Discovery of AZ0108, an orally bioavailable phthalazinone PARP inhibitor that blocks centrosome clustering

Article abstract of DOI:10.1016/j.bmcl.2015.10.079

The propensity for cancer cells to accumulate additional centrosomes relative to normal cells could be exploited for therapeutic benefit in oncology. Following literature reports that suggested TNKS1 (tankyrase 1) and PARP16 may be involved with spindle s

Full text of DOI:10.1016/j.bmcl.2015.10.079

Bioorganic & Medicinal Chemistry Letters 25 (2015) 5743–5747
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of AZ0108, an orally bioavailable phthalazinone PARP
inhibitor that blocks centrosome clustering
a
a
a
a
Jeffrey W. Johannes a, , Lynsie Almeida , Kevin Daly , Andrew D. Ferguson , Shaun E. Grosskurth ,
Huiping Guan a, Tina Howard b, Stephanos Ioannidis a, Steven Kazmirski a, Michelle L. Lamb a,
Nicholas A. Larsen a, Paul D. Lyne a, Keith Mikule a, Claude Ogoe a, Bo Peng a, Philip Petteruti a, Jon A. Read c,
Nancy Su a, Mark Sylvester a, Scott Throner a, Wenxian Wang a, Xin Wang a, Jiaquan Wu a, Qing Ye a, Yan Yu a,
Xiaolan Zheng a, David A. Scott a
⇑
a AstraZeneca R&D Boston, 35 Gatehouse Drive, Waltham, MA 02451, United States
b AstraZeneca R&D Alderley Park, Macclesfield, Cheshire SK10 4TG, United Kingdom
c AstraZeneca R&D Building 310, Milton Science Park, Cambridge CB4 0WG, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history:
The propensity for cancer cells to accumulate additional centrosomes relative to normal cells could be
exploited for therapeutic benefit in oncology. Following literature reports that suggested TNKS1 (tankyr-
ase 1) and PARP16 may be involved with spindle structure and function and may play a role in suppress-
ing multi-polar spindle formation in cells with supernumerary centrosomes, we initiated a phenotypic
screen to look for small molecule poly (ADP-ribose) polymerase (PARP) enzyme family inhibitors that
Received 4 September 2015
Revised 23 October 2015
Accepted 26 October 2015
Available online 27 October 2015
could produce
a multi-polar spindle phenotype via declustering of centrosomes. Screening of
Keywords:
Centrosome
PARP
Tankyrase
Oncology
Cell cycle
AstraZeneca’s collection of phthalazinone PARP inhibitors in HeLa cells using high-content screening
techniques identified several compounds that produced a multi-polar spindle phenotype at low nanomo-
lar concentrations. Characterization of these compounds across a broad panel of PARP family enzyme
assays indicated that they had activity against several PARP family enzymes, including PARP1, 2, 3, 5a,
5b, and 6. Further optimization of these initial hits for improved declustering potency, solubility, perme-
ability, and oral bioavailability resulted in AZ0108, a PARP1, 2, 6 inhibitor that potently inhibits centro-
some clustering and is suitable for in vivo efficacy and tolerability studies.
Ó 2015 Elsevier Ltd. All rights reserved.
One strategy for treating cancer focuses on targeting the differ-
ences between normal cells and tumor cells.1 An example of such a
difference is the propensity for cancer cells to accumulate extra
copies of centrosomes, the cellular machinery that serves as the
critical microtubule organizing center during mitosis.2 During the
S phase of mitosis, the mother centrosome is copied to yield a
daughter centrosome, and these separate to opposite ends of the
cell. Each centrosome then serves as a microtubule organizing cen-
ter for the mitotic spindle. A normal cell entering mitosis with two
centrosomes results in a bipolar spindle, which is a prerequisite for
proper segregation of sister chromatids to each daughter cell. How-
ever, cancer cells that have accumulated extra centrosomes face a
conundrum. If a cell has three or more centrosomes during mitosis,
this can result in a multi-polar mitotic spindle, leading to improper
segregation of chromosomes, and eventual mitotic catastrophe. To
overcome this obstacle, cancer cells with supernumerary centro-
somes evolve mechanisms to cluster them into two bundles. These
two centrosome clusters allow the cell to form a bipolar mitotic
spindle and undergo mitosis with proper fidelity.
