.06 (s, 3H). In a 100
mL flask with N'ꢀ(2ꢀchlorophenyl)ꢀ5ꢀ(methylsulfonyl) picolinimidohydrazide (0.63 g, 1.94 mol)
in toluene anhydrous (20 mL), maleic anhydride (0.20 g, 2.04 mol) was added. The reaction was
kept at room temperature for 1 h and then heated to reflux for 3 h. After solvent was removed the
residue was dried under high vacuum to give 3ꢀ(4ꢀ(2ꢀchlorophenyl)ꢀ5ꢀ(5ꢀ
(methylsulfonyl)pyridinꢀ2ꢀyl)ꢀ4Hꢀ1,2,4ꢀtriazolꢀ3ꢀyl)acrylic acid as yellow solid (0.75 g, 95.6%).
To a solution of 4ꢀCyanoꢀbenzoic acid methyl ester (1.0 g, 13 mmol) in 30 mL of methanol, 3
mL of H2NNH2ꢂH2O was added. The mixture was stirred at room temperature for 16 h. The solid
was collected and washed with methanol to give 4ꢀCyanoꢀbenzoic acid hydrazide as a solid (0.8
g, 80%): 1HNMR (DMSOꢀd6): δ ppm 7.93 (d, 4 H); 4.70 (s, 1 H); 3.30 (s, 2 H). In a 100 mL
flask with 3ꢀ(4ꢀ(2ꢀchlorophenyl)ꢀ5ꢀ(5ꢀ(methylsulfonyl)pyridinꢀ2ꢀyl)ꢀ4Hꢀ1,2,4ꢀtriazolꢀ3ꢀyl)acrylic
acid (0.21 g, 0.52 mmol) in dichlormethane (20 mL), oxalyl chloride (0.20 g, 1.55 mmol) and a
18
ACS Paragon Plus Environment

Page 19 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
drop of DMF were added. The reaction was kept at room temperature for 3 h. The solvent was
removed and the residue was further dried under high vacuum and cooled to ꢀ20 °C.
Dichlormethane (10 mL) was added, followed by 4ꢀcyanobenzohydrazide (0.10 mg, 0.62 mmol)
and triethylamine (0.5 mL). The reaction was kept at this temperature for 0.5 h and then room
temperature for 1 h. The solvent was removed and the resulting crude (E)ꢀN'ꢀ(3ꢀ(4ꢀ(2ꢀ
chlorophenyl)ꢀ5ꢀ(5ꢀ(methylsulfonyl)pyridinꢀ2ꢀyl)ꢀ4Hꢀ1,2,4ꢀtriazolꢀ3ꢀyl)acryloyl)ꢀ4ꢀ
cyanobenzohydrazide was used for next step. (E)ꢀN'ꢀ(3ꢀ(4ꢀ(2ꢀchlorophenyl)ꢀ5ꢀ(5ꢀ
(methylsulfonyl)pyridinꢀ2ꢀyl)ꢀ4Hꢀ1,2,4ꢀtriazolꢀ3ꢀyl)acryloyl)ꢀ4ꢀcyanobenzohydrazide was
dissolved into dichlormethane (10 mL), and was added triphenyl phosphine (0.27 g, 1.0 mmol),
carbon tetrabromide (0.34 g, 1.04 mmol) and triethylamine (0.18 mL, 1.30 mmol). The reaction
was kept at room temperature for 2 h. The final compound was purified by column
chromatography (eluting with hexane and ethyl acetate 1:3) to give (E)ꢀ4ꢀ(5ꢀ(2ꢀ(4ꢀ(2ꢀ
chlorophenyl)ꢀ5ꢀ(5ꢀ(methylsulfonyl)pyridinꢀ2ꢀyl)ꢀ4Hꢀ1,2,4ꢀtriazolꢀ3ꢀyl)vinyl)ꢀ1,3,4ꢀoxadiazolꢀ2ꢀ
yl)benzonitrile.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
66, lead compound: Mp: 257ꢀ259°C; TLC (CHCl3: MeOH, 90:10 v/v): Rf = 0.60;
UV/Vis: λmax 323 nm. 1H NMR (600 MHz, DMSO-d6): δ 8.73 (dd, J = 2.3 Hz, 0.8 Hz, 1H), 8.53
