Pharmacological Inhibition of PARP6 Triggers Multipolar Spindle Formation and Elicits Therapeutic Effects in Breast Cancer

Author Manuscript Published OnlineFirst on October 8, 2018; DOI: 10.1158/0008-5472.CAN-18-1362 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

TITLE: Pharmacological inhibition of PARP6 triggers multipolar spindle formation and elicits therapeutic effects in breast cancer

AUTHORS AND AFFILIATIONS:

Zebin Wang1*, Shaun E. Grosskurth1*, Tony Cheung1, Philip Petteruti1, Jingwen Zhang1, Xin Wang1, Wenxian Wang1, Farzin Gharahdaghi1, Jiaquan Wu1, Nancy Su1, Ryan T. Howard2, Michele Mayo1, Dan Widzowski1, David A. Scott1, Jeffrey W. Johannes1, Michelle L. Lamb1, Deborah Lawson1, Jonathan R. Dry1, Paul D. Lyne1, Edward W. Tate2, Michael Zinda1, Keith Mikule1, Stephen E. Fawell1, Corinne Reimer1 and Huawei Chen1

Oncology, IMED Biotech Unit, AstraZeneca R&D Boston, Waltham, MA, 02451, USA1

Institute of Chemical Biology, Department of Chemistry, Imperial College London, London SW7

2AZ, UK2

RUNNING TITLE:

Targeting PARP6 in breast cancer

KEYWORDS:

PARP6; PARP inhibitors; PARylation; ADP‐ribosylation; Chk1, multipolar spindle formation (MPS)

Significance:

Findings describe a new inhibitor of PARP6 and identify a novel function of PARP6 in regulating

activation of Chk1 in breast cancer cells

CORRESPONDING AUTHOR:

Huawei Chen, Oncology, IMED Biotech Unit, AstraZeneca R&D Boston, Waltham, MA, 02451 Tel.

(781)‐839‐4417; Fax. (781)‐839‐4500; E‐Mail: [email protected]

* These authors contributed equally to this work and share first authorship.

POTENTIAL CONFLICTS OF INTEREST:

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2018 American Association for Cancer Research. Author Manuscript Published OnlineFirst on October 8, 2018; DOI: 10.1158/0008-5472.CAN-18-1362 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

R.T. Howard and E. W. Tate are collaborators of AstraZeneca and receive funding through the

EPSRC Centre for Doctoral Training in Chemical Biology (Imperial College London) and the rest of the authors were employees and/or shareholders of AstraZeneca.

ABSTRACT

PARP (poly ADP‐ribose polymerase) proteins represent a class of post‐translational modification enzymes with diverse cellular functions. Targeting PARPs has proven to be efficacious clinically, but exploration of the therapeutic potential of PARP inhibition has been limited to targeting poly(ADP‐ribose) (PAR) generating PARP including PARP1/2/3 and tankyrases. The cancer‐related functions of mono(ADP‐ribose) (MAR) generating PARP, including PARP6, remain largely uncharacterized. Here, we report a novel therapeutic strategy targeting PARP6 using the first reported PARP6 inhibitors. By screening a collection of PARP compounds for their ability to induce mitotic defects, we uncovered a robust correlation between PARP6 inhibition and induction of multipolar spindle (MPS) formation, which was phenocopied by PARP6 knockdown.

Treatment with AZ0108, a PARP6 inhibitor with a favorable pharmacokinetic profile, potently induced the MPS phenotype, leading to apoptosis in a subset of breast cancer cells in vitro and antitumor effects in vivo. In addition, Chk1 was identified as a specific substrate of PARP6 and was further confirmed by enzymatic assays and by mass spectrometry. Furthermore, when modification of Chk1 was inhibited with AZ0108 in breast cancer cells, we observed marked upregulation of p‐S345 Chk1 accompanied by defects in mitotic signaling. Together these results establish proof‐of‐concept antitumor efficacy through PARP6 inhibition and highlight a novel function of PARP6 in maintaining centrosome integrity via direct ADP‐ribosylation of Chk1 and modulation of its activity.

INTRODUCTION

The presence of supernumerary centrosomes is a common feature of solid malignant

tumors, differentiating them from normal cells. Centrosomes normally function as the main

microtubule‐organizing centers (MTOCs) and play an essential role during cellular mitosis [1].

Evidence suggests that centrosome abnormalities contribute to a high degree of aneuploidy in

cancer cells and are associated with advanced tumor grade [2‐4]. A recent report indicates

centrosome amplification is sufficient to promote spontaneous tumorigenesis, although the

typical consequence of multipolar mitosis is often detrimental to dividing cells [5].

To ensure successful mitosis, aneuploid cancer cells can evolve several mechanisms to

avoid chromosome segregation errors [6]. One major mechanism involves clustering and

assembling multiple centrosomes into pseudo‐bipolar spindles through a tightly regulated

process [6, 7]. Disruption of this process results in multipolar mitosis and has been shown to be

a therapeutic strategy to specifically kill cancer cells with amplified centrosomes while sparing

normal cells [8, 9]. The tubulin‐stabilizing agent griseofulvin and its derivative GF‐15, as well as

the HSET/KIFC1 inhibitor AZ82, have been reported to induce formation of multipolar spindles

(MPS) to selectively kill cancer cells [10‐12]. Through unbiased siRNA screens, members of the therapeutically tractable poly(ADP‐ribose) polymerases (PARPs) enzyme family have also been implicated in centrosome clustering and bipolar spindle formation, providing the preliminary evidence to further investigate the pharmacological inhibition of PARPs to specifically perturb mitosis in cancer cells [8, 13].

PARPs are a family of 17 enzymes that catalyze the transfer of the ADP‐ribose from NAD+ to post‐translationally modify acceptor proteins. Depending on their catalytic activity, PARPs can

be further divided into poly(ADP‐ribose) generating PARPs such as PARPs 1‐5, catalytically inactive PARPs such as PARPs 9 and 13, and mono(ADP‐ribose) generating PARPs such as PARPs

6‐8, 10‐12, and 14‐15 [14, 15]. The post‐translational modification of substrate proteins by PARPs has been demonstrated, for example, by PARP1‐3 to regulate numerous signaling cascades including DNA damage response, chromatin remodeling, and transcriptional regulation, as well as by PARP5a‐5b for telomere maintenance, spindle assembly, vesicular movement, and degradation of the beta‐catenin destruction complex [16, 17]. Given that PARPs have functional roles in mitosis and that PARP1/2 inhibition is a successful therapeutic approach to treat homologous recombination defective tumors in the clinic [18, 19], this prompted us to perform a cell‐based MPS phenotypic screen with compounds that have structural similarity to PARP inhibitor scaffolds.

Here, we report the characterization of novel small molecule PARP inhibitors that induce

MPS formation in cancer cells and provide evidence that this phenotype is due to PARP6 inhibition. PARP6 is a mono (ADP‐ribose) generating PARP with little biological characterization.

Pharmacological inhibition of PARP6 in breast cancer cells using AZ0108 [20], an optimized inhibitor developed from phenotypic screen hits, resulted in MPS formation, impaired cell growth, and induction of apoptosis in vitro and in vivo. Using a high‐density protein array‐based ADP‐ ribosylation assay, we identified Chk1 kinase within a subset of proteins involved in regulating centrosome function that were enriched as direct PARP6 substrates. We confirmed PARP6 ADP‐ ribosylates Chk1 directly and demonstrated Chk1 S345 phosphorylation was significantly upregulated upon PARP6 inhibition, which was accompanied by de‐activation of other mitotic proteins during the G2‐M transition. Taken together, our studies demonstrate a critical role for

PARP6 in ensuring the integrity of mitosis and provide preclinical proof‐of‐concept for inhibiting

PARP6 as a novel cancer therapeutic strategy.

MATERIALS AND METHODS

Chemicals and cell lines

AZD2281/olaparib, AZ9482, and AZ0108 were synthesized by AstraZeneca [20‐22] and

diluted in dimethyl sulfoxide (Sigma‐Aldrich). Majority of cell lines were purchased from ATCC or some were obtained from DSMZ and cultured according to providers’ instructions (detailed

information in Supplemental Table S1). Cell lines were generally maintained at low passages

before use.

PARP protein enzymatic assays

All assays were performed following the BPS PARP assay kit protocols. In brief, PARP

enzymatic reactions were conducted in duplicate at room temperature for 1 hour in a 96 well

plate coated with histone substrate, except for TNKS1 and 2 (PARP5a and PARP5b), where GST‐

TNKS1 or 2, instead of histone, was coated on the glutathione plate for auto‐ADP‐ribosylation.