This observation has led to the hypothesis that disruption of the
clustering process in cancer cells with supernumerary centrosomes
could lead to mitotic catastrophe while normal cells that can
undergo mitosis independent of centrosome clustering mecha-
nisms would be unaffected.3,4 Discovery of small molecules that
could achieve this goal could lead to treatments for cancer. This
promise prompted detailed investigations of centrosome cluster-
ing to uncover the exact biochemical mechanisms and protein
components that are involved in this process. Based on a siRNA
screen against all PARP enzymes, TNKS1 (PARP5a) was shown to
Abbreviations: CL, clearance; ent, enantiomer; F%, bioavailability; calc., calcu-
lated; enz, enzyme; S2, Schneider 2; PK, pharmacokinetics; PPB, plasma protein
binding; PARP, poly (ADP-ribose) polymerase.
⇑
Corresponding author.
http://dx.doi.org/10.1016/j.bmcl.2015.10.079
0960-894X/Ó 2015 Elsevier Ltd. All rights reserved.

5744
J. W. Johannes et al. / Bioorg. Med. Chem. Lett. 25 (2015) 5743–5747
be required for the formation of bipolar spindles.5 Moreover, Kwon
et al. performed a genome wide siRNA knockdown screen in Dro-
sophila to uncover which genes may be involved in the centrosome
clustering process.6 Upon treatment with siRNAs to induce a multi-
polar spindle phenotype in Drosophila S2 cells, they identified
genes homologous to the human PARP genes TNKS1 and PARP16.
The demonstrated druggability of the PARP enzyme family7,8 along
with AstraZeneca’s experience with the PARP inhibitor olaparib
prompted us to explore the possibility that a small molecule inhi-
bitor of the PARPs could induce a multi-polar spindle phenotype
and prevent the proliferation of cancer cells.
Our initial efforts concentrated on the development of a high
content phenotypic assay to screen for small molecules that block
centrosome clustering and cause a multi-polar spindle phenotype.
This cell-based centrosome declustering assay begins with treat-
ment of HeLa cells with increasing concentrations of compound.
After 24 h, the cells are fixed and then the mitotic cells are identi-
fied by treatment with a cyclin B antibody. Next, the cells are trea-
ted with an antibody towards pericentrin to label the centrosomes.
After that, an automated process generates an overlay of the two
images and analyzes it to calculate the number of centrosomes
observed in each mitotic cell. Mitotic cells with greater than four
centrosomes9 are then assumed to have a multi-polar spindle phe-
notype. Finally, a declustering EC50 is calculated from a dose-
response curve generated from a plot of the percentage of mitotic
cells with a multi-polar spindle phenotype at eight concentrations
Figure 2. Crystal structure of AZ9482 (compound 1, yellow) bound to TNKS1 (PDB
ID: 5ECE). The 2Fo ꢀ Fc electron density is displayed at 1
r
.
concentration, followed up by screening of dozens of compounds
with a full centrosome declustering EC50. Experimental details of
this assay, including images of the multipolar spindle phenotype,
can be found in the Supporting Information.
ranging from 3 nM to 10 lM. The throughput of this assay allowed
for the screening of several hundred compounds at a single
We screened a sub-set of compounds from the AstraZeneca col-
lection containing mostly phthalazinone and quinazolinone based
NAD+ mimetic cores. The most potent compound in our initial
screen was AZ9482, a phthalazinone PARP inhibitor featuring an
O
NH
amide linkage to
a 2-piperazinyl-3-cyano-pyridine (Fig. 1).
