Towards the clinical application of functional analysis of BRCAness in gynecological cancers

Towards the clinical application of

functional analysis of BRCAness in

gynecological cancers

The efficacy of radiomimetic drugs and PARP inhibitors to induce DNA double strand

breaks

Katja Klooster 5 July 2017 Bachelor thesis

University of Amsterdam

Leiden University Medical Centre, Leiden Laboratory coordinator: L.M. van Wijk, MSc Supervisor: Dr. M.P.G. Vreeswijk

Abstract

Women with homologous recombination (HR) deficient ovarian tumors, either related or unrelated to germline mutations in BRCA1 or BRCA2, are potentially sensitive to a cancer-specific treatment with poly(ADP-ribose) polymerase (PARP) inhibitors (PARPi). PARPi treatment only leads to cell death in HR deficient cells. A functional ex vivo HR assay was developed to determine whether a tumor is HR deficient or proficient. Currently, in research setting DSBs are induced by ex vivo ionizing radiation. However, the use of ionizing radiation may not be suitable in diagnostic setting. In this study the efficiency to induce DSBs by the PARPi Olaparib and the radiomimetic agents Bleomycin, Zeocin and Phleomycin are compared. Treatments were performed in a Chinese hamster lung fibroblast V79 cell line and in tumor tissue (i.e. cervix, ovarian and endometrium). Results were obtained by staining and scoring of ƴH2AX foci, a marker for DSBs and RAD51 foci, a marker for HR. Concluded from the results, the 3 µM Olaparib 24 h treatment is the most suitable alternative to radiation in the HR assay. The 3 µM Olaparib 24 h treatment caused a similar percentage of cells with DSB induction and DSB repair by HR as the 5 Gy irradiation treatment, with a consistent number of ƴH2AX and RAD51 foci in each cell.

Introduction

Last year, 14.640 new patients were diagnosed with breast cancer and 1350 new patients were diagnosed with ovarian cancer in the Netherlands (www.cijfersoverkanker.nl). Germline breast cancer 1 (BRCA1) and breast cancer 2 (BRCA2) mutations predispose to breast and ovarian cancer. BRCA1 and BRCA2 mutations are heterozygous and during tumorigenesis the wild-type BRCA allele is lost (Lord & Ashworth, 2016). In ovarian cancer, also somatic mutations in BRCA1 and BRCA2 have been identified. Recently, the frequency of BRCA1 and BRCA2 mutation, both somatic and germline, was estimated around 20% in high-grade serous ovarian carcinomas (Konstantinolpoules, Ceccaldi,

Shapiro & D’Andrea 2015). BRCA1 and BRCA2 proteins are key players in the DNA repair pathway of homologous

recombination (HR). This pathway repairs double strand breaks (DSB) by utilizing the sequence on the sister chromatid as a template. BRCA1 protein controls signal transduction pathways in HR and it decides whether HR or another repair pathway is activated. The BRCA2 protein binds directly to DNA recombinase RAD51 and localizes it to the DSB (fig. 1) (Lord & Ashworth, 2016).

Fig. 1. DSB repair by homologous recombination (HR), facilitated by the BRCA1, BRCA2 and RAD51 proteins. A mutation causing a deficient HR pathway leads to the accumulation of DSBs, because of the inability to repair these breaks. HR deficient cells rely on other error-prone repair pathways, that could lead to mutations and deletions. These HR deficient tumor cells can be treated with PARPi or platinum based chemotherapy (Lord & Ashworth, 2016).

tumor and platinum based chemotherapy. However, cisplatin chemotherapy is toxic in both tumor and in healthy cells. This causes severe side effects for the patients. The predictive power of whether women will benefit from the cisplatin treatment is low and therefore most patients receive the same treatment (Florea & Büsselberg, 2011). Besides, many patients develop resistance to cisplatin, causing recurrence of the disease (Stordal & Davey, 2007). The use of drugs that specifically target HR deficient cells could increase the efficiency of a treatment.

Recently, a new class of drugs, so-called poly(ADP-ribose) polymerase (PARP) inhibitors (PARPi), have proven to be very successful to treat BRCA related tumors (Audeh et al., 2010; Pommier, O’Connor & de Bono, 2016; Murai et al., 2013; Ordóñez et al., 2015). PARPi inhibit PARP1 and PARP2 proteins. PARP1 and PARP2 are PAR polymerases required for the repair of DNA single strand breaks (SSB). PARP1 binds to SSBs and a poly-chain of ADP molecules bind to PARP1. This functions as a flag to recruit DNA repair proteins that can repair the DNA lesion. Besides inhibiting the function of PARP1 and PARP2, PARPi also inhibit SSB repair by a process called ‘PARP trapping’. Binding of repair proteins to the SSB is only possible when PARP1 is detached. PARPi prevent PARP1 detachment and in this way double strand breaks (DSBs) can be formed when SSBs persist to the

next replication round. In the absence of HR, the cell uses other pathways to repair DNA DSBs, such as

non-homologous end-joing (NHEJ). This is a DNA repair mechanism that joins two DNA single strands without using a template. NHEJ is an error-prone process that can lead to DNA mutations and deletions, ultimately leading to cell death. Deficiency of both PARP and BRCA therefore leads to a lethal effect, called synthetic lethality (fig. 2) (Lord & Ashworth, 2012, Frey & Pothuri, 2017). Since the tumor cells are HR deficient whereas the normal cells of the patient are HR proficient, this therapy is highly targeted to the tumor cells. PARPi treatment is therefore suitable as a cancer-specific treatment (Konstantinopoulos et al., 2015; Pommier et al., 2016).

