VEGF pathway inhibition potentiates PARP inhibitor efficacy in ovarian cancer independent of BRCA status

Poly ADP-ribose polymerase inhibitors (PARPi) have transformed ovarian cancer (OC) treatment, primarily for tumours deficient in homologous recombination repair. Combining VEGF-signalling inhibitors with PARPi has enhanced clinical benefit in OC. To study drivers of efficacy when combining PARP inhibition and VEGF-signalling, a cohort of patient-derived ovarian cancer xenografts (OC-PDXs), representative of the molecular characteristics and drug sensitivity of patient tumours, were treated with the PARPi olaparib and the VEGFR inhibitor cediranib at clinically relevant doses. The combination showed broad anti-tumour activity, reducing growth of all OC-PDXs, regardless of the homologous recombination repair (HRR) mutational status, with greater additive combination benefit in tumours poorly sensitive to platinum and olaparib. In orthotopic models, the combined treatment reduced tumour dissemination in the peritoneal cavity and prolonged survival. Enhanced combination benefit was independent of tumour cell expression of receptor tyrosine kinases targeted by cediranib, and not associated with change in expression of genes associated with DNA repair machinery. However, the combination of cediranib with olaparib was effective in reducing tumour vasculature in all the OC-PDXs. Collectively our data suggest that olaparib and cediranib act through complementary mechanisms affecting tumour cells and tumour microenvironment, respectively. This detailed analysis of the combined effect of VEGF-signalling and PARP inhibitors in OC-PDXs suggest that despite broad activity, there is no dominant common mechanistic inter-dependency driving therapeutic benefit. Supplementary Information The online version contains supplementary material available at 10.1186/s13045-021-01196-x.

2 mg/kg (10 ml/kg). DDP was given once a week for 3 or 4 cycles. Vehicles were given with the same schedules and volumes as active compounds.

Antitumor activity
Tumor growth in subcutaneous models was measured with a Vernier caliper, and tumor volume was calculated as (length × width 2 ) / 2 = [mm 3 ]. OC-PDXs were randomized to treatment at approximately 280-330 mm 3 of tumor volume; OV2022 tumors were randomised at ~100mm 3 .
Mice were euthanized when tumor reached approximately 1500 mm 3 (and never beyond if it exceeded 15% of body weight). Tumor-free mice, as assessed by visual inspection approximately 200 days after the last dose, were considered cured. Differences in subcutaneous tumour volume were analysed by ANOVA and Tukey's post-test (or t test when only two groups were to be compared) all the days of measuring.
For orthotopic models, abdominal distension and palpable tumor masses in the peritoneal cavity were indicative of tumor growth [3]. Mice were randomized to treatment groups at an advanced stage (i.e. 25-30% of the expected median survival time), regularly monitored and euthanized as soon as they presented signs of discomfort (being considered the limit of survival) [3]. Survival time (ST) was recorded and Kaplan Meier curves generated. Increment of lifespan (ILS%) was calculated as [(median survival day of treated group − median survival day of control group) / median survival day of control group] x 100. A detailed autopsy was done to record the tumor burden in the peritoneal cavity. Pictures of the peritoneal cavity or organs were taken with a macrodigital imaging system (MacroPATH; Milestone S.r.l.). Tumor dissemination in representative organs of the peritoneal cavity (liver, diaphragm, omentum, pancreas, uterus/ovary, nodes) was ranked using an arbitrary score, previously described [3]: 0 = not infiltrated; 1 = small masses; 2 = evident masses; 3 = nearly completely invaded and 4 = completely invaded. Tumor burden was then calculated by summing the scores for each animal. Complete response, confirmed macroscopically at autopsy, was the absence of tumor in animals still alive on day 200 after transplant. Kaplan Meier survival curves were analyzed by Wilcoxon rank-sum test/log-rank test. Differences in abdominal tumor burden (ascites and dissemination scores) were analyzed by one way ANOVA and Tukey's post-hoc test.

Mutational analyses
Genomic DNA was obtained from tumor fragments or malignant ascites using a Maxwell ® 16 Tissue DNA purification kit and the Maxwell ® 16 Instrument for automated purification.