N
O
N
AZ9482 exhibits very potent centrosome declustering activity in
HeLa cells (EC50 < 18 nM). Encouragingly, AZ9482 also showed a
3 nM GI50 in the DLBCL cell line OCI-LY-19 in a 3-day AlamarBlue
assay. Consistent with published siRNA knockdown experiments,
AZ9482 was a 9 nM enzyme inhibitor of tankyrase 1 (TNKS1).
This interesting biological profile prompted us to generate a
crystal structure of AZ9482 bound to TNKS1 to inform the further
optimization of the compound. We used TNKS1 protein as a PARP
enzyme for crystallography since the tankyrases have proven to be
N
N
AZ9482 (1)
N
Declustering EC50 = <18 nM
TNKS1 IC50 = 9 nM
OCI-LY-19 GI50 = 3 nM
Figure 1. Initial declustering hit from a cell-based screen of PARP inhibitors.
Table 1
Fused 1,2,4-triazole improves solubility and intrinsic metabolic stability, while methyl substitution on the piperazine rescues cell potency and further improves intrinsic
metabolism
O
NH
N
O
R2
A
N
N
F
N
R1
F
F
Compd
AZ9482 (1)
2a
2b
3a
3b
3c
3d
A
—
—
—
<0.018
11
1.53
2.42
3.2
134
83
N
H
H
0.051
984
0.79
1.03
27
CH
H
H
>11
511
1.83
1.66
14
N
N
N
H
N
H
R1
CH3 (S)
H
CH3 (R)
H
R2
CH3 (S)
5.04
612
1.31
CH3 (R)
>11
612
1.31
Declustering EC50
Aq sol. pH 7.4 (
ClogP
LogD 7.4
Hu PPB (% free)
Rat heps (
Rat IV CL (mL/min kg)
(
l
M)
<0.015
>1000
1.31
1.19
19
>11
>1000
1.31
1.21
19
lM)
l
L/min 106)
11
159
66
<2.2
97
11
Rat oral bioavailability (%)
12

J. W. Johannes et al. / Bioorg. Med. Chem. Lett. 25 (2015) 5743–5747
5745
a robust PARP enzyme for expression and crystallization.10,11 The
Table 2
Difluoro substitution on the benzylic linker improves rat CL
crystal structure of AZ9482 bound to TNKS1 is presented in
Figure 2. Similar to other PARP inhibitors,12,13 the phthalazinone
core mimics nicotinamide and binds in the same pocket as the
nicotinamide portion of the PARP enzyme cofactor NAD+. The
carbonyl and NH of the phthalazinone form hydrogen bonds with
the backbone of Gly1185. The cyano-pyridine moiety serves as
an adenine mimetic and binds between the two aromatic residues
His1201 and Phe1188. The nitrogen of the nitrile forms a hydrogen
bond with the backbone NH of Asp1198. Linking these two parts
together is a phenyl amide that interacts with the backbone NH
of Tyr1213 via the amide carbonyl.
O
NH
N
O
N
L
N
N
N
Compd*
L
Declustering Rat heps
Rat CL (mL/ Caco2 AB Papp
EC50 M)
(
l
(l
L/min 106) min kg)
(10ꢀ6 cm/s)
Intrigued by this compound’s ability to induce a multi-polar
spindle phenotype and inhibit the growth of cancer cells at low
nanomolar concentrations, we were eager to test the consequences
of this mechanism of action in vivo in a suitable xenograft model.