Fig. 2. PARP trapping by PARPi causes an accumulation of DSB in HR deficient cells when SSB are persisted to the next replication round. This causes cell death in HR deficient cells. The number of ƴH2AX foci, what is used as a marker for DNA double strand breaks, increase when DSBs occur. The accumulation of RAD51 protein at the sites of DNA DSB (RAD51 foci) is used as a marker for HR. In the absence of HR, no RAD51 foci are formed (Mukhopadhyay et al., 2010).

Moreover, a substantial number of ovarian tumors might have deficiencies in HR unrelated to germline mutations in BRCA1 or BRCA2 (i.e. BRCAness). Assessment of HR efficacy in ovarian tumors might therefore allow identification of additional cancer patients that could benefit from PARPi

treatment (Konstantinopoulos et al., 2015). In research setting, a biomarker for determining PARPi

sensitivity is searched for.

A functional ex vivo HR assay was developed to determine whether a tumor is HR proficient or deficient, and thereby which patients are susceptible to PARPi treatment. Mukhopadhyay et al. (2012) performed an ex vivo functional analysis in ascites from ovarian cancer patients. HR deficiencies was associated with higher ex vivo PARPi sensitivity and clinical platinum sensitivity. Patients with HR deficiencies had lower tumor progression rates at 6 months, lower overall and disease specific cell death rates at 12 months and higher median survival. Besides, 78.3% of the patients with no ex vivo response to PARPi, were platinum resistant. Therefore, HR deficiencies could correlate with in vitro PARPi sensitivity, clinical platinum sensitivity and improves survival outcome. HR functionality could potentially function as an important biomarker for sensitivity to PARPi treatment.

In this assay, DSBs are induced by exposing fresh tumor tissue to ex vivo ionizing radiation. The HR functionality is determined by the number of RAD51 foci (i.e. accumulation of RAD51 protein at the sites of DNA DSB) formed after DNA damage. The presence or absence of RAD51 foci is a readout for HR functionality (Naipal et al., 2014; Mukhopadhyay et al., 2010).

The DSB induction by ionizing irradiation is developed in research setting. The use of ionizing radiation may not be suitable in diagnostic setting. To optimize the diagnostic implementation of this assay, this study will look for other potential DNA DSB inducers as an alternative to irradiation. The first alternative that was examined is a PARPi. PARPi, such as Olaparib, Rucaparib and Niraparib, differ in potencies in trapping PARP-DNA complexes, and thereby in cytotoxicity (Murai et al., 2013). In this study the PARPi that will be used to induce DSB is Olaparib, because Olaparib has been used in clinical studies on patients with BRCA1 and BRCA2 mutations before (Audeh et al., 2010).

Another alternative to radiation are radiomimetic drugs. In this study the drugs that were examined are Bleomycin, Zeocin and Phleomycin, which are part of the Bleomycin family. Bleomycins are a family of glycopeptide antibiotics produced by Streptomyces verticillus, that are used as anti-tumor agents. Members of the Bleomycin family share the same core structure but differ in bound sugars and positively charged tails, which can be used to bind negatively charged DNA. Bleomycins can be activated in the presence of a reduced Fe(II) or Cu(I), oxygen and a reductant. When activated, a single Bleomycin molecule can intercalate into DNA and induce sequence-specific cleavage of DNA and oxidative stress and thereby causing SSBs or DSBs (Fig. 3). This leads to changes in chromosome morphology that causes the cytotoxic effect of the drug (Chen & Stubbe, 2005). The Bleomycin-family members differ in activity due to relatively small chemical differences. For instance, Phleomycin is thought to be more active on intracellular DNA than Bleomycin (Moore, 1989). Another difference between the drugs is the way the compounds are transported over the membrane. Zeocin is transported into the cell via the electrochemical gradient (Krol, Brozda, Skoneczny, Bretne & Skoneczna 2015) and Bleomycin via receptor binding (Chen & Stubbe, 2005).

Fig. 3. Activated bleomycin induces DSBs. Bleomycins can be activated in the presence of a reduced Fe(II) or Cu(I), oxygen and a reductant. When activated, a single Bleomycin molecule can intercalate into DNA and induce sequence-specific cleavage of DNA and oxidative stress and thereby causing SSBs or DSBs (Chen & Stubbe, 2005).