Whole Exome DNAseq
3 Two hundred nanograms of purified genomic DNA were used to construct a next generation sequencing (NGS) library. Libraries were generated using the Kapa Biosystems HyperPrep kit (following the manufacturer's protocol). All libraries were visualized on the Agilent TapeStation and the concentration was determined using the Kapa Biosystems NGS Library Quantification qPCR kit. Whole genome libraries were pooled and used for exome hybridization capture with xGen ® Exome Research Panel v1.0 (Integrated DNA Technologies). Libraries were sequenced on the Illumina HiSeq 4000 platform (2 x 150) using TruSeq SBS (sequencing by synthesis) reagents (Illumina). The NGS data was aligned and analyzed within the BCBio framework (https://bcbionextgen.readthedocs.org/en/latest/) using an AstraZeneca-developed variant calling algorithm (https://github.com/AstraZeneca-NGS/VarDict) [4]. For this study, 31 genes involved in homologous recombination repair (HRR) were analyzed (30 genes as reported in [5] plus FANCB).

Sanger sequencing
DNA (or mRNA/cDNA) were amplified with a GoTaq ® PCR Core System (Promega), using a thermocycler (2720 Thermal Cycler, Applied Biosystem) and ad hoc primers designed to detect specific mutations and amplify human but not murine variants (lack of amplification of purified mouse DNA (or mRNA/cDNA) was the benchwork technical control). PCR products were purified with Illustra™ GFX PCR and Gel Band Purification kit (GE Healthcare). and sequenced by Microsynth SeqLab (Switzerland). Electropherograms were analyzed, and mutations confirmed using Sequencer 5.1 software and FinchTV software.

Gene expression
Total RNA was extracted from tumor fragments or malignant ascites using QIAzol Lysis Reagent and Tissuelyser LT Adapter (Qiagen) and purified through RNeasy ® mini-Kit with the QIAcube ® robotic workstation (Qiagen). The concentration and purity of nucleic acid samples were determined by spectrophotometric analysis (Nanodrop ® spectrophotometer).

RNA-seq
RNA integrity was further checked using Agilent 6000 Nano Assay (Agilent Bioanalyzer ® ), considering only RIN numbers of 6 or more acceptable. RNA was quality re-validated using the Agilent TapeStation and Quant-iT RNA Assay Kit (Thermo Fisher) before submitting the totRNA for eukaryotic RNA-seq library prep and paired-end sequencing (50M reads/sample) on the Illumina HiSeq Platform at Novogene. Quality control of raw reads was done with the FastQC 0.11.9 tool. Adapters and low-quality reads were trimmed using Cutadapt 2.9 and the reads were then mapped on the Human Reference Genome GRCh38.99 using Star 2.6. The mapped reads were counted with FeatureCounts 2.0. Finally, the DeSeq2 package, available within Bioconductor, was 4 used to normalize counts, estimate biological variance and determine differential expression. Log2fold changes and adjusted p-values were generated for each class comparison. A heatmap representation of RNA-seq data was plotted using MeV version 4.9.0.

Immunohistochemistry
For immunohistochemical analyses, solid tumors fragments were collected after 28 days of therapy, 6 h after the last dose. Tumors and organs were fixed in 10% neutral buffered formalin and paraffin-

Effect of hypoxia in cell lines
To determine the effect of hypoxia on the expression of HRR proteins, the panel of ovarian cell lines (N=14) were incubated in standard media in the presence of either 20% oxygen or 0.1% oxygen for 24 h in a Baker Ruskinn (Maine, USA) InvivO 2 hypoxic chamber for 24 h. Cells were then lysed in 20 mM Tris (pH 7.5), 137 mM NaCl, 10% Glycerol, 50 mM NaF, 1 mM Na 3 VO4, 1% SDS, 1% NP40 substitute, protease inhibitors (Roche) and phosphatase inhibitor cocktails 2 and 3 (Sigma) on ice (4 o C). Lysates were sonicated, clarified by centrifugation. Equal amounts of protein were then Western blotted for PARP (CST cat 9542L), RAD51 (CST cat D4B10), H2AX (CST cat 9718), Vinculin (Abcam cat ab18058), and detected using the appropriate goat anti-mouse or -rabbit 6 HRP conjugated secondary antibodies and chemiluminescence reagent. Signal was imaged on a Gbox (Syngene, Maryland USA).

Cell proliferation assay to assess cediranib-mediated sensitization of cells to olaparib
Cells were plated in 96-well plates, allowed to attach overnight and were then dosed using the HP D300e Digital Dispenser (HP Life Science Dispensing) at the indicated concentrations of cediranib and olaparib. Cells were allowed to proliferate for 7 days at 37 o C in standard culture media in the presence or absence of the relevant compound. To determine proliferation a live cell count posttreatment was determined using a Sytox Green endpoint as previously described [7], and a

Assessment of regulation of RAD51 expression in SKOV3
To analyze changes in protein expression following treatment with cediranib, SKOV3 cells were