However, as outlined in Table 1, compound 1 has poor aqueous sol-
1
4
5
6
7
8
9
10
11
CH2
O
NH
S
CF2
CHCH3
CHOH
CHF
CHNH2
<0.018
9.18
9.33
0.311
0.573
0.027
0.115
0.140
0.057
134
39
31
208
109
109
43
192
39
83
21
13
12
97
25
122
86
5y
ubility (11 lM) and DMPK properties. When subjected to human
microsomal (hu mics) or rat hepatocycte (rat heps) incubations
(see Supporting Information Table S1), compound 1 is rapidly
metabolized resulting in very high intrinsic clearances (CLint). This
high intrinsic hepatic clearance is also reflected in vivo, where the
rat clearance (CL) following IV bolus was measured at 83 mL/
min kg. Moreover, the low solubility likely resulted in poor fraction
absorbed and high plasma clearance led to low bioavailability
(12%) following oral dosing in rat. With this data in hand, we
focused our medicinal chemistry efforts on improving the proper-
ties of our analogs with an ultimate goal to increase water solubil-
ity. Our designs also concentrated on mitigating the in vivo
clearance liability of the series while maintaining potency in the
centrosome declustering assay.
The amide linkage in compound 1 served as a handle for rapid
generation of analogs, wherein a large number of diverse piperazi-
nes were incorporated. From this set, we uncovered compound 2a,
a compound featuring a fused triazolopiperazine moiety which
showed greatly improved physicochemical properties compared
to initial hit 1 (Table 1). Compound 2a was 51 nM in the decluster-
30
0.2
*
Data for most cell potent, chirally pure enantiomer where applicable.
39% mass recovery.
y
intrinsic clearance of <2.2 l
L/min 106 cells. In spite of this, the
in vivo PK of compound 3a still showed poor clearance in rat
(CL = 97 mL/min kg).
Other metabolites produced in vivo in rat from 2a included
hydroxylation of the benzylic methylene linker between the
phthalazinone core and the phenyl ring. Following this observa-
tion, we designed and synthesized a series of compounds featuring
alterations and substituents at the benzylic methylene linker
(Table 2, position L in the structure). We chose to make these
alterations to 1, since many of the substituents would create a
stereogenic center at position L and the 3-cyano-2-piperazinyl-
pyridine moiety was commercially available, simplifying the
synthesis and chiral purification of these compounds. Changing
the CH2 linker to O or NH caused a large reduction in declustering
potency. Nevertheless, compounds 4 and 5 had improved rat heps,
and for compound 4, this manifested itself in a greatly improved
rat in vivo clearance of 21 mL/min kg, an almost 4-fold improve-
ment over 1. Encouraged by this, we prepared additional analogs
such as sulfur linked compound 6, which retained some decluster-
ing cell potency but showed no improvement rat hepatocytes or
in vivo clearance. Interestingly, CF2 linked compound 7 had greatly
improved in vivo clearance and still retained some declustering
cell potency.14 We also explored compounds that kept the carbon
linker while introducing a substituent such as CH3, OH, F, and
NH2 (compounds 8–11). In these cases, only one enantiomer was
appreciably active in the declustering cell assay. Of this set, only
compound 11 retained cell potency while showing an improve-
ment in both in vitro and in vivo PK. Examining 7 versus 11 with
the goal of optimizing this series to an orally bioavailable com-
pound, we chose to progress with compound 7, since it retained
permeability while compound 11 had very low permeability in a
Caco2 permeability assay.15,16
ing assay and also had an aqueous solubility of 984 lM. More
importantly, 2a had greatly reduced lipophilicity (calculated
reduction in ClogP) and correspondingly low CLint in rat heps.
Unexpectedly, this compound did not show a corresponding
improvement in rat CL in vivo; its measured rat clearance was very
high at 159 (mL/min kg). Interestingly, close analog 2b, which has
swapped nitrogen for carbon to create a fused imidazopiperazine,
was not potent in our declustering assay. We surmised that this
nitrogen in compound 2a serves as a hydrogen bond acceptor for
the backbone NH of Asp1198, taking the place of the nitrile moiety
in 1.