In this study Bleomycin, Zeocin, Phleomycin and Olaparib will be compared to ionizing radiation. The drugs will be tested in Chinese hamster lung fibroblast V79 cell line and in tumor tissue (i.e. cervix, ovarian and endometrium). The amount of DNA damage is visualized by the number of ƴH2AX foci, what is used as a marker for DSBs. (Liddle et al., 2014). The number of ƴH2AX foci after treatment is used to determine which drug induces a similar number of DSBs in tumor cells compared to ionizing radiation.

Methods

Cell culture

Chinese hamster lung fibroblast V79 cell line was cultured from cells frozen in liquid nitrogen (ampule 10-09-2012). Cells were cultured in a 10 cm (P9) plastic dish (Greiner) in DMEM-F12 medium (Gibco) supplemented with 10% Fetal Bovine Serum (Bodinco), 5% penicillin (100 U/ml) and streptomycin (0.1 mg/ml). Cells were maintained at 37°C, 5% CO2 humidified atmosphere.

Subcultures were made from 70-80% confluent cell cultures, trypsin (1x) was used to detach cells from the plastic and fresh medium was added. 50,000 cells were seeded on 18 mm coverslip and placed in 12-well plates. After o/n incubation cells were treated with one of the treatments (see below). EdU (0.02 mM, Click-it kit Invitrogen) was added two hours before fixation. The cells were fixed in 4% paraformaldehyde and 0.5% triton for 15 minutes and stored in PBS at 4°C followed by IF staining.

Tumor tissue culture and treatment

Fresh tumor tissue was received from the department of pathology at the Leiden University Medical Centre (LUMC), Leiden, derived from patients undergoing primary surgery. None of the patients had been screened for BRCA mutations or selected on family history of breast or ovarian cancer. All tested tumors were HR proficient, based on earlier RAD51 foci scoring data of the same tumors. Viable frozen tumor tissue was cultured from liquid nitrogen. Tissue was incubated in DMEM-F12 medium (Gibco) supplemented with 10% Fetal Bovine Serum (Bodinco), 5% penicillin (100 U/ml) and streptomycin (0.1 mg/ml). Cells were maintained at 37°C, 5% CO2 humidified atmosphere on a

rotating platform (60 rpm). Tissue samples were embedded in 4 % low-melting point agarose gel (ThermoFisher) and sliced using a Leica VT1200S vibratome (0.4 mm/sec, 300 µm, ampl. 3 mm). The slices were treated with one of the treatments (see below). EdU (0.02 mM, Click-it kit Invitrogen) was added two hours before fixation. The slices were fixated in 10% formalin o/n and stored in 70% ethanol at RT. The tumor slices were embedded in paraffin (ThermoFisher Excelsior™ AS Tissue Processor) and sliced in 5 µm slices using a Microm (HM 325). The thin slices were placed on Starfrost microscope slides (76 mm x 26 mm, Knittel) and dried in a 60 °C incubator o/n followed by IF staining.

Treatments

Cells and tumor tissue were treated with 100 or 500 µg/ml Bleomycin (stock 6 mg/ ml, Bleomedac), 100 or 500 µg/ml Zeocin (stock 100 mg/ml, Invitrogen), 100 or 500 µg/ml Phleomycin (stock 100 mM, Bio-Connect) or 3, 10 or 20 µM Olaparib (stock 5.5 mM, AstraZeneca) for 1, 2 or 4 hours. Cells and tumor tissue were irradiated with an X-ray machine (YXLON) in culture medium with a dose of 5 Gy (200 kV, 4mA, 1.5Gy/min). The number of treatments executed per tumor was based on the obtained number of slices.

Immunofluorescence (IF)

Two tumor tissue microscope slides were made, one was used for IF staining and one for Hematoxylin–eosin (HE) staining. The IF microscope slides were deparaffinized using xylene and hydrated with declining concentrations of ethanol. Target antigen retrieval was performed using DAKO Antigen Retrieval buffer (pH 9, 10% 10x in PBS). Cells were permeabilized using PBS with 0.2% TritonX-100 for 20 minutes. DNase (1,000 U/mL; Roche Diagnostics, cat. 04536282001) incubation was performed at 37°C for 1 hour in a SQ-wet chamber. Blocking was achieved using PBS with 2% FBS and 1% BSA. The coverslips with V79 cells were washed in 3.0% BSA (in PBST) before staining. The cells were stained on the coverslips and the tissue samples on microscope slides. ƴH2AX staining was performed using primary antibody mouse (Millipore DAM1567248) 1:500 and secondary antibody Alexa Fluor 555 goat anti-mouse (Life Tech A21424) 1:1000. Geminin staining was performed using primary antibody rabbit (Proteintech Europe, cat. 10802-1-AP) 1:400 and secondary antibody Alexa fluor 488 goat anti-rabbit (Life Tech A11034) 1:1000. EdU staining was performed using Alexa Fluor 647 (ThermoFisher Scientific, Click-iT™ EdU Alexa Fluor™ 647 Imaging Kit, cat. C10340). Tumor slices were also stained for RAD51 using primary antibody mouse, Gene Tex RAD51 antibody GTX70230, 1:400 and the same secondary antibody as ƴH2AX staining. The coverslips were mounted on 24x60mm microscope slides with ProLong Gold Antifade Mountant with DAPI (cat. P36935, ThermoFisher Scientific).