Reaction buffer (50 μL, Tris‐HCl, pH 8.0) containing NAD+, biotinylated NAD+, activated DNA, a

PARP enzyme (Supplemental Table S2 for specific enzyme amounts and NAD+ concentrations)

and the test compound were added to start the reaction. 50 μL of Streptavidin‐horseradish

peroxidase was added to each well and the plate was incubated at room temperature for 30

minutes. After adding 100 μL of developer reagents, luminescence was measured using a BioTek

Synergy™ 2 microplate reader. Percent inhibition were calculated for each compound

concentration with the following equation: (reaction signal – negative control) / (enzyme alone

positive control – negative control). IC50s were calculated by fitting the percentage inhibition to

a four‐parameter logistic regression equation with an AstraZeneca in‐house calculation engine.

Immuno‐fluorescent multipolar spindle assay

Indicated cell lines were plated in 96‐well plates at 7,000 cells per well and incubated at 37°C

overnight. The cells were treated with compounds ranging from 0 to 11 M for 48 hours. The cells were fixed with 4% formaldehyde at room temperature for 10 minutes followed with ice‐ cold methanol fixation for another 10 minutes. After four washes in PBS, the cells were incubated in blocking buffer for 1 hour at room temperature to reduce non‐specific binding. The cells were labelled with primary antibodies, 1:2000 dilution of anti‐cyclin B antibody (Thermo Fisher) and

1:4000 dilution of anti‐pericentrin antibody (Abcam), for 16 hours at 4°C. After washing with PBS four times, the cells were labelled with secondary antibodies, 1:200 Alexa Fluor 488 anti‐rabbit antibody and Alexa Fluor 594 anti‐mouse antibody, for 1 hour at room temperature. After washing with PBS twice, the cells were stained with Hoechst dye for 10 minutes at room temperature. The cells were washed twice with PBS and then images were acquired by

ImageXpress Micro (Molecular Devices) or Operatte (PerkinElmer) High Content Screening

System. Cyclin B label was used for scoring the mitotic cells and pericentrin was used for scoring the centrosome number in each mitotic cell. The percent increase of mitotic cells with greater than 2 centrosomes compared to the DMSO control was used to calculate EC50 values. For CHK1

and pericentrin staining, CHK1 antibody (1:100, Cell signaling) and pericentrin (Abcam) antibodies

were used to stain HCC1806 cells after AZ0108 treatment following the same protocol. Images

were taken with Operetta (PelkinElmer) High Content Screening System, and CHK1 intensity was

analyzed with Harmony analysis software (PerkinElmer).

Breast cancer cell line proliferation assays

Cell lines were sourced as previously described [23, 24] and were cultured in RPMI 1640

medium supplemented with 10% FBS (GIBCO), 2 mM L‐glutamine, 100 units/mL penicillin and 0.1

mg/mL streptomycin at 37°C in 5% CO2. Cells were treated with compounds diluted in DMSO.

Cell proliferation was determined by two methods, MTS and Sytox Green assay. Briefly, cells were

seeded in 96‐well plates (at a density to allow for logarithmic growth during the 72‐hour assay)

and incubated overnight at 37°C, 5% CO2. Cells were then exposed to concentrations of PARP inhibitors ranging from 0.003 to 30 μM for 72 hours. For the MTS endpoint, cell proliferation was measured by the CellTiter AQueous Non‐Radioactive Cell Proliferation Assay (Promega) reagent in accordance with the manufacturer's protocol. Absorbance was measured with a Tecan Ultra instrument. For the Sytox Green endpoint, Sytox Green nucleic acid dye (Invitrogen) diluted in

TBS‐EDTA buffer was added to cells (final concentration of 0.13 μM) and the number of dead cells was detected using an Acumen Explorer. Cells were then permeabilized by the addition of saponin (0.03% final concentration, diluted in TBS‐EDTA buffer), incubated overnight, and a total cell count was measured. Predose measurements were made for both MTS and Sytox Green endpoints, and the concentration needed to reduce 50% of the growth of treated cells relative to the untreated cells (GI50) were determined using absorbance readings (MTS) or live cell counts

(Sytox Green). All cell line panel MTS results are summarized in Supplemental Table S3.

In vivo efficacy study

Female C.B.‐17 SCID (severe combined immunodeficient) mice were purchased from

Charles River Laboratories (Wilmington, MA). Mice were housed under pathogen‐free conditions

in individual ventilated cages (IVC) at our AAALAC (Association for the Assessment and

Accreditation of Laboratory Animal Care) accredited facility in Waltham, MA. Animal studies were

approved by the internal IACUC (Institutional Animal Care and Use Committee) and reported

following the ARRIVE (Animal Research: Reporting In Vivo experiments) guidelines [25].

MDA‐MB‐468 (10x 10^6) or HCC1806 (2x 10^6) tumor cells in serum‐free medium with matrigel (1:1 ratio) were injected subcutaneously in the right flank of 5‐6 week old mice in a volume of 0.1 mL. Tumor volumes (measured by caliper), animal body weight, and tumor condition were recorded twice weekly for the duration of the study. The tumor volume was calculated using the formula: length (mm) x width (mm)2/0.52. Mice were randomized into control and treatment groups based on tumor volumes using stratified sampling. When the mean tumor size reached 225 mm3, treatment was initiated via oral gavage with either vehicle (15% captisol) or AZ0108 10 mg/kg daily for 4 weeks. Growth inhibition was assessed by comparison of the differences in tumor volume between control and treated groups. Data were log transformed to remove any size dependency before statistical evaluation. Statistical significance was evaluated using a one‐tailed, 2‐sample t test.

ProtoArray screen for ADP‐ribosylation substrates

ProtoArray version 5.0 (Life Technologies Termo Fisher) are a human protein microarray with 9,000+ full‐length human proteins printed in tandem duplicates plus controls. ADP‐ ribosylation experiments were performed by Life Technologies as follows. First, arrays were blocked in BPS BioSci buffer (#80602 BPS Bioscience) and BSA. Then, the arrays were incubated in buffer, NAD/biotinylated‐NAD mixture, +/‐ 250 nM PARP1 or 125‐250 nM PARP6 enzyme

(#80501 and #80506 BPS Bioscience), +/‐ 0.8 M AZ0108, and activated DNA. Arrays were

washed; ADP‐ribosylation from biotinylated‐NAD was identified with Alexa Fluor 647‐conjugated

Streptavidin. Data acquisition was performed with a GenePix Pro 6.0 by Life Technologies and

mapping alignment was performed with Alexa Anti Mouse Fluor Antibodies (8 positive control

spots in tandem duplicates per block). Negative control spots include enzyme buffer and BSA (12

tandem duplicate spots per block).

Statistics were performed in R (http://www.R‐project.org) with additional visualizations

generated in Spotfire (TIBCO Software). First, Z‐scores were calculated with the raw signals as in other ProtoArray ADP‐ribosylation experiments [26, 27]. Note the high Z‐score values in the negative arrays were from PCCA (propionyl Coenzyme A carboxylase, alpha), a protein that has a biotin carboxylation domain and a biotinyl‐binding domain. To identify potential differences between negative arrays and various PARP array experiments, Mann‐Whitney U tests were performed and p‐values were calculated. In addition, logistic regression analysis was performed between the various individual PARP and negative arrays to identify spots that correlated to a specific PARP enzyme. For DAVID Functional Annotation Clustering, the on‐line tools

(http://david.abcc.ncifcrf.gov/) were used to perform functional enrichment analysis on the 230

unique PARP6 substrates [28, 29]. Functional clusters with enrichment scores >1.3 were

considered significantly over‐represented for PARP6 ADP‐ribosylation. All ProtoArray results are

provided in Supplemental Table S4.

In vitro ADP‐ribosylation assay

Recombinant GST‐PARP6 protein was produced in sf21 cells and purified. For the Chk1 in

vitro ADP‐ribosylation assay, 70 nM GST‐PARP6, 250 nM GST‐Chk1 (BPS Bioscience Cat#400439)

or 100 nM BSA (Biorad Cat#5000207), and 10M biotin labeled NAD+ (BPS Bioscience Cat#80610)

were diluted in 250 mM Tris‐HCl (pH 8.0), 15mM MgCl2, 5mM DTT, and 0.05% Triton X‐100.