(dd, J = 8.3 Hz, 0.8 Hz, 1H), 8.47 (dd, J = 8.4 Hz, 2.3 Hz, 1H), 8.20 (m, 2H), 8.08 (m, 2H), 7.83ꢀ
7.80 (m, 2H), 7.73ꢀ7.70 (m, 1H), 7.63 (dt, J = 6.7 Hz, 1.4 Hz, 1H), 7.45 (d, J = 16.4 Hz, 1H),
7.13 (d, J = 16.8 Hz, 1H), 3.35 (s, J = 6.99 Hz, 3H). 13C NMR (600MHz, DMSOꢀd6): δ 163.6,
163.5, 152.8, 152.5, 149.8, 147.9, 137.8, 137.1, 134.1, 133.8, 132.6, 132.5, 131.4, 130.7, 130.4,
129.2, 128.4, 128.0, 127.4, 123.8, 121.2, 118.5, 117.0, 114.7, 15.61. HRMS (m/z): [MH]+ calcd.
for C25H17ClN7O3S+, 530.08; found, 530.08.
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 43
1
2
3
4
5
6
7
8
9
1, starting molecule: Mp: 169ꢀ171°C; TLC (CHCl3: MeOH, 90:10 v/v): Rf =0.59;
UV/Vis: λmax 230, 251 nm. 1H NMR (600 MHz, CDCl3): δ 8.55 (d, J = 6.2 Hz, 2H); 7.90 (d, J =
4.5 Hz, 2H); 7.32 (d, J = 6.2 Hz, 2H); 7.25 (d, J = 4.5 Hz, 2H); 7.15 (d, J = 8.96 Hz, 2H); 6.99
(d, J = 8.96 Hz, 2H); 4.73 (s, 1H); 3.85 (s, 1H); 2.39 (s, 1H). 13C NMR (600MHz, CDCl3): δ
174.8, 168.7, 161.0, 153.2, 152.4, 150.3, 141.8, 133.8, 129.6, 128.3, 127.4, 125.5, 123.5, 121.4,
115.5, 55.6, 41.0, 26.9, 21.6. HRMS (m/z): [MH]+ calcd. For C25H23N6O2S+, 471.16; found,
471.14.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Plasmids, constructs, cell lines and conditioned media
The L Wnt3aꢀexpressing cells were purchased from ATCC (American Type Culture
Collection) and, including STꢀLuc/Ren HEK293 cells (see below), maintained according to the
supplier’s recommendations. A stable HEK293 cell line containing SuperTOPꢀFlash plasmid
(STꢀLuc HEK293) (7 X TCF binding sites promoter)52 was kindly provided by V. Korinek. To
create SuperTOPꢀLuciferase/Renilla HEK293 cells (STꢀLuc/Ren HEK293), the pRLꢀTK
(Renilla, Promega) cassette was subcloned into pPUR (Promega) giving rise to the construct
pRLꢀTKꢀpuro. Linearized pRLꢀTKꢀpuro was transfected (FuGENE6, Roche) into STꢀLuc
HEK293 before selection (2.5 ꢁg/mL Puromycin, Sigma). Wnt3a containing conditioned media
(Wnt3aꢀCM) from L Wnt3a expressing cells was collected as described by ATCC.
Transfection and luciferase assays
40,000 STꢀLuc/Ren HEK293 cells were seeded in 96ꢀwell plates coated with polyꢀL
lysine. 24 hours after seeding, the cells were incubated for an additional 24 hours with various
compound concentrations in 50 % Wnt3aꢀCM. After compound exposures, the cells were lysed
20
ACS Paragon Plus Environment

Page 21 of 43
Journal of Medicinal Chemistry
1
2
3
4
and the firefly luciferase and Renilla activities were measured on a GloMax® Luminometer
5
6
(Promega) using DualꢀGlo Luciferase Assay System (Promega).