Microscopy

The results of the staining were visualized using the fluorescence microscope (Zeiss AXI0, 63x objective, Geminin: 488 nm, ƴH2AX/RAD51: 555 nm, EdU: 647 nm, DAPI: 460 nm). At least 50 geminin positive tumor cells were scored per coverslip and tumor cells were selected based on morphology using DAPI staining. The percentages of cells with ≥7 RAD51 foci and ≥7 ƴH2AX foci was calculated, as well as the percentage of EdU positive cells to determine the reliability of the results.

Results

The V79 cell line was used to determine whether the used drugs were still active. The cell line was also used to determine the combination of concentration and treatment time in which the drugs induced a comparable number of ƴH2AX foci as 5 Gy ionizing radiation. In each sample, 50 geminin positive cells were counted and scored. Based on the findings in the untreated controls (fig. 4A) the number of background ƴH2AX foci was set on <7. ≥7 ƴH2AX foci per nucleus was set as DSB induction. The percentage of cells with <7 or ≥7 ƴH2AX foci was calculated for the 50 geminin positive cells per treatment. All treatments with both the radiomimetic drugs or Olaparib showed DSB formation compared to the untreated samples (fig. 4A and B), indicating that the compounds were still active. For most treatments, a longer treatment time or a higher concentration had no effect on the DSB formation (supplementary fig. 1 and 2). The scoring results showed that irradiation with 5 Gy induced the most ƴH2AX foci per cell (22.66 foci/cell). 100 µg/ml Bleomycin 1h treatment induced a similar number of foci per cell (21.88 foci/cell) compared to 5Gy ionizing radiation

.

20 µM Olaparib 24h treatment induced a similar percentage of cells with ≥7 ƴH2AX foci per cell (p=0.9976) (fig. 4B). The number of foci formed with 100 µg/ml Bleomycin 1h differed from cell to cell, in contrast to cells treated with 3µM Olaparib 24 h and 5 Gy irradiation (fig. 5). This makes the 100 µg/ml Bleomycin 1h treatment less suitable to determine HR function in the HR assay with RAD51 foci scoring. The V79 Chinese Hamster lung fibroblast cell line was tested positive for mycoplasma. It cannot be excluded that this could have influenced the results.

Fig. 4 The number of ƴH2AX foci per cell after different treatments. All radiomimetic drug and Olaparib treatments induced DSBs compared to the 0 Gy control. A. The number of ƴH2AX foci per cell after treatment with the radiomimetic drugs Bleomycin, Zeocin and Phleomycin in V79 cell line. The tested concentrations differed between 100 and 500 µg/ml with a treatment of 1 h. 100 µg/ml Bleomycin 1h treatment induces a similar number of foci per cell (21.88 foci/cell) compared to 5Gy ionizing radiation. B. The number of ƴH2AX foci per cell after treatment with the PARPi Olaparib in V79 cell line. The tested concentrations differed between 3, 10 and 20 µM Olaparib with a treatment of 24 h. 20 µM Olaparib 24h treatment induces a similar percentage of cells with ≥7 ƴH2AX foci per cell (p=0.9976).

Fig 5. Microscopy results after IF staining for the 0 Gy 30 min, 5 Gy 30 min, 100 µg/ml Bleomycin 1h, 0 µM Olaparib 24 h and the 3 µM Olaparib 24 h treatment. DAPI staining was used to select for tumor cells in tumor tissue, based on morphology. No cells were tumor cells in V79 cell line. Geminin staining was used to select for replicating cells. In the V79 cell line, almost all cells were replicating. In the V79 cell line, ƴH2AX foci were formed in all treated samples and no ≥7 ƴH2AX foci per nucleus were formed in the control samples. The 5 Gy 30 min and the 3 µM Olaparib 24 h treatment induce a comparable number of ƴH2AX foci in each cell. The number of ƴH2AX foci formed with 100 µg/ml Bleomycin 1h differed from cell to cell. Exposure time set on: 100 ms for DAPI, 100 ms for geminin, 400 ms for ƴH2AX, and 100 for EdU.

The same concentrations as in the V79 cell line were used in the tumor tissues. In the V79 cell line, the drugs functioned properly with a treatment time of 1 h. In the tumor tissue the treatment time was prolonged to a minimum of 2 h, since tumor tissue contains more cell layers than a cell line. The treatments were performed on three endometrium, four ovarian and one cervix tumor. Preferably, more than 50 geminin positive cells were scored. When between 30 and 50 geminin positive cells were scores, it is noted in the figures. Samples with less than 30 geminin positive cells were excluded from the results. For each geminin positive cell, it was noted whether the cell was EdU negative or positive. EdU is marker for proliferation.