Having identified compound 2a, which featured improved
CLint and solubility, we sought to fix the poor in vivo clearance
of this series. Guided by the observation of oxidative metabolism
of 2a at the methylene groups alpha to the amide nitrogen
in vivo, we explored methyl group substitution at these positions
(Table 1). Methyl substitution at the pseudo-benzylic methylene
between the piperazine nitrogen and the triazole was not toler-
ated; compounds 3c and 3d lost cell potency significantly. Sur-
prisingly, addition of a methyl group at the opposite methylene
alpha to the amide nitrogen was tolerated, provided the com-
pound had (S) stereochemistry. (S)-methyl compound 3a had
improved declustering potency comparable to 1, while (R)-methyl
Combining the difluoro linker with the (S)-methyl fused
triazolopiperazine adenine mimetic led to compound 14a (Table 3)
which had improved cell potency (216 nM) relative to compound 7.
More importantly, this compound had a clearance of 33 mL/min kg
with an oral bioavailability of 43%, demonstrating that the
beneficial effects of the CF2 linker on PK carried over to the more
potent fused triazolopiperazine sub-series. Compound 14a also
had favorable solubility, and its overall potency, property and PK
profile served as the basis for the final phase of optimization.
compound 3b had an EC50 of >11 lM in the declustering assay.
Moreover, compound 3a was in a much better physicochemical
property space and DMPK space. Compound 3a had a measured
aqueous solubility of >1000
ubility of only 11 M. Most importantly, (S)-methyl compound 3a
was more stable in rat hepatocytes compared to 2a, with an
lM compared to 1, which had a sol-
l

5746
J. W. Johannes et al. / Bioorg. Med. Chem. Lett. 25 (2015) 5743–5747
Table 3
Modification of the 1,2,4-triazole 3-substituent improves bioavailability leading to AZ0108 (compound 14f)
O
NH
N
O
F
N
N
F
N
N
R3
R3
R3 Hammett rp
Decl. EC50
(lM)
Aq sol. pH 7.4 (lM)
LogD 7.4
Rat CLy
Rat oral F%
Caco2 AB Papp
Caco2 efflux ratio
à
Compd
14a
14b
14c
14d
14e
14f (AZ0108)
CF3
iPr
cyBu
tBu
C(Me)2F
C(Me)F2
0.54
ꢀ0.15
ꢀ0.14
ꢀ0.20
ꢀ0.01*
0.21*
0.216
0.102
0.049
0.087
0.043
0.053
268
>928
619
473
436
2.3
2.1
2.4
2.5
2.4
2.7
33
23
36
60
38
17
43
3
5
15
57
79
9.3
4.0
4.8
6.2
8.4
3.6
13.5
11.0
5.5
4.2
2.3
238
18.5
*
y
Calculated using ACD labs.
Unit: mL/min kg.
à
Unit: 10ꢀ6 cm/s.
in the hydrophobic cleft between aromatic residues His1202 and
Phe1188 in the adenine pocket. Both compounds 1 and 14a mimic
the NAD+ cofactor very closely, and based on this, we hypothesize
that these compounds would have a similar binding mode in other
PARPs.
In considering the further optimization of 14a, we hypothesized
that modulation of the electronic nature of the substituent at R3
(Table 3) would influence the strength of the backbone hydrogen
bond between the triazole and Asp1198. We also strove to keep
the R3 group hydrophobic, consistent with the structure-activity
relationships we had observed. To test this hypothesis, a series of
compounds (14b–d in Table 3) were made featuring a lipophilic,
electron donating group at R3 as approximated by Hammett
rp val-
ues.17 Introduction of donating lipophilic groups such as isopropyl,
cyclobutyl, or t-butyl (compounds 14b–d) improved declustering
cell potency as desired, but they also exhibited a sharp drop in oral
bioavailability in rat. Bioavailability appeared to drop after cross-
ing a threshold Caco2 efflux ratio of approximately 10. For exam-
ple, iPr analog 14b had only 3% oral bioavailability, with a Caco2
efflux ratio of 13.5. The most potent compound in this set was
the cyclobutyl analog 14c, which was 49 nM in the declustering
assay, but it had only 5% bioavailability and a Caco2 efflux ratio
11.0.