Reactions were incubated at 30oC for 3 hours, terminated by adding SDS sample buffer and separated by Bis‐Tris gel (Life Technologies). The biotin signal was detected by using HRP

(horseradish peroxidase)‐conjugated streptavidin (CST Cat#3999) and visualized using PIERCE

SuperSignal West Duro Extend Duration Substrate (cat# 34076).

RESULTS

PARP6 suppression is responsible for inducing MPS formation and centrosome defects

The recent clinical success of the PARP inhibitors olaparib (AZD2281), niraparib, and rucaparib highlights the utility of targeting the PARP enzyme family for treating human malignancies [30‐32]. To further explore the therapeutic potential of inhibiting the PARP family of proteins, we performed a phenotypic cell‐based compound screen using a subset of

AstraZeneca’s collection of PARP inhibitors, predominantly from the phthalazinone series. We chose to evaluate mitotic defects since the biological function of certain PARPs had been implicated in mitosis [33, 34]. The goal of the screen was to identify compounds that could induce the formation of MPS, a cytotoxic phenotype for cancer cells [6, 11]. We established a quantitative immunofluorescent assay in HeLa cells to measure MPS induction based on the percent increase in mitotic cells with increased centrosome number (≥ 4 centrosomes per cell), and an EC50 was determined for all compounds in this subset. One of the initial hits, compound

AZ9482, demonstrated strong MPS induction activity with excellent potency (Fig. 1A‐B).

Subsequent optimization led to AZ0108 with improved physical properties and selectivity [20].

AZ0108 potently induced the MPS phenotype with an EC50 of 46 nM (Fig. 1C‐E). In contrast,

PARP1/2 inhibitor AZD2281 was largely inactive in the same assay even at a high concentration of 3.7 µM (Fig. 1E). This result suggested that inhibition of other PARPs besides PARP1/2 may be contributing to the MPS phenotype. To identify the accountable PARP(s), selectivity profiling for

AZ0108 was performed in all available PARP enzymatic assays and the data demonstrated that in addition to having PARP1/2 activity, AZ0108 also potently inhibited PARP6 (Supplemental Fig. S1)

[20]. Furthermore, we observed a strong correlation (R2 = 0.76) between MPS induction potency and PARP6 inhibitory activity for all compounds from the phthalazinone series tested (Fig. 1F). In contrast, no such correlation was observed for other PARP enzymes we examined, including

PARP1‐3, TNKS1, and TNKS2 (Supplemental Fig. S2). These observations suggest PARP6 inhibition may be accountable for MPS induction. Consistent with our expectation, PARP6 knockdown induced MPS formation, closely phenocopying the inhibitory effect of PARP6 inhibitors (Fig. 1G‐

H). Taken together, our data establishes the MPS phenotype induced by these PARP compounds is through PARP6 inhibition and suggests a role for PARP6 in maintaining centrosome integrity.

PARP6 inhibition impairs proliferation and induces apoptosis in breast cancer cell lines

To evaluate the antitumor potential of the PARP6 inhibitor AZ0108, growth inhibition was examined in a panel of breast cancer cell lines since centrosome abnormalities have been previously described in malignant breast tissues [35]. Designating GI50 < 1M as an arbitrary cut‐ off for defining sensitivity, the waterfall plot for AZ0108 clearly indicates that it is not pan‐ antiproliferative across the entire panel, as a number of breast cancer lines are insensitive to treatment (Fig. 2A, upper panel). The sensitivity profile of AZ0108 differs drastically from that of

PARP1/2 inhibitor AZD2281 which is inactive for all 18 lines that we have data on (Fig. 2A, lower panel). This observation suggests a unique dependency for PARP6 in select breast cancer cell lines.

Perturbations causing centrosome abnormalities have been previously shown to induce apoptosis in cancer cells [11] and we speculated that PARP6 inhibition would induce apoptosis to drive antitumor activity. To further evaluate the phenotypic consequence of PARP6 inhibition, the induction of apoptosis by AZ0108 in two sensitive cell lines, HCC1806 and MDA‐MB‐468, was followed longitudinally by monitoring the increase of caspase signal. AZ0108 induced strong apoptosis in both cell lines after ~2 days (Fig. 2B). The PARP6 dependency in AZ0108 sensitive breast cancer cell lines was further examined using PARP6 shRNA and siRNA. Consistent with the effect of AZ0108 treatment, PARP6 knockdown impaired the cell viability of HCC1806 and MDA‐

MB‐468 cancer lines (Fig. 2C). It is worth noting that, similar to what we have observed previously in HeLa cells, PARP6 suppression either by AZ0108 treatment or by siRNA mediated knockdown efficiently induced MPS formation in HCC1806 cells (Fig. 2D‐E). When additional breast cancer cell lines were examined, marked induction of MPS was observed in BT‐549 and MDA‐MB‐157 breast cancer cells, which showed moderate to high sensitivity to AZ0108 (Figure 2A,

Supplemental Fig. S3). In contrast, minimal increase of mitotic cells with MPS was seen in MCF7 cells following AZ0108 treatment, consistent with insensitivity of this cell line to AZ0108

(Supplemental Fig. S3). The MPS phenotype was also not observed with PARP1/2 inhibitor

AZD2281 or PARP1 knockdown (Supplemental Fig. S4), nor with AZ0108 in the immortalized noncancerous mammary gland epithelial cell line MCF10A (Supplemental Fig. S5). Altogether, these results suggest there is potential utility of a PARP6 inhibitor for MPS induction and apoptotic cell killing in breast cancer cells.

In an effort to identify potential biomarkers associated with AZ0108 sensitivity (or insensitivity), given that clinically relevant genetic alterations or subtypes in breast cancer did not

predict response, we found that high expression and genetic amplification of centrosome‐related proteins KIAA1429 (CENP‐A centromere complex binding protein) [36] and SCYL1/TEIF

(centrosomal linking protein) [37] appeared to be linked to insensitivity (Supplemental Table S5).

While the associations are circumstantial on their own, intriguingly we identified SCYL1 as a substrate for PARP6 from ProtoArray studies (see below; Supplemental Table S4). Further work is needed to establish if these are bona fide insensitivity biomarkers and to define the molecular mechanism of how their elevated expression levels provide protection from PARP6 inhibition.

PARP6 inhibitor AZ0108 induces centrosome defects and demonstrates antitumor activity in breast cancer xenograft models

To evaluate the antitumor activity in vivo, we explored the effects of dosing AZ0108 in

MDA‐MB‐468 and HCC1806 breast cancer xenograft models. Significant antitumor activity (91%

TGI (tumor growth inhibition), p < 0.01) was observed in MDA‐MB‐468 model with daily oral dosing of 10 mg/kg of AZ0108, whereas the same treatment regimen gave lesser activity in

HCC1806 xenograft tumors (32% TGI, p < 0.05) (Fig. 3A). It is noted that HCC1806 is an aggressive tumor with a fast tumor doubling time. Treatment with AZ0108 was well tolerated with little body weight loss when compared with vehicle control animals (Fig. 3A).

To investigate whether antitumor activity of AZ0108 was associated with the induction of mitotic defects in vivo, samples of MDA‐MB‐468 and HCC1806 xenografts tumors were collected

48 hours after a single dose of 10 mg/kg AZ0108 and examined by ‐tubulin immunohistochemistry for mitosis effects. Treated tumor samples exhibited an increase in number of cells showing mitotic defects, including MPS, disorganized spindles, and mitotic failure

(Fig. 3B). The percent change of aberrant mitotic nuclei with AZ0108 treatment compared to

vehicle control was rather remarkable, increasing from 8% to 32% for MDA‐MB‐468 tumors and

from 23% to 63% for HCC1806 tumors (Fig. 3C).

Identification of Chk1 as a physiological substrate of PARP6

To gain molecular insights into MPS induction by PARP6 inhibition, we sought to identify

direct substrates of PARP6. Currently, there is no known PARP6 substrate reported. In addition,

very few approaches are effective in identifying substrates of MAR‐generating PARPs such as

PARP6 due to the lack of a suitable MAR‐specific antibody. Hence, we decided to use high density

protein microarrays (ProtoArray) in conjunction with biotinylated NAD+ as a donor substrate for

Streptavidin‐based detection of novel ADP‐ribosylation substrates, bypassing the requirement of

a MAR‐specific antibody [27]. Protein microarrays spotted with 9,000+ full‐length human

proteins were incubated with full‐length PARP6 and the ADP‐ribosylated proteins were identified

with Alexa Fluor 647‐conjugated Streptavidin (Z‐score difference >2.5 between the intensities of

PARP6 treated arrays and control, Supplemental Table S4). DAVID functional enrichment analysis

(https://david.ncifcrf.gov) of the ADP‐ribosylated proteins shows a statistically significant over‐

representation for kinases, cytoskeletal proteins, proteins at cell junctions, and cell cycle proteins

(Table 1, showing the first biological sub‐cluster for each significant cluster). Of particular interest, a biological sub‐cluster under cytoskeletal proteins involved in regulating centrosome function is significantly enriched, which may be proteins of interest accountable for the MPS phenotype induced by PARP6 inhibition.

