PARP inhibitors in the treatment of ARID1A mutant ovarian clear cell cancer: PI3K/Akt1-dependent mechanism of synthetic lethality

Introduction Poly(ADP-ribose) polymerase (PARP) is a nuclear enzyme involved in the repair of DNA single-strand breaks (SSB). The recent development of poly(ADP-ribose) polymerase inhibitors (PARPi) results from over 45 years of studies. When the activity of PARP1 or PARP2 is compromised, DNA SSB lesions are unresolved and can be converted to DNA double-strand breaks (DSBs) by the cellular transcription mechanisms. ARID1A (also called BAF250a) is an important component of the mammalian Switch/Sucrose Non-Fermentable (SWI/SNF) chromatin-remodeling complex. ARID1A gene demonstrates >50% of mutation rate in ovarian clear-cell carcinomas (OCCC). Mutated or downregulated ARID1A significantly compromises the Homologous Recombination Repair (HRR) of DNA DSB. Results The present study demonstrated that downregulated or mutated ARID1A attenuates DNA HRR through stimulation of the PI3K/Akt1 pathway and makes tumor cells highly sensitive to PARPi and PARPi/ionizing radiation (IR) combination. We showed that PI3K/Akt1 pathway plays an important role in the sensitization of cancer cell lines with compromised function of ARID1A to PARPi treatment. Discussion We believe that using of PARPi monotherapy or in combination with radiation therapy is an appealing strategy for treating ARID1A-mutated cancers, as well as many other types of PI3K/Akt1-driven cancers.


Introduction
The recent development of poly(ADP-ribose) polymerase inhibitors (PARPi) results from over 45 years of studies. Poly (ADP-ribose) polymerase (PARP) is an enzyme essential for the DNA single-strand breaks (SSB) repair (1)(2)(3)(4). When the activity of PARP1 or PARP2 is compromised, DNA SSB lesions are unresolved and can be converted to DNA double-strand breaks (DSBs) by the cellular transcription mechanisms (5). Breast cancer susceptibility protein 1 (BRCA1) plays a significant role in the error-free homologous recombination repair (HRR) of DNA DSBs. Loss of BRCA1 activity leads to a switch from error-free HRR to errorprone non-homologous end-joining (NHEJ) repair of DSB DNA which resulted in genomic instability (6)(7)(8). The PARP/BRCA genetic interaction was described as the synthetic lethality effect: both genes when individually downregulated do not affect the cell survival, but a contemporary loss of both genes' activities leads to cell death (9)(10)(11). Mechanisms of synthetic lethality can be applied to the treatment of cancer (12) and the PARP/BRCA genetic interaction demonstrated an effective approach for the treatment of ovarian cancer (OC) (13)(14)(15)(16). However, inherited BRCA1/2 mutations are present in 13-15% of OC (17), most frequently (18% in high-grade serous carcinomas) and less commonly for other histologic subtypes (18) and many cancer patients with BRCA1/2-wt tumors, sensitization to the different DNA-damaging agents with PARPi is less effective. Enhancing the efficacy of PARPi in the treatment of those OC without pathogenic mutations in homologous recombination deficiency proteins is an unmet clinical need.
ARID1A (also called BAF250a) is an important component of the mammalian Switch/Sucrose Non-Fermentable (SWI/SNF) chromatin-remodeling complex that regulates gene expression by controlling gene accessibility. ARID1A has been identified as a tumor suppressor gene with one of the highest mutation rates across different cancer types. Among OCs, ARID1A is mutated in >50% of clear-cell carcinomas (OCCC) and >30% of endometrioid carcinomas (19)(20)(21). Different reports demonstrated that ARID1A deficiency sensitizes cells to the PARPi (22,23), however, the mechanism of this effect was not described. Previous investigations showed that ADIR1A protein plays a significant inhibitory role in the PI3K-Akt1 pathway (24)(25)(26). Furthermore, ARID1A mutations were discovered to occur frequently in a synergistic fashion with mutations in PIK3CA (26-28). Overactivation of the PI3K-Akt1 pathway suppresses the HRR of DNA DSBs by different mechanisms. As previously shown, Akt overactivation can suppress HRR via p70S6 kinase-dependent downregulation of MRE11 (29). It has also been demonstrated that stimulation of the PI3K-Akt1 pathway suppresses HRR via cytoplasmic retention of Rad51 and BRCA1 proteins (30, 31). These findings indicate that tumors with mutant ARID1A or ARID1A protein deficiency may depend more on the overactivated PI3K/ AKT pathway than on homologous recombination for DNA repair. In the present work, we have demonstrated that downregulated or mutated ARID1A attenuates DNA HRR by stimulating the PI3K/ Akt1 pathway and renders tumor cells highly sensitive to PARPi.
We have also shown that the use of PARPi in combination with a DNA-damaging agent (for example -ionizing radiation (IR)) leads to a synthetic lethality effect, resulting in a significantly more pronounced therapeutic effect compared to monotherapy with PAPRi or a DNA-damaging agent. The results of our study provide a new strategy for using PARPi alone or in combination with DNA-damaging agents (radio or chemotherapy) to treat ARID1A-mutated cancers, as well as many other PI3K/Akt1induced cancers.