The results of the ƴH2AX foci and RAD51 foci in one ovarian tumor (OC156) are shown below (fig. 6A and B). Again all tested treatments induced DSBs compared to the 0 Gy control, visible in the percentage of cells with ≥7 ƴH2AX foci per nucleus. Besides, all tested treatments induced DSB repair compared to the 0 Gy control, visible in the percentage of cells with ≥7 RAD51 foci per nucleus. The 3 µM Olaparib 24 h showed a similar percentage of cells with ≥7 ƴH2AX foci to the 5 Gy 2 h treatment, making the 3 µM Olaparib 24 h treatment the most suitable as an alternative to irradiation in this tumor when looking at DSB induction (p=0.9976) (fig. 6A). No large differences in RAD51 staining was observed between different treatments. Almost all treatments show a comparable percentage of cells with ≥7 RAD51 foci per nucleus. The 500 µg/ml Bleomycin 2h has the exact same percentage of cells with ≥7 RAD51 foci per nucleus as the 5 Gy , making this the most suitable alternative to irradiation in this tumor when looking at DSB repair (p>0.9999) (fig. 6B).

The IF staining results of OC156 ovarian tumor are shown below, for both the ƴH2AX and RAD51 staining (fig. 7A and B). The results show the 0 Gy 2h, the 5 Gy 2h and the 3 µM Olaparib 24 h

treatment. The number of ƴH2AX and RAD51 foci is consistent in each cell in the 5 Gy 2h and the 3 µM Olaparib 24 h treatment.

Fig. 6 The results of ƴH2AX foci and RAD51 foci in one ovarian tumor (OC156). All tested treatments induced DSBs and DSB repair compared to the 0 Gy control, visible in the percentage of cells with ≥7 ƴH2AX foci and RAD51 foci per nucleus. A. The 3 µM Olaparib 24 h shows a similar percentage of cells with ≥7 ƴH2AX foci to the 5 Gy 2 h treatment, making the 3 µM Olaparib 24 h treatment the most suitable as an alternative to irradiation in this tumor when looking at DSB induction (p=0.9976) . B. The 500 µg/ml Bleomycin 2h has the exact same percentage of cells with ≥7 RAD51 foci per nucleus as the 5 Gy , making this the most suitable alternative to irradiation in this tumor when looking at DSB repair (p>0.9999).

Fig 7. The IF staining results of OC156 ovarian tumor for the ƴH2AX (A) and RAD51 (B) staining. The results show the 0 Gy 2h, the 5 Gy 2h and the 3 µM Olaparib 24 h treatment. The number of ƴH2AX and RAD51 foci is consistent in each cell in the 5 Gy 2h and the 3 µM Olaparib 24 h treatment. Exposure time for ƴH2AX staining is set on 50 ms for DAPI, 150 ms for geminin, 300 ms for ƴH2AX and 20 ms for

A B

A

EdU. Only the exposure time of the 0 Gy treatment was set on 50 ms for ƴH2AX. Exposure time for RAD51 staining was set on 70 ms for DAPI, 250 ms for geminin, 400 ms for RAD51 and 70 ms for EdU.

Results of RAD51 and ƴH2AX staining differed between tumors and within tumors. Two endometrial tumors (ET034 and ET046) showed similar induction of ƴH2AX foci as a result of Bleomycin treatments (supplementary fig 3). Both 100 and 500 µg/ml Bleomycin 2 h treatment showed many cell with ≥7 ƴH2AX foci per nucleus, compared to the 0 Gy 2 h and 5 Gy 2 h treatment. The third endometrium tumor (ET024B) showed DSB induction with all treatments, with the highest induction in the 500 µg/ml Bleomycin and µg/ml Phleomycin 2 h treatments (supplementary fig 4A). Proper RAD51 foci formation was visible in the 500 µg/ml Bleomycin 2h and the 3 µM Olaparib 24 h treatments (supplementary fig 4B). The cervix tumor (CT043) only showed ƴH2AX foci formation with Olaparib treatments (with all concentrations) and not with radiomimetic drug treatments (supplementary fig 5A). However, both the radiomimetic drug and Olaparib treatments showed a comparable percentage of cells with ≥7 RAD51 foci compared to the 5 Gy 2 h treatment, with the 3, 10 and 20 µM Olaparib 24 h being the most similar. Beside the OC156 tumor, three other ovarian tumors were tested. One ovarian tumor (OC141) showed a high percentage of cells with ≥7 ƴH2AX foci per nucleus, even in the 0 Gy 2h sample (supplementary fig. 6A). 3 and 10 µM Olaparib induced RAD51 foci formation in comparison to the 0 Gy 2h treatment (supplementary fig. 6B). Also OC192 ovarian tumor showed much ≥7 ƴH2AX foci formation in all samples, including the controls (supplementary fig. 7A). A high percentage of cells showed less than 7 RAD51 foci for each treatment (supplementary fig. 7B). In these samples no ƴH2AX foci data was obtained for the last ovarian tumor (OC193). The RAD51 foci formation was comparable between the 5 Gy 2 h and the 100 µg/ml Bleomycin 2h treatment (supplementary fig. 8).