Since AZ0108 also inhibits PARP1/2, a PARP1 ProtoArray experiment was performed in parallel to identify PARP1 substrates. A similar number of proteins, 194 and 204, were found to be PARylated by PARP1 or ADP‐ribosylated by PARP6, respectively. However, there is little

overlap of the putative substrates identified in this study between PARP6 and PARP1, or with substrates that have been reported for other PARP proteins including PARP2, PARP10, and

PARP14 using the same ProtoArray technology (Fig. 4A, Supplemental Table S4) [26, 27].

Among the enriched centrosomal proteins, Chk1 was prioritized for subsequent analysis since it was uniquely identified through the PARP6 ProtoArray and has also been shown to prevent premature activation of the cyclin‐B‐Cdk1 complex in initiating mitosis [23], in addition to its role in DNA damage induced centrosome amplification [38]. The level of Chk1 ADP‐ ribosylation was diminished when the amount of PARP6 protein used in the ProtoArray experiment was reduced, and was nearly abolished when AZ0108 was included, suggesting a specific PARP6‐medidated ADP‐ribosylation of Chk1 (Fig. 4B, upper panel). In comparison, SFRS1, a previously reported PARP1 substrate, was PARylated by PARP1 but not by PARP6, suggesting

ADP‐ribosylation in this ProtoArray experiment is specific (Fig. 4B, lower panel). The PARP6 mediated ADP‐ribosylation of Chk1 was further confirmed by an in vitro enzymatic assay where

PARP6 effectively used GST‐Chk1 as an acceptor substrate in a ADP‐ribosylation reaction with biotinylated NAD+ as the donor substrate (Fig. 4C). Again, the modification of Chk1 was reduced when AZ0108 was included in the enzymatic assay.

To determine whether Chk1 is ADP‐ribosylated by PARP6 intracellularly, macrodomain mAf1521 pulldown followed by mass spectrometry analysis was performed on cell lysates prepared from MDA‐MB‐468 cells under different conditions, i.e. DMSO control, nocodazole treatment (microtubule inhibitor to induce prometaphase arrest), and co‐treatment of nocodazole with AZ0108. Since no antibody is currently available to effectively recognize MAR post‐translational modification, the macrodomain mAf1521 was used as a tool to pulldown

mono‐ and poly‐ADP‐ribose unit(s) containing proteins [39]. Gel electrophoresis coupled with silver staining of the macrodomain‐pulldown material identified a clear difference in staining intensity within the 50‐75 kDa region (band highlighted in Fig. 4D, upper panel) between the different treatment conditions, showing increased staining with nocodazole treatment (Fig. 4D, compare Lane 2 vs. 1) that was reversed with AZ0108 co‐treatment (Fig. 4D, compare Lane 3 vs.

2). Mass spectrometry analyses were performed on the highlighted band excised from the gel and a handful of protein candidates including Chk1 were identified (Fig. 4D, lower panel).

Consistent with our expectation, Chk1 was only present in the band under the treatment condition with nocodazole alone. Since the samples were enriched for ADP‐ribosylated proteins with macrodomain pulldown, these results indicated that Chk1 is ADP‐ribosylated when cell cycle was perturbed to induce prometaphase arrest and this ADP‐ribosylation was prevented with

AZ0108 treatment, suggesting that Chk1 ADP‐ribosylation in cells is likely mediated by PARP6. In summary, the results from ProtoArray screening, biochemical confirmation using isolated proteins, and cellular investigation through mass spectrometry analyses provide solid evidence that Chk1 is a physiological substrate of PARP6.

PARP6 inhibition by AZ0108 results in elevated p‐S345 Chk1 and reduced mitotic signaling in vitro and in vivo

Previous reports noted that at interphase, Chk1 is able to prevent premature activation of the cyclin‐B‐Cdk1 complex, and overexpression of centrosome‐targeted Chk1 prohibited centrosome separation as well as inducing polyploidization [23, 40]. Given the importance of

Chk1 function in regulating mitosis and our findings that Chk1 is a substrate of PARP6, we set out

to investigate the effects of PARP6 inhibition on the G2‐M phase transition in HCC1806 cells following sequestration at prometaphase using nocodazole. Activation of mitosis‐related proteins such as FoxM1, Aurora kinases B and C, and Histone H3 were observed when cells were entering mitosis (the 0 hour time point for phosphorylated proteins, and in the case of FoxM1, the slower migrating band corresponding to the phosphorylated, active form), all of which are known to be phosphorylated (i.e. activated) during the G2‐M transition (Fig. 5A, under DMSO).

Interestingly, when AZ0108 was co‐treated with nocodazole marked activation of Chk1

(upregulated p‐S345 Chk1 level) and deactivation of FoxM1, Aurora kinases, and Histone H3, were observed (Fig. 5A, under AZ0108). However, a similar effect was not observed with PARP1/2 inhibitor AZD2281 (Fig. 5A, under AZD2281). Taken together, these data strongly suggest that

Chk1 activation is regulated by PARP6 via ADP‐ribosylation during mitosis and is accompanied by inhibition of activated mitotic signaling.

To examine the impact of PARP6 inhibition on Chk1 activation under endogenous settings without cell cycle perturbation, MDA‐MB‐468 and HCC1806 breast cancer cells were treated with

AZ0108 and levels of phosphorylated and total Chk1 were evaluated (Fig. 5B). AZ0108 treatment resulted in dose‐dependent increase of p‐S345 Chk1 without altering the total protein expression in both cell lines. Profound induction was achieved with 300 nM AZ0108 at 24 hours with no further increase observed when incubation time was extended to 48 hours. Consistent with our previous observations in HCC1806 cells under nocodazole perturbation conditions (Fig. 5A), p‐

S10 histone H3 was reduced with AZ0108 treatment in a concentration‐dependent manner in both cell lines (Fig. 5B). Considering the role of Chk1 in G2/M checkpoint and centrosome maintenance, we further examined if AZ0108 treatment leads to accumulation of Chk1 to the

centrosomes. Co‐staining of CHK1 with centrosome marker pericentin revealed that increasing amount of Chk1 is localized to the centrosome following AZ0108 treatment, while the overall intensity of Chk1 within the cells did not change (Supplemental Fig S6). These observations provide further evidence that ADP‐ribosylation mediated by PARP6 affects the activation and localization of Chk1 in cancer cells.

To extend the impact of Chk1 phosphorylation findings in vivo, tumors were examined for pharmacodynamic effects on nuclear p‐S345 Chk1 by immunohistochemistry following a single oral administration of 10mg/kg AZ0108 to female C.B.‐17 SCID mice bearing HCC1806 sub‐ cutaneous xenografts. Staining for nuclear p‐S345 Chk1 was significantly higher for AZ0108‐ treated tumors relative to vehicle controls by immunohistochemistry (Fig. 5C). When quantified, an approximately six‐fold increase of nuclear p‐Chk1 staining was observed in the treatment group at 6 hours (Fig. 5D). The increase of p‐Chk1 staining was still evident at 24 h but largely dissipated when the drug was cleared from the system (48h). Taken together, learnings from these molecular investigations provide a mechanistic description of PARP6 regulating Chk1 activation in mitosis and may account for the MPS phenotype associated with PARP6 inhibition from AZ0108 treatment.

Discussion

We uncovered PARP6 as a molecular target for a subset of PARP compounds identified through a phenotypic screen for MPS formation in cancer cells. Initial evidence came from the observation of a robust correlation between potency of PARP6 enzyme inhibition and activity in

MPS induction. This correlation was associated with PARP6 but not with other PARP enzymes

examined, including PARP1, PARP2, PARP3, TNKS1, and TNKS2. We further supported this finding

with PARP6 knockdown, which phenocopied the induction of MPS formation in cancer cell lines.