Materials and methods
Cell culture, apoptosis, and clonogenic assays OC cell lines CAOV-3, OVCA-429, SKOV-3, and TOV-21G were obtained from American Type Culture Collection (ATCC) and grown according to ATCC recommendations. The base medium for CAOV-3 -DMEM (cat.# 12491015, ThermoFisher), OVCA-429 -DMEM (with high glucose) (cat.# 11960044, ThermoFisher), and SKOV-3 was grown in McCoy's 5a medium (cat.# 16600082, ThermoFisher). Medium for all three cell lines contained 10% fetal bovine serum (FBS) (cat.# A5256701, ThermoFisher), and 1% Penicillin/Streptomycin. The base medium for the TOV-21G cell line was a 1:1 mixture of MCDB 105 medium containing a final concentration of 1.5 g/L sodium bicarbonate and Medium 199 containing a final concentration of 2.2 g/L sodium bicarbonate with 15% of FBS and 1% Penicillin/ Streptomycin (cat.# 15140122, ThermoFisher). For the clonogenic assay, cells were seeded into a 6-well plate or 60-mm culture dish. After an incubation period of 2 weeks, the colonies were fixed with methanol and stained with crystal violet. Cell apoptosis was determined through APC Annexin V/Propidium Iodide (PI) assay staining by using the APC-Annexin V Apoptosis Detection Kit (cat.# 640932, BioLegend) according to the manufacturer's protocol. At specific time points, cells were stained and analyzed by flow cytometry (FACS Canto II flow cytometer, BD Biosciences). Fluorescence-based DSB repair assay, adenovirus treatment, and flow cytometry All cell lines used in our study were stably transfected with the pDR-GFP reporter construct. The reporter constructs and the fluorescence-based assay for measuring the frequency of HRR at a single chromosomal DSB have been described previously (32,33). Infection with an I-SceI expression adenovirus (Ad-SceI-NG) produces a DNA DSB in the SceGFP sequence that can be repaired by 2 general mechanisms: error-free HRR or error-prone non-homologous end joining repair (NHEJ). In this assay, a functional GFP sequence can only be restored if the DSB is repaired by HRR in an error-free manner using the downstream GFP fragment (iGFP) as a template ( Figure 1C). The percentage of GFP-positive cells after infection with Ad-SceI-NG represents the  level of HRR in the test. The Ad-SceI-NG adenovirus was a generous gift from Dr. Kristoffer Valerie (32,33). Adenovirus was added to the culture medium at 30 virus particles/cell and incubated while rocking for 4 h at 37°C. The virus was then removed, and a fresh medium was applied. After 48 hours of infection, flow cytometry was used to determine the fraction of GFP-positive cells in each sample. The amount of HRR was calculated as a percentage of the GFP-positive cells in 30,000 cells counted.

Western blotting
Proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes. The membranes were exposed to antibodies at specific dilutions. Specific protein bands were detected using infrared-emitting conjugated secondary antibodies: anti-Rabbit DyLight ™ 800 4XPEG Conjugate (1:10,000 dilution, cat.# 5151, Cell Signaling), and the ChemiDoc ™ MP Imaging System (Bio-Rad).