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
TNKS1, TNKS2 and PARP1 in vitro biochemical assays
66 inhibitory activity at various doses (duplicates) was tested twice against TNKS1
(80564), TNSK2 (80576) and PARP1 (80551) Chemiluminescent Assay Kits (BPS Bioscience,
Nordic Biosite). The procedures were performed according to the protocols from BPS
Bioscience and the luminescence was measured on a GloMax® Luminometer (Promega).
IC50 calculations
XLfit (idbs) was used to determine the IC50ꢀvalues in inhibition experiments. The
following formula was chosen to fit the data points (Langmuir Binding Isotherm):
fit = ((A+(B*x))+(((CꢀB)*(1ꢀexp((ꢀ1*D)*x)))/D)), res = (yꢀfit)
HLM analysis
The human liver microsome analysis (HLM) was performed according to the standard
protocols of Medicilon/MPI Preclinical Research (Shanghai, China) or Cyprotex (Macclesfield,
United Kingdom).
Chemoproteomics target profiling assay
Chemoproteomics target profiling assays were carried out as described previously30 with
minor modifications: Compound 161 was immobilized on sepharose resin and the affinity matrix
was washed in lysis buffer. Aliquots of HEK293 cell lysate (5 mg total protein each) were preꢀ
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 43
1
2
3
4
5
6
7
8
9
incubated with 20 ꢁM of the tankyrase inhibitor 66 or the inactive 25 (all containing 1% DMSO)
or with vehicle control at 4 °C. 35 ꢃL 161 affinity matrix per sample was added and incubated
for two hours, before being washed extensively and eluted. Separation of eluted proteins by
SDSꢀPAGE, processing of gels, labeling with TMT tags (ThermoꢀFisher Scientific), LCꢀMS/MS
analysis on a nanoꢀLC system (Eksigent 1D+) coupled to an LTQꢀOrbitrap Velos mass
spectrometer (Thermo Fisher Scientific) and peptide and protein identification were all done as
described.30
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Reporter ion intensity for the vehicle control was defined as 0% competition. A protein
competed to a larger extend than defined by the 3 times standard deviation interval was
considered to be specifically bound. Competition data was analyzed and plotted using Tableau
software.
Kinase, phosphatase and GPCR assays
The kinase, phosphatase and GPCR assays were performed as described in
KinaseProfiler™ Service and GPCRProfiler® Assay Protocols (Millipore).
Modeling of metabolic stability in human liver microsomes
During synthesis of new compounds two types of in silico approaches were used:
Ranking of sites of metabolism (SOM) in cytochromes P450 and estimation of halfꢀlife times in
human liver microsomes. Five software packages, based on different approaches, have been used
in the study: MetaPrint2D, SmartCYP 2.3, MetaDrug, MetaSite 3.1.2 and SOME 1.3.35ꢀ39 Each
program calculated the individual scores for nonꢀhydrogen atoms. The top three calculated SOM
atoms were used in SOM consensus score calculations. Descending rank values were assigned to
22
ACS Paragon Plus Environment

Page 23 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
each of the top three predicted SOMs (from most unstable to least unstable): 3 for the first SOM,
2 for the second SOM and 1 for the third SOM.
7
8
9
A second in silico approach was used to evaluate a quantitative correlation between in
silico predictions and metabolism in human liver microsomes. The programs used estimated
metabolic stability against members of the cytochrome P450 family. We hypothesized that an
overall metabolic stability in HLM assays correlates with the scores of the highest ranked SOMs
as these atoms are supposed to kinetically define the fastest metabolism reaction. This correlation
was observed with all programs except Metadrug. The metabolism rate consensuses scores
(MRCS), based on the first site of metabolism from each program, were estimated by the
following formula:
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Predictions of metabolic modifications
Three software packages were used in the analysis and identification of metabolic top
SOMs: MetaDrug, MetaSite 3.1.2 and SOME 1.3. The main metabolic reactions were suggested
by all three programs: Oxidation of sulfur, Nꢀoxidation of nitrogen in position 4 of region E, Oꢀ
dealkylation of methoxy groups, aliphatic hydroxylations of carbons in region C and in the
methyl group of region A. The aromatic hydroxylation was predicted as a metabolism reaction
for the aromatic carbons. MetaDrug and MetaSite prediced dealkylation of the sulfur by breaking
a bond between carbon and sulfur in the region C in addition to oxidation.