We further characterized AZ0108, the first potent and orally available PARP6 inhibitor, identified through optimization of phenotypic screening hits. While AZ0108 also carries PARP1/2 activity, several lines of evidence support that AZ0108 is still a suitable PARP6 probe compound.

Firstly, AZ0108 resides nearly on the regression line on the correlation plot between PARP6 inhibition and MPS induction, suggesting PARP6 inhibition is a driver for the MPS phenotype.

Secondly, PARP1/2 inhibitor AZD2281 (olaparib) which lacks PARP6 activity does not induce MPS formation even at 3.7 µM concentration, compared to the significant MPS effect with AZ0108 at concentrations as low as 0.046 µM. Although there is a report that a PARP1/2 inhibitor can induce the MPS phenotype when used at a high concentration [35], we believe this is likely to be an off‐ target effect. This notion is further corroborated by the observation that PARP1 knockdown failed to induce the MPS phenotype. Lastly, similar to AZ0108, PARP6 knockdown led to impaired viablility in cancer cells. Thus, AZ0108 is a suitable small molecule probe to investigate the phenotypic consequence of MPS induction in cancer models, allowing further evaluation of

PARP6 as a potential therapeutic target.

AZ0108 has a distributive antitumor response in a panel of breast cancer lines, with activity associated with strong induction of the MPS phenotype and cancer cell apoptosis. The sensitivity profile of AZ0108 is differentiated from that of PARP1/2 inhibitor AZD2281, further underscoring the different phenotypic consequences of inhibiting PARP6 or PARP1/2. The

antitumor activity of AZ0108 is translatable in vivo, with efficacy observed in two breast cancer

xenograft models, MDA‐MB‐468 and HCC1806. The treated tumors exhibited a statistically

significant increase in aberrant mitotic cells. While these observations are encouraging, several gaps remain. Additional work around biomarkers such as the centrosome‐related proteins

KIAA1429/CENP‐A and SCYL1/TEIF that we found to be associated with breast cancer cell line insensitivity to AZ0108 would be important to identify potential models and patients that would most likely benefit from PARP6 inhibitors. Additional pharmacological investigation regarding the extent and duration of PARP6 inhibition necessary for optimal MPS induction and cancer cell apoptosis, as well as rational combinations to augment antitumor activity, would also be beneficial. Toxicology studies will be required to determine the potential for mechanism‐based toxicity in normal tissues. Lastly, further chemistry efforts to design an even more selective

PARP6 inhibitor would be desirable for realizing optimal therapeutic benefit while minimizing any off‐target toxicity.

The other notable learning from our study is the identification of Chk1 as a PARP6 substrate, with its activation putatively modulated through PARP6‐mediated ADP‐ribosylation.

Chk1 plays prominent roles in mediating DNA damage response signaling and regulating cell cycle checkpoints during the S, G2, and M phases of the cell cycle. Chk1 activation is tightly regulated for proper M phase entry and progression, and inhibition of Chk1 has been shown to lead to various mitotic abnormalities and kinetochore defects [23, 24]. Initially identified from

ProtoArray, Chk1 was further confirmed as a PARP6 substrate by a biochemical ADP‐ribosylation assay using isolated recombinant PARP6 and Chk1 proteins. Additional support was provided by indirect evidence from a cellular experiment in which the ADP‐ribosylation of Chk1, induced by mitotic arrest, was reversed by PARP6 inhibitor AZ0108. In breast cancer models both under in vitro and in vivo conditions, AZ0108 treatment induces robust phosphorylation of S345 of Chk1,

a well‐established Chk1 activation biomarker associated cell cycle checkpoints and mitotic defects including centrosome amplification[38], and de‐activation of other key mitosis proteins including FoxM1, histone H3, and Aurora kinases. These observations support a novel mechanism of PARP6 regulating Chk1 activation in mitosis, and perhaps in other phases of cell cycle as well, via ADP‐ribosylation and thus provide a mechanistic basis for MPS induction by PARP6 inhibition.

It is possible that deregulated Chk1 phosphorylation induced by PARP6 inhibition alters the Chk1‐

CDK1 pathway during mitosis and therefore affects centrosome clustering. Interestingly, Chk1 activity was reported to be involved in correcting merotelic kinetochore attachment to ensure proper spindle checkpoint signaling by regulating Aurora‐B, MCAK, Kif2b and Hee1 [41], which also suggest that Chk1 activity could directly impact the centrosome clustering process. This notion is consistent with the observation that the expression level of KIAA1429/CENP‐A, one of the key drivers of centrosome clustering, is a negative correlate for AZ0108 sensitivity. However, the association of Chk1 activation with centrosome amplification would suggest another possible mechanism for PARP6‐induced MPS phenotype [38]. Further mechanistic studies are required to define the contribution of Chk1‐mediated defective centrosome clustering process, aberrant centrosome amplification, and perhaps other mechanisms, to MPS formation when PARP6 is inhibited. It is unclear whether the interplay between Chk1 phosphorylation and ADP‐ ribosylation results from direct competition for the site of modification or if ADP‐ribosylation creates a structural hindrance that interferes with the interaction of Chk1 with its modifying kinase and phosphatase. Additional mechanistic work is warranted to further delineate the molecular mechanism of how PARP6‐catalyzed Chk1 ADP‐ribosylation affects its phosphorylation.

Collectively, our findings provide strong evidence supporting a novel therapeutic strategy by inhibiting PARP6. We identified and characterized the first small molecule inhibitor for PARP6,

AZ0108. Both compound and genetic inhibition of PARP6 function caused mitotic defects in breast cancer cells. PARP6 was shown to directly ADP‐ribosylate proteins that function as kinases, cytoskeletal proteins, and centrosomal proteins such as Chk1, providing a mechanistic basis for the mitotic defects observed in breast cancer cells with AZ0108.

ACKNOWLEDGEMENTS

R.T.H. and E.T. thank AstraZeneca and the Engineering and Physical Sciences Research Council of the UK for support through the Imperial College Centre for Doctoral Training (grant

EP/L015498/1). We thank Jane Cheng for in vivo technical assistance and the anonymous reviewers for insightful comments.

References

1. Bornens, M., Centrosome composition and microtubule anchoring mechanisms. Curr Opin Cell Biol, 2002. 14(1): p. 25-34. 2. Fukasawa, K., Centrosome amplification, chromosome instability and cancer development. Cancer Lett, 2005. 230(1): p. 6-19. 3. Fukasawa, K., Oncogenes and tumour suppressors take on centrosomes. Nat Rev Cancer, 2007. 7(12): p. 911-24. 4. D'Assoro, A.B., W.L. Lingle, and J.L. Salisbury, Centrosome amplification and the development of cancer. Oncogene, 2002. 21(40): p. 6146-53. 5. Levine, M.S., et al., Centrosome Amplification Is Sufficient to Promote Spontaneous Tumorigenesis in Mammals. Dev Cell, 2017. 40(3): p. 313-322.e5. 6. Godinho, S.A. and D. Pellman, Causes and consequences of centrosome abnormalities in cancer. Philos Trans R Soc Lond B Biol Sci, 2014. 369(1650). 7. Quintyne, N.J., et al., Spindle multipolarity is prevented by centrosomal clustering. Science, 2005. 307(5706): p. 127-9. 8. Kwon, M., et al., Mechanisms to suppress multipolar divisions in cancer cells with extra centrosomes. Genes Dev, 2008. 22(16): p. 2189-203. 9. Fielding, A.B., et al., A critical role of integrin-linked kinase, ch-TOG and TACC3 in centrosome clustering in cancer cells. Oncogene, 2011. 30(5): p. 521-34.