Figure 3. Crystal structure of compound 14a (magenta) bound to TNKS1 (PDB ID:
5EBT). The 2Fo ꢀ Fc electron density is displayed at 1
r
.
We next obtained a crystal structure of 14a bound to TNKS1
(Fig. 3). As expected, the phthalazinone core of compound 14a
binds in a similar manner as 1. The switch to a CF2 linker causes
a slight increase in the dihedral angle between the phthalazinone
core and the linker-phenyl bond (ꢀ93° for the CF2 linker in com-
pound 14a vs ꢀ83° for the CH2 linker in 1). The amide linkage
engages Tyr1213 just as compound 1 does. The structure shows
that the (S)-methyl of the piperazine fits into a small hydrophobic
pocket near the D-loop. The 1,2,4-triazole moiety forms a hydrogen
bond to the backbone NH of Asp1198 mimicking the nitrile of
AZ9482 as we had hypothesized. The trifluoromethyl group lies
Clearly, alkyl groups at R3 gave the improvement in potency
that we sought, but at the expense of higher efflux ratios and poor
bioavailability. We suspected that higher efflux arose from a more
electron rich triazole due to the strongly electron donating charac-
ter of these alkyl groups (negative Hammett rp values). A similar
trend in the bioavailability of alkyl versus trifluoromethyl triazoles
was observed during the optimization of sitagliptin.18,19 In order to
tune the electronics of the group at R3, addition of a single fluorine
atom to 14b gave rise to fluoro-isopropyl analog 14e, which has a
near electron-neutral predicted Hammett rp value. Gratifyingly,
introduction of this fluorine atom led to a 5-fold improvement in
declustering potency compared to 14a while maintaining good
rat pharmacokinetics. The difluoroethyl analog 14f (AZ0108),
Table 4
PARP family enzyme and cell data for and AZ9482 (1), AZ0108 (14f) and olaparib
AZ9482 (1)
AZ0108 (14f)
Olaparib
which has a predicted rp value that falls between the highly with-
PARP1 Enz IC50
PARP2 Enz IC50
PARP3 Enz IC50
TNKS1 Enz IC50
TNKS2 Enz IC50
PARP6 Enz IC50
Declustering EC50
DLD-1 Wnt IC50
OCI-LY-19 GI50
(
(
(
(
(
l
l
l
l
l
M)
M)
M)
M)
M)
0.001
0.001
0.046
0.009
0.16
<0.03
<0.03
2.8
3.2
>3
0.083
0.053
>3
0.001
0.003
0.046
1.9
1.7
1.8
>11
>3
0.849
drawing trifluoromethyl and the electro-neutral fluoroisopropyl,
showed the best balance of potency and pharmacokinetics. In addi-
tion to excellent declustering potency and low in vivo rat Cl, 14f
had good solubility and the highest permeability of all the analogs
tested, with correspondingly good oral bioavailability.
(l
M)
M)
M)
M)
0.64
(
l
<0.018
0.222
0.003
(
l
Overall, 14f shows several notable improvements compared to
1. Both compounds inhibit centrosome clustering at in HeLa cells
(
l
0.017

J. W. Johannes et al. / Bioorg. Med. Chem. Lett. 25 (2015) 5743–5747
5747
at <100 nM, and show a very potent GI50 against OCI-LY-19 cells. In
spite of having higher logD, 14f is more soluble than compound 1
and has much better human in vitro pharmacokinetic properties.
Across mouse, rat and dog, 14f has an excellent in vivo pharma-
cokinetic profile. (See the Supporting Information Table S1 for fur-
ther details on the DMPK of compounds 1 and 14f).