RNA extraction, real-time quantitative PCR
Total RNA was isolated from the cultured cells following the manufacturer's instructions with the RNeasy kit (cat.# 74004, QIAGEN). The RNA concentration was measured using a NanoDrop ND-1000 spectrophotometer (Thermo Scientific, Wilmington DE). RNA purity was assessed by the ratios of A260/ A280 and A260/A230. RNA integrity was evaluated by the ratio of 28S/18S ribosomal RNA (rRNA) and the RNA integrity number (RIN) using an Agilent 2100 BioAnalyzer (Agilent Technologies, Wilmington DE). cDNA synthesis and genomic DNA elimination were performed by using RT 2 First Strand Kit (cat.# 330404, QIAGEN). Samples were amplified using RT 2 SYBR ® Green qPCR Mastermix probes with ROX (carboxy-X-rhodamine) passive reference dye from QIAGEN (cat.# 330529)on the QuantStudio 3 RT-PCR machine (Applied Biosystems). The realtime PCR data were normalized by ROX passive reference. The specificity of amplicons was verified by melting curve analysis and C t values of all amplicons were normalized using mRNA expression as an internal control. The 2 -DDCt method was used for the calculation of relative mRNA expression levels. The following RT 2 qPCR Primer Assays (QIAGEN) were used: RT² qPCR Primer Assay for Human ACTB (GeneGlobe ID: PPH00073G-200), RT² qPCR Primer Assay for Human ARID1A (GeneGlobe ID: PPH13453B-200).

Statistical analysis
Unless indicated otherwise, data are demonstrated as mean ± standard deviation (SD). The significance of the difference between groups was determined by paired or unpaired two-tailed Student's t-test or the one-way analysis of variance (ANOVA). The Data Analysis Toolpak for Excel was used for the calculation of the Student's t-test or the ANOVA. Differences were considered significant for p-values < 0.05.

Results
Mutation or downregulation of ARID1A expression stimulates PI3K/Akt1 pathway and reduces HRR of DNA DSB Four different OC cell lines were used in our study. Three cell lines contain a wild type of ARID1A gene: CAOV-3, OVCA-429, and SCOV-3; whereas the TOV-21G cell line contains a mutant ARID1A gene leading to loss of the protein expression ( Figure 1A). Since the ARID1A protein is involved in the regulation of the PI3K/ Akt1 pathway, we tested the protein expression level and phosphorylation status of Akt1. All four cell lines demonstrated similar levels of total Akt1, however, TOV-21G showed a significant increase in Akt-1 phosphorylation at S473 and T308 ( Figure 1A). In TOV-21G cells the ARID1A mutation coexists with a PIK3CA mutation which can additionally stimulate PI3K/Akt1 pathway. Hence, as a next step, we tested if ARID1A protein expression alone can modulate the activity of the PI3K/Akt1 pathway. All three cell lines with wild-type ARID1A were transfected with 20nM of siRNA against ARID1A or 20nM of negative control siRNA. As shown in Figure 1B, downregulation of ARID1A protein expression stimulated phosphorylation of Akt-1 protein on S473 and T308. Both experiments demonstrated a strong connection between the level of ARID1A protein expression and PI3K/Akt1 pathway activation evaluated by Akt-1 protein phosphorylation.
It was previously demonstrated that stimulation of the PI3K/ Akt1 pathway suppresses the HRR of DNA DSB by various mechanisms (29)(30)(31). To estimate the impact of ARID1A on the level of DNA HRR, all cell lines were stably transfected with the DR-GFP reporter construct. Infection of stably transfected cell lines with an I-SceI expression adenovirus (Ad-SceI-NG) generates a DSB in the SceGFP sequence that can be repaired by 2 general mechanisms: HRR or NHEJ. In this assay, a functional GFP sequence can only be restored if the DSB is repaired in an errorfree manner using the downstream GFP fragment (iGFP) as a template for HRR ( Figure 1C). The percentage of GFP-positive cells after infection with Ad-SceI-NG represents the level of DNA HRR in the test ( Figure 1D). Transfection with the negative control siRNA did not significantly affect the level of DNA HRR for all three cell lines with wild type of ARID1A ( Figures 1D, E). However, the reduction of ARID1A expression by siRNA transfection significantly reduces the relative level of DNA HRR: CAOV-3 -  Figure 1E).
The coexistence of ARID1A and PIK3CA mutations in TOV-21G cells leads to PI3K/Akt1 pathway overactivation ( Figure 1A) and may result in a substantial decline in DNA HRR activity. To test the correlation between the level of DNA HRR and PI3K/Akt1 pathway activity, TOV-21G cells stably transfected with DR-GFP reporter construct were treated with a potent inhibitor of PI3K LY294002 or selective Akt inhibitor MK-2206. Both inhibitors dramatically blocked Akt1 phosphorylation on S473 and T308 ( Figure 2A)