10. Pannu, V., et al., Centrosome-declustering drugs mediate a two-pronged attack on interphase and mitosis in supercentrosomal cancer cells. Cell Death Dis, 2014. 5: p. e1538. 11. Raab, M.S., et al., GF-15, a novel inhibitor of centrosomal clustering, suppresses tumor cell growth in vitro and in vivo. Cancer Res, 2012. 72(20): p. 5374-85. 12. Wu, J., et al., Discovery and mechanistic study of a small molecule inhibitor for motor protein KIFC1. ACS Chem Biol, 2013. 8(10): p. 2201-8. 13. Chang, P., M. Coughlin, and T.J. Mitchison, Tankyrase-1 polymerization of poly(ADP- ribose) is required for spindle structure and function. Nat Cell Biol, 2005. 7(11): p. 1133- 9. 14. Gibson, B.A. and W.L. Kraus, New insights into the molecular and cellular functions of poly(ADP-ribose) and PARPs. Nat Rev Mol Cell Biol, 2012. 13(7): p. 411-24. 15. Hottiger, M.O., et al., Toward a unified nomenclature for mammalian ADP- ribosyltransferases. Trends Biochem Sci, 2010. 35(4): p. 208-19. 16. Vyas, S. and P. Chang, New PARP targets for cancer therapy. Nat Rev Cancer, 2014. 14(7): p. 502-9. 17. Schreiber, V., et al., Poly(ADP-ribose): novel functions for an old molecule. Nat Rev Mol Cell Biol, 2006. 7(7): p. 517-28. 18. Helleday, T., H.E. Bryant, and N. Schultz, Poly(ADP-ribose) polymerase (PARP-1) in homologous recombination and as a target for cancer therapy. Cell Cycle, 2005. 4(9): p. 1176-8. 19. Farmer, H., et al., Targeting the DNA repair defect in BRCA mutant cells as a therapeutic strategy. Nature, 2005. 434(7035): p. 917-21. 20. Johannes, J.W., et al., Discovery of AZ0108, an orally bioavailable phthalazinone PARP inhibitor that blocks centrosome clustering. Bioorg Med Chem Lett, 2015. 25(24): p. 5743- 7. 21. Menear, K.A., et al., 4-[3-(4-cyclopropanecarbonylpiperazine-1-carbonyl)-4- fluorobenzyl]-2H-phthalazin- 1-one: a novel bioavailable inhibitor of poly(ADP-ribose) polymerase-1. J Med Chem, 2008. 51(20): p. 6581-91. 22. Menear, K.A., et al., Novel alkoxybenzamide inhibitors of poly(ADP-ribose) polymerase. Bioorg Med Chem Lett, 2008. 18(14): p. 3942-5. 23. Kramer, A., et al., Centrosome-associated Chk1 prevents premature activation of cyclin- B-Cdk1 kinase. Nat Cell Biol, 2004. 6(9): p. 884-91. 24. Wilsker, D., et al., Essential function of Chk1 can be uncoupled from DNA damage checkpoint and replication control. Proc Natl Acad Sci U S A, 2008. 105(52): p. 20752-7. 25. Kilkenny, C., et al., Improving bioscience research reporting: The ARRIVE guidelines for reporting animal research. J Pharmacol Pharmacother, 2010. 1(2): p. 94-9. 26. Troiani, S., et al., Identification of candidate substrates for poly(ADP-ribose) polymerase- 2 (PARP2) in the absence of DNA damage using high-density protein microarrays. FEBS J, 2011. 278(19): p. 3676-87. 27. Feijs, K.L., et al., ARTD10 substrate identification on protein microarrays: regulation of GSK3beta by mono-ADP-ribosylation. Cell Commun Signal, 2013. 11(1): p. 5. 28. Huang da, W., B.T. Sherman, and R.A. Lempicki, Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc, 2009. 4(1): p. 44-57. 29. Huang da, W., B.T. Sherman, and R.A. Lempicki, Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res, 2009. 37(1): p. 1-13.

30. Kaufman, B., et al., Olaparib monotherapy in patients with advanced cancer and a germline BRCA1/2 mutation. J Clin Oncol, 2015. 33(3): p. 244-50. 31. Mirza, M.R., et al., Niraparib Maintenance Therapy in Platinum-Sensitive, Recurrent Ovarian Cancer. N Engl J Med, 2016. 375(22): p. 2154-2164. 32. Swisher, E.M., et al., Rucaparib in relapsed, platinum-sensitive high-grade ovarian carcinoma (ARIEL2 Part 1): an international, multicentre, open-label, phase 2 trial. Lancet Oncol, 2017. 18(1): p. 75-87. 33. Chang, P., M. Coughlin, and T.J. Mitchison, Interaction between Poly(ADP-ribose) and NuMA contributes to mitotic spindle pole assembly. Mol Biol Cell, 2009. 20(21): p. 4575- 85. 34. Perdoni, F., et al., Distribution of centromeric proteins and PARP-1 during mitosis and apoptosis. Ann N Y Acad Sci, 2009. 1171: p. 32-7. 35. Lingle, W.L., et al., Centrosome hypertrophy in human breast tumors: implications for genomic stability and cell polarity. Proc Natl Acad Sci U S A, 1998. 95(6): p. 2950-5. 36. Obuse, C., et al., Proteomics analysis of the centromere complex from HeLa interphase cells: UV-damaged DNA binding protein 1 (DDB-1) is a component of the CEN-complex, while BMI-1 is transiently co-localized with the centromeric region in interphase. Genes Cells, 2004. 9(2): p. 105-20. 37. Gong, Y., et al., Localization of TEIF in the centrosome and its functional association with centrosome amplification in DNA damage, telomere dysfunction and human cancers. Oncogene, 2009. 28(12): p. 1549-60. 38. Bourke, E., et al., DNA damage induces Chk1-dependent centrosome amplification. EMBO Rep, 2007. 8(6): p. 603-9. 39. Jungmichel, S., et al., Proteome-wide identification of poly(ADP-Ribosyl)ation targets in different genotoxic stress responses. Mol Cell, 2013. 52(2): p. 272-85. 40. Loffler, H., et al., Chk1-dependent regulation of Cdc25B functions to coordinate mitotic events. Cell Cycle, 2006. 5(21): p. 2543-7. 41. Petsalaki, E. and G. Zachos, Chk1 and Mps1 jointly regulate correction of merotelic kinetochore attachments. J Cell Sci, 2013. 126(Pt 5): p. 1235-46.

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2018 American Association for Cancer Research. Downloaded from Table 1. Summary of the top 15 biological functional clusters from DAVID Functional Enrichment for in vitro PARP6 ADP‐ribosylated Author manuscriptshavebeenpeerreviewedandacceptedforpublicationbutnotyetedited. Author ManuscriptPublishedOnlineFirstonOctober8,2018;DOI:10.1158/0008-5472.CAN-18-1362 proteins. Highlighted in blue is a cytoskeletal sub‐cluster for centrosomal proteins.

Enrichment Cluster DAVID Functional Term ADP‐ribosylated Proteins P‐Value Benjamini Score cancerres.aacrjournals.org AAK1, ABL1, ABL2, ADRBK1, AKT3, BMX, BTK, CAMK2A, CDK7, CHEK1, CSNK1G1, CSNK1G2, DAPK3, EPHA1, EPHA2, EPHA3, EPHB1, FES, GRK6, 1 15.33 Protein kinase MAP2K6, MAP4K5, MAPK1, MATK, NUAK1, PAK1, PIM1, PRKCA, PRKCI, 7.1E‐25 4.3E‐22 PRKG2, PTK2, RIOK1, RPS6KB1, RPS6KB2, SCYL1, SRPK1, STK25, STK33, STK40, TEC, TTBK2 ACAP1, ADRBK1, AKT3, ANLN, ARHGEF1, BMX, BTK, DNM2, HOMER1, 2 5.41 Pleckstrin Homology‐Type 1.9E‐07 8.4E‐06 HOMER2, HOMER3, NUP50, PLCG2, RALGPS1, RALGPS2, SH3BP2, TEC ABL1, ABL2, BMX, BTK, EPHA1, EPHA2, EPHA3, EPHB1, FES, MAP2K6, MATK, 3 5.13 Tyrosine‐Protein Kinase 2.2E‐09 5.5E‐08 PTK2, TEC ABL1, ABL2, ABLIM1, ADD2, ANLN, BAG1, BMX, CDC42EP3, CEP57, CHEK1, on September 30, 2021. © 2018American Association for Cancer CKAP2, CSPP1, DNALI1, DNM2, DYNC1I1, EPB49, FSD1, GSN, HIP1R, 4 4.23 Cytoskeleton HOMER1, HOMER2, HOMER3, LMNA, MAPK1, MAPRE1, MYOT, NME2, 1E‐07 2.5E‐05 Research. ODF2, PDLIM5, PTK2, SCYL1, SEPT9, TBCC, TMOD2, TPPP, TTBK2, TWF1, TWF2, USP2 Centrosome BMX, CEP57, CHEK1, CSPP1, FSD1, MAPRE1, NME2, ODF2, SCYL1, USP2 0.00029 0.0103 ABLIM1, ADD2, ANLN, ANXA2, BAIAP2, CDC42EP3, DNM2, DYNC1I1, EPB49, 5 3.82 Cytoskeletal Protein Binding FMNL1, GSN, HIP1R, HOMER2, MAPRE1, MYOT, NME2, PDLIM5, TMOD2, 1E‐06 2.5E‐05 TPPP, TWF1, TWF2 6 3.53 AGC‐Kinase C‐Terminal ADRBK1, AKT3, GRK6, PRKCA, PRKCI, PRKG2, RPS6KB1, RPS6KB2 1.8E‐06 0.00018 7 2.90 Non‐Membrane Spanning Protein Tyrosine Kinase Activity ABL1, ABL2, BMX, BTK, FES, MATK, PTK2, TEC 3.4E‐07 9E‐06 BTK, C7ORF16, DAPK3, MAP2K6, MAP4K5, MAPK1, NFKBIA, PAK1, PDE6H, 8 2.49 Protein Kinase Cascade 0.00017 0.0422 PIK3R1, PRKCA, RPS6KB1, RPS6KB2, SRPK1, TEC 9 1.90 Phosphoprotein Phosphatase Inhibitor PPP1R8, PPP1R2P9, C7ORF16, PPP1R10 0.00379 0.0735 10 1.79 Regulation of Protein Polymerization ADD2, ARF6, CDKN1B, EPB49, GSN, MAPRE1, TPPP 0.00017 0.0364 CAMK2A, DNM2, HOMER1, HOMER2, HOMER3, LIN7B, MPP7, NPHP1, PAK1, 11 1.59 Cell Junction 0.00353 0.046 PTK2, RPS6KB1, STXBP5 12 1.57 D111/G‐Patch AGGF1, GPKOW, RBM17 0.0233 0.351 13 1.56 Phosphoinositide Binding ANXA2, PICALM, PIK3R1, HIP1R, NCF1, TULP3 0.00553 0.0945 ABL1, ABL2, AKT3, CAMK2A, CDKN1B, MAPK1, PAK1, PIK3R1, PLCG2, PRKCA, 14 1.55 ErbB Signaling Pathway 3.2E‐09 3.1E‐07 PTK2, RPS6KB1, RPS6KB2 ANLN, CDCA3, CDK7, CDKN1B, CHAF1B, CHEK1, CKAP2, FSD1, MAPK1, 15 1.51 Cell Cycle 0.00127 0.0187 MAPRE1, NCAPG, PIM1, SEPT9