Statistical analysis

ANOVA analysis in Graphpad Prism 7 was used to compare the results of each treatment to the results of the 5 Gy treatment per tumor. Both the percentages of cells with ≥7 ƴH2AX foci per nucleus and the percentage of cells with ≥7 RAD51 foci per nucleus were analyzed (α = 0.05). The standard deviation was set on 5%, meaning that a p-value is significant when 2.5 cells more or 2.5 cells less are in the ≥7 foci category compared to the 5 Gy treatment. The aim of the study is to find a treatment which causes a clear induction of DSBs. However, ANOVA analysis can only be used to determine what treatment shows the highest p-value when comparing percentage of cells with ≥7 ƴH2AX or RAD51 foci of a treatment to 5 Gy irradiation, and thereby showing the least difference to the 5 Gy treatment.

For each geminin positive cell, it was noted whether the cell was EdU negative or positive. EdU is marker for proliferation. Many of the tested tumors showed a low average percentage of EdU positive cells. A cell needs to be proliferating to form ƴH2AX and RAD51 foci. These EdU negative tumors showed inducing of ƴH2AX and RAD51 foci. Since most of the cells were not proliferating, these results are less reliable. A threshold with an average of at least 50% EdU positive cells per tumor was maintained, to only draw conclusions from reliable results and still have enough data left to draw a conclusion. The results of tumors with less than 50% EdU positive cells were excluded (i.e. CT043, OC141, OC192, OC193 tumors) from conclusions based on calculated p-values. In addition, there was no link between thawing the tumor tissue from ampules from liquid nitrogen and a lower percentage of EdU positive cells (supplementary fig 9).

Discussion

The aim of this study was to optimize the diagnostic implementation of the HR assay by focus on other potential DNA DSB inducers as an alternative to irradiation. The drugs Bleomycin, Zeocin, Phleomycin and Olaparib were tested in different concentration and with different treatment times in a V79 Chinese hamster lung fibroblast cell line and in tumor tissue. DNA DSBs were visualized by staining with yH2AX and accumulation of RAD51 was used as a marker for HR. In each sample 50 geminin positive tumor cells were scored as a cell with <7 or ≥7 yH2AX or RAD51 foci per nucleus. Samples with less than 30 geminin positive cells were excluded from the results. Also samples of tumors with an average of less than 50% EdU positive cells were excluded from the results, to draw

conclusions from more reliable results.

The percentage of geminin positive cells with ≥7 yH2AX and RAD51 foci per nucleus was compared between each treatment and the 5 Gy treatment. For each comparisons, a p-value was calculated. Since as little as possible difference between a treatment and irradiation is desired, a p-value close to 1 is preferable. Fig. 8 shows the treatments with the highest p-p-value for both yH2AX and RAD51 foci formation compared to 5 Gy sample in the same tumor. The 500 µg/ml Bleomycin 2 h treatment shows the most p-values close to 1. Therefore, based on p-values the 500 µg/ml Bleomycin 2 h treatment could be determined as the most comparable alternative to 5 Gy 2h radiation treatment.

Fig. 8 the treatments with the highest p-value of both yH2AX (A) and RAD51 (B) foci formation compared to 5 Gy sample in the same tumor. The 500 µg/ml Bleomycin 2 h treatment shows the most p-values close to one.

However, both ionizing radiation and Olaparib treatment induced a consistent number of yH2AX and RAD51 foci per nucleus in all cells. The number of foci formed with Bleomycin treated cells however, differed from cell to cell. Therefore, this drug not suitable to use in the HR assay. It might influence the determination whether a cell is HR proficient or deficient. Therefore, the 3 µM Olaparib 24 h treatment is determined as the most comparable alternative to 5 Gy irradiation based

on the results obtained in this study.

A possible explanation for the differences in number of foci per cell for Bleomycin treatment could be that the drugs do not directly interfere with DNA replication. (Chen & Stubbe, 2005). This could also explain why it was possible to score yH2AX foci in geminin negative cells. Markers for other cell phases could be used to determine how these drugs function. Besides, longer treatment times could be tested, because this might result in a more consistent number of foci per nucleus. The samples treated with Zeocin and Phleomycin showed inconsistent results in the yH2AX and RAD51 foci scoring data. Little is known about the function of Zeocin and Phleomycin, it is therefore difficult to determine the cause of the inconsistent foci induction. In further research,

higher concentrations of the drugs could be tested, since the drug did not induce as many DSB as expected (Moore, 1989; Krol et al., 2015; Chen & Stubbe, 2005). The difference in potential to induce DSBs by Bleomycin and Zeocin could be that Zeocin induces mostly SSBs and Bleomycin induces a lot

of DSBs as well (Moore, 1989). Since only replicating cells are capable to perform HR it is essential that the concentration of

the drug is high enough to induce DNA DSBs but does not induce cytotoxic effects e.g. an apoptotic response. In this study, the treatments at the used concentrations did not lead to changes in the frequency of replicating tumor cells, what was measured by the incorporation of EdU (Chen and Stubbe, 2005). This study shows that geminin positivity does not guarantee proliferation. The apoptotic state of a cell might be determined based on results from HE staining (Naipal et al., 2014; 2016). Therefore, the HE staining made from tested samples in this study could be used in further research to determine the right concentration to induce RAD51 foci.