ARID1A status and sensitivity to PARPi
All cell lines were subjected to the clonogenic assay with two different PARPi ABT-888 and Olaparib. Survival fraction was estimated for 4 different groups: 1) Non-treated control; 2) Vehicle control (with DMSO); 3) 10mM of ABT-888; 4) 1mM of Olaparib. All cell lines were incubated with the PARPi or vehicle for 24h, and then culture media were replaced with fresh media without drug or vehicle. Only TOV-21G cells demonstrated a significant decrease in clonogenic survival with ABT-888 and Olaparib treatment ( Figures 3A, B): ABT-888 treatment -0.33±0.025, p<0.001; Olaparib treatment -0.16±0.023, p<0.001. Similar results were obtained when apoptosis was assessed after treatment with ABT-888 and Olaparib ( Figure 3C). Levels of apoptosis were measured by the APC AnnexinV/PI assay after 48 h of incubation with PARPi. As in the clonogenic experiment, only TOV-21G cells show a significant increase in apoptosis after treatment with PARPi ( Figure 3D): ABT-888 -18.75±0.56% (p<0.001) and Olaparib -22.12±1.33% (p<0.001) vs. Vehicle control -9.8±0.66%.
ARID1A status and sensitization to ionizing radiation by PARPi  (Figures 5A, B). Similar results were obtained when cell killing was measured by apoptosis ( Figure 5C).

Inhibition of PI3K/Akt1 pathway by LY294002 or MK-2206 attenuates sensitivity of TOV-21G cells to PARPi
In experiments in Figure 2, we demonstrated that incubation of TOV-21G cells with a potent inhibitor of PI3K LY294002 or selective Akt inhibitor MK-2206 blocked Akt-1 phosphorylation on S473 and T308 (Figure 2A) and significantly increased DNA HRR level. We next tested how inhibition of the PI3K/Akt1 pathway affects the sensitivity of the TOV-21G cell line to PARPi and PARPi/IR combination. TOV-21 cells were pre-treated with 50mM LY294002 or the same volume as a vehicle (DMSO). After 6 hr media was replaced with the fresh media containing either vehicle (DMSO), 10 mM ABT-888, or 1 mM Olaparib, and cells were incubated for another 6 h, and then cells were irradiated at 2 Gy. Survival fraction was estimated by clonogenic assay  Figure 6D).