Author Manuscript Published OnlineFirst on October 8, 2018; DOI: 10.1158/0008-5472.CAN-18-1362 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Figure Legends

Figure 1. Identification of small molecule PARP6 inhibitors capable of inducing MPS formation

in HeLa cells.

(A) Dose response of AZ9482 in promoting MPS positive cells. Mitotic cells with at least four

centrosomes were scored as MPS positive.

(B) Chemical structure of AZ9482.

(D) Chemical structure of AZ0108.

(E) Representative images of HeLa cells treated with DMSO, 3.7 M AZD2281, 0.04 M AZ0108 and 0.4 M AZ0108. Cyclin B stains for mitotic cells (red) and pericentrin stains for centrosomes

(green). Asterisk (*) indicates representative cell in enlarged box for each respective panel.

(F) Scatter plot correlating MPS induction EC50 versus PARP6 enzyme inhibition IC50 for

compounds (Pearson R2=0.76). Highlighted compounds are: Dark blue – AZ0108, Light blue –

AZ9482, Red – AZD2281.

(G) Representative immunofluorescent images taken 72 hours after transfection of siRNAs

targeting GAPDH and PARP6 demonstrating PARP6 knockdown triggers MPS formation.

Hoechst, DNA marker (blue); pericentrin, centrosome marker (green); and cyclin B, mitosis

marker (red). Asterisk (*) indicates representative cell in enlarged box for each respective

panel.

(H) Quantification of percentage of mitotic cells with MPSphenotype (>2 centrosomes per nuclei)

in (G), where double asterisks (**) represent raw p‐value <0.001.

Figure 2. PARP6 inhibitor AZ0108 impairs proliferation, induces apoptosis, and promotes MPS

in breast cancer cells.

(A) Upper panel, GI50 waterfall plot for 3‐day AZ0108 treatment across a panel of 22 breast

cancer cell lines. Cell lines are categorized as sensitive (GI50 ≤ 1 M), marginal (1 M < GI50 <

5M), and insensitive (GI50 ≥ 5M) based on the GI50 values; lower panel, AZD2281 sensitivity is

assessed in the same assay format across the breast cancer cell line panel (NA indicates data not available).

(B) Confluence and caspase 3/7 activation over time in the presence of DMSO, 30 nM AZ0108 or 30nM AZ9482 captured by high‐throughput time‐lapse imaging system Incucyte Zoom.

(C) Upper panel, PARP6 mRNA knockdown efficiency were evaluated at 24 hours post‐ transfection of indicated siRNA or shRNA. Lower panel, cell viability was evaluated at 6 days post‐transfection for breast cancer cell lines following indicated gene perturbation. Data were presented as mean with standard error. Statistical significance comparing the means of the control and the PARP6 targeted knockdown were determined by a two‐way Student’s T‐test, where an asterisk (*) represents raw p‐values < 0.05.

(D) Representative images of HCC1806 cells following PARP6 mRNA knockdown (72 hours post‐ transfection) or AZ0108 or AZD2281 treatment (48 hours). Cyclin B staining (red) indicates mitotic population. Pericentrin (green) was used to label centrosome and Hoechst dye was used to stain DNA (blue). Scale bar represents 20m.

(E) Quantification for percentage of cells with MPS phenotype (>2 centrosomes per nuclei) in

(D), where double asterisks (**) represent raw p‐values <0.001.

Figure 3. AZ0108 is efficacious in inhibiting tumor growth and causes mitotic defects in vivo.

(A) MDA‐MB‐468 and HCC1806 xenografts were established, randomized, and then dosed daily

(p.o.) with 10mg/kg AZ0108. Tumor volume were measured and graphed as mean with standard

error (SEM). Percent body weight change is relative to animal weight at the start of dosing.

Statistical significance comparing the means of the vehicle and the AZ0108 treated animals were

determined by a two‐way Student’s t‐test, where an ampersand (&) or asterisk (*) represents

raw p‐values < 0.05 or < 0.01, respectively.

(B) PD samples were taken 48 hours post treatment. ‐tubulin immunohistochemistry staining

was performed to indicate mitosis effects of AZD0108 in MDA‐MB‐468 and HCC1806 tumors.

Representative normal and aberrant mitotic MDA‐MB‐468 cells are shown in the top panel.

(white filled arrows for normal mitotic cells, gray filled arrows for intermediate cells, and black

filled arrows for aberrant mitotic cells).

respective time points (in hours) post single dose treatment of AZ0108 in MDA‐MB‐468 and

HCC1806 tumors. The mean and standard deviation are plotted. Statistical significance between

vehicle and AZ0108 were determined by a test of equal proportions, where an asterisk (*)

represent raw p‐values < 0.01.

Figure 4. Identification of Chk1 as a specific substrate of PARP6.

(A) Venn diagram summary of proteins ADP‐ribosylated by PARP1, PARP2, PARP6, PARP10, and

PARP14 highlighting the low proportion of proteins ADP‐ribosylated by multiple PARPs.

(B) Z‐score bar charts with individual protein spot scatter plots for Chk1 and SFRS1 across different ProtoArray experimental conditions. Values for the ADP‐ribosylation of Chk1 by PARP6

and SFRS1 by PARP1 are statistically different from the no PARP enzyme condition with raw p‐ values < 0.05 (*) using a Mann‐Whitney U Test.

NAD+ as donor, GST‐Chk1 as substrate, with and without AZ0108. ADP‐ribosylation was detected with the Streptavidin Conjugated system.

(D) Upper panel, silver staining of macrodomain‐pulldown materials prepared from MDA‐MB‐

468 cells treated with DMSO control, with nocodazole for mitotic synchronization, or with nocodazole plus 0.3 M AZ0108. Lower panel, identification of candidate proteins from the highlighted band from mass spectrometry analysis.

Figure 5. PARP6 inhibition by AZ0108 results in elevated p‐S345 Chk1 and reduced mitotic signaling in vitro and in vivo

(A) Western blots analysis of mitotic proteins and their activation status in HCC1806 cells that were mitotically synchronized with 40 ng/mL nocodazole for 16 hours and released for 0, 3 or 30 hours while treated with DMSO, 0.3 µM AZ0108, or 0.3 µM AZD2281.