AZ9482 (1) and AZ0108 (14f) have different inhibition profiles
of the PARP enzyme family as shown in Table 4. Compound 1 is
a multi-PARP enzyme inhibitor, with significant inhibition of
PARPs 1, 2, 3 and TNKS1/2 as well as PARP6 (a mono(ADP-ribosyl)
transferase).20 Compound 1 was also active in a Wnt pathway
PARP enzymes. Experimental details on the preparation of
AZ9482 and AZ0108 can be found in the Supporting Information.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2015.10.
079.
References and notes
1. Hanahan, D.; Weinberg, R. A. Cell 2011, 144, 646.
2. Nigg, E. A. Nat. Rev. Cancer 2002, 2, 815.
reporter assay, producing
a DLD-1 firefly luciferase IC50 of
222 nM, presumably due to the TNKS1/2 activity.21 In contrast to
1, compound 14f is more selective in its enzyme inhibition profile
and effects on cellular pathways and phenotypes. Specifically, 14f
inhibits PARPs 1, 2, and 6 with approximately 100-fold selectivity
against PARP3 and TNKS1. Consistent with this lack of potency
towards tankyrase, 14f is not active in a DLD-1 Wnt luciferase
reporter assay. In contrast, olaparib22 is a very potent PARP1 inhi-
bitor, which also carries some PARP2 and PARP3 activity, but is
selective against the tankyrases and PARP6. Olaparib has an IC50
3. Krämer, A.; Maier, B.; Bartek, J. Mol. Oncol. 2011, 5, 324.
4. Gergely, F.; Basto, R. Genes Dev. 2008, 22, 2291.
5. Chang, P.; Coughlin, M.; Mitchison, T. J. Nat. Cell Biol. 2005, 7, 1133.
6. Kwon, M.; Godinho, S. A.; Chandhok, N. S.; Ganem, N. J.; Azioune, A.; Thery, M.;
Pellman, D. Genes Dev. 2008, 22, 2189.
7. Amé, J. C.; Spenlehauer, C.; de Murcia, G. BioEssays 2004, 26, 882.
8. Rouleau, M.; Patel, A.; Hendzel, M. J.; Kaufmann, S. H.; Poirier, G. G. Nat. Rev.
Cancer 2010, 10, 293.
9. Setting the cut-off at >4 centrosomes gave a bigger assay window compared to
using a cut-off of >3 or >2 centrosomes.
10. 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.
of >3 lM in the DLD-1 Wnt luciferase reporter assay as expected,
but also shows little effect in the declustering assay. These data
suggest that PARP6 enzyme inhibition may contribute to a declus-
tering phenotype in HeLa cells. Additionally, the data imply that
small molecule inhibition of the PARP catalytic domain of the tan-
kyrases does not contribute to this phenotype. However, since
these compounds are not perfectly selective for PARP6, it cannot
be ruled out that inhibition of the PARP catalytic domain of other
PARP family enzymes also contributes to the phenotype. In depth
details of the biological effects of PARP6 enzyme inhibition both
in vitro and in vivo by AZ9482 (1) and AZ0108 (14f) will be
reported in a separate publication.23
In conclusion, using a cell-based phenotypic screen for com-
pounds that can inhibit centrosome clustering in cancer cells, we
screened a subset of the AstraZeneca collection containing known
PARP inhibitor scaffolds. This screen identified AZ9482 (1) as a
potent inhibitor of centrosome clustering. In order to test the
effects of centrosome declustering in vivo, we optimized 1 using
structure and property guided medicinal chemistry design to pro-
duce AZ0108 (14f), an orally bioavailable, PARP1,2,6 inhibitor that
causes a multi-polar spindle phenotype at double digit nM concen-
trations. Important highlights of the optimization include: (1)
identification of a triazolopiperazine with improved physicochem-
ical properties, (2) chiral methyl substitution on the piperazine to
improve potency, (3) blocking metabolism and improving pharma-
cokinetics by difluorination of the methylene linker, and (4) tuning
the electronics of the triazole moiety to improve potency and oral
bioavailability. Both AZ9482 (1) and AZ0108 (14f) have been uti-
lized as in vitro tools24 and in vivo probes25 to investigate the bio-
logical consequences of inhibiting centrosome clustering through
11. http://www.thesgc.org/structures/2rf5#mandm.