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 43
1
2
3
4
Optimization of cytochromes P450 inhibition
5
6
7
8
LogP calculations were performed using the program ALOGP2.1. The program
calculates average LogP values based on several methods along with standard deviation
estimations.33, 34
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Pharmacokinetical analysis
The pharmacokinetical (PK) analyses, per oral injections (p.o.), intra peritoneal (i.p.) and
intravenous injections (i.v.) were performed according to the standard protocols of
Medicilon/MPI Preclinical Research Shanghai, China. A total of 160 male and female ICR mice
from SinoꢀBritish SIPPR/BK Lab Animal Ltd, Shanghai (18.2ꢀ26.0 grams) were used in the
study. The feed were provided ad libitum throughout the inꢀlife portion with the exception of the
overnight fasting period (10ꢀ15 hours) prior to oral administration. 66 was administered via
intravenous injection (i.v.) or oral administration (p.o.) or intraperitoneal injection (i.p.),
respectively. The i.v. and i.p formulations were made in 5% DMSO, 50% PEG400, 45% saline.
The p.o. formulation was made in 10% NMP, 60% PEG400, 30% saline. Blood samples
(approximately 500 ꢁL) were collected via cardiac puncture after euthanasia by carbon dioxide
inhalation at 5 min, 15 min, 30 min, 1, 2, 4, 8, and 24 hours post dose. Blood samples were
placed into tubes containing sodium heparin and centrifuged at 8000 rpm for 6 minutes at 4°C to
separate plasma from the samples. Analyses of the plasma samples were conducted by the
Analytical Sciences Division of Medicilon Preclinical Research (Shanghai) LLC. The
concentrations of 66 in plasma were determined using a high performance liquid
chromatography/mass spectrometry (HPLCꢀMS/MS). No interfering peaks were detected at the
24
ACS Paragon Plus Environment

Page 25 of 43
Journal of Medicinal Chemistry
1
2
3
4
retention time of the analyte or the IS indicating that the method was specific for both 66 and the
5
6
IS.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Crystallography, Transformation and expression
Plasmid pNICꢀBsa4 contains the catalytic PARPꢀdomain of human tankyrase 2 (L940
–
G1162) with an Nꢀterminal His6 tag. The plasmid was transformed into E. coli Rosetta II. Colonies
were inoculated in LB medium over night (o/n) at 37 ˚C, 200 rpm, containing 35 ꢁg/mL CAM
and 50 ꢁg/mL KAN. Fresh o/n cultures were used to inoculate 1 L TB medium supplemented
with 35 ꢁg/mL CAM, 50 ꢁg/mL KAN and 130 ꢁL antifoam (addition of 100 phosphate buffer to
TB medium was added the same day (to keep it sterile for as long as possible)). The culture was
grown at 37 ˚C, 200 rpm until OD600 reached ~1.5. The culture was downꢀtempered to roomꢀ
temperature and expression of the catalytic PARPꢀdomain of TNKS2 was induced by addition of
0.5 mM IPTG and growth continued overnight at 18 ˚C, with full airflow in a Lex Harbinger
Fermentor. Cells were harvested by centrifugation at 6,000 x g for 30 min, 4 ˚C. The resulting
cell pellet was resuspended in resolubilization buffer (50 mM TrisꢀHCl, 5 % glycerol)
supplemented with 1 mM freshly added PMSF, distributed into 50 mL tubes, and centrifuged at
4000 rpm for 15 min, 4 ˚C. Cell pellets were stored at ꢀ20 ˚C until further use.