(B) Representative western blots of p‐S345 and total Chk1 as well as p‐S10 and total histone H3 for MDA‐MB‐468 and HCC1806 cells following 24 or 48 hours of AZ0108 treatment at 0, 30, and

300 nM (upper panel). Quantification of western blot intensity of p‐Chk1 and p‐S10 histone H3 normalized against total protein (lower panel, N=3).

(C) IHC staining of p‐S345 Chk1 for PD samples from HCC1806 tumors for 6, 24, and 48 hours post single dose of 10 mg/kg AZ0108 treatment.

(D) Quantification of (C). The mean and standard deviation are plotted. Statistical significance between vehicle and AZ0108 at the same time point were determined by a test of equal proportions, where an asterisk (*) represents raw p‐value < 0.01.

Fig.1

A B E 150 O DMSO AZD2281 3.7μM NH 100 N O

AZ9482 N 50 N N Increase of MPS positive cells (%) AZ9482 0 N HOECHST -3 -2 -1 0 1 2 PERICENTRIN 10µm CYCLIN B Concentration (log10 , μ M)

C D * 150 O NH 100 N * O * AZ0108 F N N * F N 50 N F

Increase of MPS F positive cells (%) AZ0108 0 -3 -2 -1 0 1 2

Concentration (log10 , μ M)

F G H siGAPDH PARP6 MPS induction Scatter Plot *

10µm siPARP6 * Cellular inductionMPS (µM)potency Percent mitotic cells with MPS siGAPDH siPARP6

BPS PARP6 IC50 (µM) *

HOECHST PERICENTRIN CYCLIN B

PARP6 mRNA 3 Day Assay

Viability (%) Knockdown (%) AZD2281 100 100 GI50 , μM 60 60 20 20 40 80 40 80 0.66 0.33 0.03 0.06 0.01 0 0 6.6 3.3 0.1 33 10 1 GAPDH siRNA MDA-MB-468 HCC1806 ZR-75-1 AZ0108 sensitivityprofileinbreastcancerlines

PARP6 MDA-MB-157 siRNA * Am05 HCC1806 * Downloaded from MDA-MB-468 Author manuscriptshavebeenpeerreviewedandacceptedforpublicationbutnotyetedited. PARP6 siRNA

* SK-BR-3 DH05 Author ManuscriptPublishedOnlineFirstonOctober8,2018;DOI:10.1158/0008-5472.CAN-18-1362 SUM52PE * T-47D Control shRNA BT-20 MDA-MB-453 MDA-MB-436 shRNA cancerres.aacrjournals.org PARP6 * 1053 BT-549 HCC1937 CAMA-1 shRNA PARP6 * 1297 MCF7/mdr+ HCC1187 Marginal > 1μM and < 5μM Sensitive <1μM NA Insensitive >5μM shRNA PARP6

* MDA-MB-231 944 HCC1395 * HCC1419 HCC1569 HCC1954 0.2μM 0.5μM

MDA-MB-468 HCC1806 E Cell Confluence (%) Cell Confluence (%) 100 120 20 40 80 60 10 20 30 40 50 60 70 0 0 ieEasd(or)Time Elapsed(hours) Time Elapsed(hours)

Percent mitotic cells with MPS 250 250 Growth 50 siCONT 50 75 siPARP6 75 100 100

DMSO AZ0108 0.4 Caspase Positive Caspase Positive Cell Area (%) Cell Area (%) 15 20 25 30 35 10 10 15 20 25 30 0 5 0 5 AZ0108 1.2 μ M DMSO 30nM AZ9482 30nM AZ0108 DMSO 30nM AZ9482 30nM AZ0108

AZD2281 1.2 M Apoptosis 250 250 μ M 50 50

μ 75 75 Fig.2 100 100 Author Manuscript Published OnlineFirst on October 8, 2018; DOI: 10.1158/0008-5472.CAN-18-1362 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Fig.3

A B

MDA-MB- 468 Tumor Volume Normal Mitosis Aberrant Mitosis 500 Metaphase Metaphase Disorganized Multipolar Vehicle Anaphase (top-view) (side-view) Spindle Spindle Mitotic Failure AZ0108 - 10mpk 400

300 Mean ± SEM 15 µm

Tumor Volume (mm³) Volume Tumor 200 * 10 15 2520 30 4035 45 Days Post Implant

MDA-MB- 468 Body Weight 10 Vehicle AZ0108 - 10mpk 5 MDA-MB-468

40 µm 0

Mean ± SEM -5

% Body Weight Change % Body Weight -10 10 15 2520 30 4035 45 Days Post Implant

HCC1806 Tumor Volume HCC1806 1500 Vehicle 1200 AZ0108 - 10mpk

900 Vehicle 48 Hours AZ0108 - 10mpk 48 Hours & 600 Mean ± SEM 300 Tumor Volume (mm³) Volume Tumor C MDA-MB-468 HCC1806 8 13 18 23 28 70 70 Days Post Implant * 60 60

HCC1806 Body Weight 50 50 10 Vehicle 40 * 40 AZ0108 - 10mpk 5 30 30

0 20 20 Mean ± St.Dev. (%) Mean ± St.Dev. (%) Mean ± St.Dev.

Aberrant Mitotic Nuclei 10 Aberrant Mitotic Nuclei 10 Mean ± SEM -5 * 0 0

% Body Weight Change % Body Weight AZ0108 AZ0108 -10 Vehicle Vehicle 8 13 18 23 28 48 Hours 10mpk 6 Hours 10mpk Days Post Implant 48 Hours 48 Hours

Fig.4

A C Streptavidin Coomassie Blue Detection PARP6 + + + + + + + + biotin-NAD+ - + + + - + + + BSA - + -- - + -- PARP6 GST-Chk1 -- + + -- + + 204 proteins AZ0108 --- + --- + PARP1 250KD 194 proteins 150KD

100KD PARP6 PARP6 75KD GST-Chk1 GST-Chk1 PARP10 PARP14 12 proteins 58 proteins 50KD PARP2 37KD 46 proteins 25KD 20KD 15KD

B D IP: Macrodomain mAf1521 ProtoArray Protein Z-Score Value 4

3 * 2

1 Chk1 (CHEK1)

2 * Lane 1 Lane 2 Lane 3 Full Protein Name MW [kDa] SFRS1 1 - NPM1 - Nucleophosmin, isoform 2 29.4 - ANXA2 - Annexin A2 38.6 - Chk1 - S/T-protein kinase Chk1 54.4 - DSC1 - Desmocolin-1, isoform 1B 93.8 0 - MYO5B - Myosin 5B 82.4 PARP6 No PARP PARP1 PARP6 PARP6 125nM & control 250nM 250nM 125nM 800nM AZ0108

Fig.5 A B 24h 48h MDA-MB-468 HCC1806 MDA-MB-468 HCC1806 Hours after DMSO AZ0108 AZD2281 nocodazole release 0 3 30 0 3 30 0 3 30 Chk1 Chk1 p-S345 p-S345 Total Chk1 Chk1 Histone H3 Histone H3 p-S10 p-S10

FoxM1 Histone H3 AurA p-T288 AurB p-T232 Vinculin AurC p-T198

GAPDH 1.00 Chk1 0.88 0.93 0.85 0.57 0.45 0.39 α-tubulin p-S345/Total 0.29 0.21 0.26 0.06 0.07

Histone H3 0.92 1.00 0.99 0.57 0.66 0.68 p-S10/Total 0.39 0.44 0.33 0.17 0.19 0.23

AZ0108

C 6 Hr 6 Hr 24 Hr 48 Hr HCC1806

50 mm

Vehicle AZ0108 mpk

D HCC1806 25 * 20

Mean ± St.Dev. (%) Mean ± St.Dev. 5

p-S345 Chk1 Positive Nuclei 0 Vehicle 6 24 48 Hours Downloaded from cancerres.aacrjournals.org6 Hours AZ0108on September - 10mpk 30, 2021. © 2018 American Association for Cancer Research. Author Manuscript Published OnlineFirst on October 8, 2018; DOI: 10.1158/0008-5472.CAN-18-1362 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Pharmacological inhibition of PARP6 triggers multipolar spindle formation and elicits therapeutic effects in breast cancer

Zebin Wang, Shaun E Grosskurth, Tony Cheung, et al.

Cancer Res Published OnlineFirst October 8, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-18-1362

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2018/10/06/0008-5472.CAN-18-1362.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet been Manuscript edited.

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2018/10/06/0008-5472.CAN-18-1362. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.