ƴH2AX and RAD51 foci were counted manually, what might cause inconsistencies, because researchers might apply different criteria when scoring. For diagnostic purpose it would be preferred if we could apply automatic scoring. However, there are some technical issues why automatic scoring was not performed. The Starfrost microscope glasses, that are used because the adherence of the paraffin slices is better, give more background foci than normal microscope slices. Furthermore, because tumor tissue consist of different cell layers, it is currently not possible to automatically count foci within the whole nucleus of the cell. Therefore, automatic foci counting was no performed in this study.

Since the 3 µM Olaparib 24 h treatment was determined as the most suitable alternative to irradiation in the HR assay, this treatment should be optimized for diagnostic purposes. For instance, lower concentrations could be tested to lower the costs and toxicity. Besides, shorter treatment times could be tested, as the aim of the study is to make the HR assay as facile as possible for diagnosis. Eventually, tumor samples with differences in size, will be incubated in medium with a specific dose of the drug for a given time. The right combination of concentration and incubation time should be found for this application of the drug.

All treatments were performed using HR proficient tumors. Since the HR assay will also be applied on HR deficient tumors, further research should include these type of tumors. In first instance, HR assays should be executed in parallel, one assay using Olaparib treatment as a DSB inducer and one assay using the 5 Gy irradiation treatment. Only when the outcome of the analyses are concordant (i.e. sample is HR proficient or deficient), Olaparib can be used as a substitute for ionizing radiation. When using Olaparib, yH2AX staining should be included to check for induction of

DNA DSBs.

After further research will determine the most suitable concentration and incubation time for Olaparib, the drug could be used as a DSB inducer as an alternative to irradiation in the ex vivo HR assay. DSB induction by Olaparib, followed by staining for the HR functionality marker RAD51, can be used in a diagnostic setting to determine which women are susceptible to PARPi treatment.

References

Audeh, M. W., Carmichael, J., Penson, R. T., Friedlander, M., Powell, B., Bell-McGuinn, K. M., ... & Lu, K. (2010). Oral poly (ADP-ribose) polymerase inhibitor olaparib in patients with BRCA1 or BRCA2 mutations and recurrent ovarian cancer: a proof-of-concept trial. The Lancet, 376(9737), 245-251. Chen, J., & Stubbe, J. (2005). Bleomycins: towards better therapeutics. Nature Reviews Cancer, 5(2),

102-112.

Florea, A. M., & Büsselberg, D. (2011). Cisplatin as an anti-tumor drug: cellular mechanisms of activity, drug resistance and induced side effects. Cancers, 3(1), 1351-1371.

Frey, M. K., & Pothuri, B. (2017). Homologous recombination deficiency (HRD) testing in ovarian cancer clinical practice: a review of the literature. Gynecologic Oncology Research and Practice, 4(1), 4.

Hennessy, B. T., Timms, K. M., Carey, M. S., Gutin, A., Meyer, L. A., Flake, D. D., ... & Li, Y. (2010). Somatic mutations in BRCA1 and BRCA2 could expand the number of patients that benefit from poly (ADP ribose) polymerase inhibitors in ovarian cancer. Journal of Clinical Oncology, 28(22), 3570-3576. Konstantinopoulos, P. A., Ceccaldi, R., Shapiro, G. I., & D'Andrea, A. D. (2015). Homologous recombination deficiency: exploiting the fundamental vulnerability of ovarian cancer. Cancer discovery, 5(11), 1137-1154.

Krol, K., Brozda, I., Skoneczny, M., Bretne, M., & Skoneczna, A. (2015). A genomic screen revealing the importance of vesicular trafficking pathways in genome maintenance and protection against genotoxic stress in diploid Saccharomyces cerevisiae cells. PloS one, 10(3), e0120702.

Liddle, P., Lafon-Hughes, L., Di Tomaso, M. V., Reyes-Ábalos, A. L., Jara, J., Cerda, M., ... & Folle, G. A. (2014). Bleomycin-induced [gamma] H2AX foci map preferentially to replicating domains in CHO9 interphase nuclei. Chromosome research, 22(4), 463.

Lord, C. J., & Ashworth, A. (2012). The DNA damage response and cancer therapy. Nature, 481(7381), 287-294.

Lord, C. J., & Ashworth, A. (2016). BRCAness revisited. Nature Reviews Cancer, 16(2), 110-120.