Crystallography, Extraction and purification
A 10 times weight volume lysis buffer (30 mM TrisꢀHCl (pH 7.6), 300 mM NaCl, 10%
glycerol, 0.5 mM DTT and 10 mM imidazole) supplemented with 1 mM PMSF and 1 ꢁg/mL
DNAse was added to 17ꢀ22 g wet weight of cells. Cells were lysed using a High Pressure
Homogenizer C3 (Avestin) at a pressure between 1000 and 1500 bar. Cell debris was removed
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 43
1
2
3
4
5
6
7
8
9
by centrifugation at 18,000 rpm for 35 minutes, 4 ˚C. Protein purification was performed using
ÄKTAprime Plus (for IMAC) and ÄKTApurifier (GE Healthcare). Prior to purification, IMAC
column was equilibrated using wash buffer 1 (30 mM TrisꢀHCl, 500 mM NaCl, 10 % glycerol,
10 mM imidazole and 0.5 mM DTT.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The lysate was loaded onto the column, usually at a volume of 200 mL and run through
the column once or twice at 4ꢀ5 ml/min. The column was then washed with wash buffer 1 (2ꢀ3
VCOL) followed by wash buffer 2 (30 mM TrisꢀHCl, 500 mM NaCl, 10% glycerol, 25 mM
imidazole and 0.5 mM DTT) (~5 VCOL). Bound protein was eluted with IMAC elution buffer (30
mM TrisꢀHCl, 500 mM NaCl, 10% glycerol, 500 mM imidazole and 0.5 mM DTT) at a flow rate
of 4ꢀ5 ml/min in 5 mL fractions. The 5 mlꢀfractions were analyzed by SDSꢀPAGE. Target
fraction was pooled and concentrated down to 1 mL before applied to the GF column (30 mM
TrisꢀHCl, 300 mM NaCl, 10% glycerol and 0.5 mM DTT). Fresh DTT was added at a final
concentration of 2 mM. Purified protein was analyzed again using SDSꢀPAGE and target
fraction was concentrated to 20ꢀ30 mg/mL using a VivaSpin 6 10,000 MWCO centrifugal filter
device (Sartorius Stedim Biotech). Concentration was measured using the NanoDrop 2000
Spectrophotometer, 1:10 dilutions (Thermo Scientific). Protein solution was flashꢀfreezed with
liquid nitrogen and stored at – 80 ˚C until further use.
Crystallography, Crystallization
Crystals were obtained by the sitting drop vapor diffusion method in a 24ꢀwell plate. A
volume of 1 ꢁL protein solution, including 66 at a ratio 10:1 (ligand:protein), was dissolved at 60
˚C, 5 min and resuspended by vortexing. 1 ꢁL of a 0.15 M 66 solution was added to a 50 ꢁL
protein solution concentrated to 15 mg/mL and left to incubate 10 min on ice. The protein: 66
26
ACS Paragon Plus Environment

Page 27 of 43
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
solution was mixed 1:1 with the well solution (100 mM NaOAc pH 4.5 and 16ꢀ26% PEG 3350).
All steps were performed in the cold room. The plates were left to incubate at 4 ˚C. Crystals
appeared after ~ 1 week and continued to grow for ≥ 2 weeks more. Crystals were quickly
transferred to a cryo solution containing 30% ethylene glycol mixed with reservoir solution and
frozen in liquid nitrogen.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The TNKS2946ꢀ1162 domain is highly temperature sensitive in the presence of 66 and had
to be kept cold throughout the entire procedure crystallization setup. Heavy precipitate (66) was
observed immediately in the presence of the crystallization conditions. Crystals only appeared
when the protein concentration was above 15 mg/mL. It was usually necessary to add additional
glycerol in the reservoir chamber to push the vapor diffusion experiment further. Compared to
Karlberg and coworkers,47 it was not necessary to use an in situ proteolysis crystallization
strategy.