Discussion
PARPi are the first drugs designed to use a synthetic lethality approach. The highest efficacy for PARPi has been demonstrated for OC patients with BRCA mutations. PARPi activity was also demonstrated for BRCAwt cancers due to mutations in genes critical for DNA repair (e.g., ATM, BARD1, BRIP1, CHEK2, NBN, PALB2, RAD51C, and RAD51D) (34,35). The idea that DNA HRR can be compromised by many different factors generated great interest in enhancing individualized profiling of homologous recombination deficiency (HRD) of OC to predict sensitivity to therapy with PARPi (34,36,37). Recently, DNA-based homologous recombination deficiency (HRD) score was developed on the basis of loss of heterozygosity (LOH), telomeric allelic imbalance (TAI), and large-scale state transitions (LST) (37)f. HRD is the first phenotypically defined predictive marker for therapy with PARPi in OC. As HRD assays are increasingly used for treatment planning, our goal is to characterize the molecular mechanisms behind the onset of HRD in patients with ovarian cancer and to identify novel predictive markers for treatment with PARPi-based therapy. Our previous work showed that cancer cells with HRD compensated this deficiency by activation of error-prone NHEJ (33).
A close correlation between ARID1A mutations and the activity of the PI3K/Akt1 signaling pathway has been previously described A B FIGURE 4 Correlation of ARID1A expression status with sensitization to IR by PARPi treatment. (A) Cell lines were pre-treated with 10mM ABT-888, 1mM Olaparib, or a vehicle (DMSO) and were subjected to the clonogenic assay with different doses of IR. Results of the clonogenic analysis were normalized to nonirradiated vehicle control and presented as the mean ± SD for quadruplicate samples. (B) Cell lines were treated as described in Figure 4A and apoptosis was assessed 48 hr after 2 Gy IR as described in Figure 3. Columns represent the means ± SD values for apoptotic cells obtained from three individual experiments . For (A, B), the P-values were calculated with the ANOVA test and shown as **p < 0.05 and ***p < 0.001. (38). In endometrial cancer (39), ovarian clear cell carcinoma (28), colon cancer (40), and gastric cancer (24), loss of ARID1A expression stimulates Akt1 phosphorylation. ARID1A protein activates SWI/SNF complex which inhibits PIK3CA and PDK1 transcription (24). Also, ARID1A mutations frequently occur with mutations in PIK3CA (26-28, 39, 41), leading to stimulation of the PI3K/Akt1 pathway synergistically. Hence, cancer cells with ARID1A mutation or deficiency depend more on the PI3K/Akt1 pathway than the cells expressing normal ARID1A. Previous investigations demonstrated that stimulation of the PI3K/Akt1 pathway suppresses the HRR of DNA DSBs by multiple mechanisms (29)(30)(31).
Our data show a high correlation between ARID1A protein expression level, PI3K/Akt1 pathway activity, and DNA HRR efficacy. Inhibition of ARID1A expression or its functional impairment due to mutation leads to stimulation of the PI3K/ Akt1 pathway as shown by the increase in phosphorylation on S473 and T308 of Akt1 and subsequent impairment of DNA HRR mechanisms. In contrast, inhibiting the PI3K/Akt1 pathway by the potent inhibitor of PI3K LY294002 or selective Akt inhibitor that ARID1A mutation stimulates PI3K/Akt1 pathway, attenuates DNA HRR, and makes tumor cells highly sensitive to PARPi and PARPi/IR combination. Our study contains several limitations. First, only IR was used as a DNA-damaging agent. Further research into chemotherapeutic drugs as DNA-damaging agents is needed. Second, in our study, we use only the siRNA approach to decrease the expression of ARID1A in several cancer cell lines. In future studies, cell lines with knockout ARID1A and members of the PI3K/AKT pathway provide us with more information. Third, studying cell lines in vitro gives us only proof of principle. Animal model studies are needed to obtain more data.
The effect of ARID1A mutation can be compared with another commonly mutated gene in human cancers, phosphatase and tensin homolog (PTEN) (42). Both ARID1A and PTEN proteins share a common function of inhibition of the PI3K/Akt1 pathway. As for ARID1A, cells with mutated or depleted PTEN show over-activation of the PI3K/Akt1 pathway and impairment of DNA HRR (43-46). Consistently, PI3K/Akt1 inhibition restored DNA HRR efficiency in PTEN-depleted cells (47). As for ARID1A, the DNA HRR inefficiency caused by PTEN depletion or mutation, sensitizes tumor cells to PARPi, both in vitro and in vivo (43, 47).
Activation of the PI3K/Akt1 pathway can occur not only by mutations in PI3K, inactivating mutations or the loss of PTEN and ARID1A, but also mutations in other upstream oncogenes (e.g. RAS) (38,48). According to current research, over 25% of lung adenocarcinoma mutations are RAS mutations (49), and the Kristin isoform of Ras (K-Ras) is mutated in over 90% of pancreatic ductal adenocarcinomas (50). As mutations in RAS are common to a variety of cancers many different strategies that aim to downregulate constitutively active PI3K/Akt1 pathway have been explored (51,52). We propose that another appealing strategy for treating cancers with RAS mutations, as well as many other types of PI3K/Akt1-driven cancers, is the use of PARPi in monotherapy or in combination with chemo-or radiation therapy.
ARID1A mutations, although uncommon in high-grade serous carcinoma of the ovary, are relatively common in cancers with clear cell and endometrioid histology (21,53). As these subtypes of OC are relatively chemotherapy resistant (54), finding effective therapies is an unmet need. The use of PARPi in combination with different DNAdamaging agents can significantly reduce the resistance of these cancer subtypes to standard chemotherapy. In some cases, PARPi can also be used in combination with RT. The safety of treatment with PARPi in combination with RT in OC has been already demonstrated in several early-phase clinical trials (55,56). Targeting this therapy to ARID1A mutant/deficient malignancies should be strongly considered in future clinical trials of enhancing PARPi efficacy for women without germline mutations in HR deficiency pathways and those with BRCAwt tumors with few therapeutic options.

Conclusions
The present study demonstrates that mutated or downregulated ARID1A compromises the HRR of DNA DSBs through stimulation of the PI3K-Akt1 pathway. Attenuation of DNA HRR in ARID1Amut tumor cells sensitizes them to PARPi and PARPi/IR combination by a mechanism of synthetic lethality.

Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Funding
This project was supported by VCU's internal VETAR (Value and Efficiency Teaching and Research) Grand. Services and products in support of the research project were generated by the Virginia Commonwealth University Flow Cytometry Shared Resource, supported, in part, with funding from NIH-NCI Cancer Center Support Grant P30 CA016059.