12. Kirby, C. A.; Cheung, A.; Fazal, A.; Shultz, M. D.; Stams, T. Acta Crystallogr., Sect.
F: Struct. Biol. Cryst. Commun. 2012, 68, 115.
13. Narwal, M.; Venkannagari, H.; Lehtiö, L. J. Med. Chem. 2012, 55, 1360.
14. Purser, S.; Moore, P. R.; Swallow, S.; Gouverneur, V. Chem. Soc. Rev. 2008, 37,
320.
15. Mandagere, A. K.; Thompson, T. N.; Hwang, K. K. J. Med. Chem. 2002, 45, 304.
16. Skolnik, S.; Lin, X.; Wang, J.; Chen, X. H.; He, T.; Zhang, B. J. Pharm. Sci. 2010, 99,
3246.
17. Hansch, C.; Leo, A.; Taft, R. W. Chem. Rev. 1991, 91, 165.
18. Kim, D.; Wang, L.; Beconi, M.; Eiermann, G. J.; Fisher, M. H.; He, H.; Hickey, G. J.;
Kowalchick, J. E.; Leiting, B.; Lyons, K.; Marsilio, F.; McCann, M. E.; Patel, R. A.;
Petrov, A.; Scapin, G.; Patel, S. B.; Roy, R. S.; Wu, J. K.; Wyvratt, M. J.; Zhang, B.
B.; Zhu, L.; Thornberry, N. A.; Weber, A. E. J. Med. Chem. 2005, 48, 141.
19. Kim, D.; Kowalchick, J. E.; Edmondson, S. D.; Mastracchio, A.; Xu, J.; Eiermann,
G. J.; Leiting, B.; Wu, J. K.; Pryor, K. D.; Patel, R. A.; He, H.; Lyons, K. A.;
Thornberry, N. A.; Weber, A. E. Bioorg. Med. Chem. Lett. 2007, 17, 3373.
20. Tuncel, H.; Tanaka, S.; Oka, S.; Nakai, S.; Fukutomi, R.; Okamoto, M.; Ota, T.;
Kaneko, H.; Tatsuka, M.; Shimamoto, F. Int. J. Oncol. 2012, 41, 2079.
21. Huang, S. M.; 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.
22. Menear, K. A.; Adcock, C.; Boulter, R.; Cockcroft, X. L.; Copsey, L.; Cranston, A.;
Dillon, K. J.; Drzewiecki, J.; Garman, S.; Gomez, S.; Javaid, H.; Kerrigan, F.;
Knights, C.; Lau, A.; Loh, V. M.; Matthews, I. T.; Moore, S.; O’Connor, M. J.;
Smith, G. C.; Martin, N. M. J. Med. Chem. 2008, 51, 6581.
23. Wang, Z.; Petteruti, P.; Wang, X.; Wang, W.; Gharahdaghi, F.; Wu, J.; Su, N.;
Cheung, T.; Mayo, M.; Scott, D.; Johannes, J.; Lamb, M.; Lawson, D.; Collins, M.;
Nadella, P.; Reimer, C.; Lyne, P.; Zinda, M.; Mikule, K.; Fawell, S.; Grosskurth, S.
Submitted manuscript.
24. Grosskurth, S.; Petteruti, P.; Wang, X.; Mikule, K. Mol. Cancer Ther. 2013, 12,
A134.
25. Mayo, M. F.; Grosskurth, S.; Wang, X.; Petteruti, P.; Nadella, P.; Reimer, C.;
Mikule, K. Mol. Cancer Ther. 2013, 12, A221.