Crystallography, Data collection and refinement
Diffraction data were collected at 100 K on the end stations X06DA at the Swiss Light
Source (SLS). The diffraction data were processed and scaled with XDS.53 Phases were obtained
by molecular replacement using the program PHASER54 and a search model of the tankyrase 2
PARP domain (PDB ID 3MHJ).24 Model building was performed using Coot55 and model
refinement was performed with phenix.refine.56 For reflection file handling, programs from the
CCP4 package were used.57 All structural figures in this paper were prepared with Pymol
(http://www.pymol.org). The atomic coordinates and experimental data (code 4HYF) have been
deposited in the Protein Data Bank.
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 43
1
2
3
4
5
6
Bioinformatics and alignment of the human PARP domains were performed with default
settings in MAFFT58 and edited by hand in Jalview.59
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACKNOWLEDGMENT
This work was supported by the CRI program of the Research Council of Norway, and by
Helse Sør Øst, grant 2010031. We thank TC Scientific Inc for excellent compound synthesis and
Marcus Bantscheff for protein mass spectrometry during affinityꢀcapture experiments. We thank
Herwig Schuler for providing expression plasmids and information needed for the production of
recombinant proteins. The authors also wish to express gratitude for beam line support by the
staff at the (Swiss Light Source) and the Berliner ElektronenspeicherringꢀGesellschaft fur
Synchrotronstrahlung (BESSY) for making our experiments possible. Initial crystal tests and
opmtimization was performed at BESSY. A special thanks to the Blix foundation for supporting
the laboratory with additional equipment to finish the study.
ANCILLARY INFORMATION
Supporting Information Available: Supplementary Figures and Chemistry. This material is
available free of charge via the Internet at http://pubs.acs.org.
28
ACS Paragon Plus Environment

Page 29 of 43
Journal of Medicinal Chemistry
1
2
3
4
REFERENCES
5
6
7
8
(1) Hottiger, M. O.; Hassa, P. O.; Luscher, B.; Schuler, H.; KochꢀNolte, F. Toward a unified
nomenclature for mammalian ADPꢀribosyltransferases. Trends Biochem. Sci. 2010, 35, 208ꢀ219.
(2) Guettler, S.; LaRose, J.; Petsalaki, E.; Gish, G.; Scotter, A.; Pawson, T.; Rottapel, R.;
Sicheri, F. Structural basis and sequence rules for substrate recognition by Tankyrase explain the
basis for cherubism disease. Cell 2011, 147, 1340ꢀ1354.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(3) Hsiao, S. J.; Smith, S. Tankyrase function at telomeres, spindle poles, and beyond.
Biochimie 2008, 90, 83ꢀ92.
(4) Sbodio, J. I.; Chi, N. W. Identification of a tankyraseꢀbinding motif shared by IRAP,
TAB182, and human TRF1 but not mouse TRF1. NuMA contains this RXXPDG motif and is a
novel tankyrase partner. J. Biol. Chem. 2002, 277, 31887ꢀ31892.
(5) Seimiya, H.; Smith, S. The telomeric poly(ADPꢀribose) polymerase, tankyrase 1,
contains multiple binding sites for telomeric repeat binding factor 1 (TRF1) and a novel
acceptor, 182ꢀkDa tankyraseꢀbinding protein (TAB182). J. Biol. Chem. 2002, 277, 14116ꢀ14126.
(6) Seimiya, H.; Muramatsu, Y.; Smith, S.; Tsuruo, T. Functional subdomain in the ankyrin
domain of tankyrase 1 required for poly(ADPꢀribosyl)ation of TRF1 and telomere elongation.
Mol. Cell Biol. 2004, 24, 1944ꢀ1955.
(7) De Rycker M.; Price, C. M. Tankyrase polymerization is controlled by its sterile alpha
motif and poly(ADPꢀribose) polymerase domains. Mol. Cell Biol. 2004, 24, 9802ꢀ9812.
(8) Ha, G. H.; Kim, H. S.; Go, H.; Lee, H.; Seimiya, H.; Chung, D. H.; Lee, C. W.
Tankyraseꢀ1 function at telomeres and during mitosis is regulated by Poloꢀlike kinaseꢀ1ꢀ
mediated phosphorylation. Cell Death. Differ. 2012, 19, 321ꢀ332.
29
ACS Paragon Plus Environment