Mukhopadhyay, A., Elattar, A., Cerbinskaite, A., Wilkinson, S. J., Drew, Y., Kyle, S., ... & Curtin, N. J. (2010). Development of a functional assay for homologous recombination status in primary cultures of epithelial ovarian tumor and correlation with sensitivity to poly (ADP-ribose) polymerase inhibitors. Clinical cancer research, 16(8), 2344-2351.

Mukhopadhyay, A., Plummer, E. R., Elattar, A., Soohoo, S., Uzir, B., Quinn, J. E., ... & Edmondson, R. J. (2012). Clinicopathological Features of Homologous Recombination–Deficient Epithelial Ovarian Cancers: Sensitivity to PARP Inhibitors, Platinum, and Survival. Cancer research, 72(22), 5675-5682. Murai, J., Huang, S. Y. N., Renaud, A., Zhang, Y., Ji, J., Takeda, S., ... & Pommier, Y. (2014). Stereospecific PARP trapping by BMN 673 and comparison with olaparib and rucaparib. Molecular cancer therapeutics, 13(2), 433-443.

Naipal, K. A., Verkaik, N. S., Ameziane, N., van Deurzen, C. H., Ter Brugge, P., Meijers, M., ... & Hoeijmakers, J. H. (2014). Functional ex vivo assay to select homologous recombination–deficient breast tumors for PARP inhibitor treatment. Clinical Cancer Research, 20(18), 4816-4826.

Naipal, K. A., Verkaik, N. S., Sánchez, H., van Deurzen, C. H., den Bakker, M. A., Hoeijmakers, J. H., ... & van Gent, D. C. (2016). Tumor slice culture system to assess drug response of primary breast cancer. BMC cancer, 16(1), 78.

Ordóñez, J. L., Amaral, A. T., Carcaboso, A. M., Herrero-Martín, D., del Carmen García-Macías, M., Sevillano, V., ... & Rodrigues, T. (2015). The PARP inhibitor olaparib enhances the sensitivity of Ewing sarcoma to trabectedin. Oncotarget, 6(22), 18875.

Pommier, Y., O’Connor, M. J., & de Bono, J. (2016). Laying a trap to kill cancer cells: PARP inhibitors and their mechanisms of action. Science translational medicine, 8(362), 362ps17-362ps17.

Stordal, B., & Davey, M. (2007). Understanding cisplatin resistance using cellular models. IUBMB life, 59(11), 696-699.

Supplementary

Fig 1. The number of ƴH2AX foci per cell after treatment with the radiomimetic drugs Bleomycin, Zeocin and Phleomycin in V79 cell line. The tested concentrations differed between 100 and 500 µg/ml and treatment times differed between 1 h (A) and 2 h (B). In most treatments, a longer treatment time or a higher concentration did not induce more DSBs.

Fig 2. Percentages of cells with <7 ƴH2AX foci per nucleus or ≥7 ƴH2AX foci per nucleus after treatment with Olaparib. The concentrations differed between 3, 10 and 20 µM Olaparib and treatment times differed between 24 h and 48 h.In most treatments, a longer treatment time or a higher concentration did not induce more DSBs.

Fig 3. Two endometrium tumors, ET034 (A) and ET046 (B), showed similar inducing of ƴH2AX foci as a result of Bleomycin treatments. Both 100 and 500 µg/ml Bleomycin 2 h treatment showed many cell with ≥7 ƴH2AX foci per nucleus, compared to the 0 Gy 2 h and 5 Gy 2 h treatment.

A B

Fig. 4 Results scoring ƴH2AX and RAD51 foci in endometrium tumor (ET024B). A. Induction of DSB was visible with all treatments, and the most with the 500 µg/ml Bleomycin and µg/ml Phleomycin 2 h treatments B. Proper RAD51 foci formation was visible in the 500 µg/ml Bleomycin 2h and the 3 µM Olaparib 24 h treatments.

Fig. 5 A. Cervix tumor (CT043) showed ƴH2AX foci formation with Olaparib treatments (with all concentrations) and not with radiomimetic drug treatments. B. All treatments had a comparable percentage of cells with ≥7 RAD51 foci compared to the 5 Gy 2 h treatment, with the 3, 10 and 20 µM Olaparib 24 h being the most similar.

Fig. 6 A. Ovarian tumor (OC141) showed a high percentage of cells with ≥7 ƴH2AX foci per nucleus, even in the 0 Gy 2h sample. B. 3 and 10 µM Olaparib induced RAD51 foci formation in comparison to the 0 Gy 2h treatment.

A B

A B

A B

A

Fig. 7 A. OC192 ovarian tumor showed much ≥7 ƴH2AX foci formation in all samples, including the controls. B. The RAD51 foci inducting showed a high percentage of <7 RAD51 foci for each treatment.

Fig. 8 The RAD51 foci formation was comparable between the 5 Gy 2 h and the 100 µg/ml Bleomycin 2h treatment.

Fig. 9 The results of tumors with less than 50% EdU positive cells were excluded from the final conclusion of the best alternative to irradiation. There was no link between thawing the tumor tissue from ampules from liquid nitrogen and a lower percentage of EdU positive cells.