Abstract
PARP inhibitors are approved for the treatment of high-grade serous ovarian cancers (HGSOC). Therapeutic resistance, resulting from restoration of homologous recombination (HR) repair or replication fork stabilization, is a pressing clinical problem. We assessed the activity of prexasertib, a checkpoint kinase 1 (CHK1) inhibitor known to cause replication catastrophe, as monotherapy and in combination with the PARP inhibitor olaparib in preclinical models of HGSOC, including those with acquired PARP inhibitor resistance.
Prexasertib was tested as a single agent or in combination with olaparib in 14 clinically annotated and molecularly characterized luciferized HGSOC patient-derived xenograft (PDX) models and in a panel of ovarian cancer cell lines. The ability of prexasertib to impair HR repair and replication fork stability was also assessed.
Prexasertib monotherapy demonstrated antitumor activity across the 14 PDX models. Thirteen models were resistant to olaparib monotherapy, including 4 carrying BRCA1 mutation. The combination of olaparib with prexasertib was synergistic and produced significant tumor growth inhibition in an olaparib-resistant model and further augmented the degree and durability of response in the olaparib-sensitive model. HGSOC cell lines, including those with acquired PARP inhibitor resistance, were also sensitive to prexasertib, associated with induction of DNA damage and replication stress. Prexasertib also sensitized these cell lines to PARP inhibition and compromised both HR repair and replication fork stability.
Prexasertib exhibits monotherapy activity in PARP inhibitor–resistant HGSOC PDX and cell line models, reverses restored HR and replication fork stability, and synergizes with PARP inhibition.
The CHK1 inhibitor prexasertib is currently in clinical development as a single agent and in combination with both cytotoxic and targeted agents. In high-grade serous ovarian cancer (HGSOC), prexasertib monotherapy has demonstrated activity in patients with BRCA wild-type, platinum-resistant disease. Here, we tested the activity of prexasertib in both BRCA wild-type and BRCA1-mutated, PARP inhibitor–resistant HGSOC patient-derived xenograft (PDX) models and cell lines. These preclinical models were routinely sensitive to prexasertib, likely related to induction of DNA damage and replication stress. Additionally, prexasertib impaired both homologous recombination repair and replication fork stability, addressing 2 major mechanisms of resistance to PARP inhibition. Consequently, prexasertib was synergistic with olaparib in multiple models. These data suggest that prexasertib, either as monotherapy or in combination with olaparib, has broad activity against HGSOCs, including those harboring BRCA1 mutation with acquired PARP inhibitor resistance, informing additional ovarian cancer populations for clinical trials.
Introduction
High-grade serous ovarian cancer (HGSOC) is the most frequent and aggressive subtype of epithelial ovarian cancer (EOC), with high recurrence rates despite initial responses to platinum-based chemotherapy (1, 2). Approximately 50% of the HGSOCs exhibit genetic and epigenetic alterations in gene members of the homologous recombination (HR) DNA repair pathway including BRCA1/2 and other Fanconi Anemia/BRCA pathway genes (3, 4). Based on the synthetic lethal interaction of defective HR repair and PARP inhibition, HR-deficient EOCs are highly sensitive to PARP inhibitors (5). Currently, olaparib, rucaparib, and niraparib are all FDA-approved for HGSOC (6); in the recurrent disease setting, these agents have been primarily used in patients with tumors harboring BRCA alterations with substantial response rates. Additionally, these agents are now commonly considered in the maintenance setting, after a response to a platinum-based chemotherapy, where PARP inhibition has significantly increased progression-free survival and the chemotherapy-free interval for many patients (7, 8).
With more common PARP inhibitor use, acquired resistance has emerged as an unmet medical need. The major mechanisms of PARP inhibitor resistance include restoration of HR repair or stabilization and protection of replication forks (9–12). There is a critical need for combinations utilizing agents that can inhibit HR and/or reverse replication fork stability and thereby sensitize HGSOCs to PARP inhibition. Such approaches may also extend the use of PARP inhibitors to include HR-proficient, BRCA wild-type tumors.
Because of the endogenous DNA damage caused by HR deficiency, HGSOCs tumors require an intact ataxia telangiectasia and Rad3-related (ATR)–checkpoint kinase 1 (CHK1) pathway to allow time for alternative DNA repair pathways to function and maintain genome integrity. Additionally, HGSOCs often demonstrate genomic alterations suggestive of replicative stress. For example, ∼20% of HGSOCs may express high levels of cyclin E (4, 13), often as a result of gene amplification. Tumors with replication stress are also dependent on the ATR–CHK1 pathway, in this case for maintaining replication fork stability (14). Based on its role in DNA-damage induced checkpoint control (15, 16) and control of replication origin firing (17–19), CHK1 is an attractive therapeutic target for HGSOC. CHK1 deficiency can exacerbate both DNA damage and replication stress, leading to cell death (16, 20–23). Additionally, the roles of CHK1 in both replication fork stability (18, 24) and HR repair (23, 25) suggest that CHK1 deficiency may also reverse PARP inhibitor resistance.
Prexasertib (LY2606368) is a potent inhibitor of CHK1, and to a lesser extent, CHK2, and is currently being tested in phase 1 and 2 clinical trials (26, 27). Among patients with HGSOC, preliminary activity has been demonstrated in BRCA wild-type, platinum-resistant and refractory disease (27). As a single agent, prexasertib causes replication catastrophe, DNA double-strand breaks, HR defects, and apoptosis (28, 29). We therefore investigated the efficacy of prexasertib as a monotherapy and in combination with PARP inhibition in 14 HGSOC patient-derived xenograft (PDX) models, and in a panel of ovarian cancer cell lines. Prexasertib showed broad preclinical monotherapy activity and was synergistic with PARP inhibition in several models, including those representing BRCA1-mutant, PARP inhibitor–resistant disease, where there was reversal of both restored HR repair and of replication fork stability.
Materials and Methods
Establishment and treatment of PDXs
Tumor ascites was collected from patients with suspected or established ovarian cancer at the Brigham and Women's Hospital or the Dana-Farber Cancer Institute (DFCI) under IRB-approved protocols conducted in accordance with the Declaration of Helsinki and the Belmont Report. Written informed consent was obtained from patients when required. Fourteen ascites-derived ovarian PDX models were established, luciferized, and propagated at the DFCI and have been described previously (30). All animal studies were performed in accordance with DFCI Institutional Animal Care and Use Committee guidelines per DFCI-approved animal protocols. Treatment studies in mice bearing PDX tumors were performed as described previously (30, 31). Approximately 2 to 10 × 106 luciferized PDX cells derived from mouse ascites were injected intraperitoneally into 8-week-old female NSG mice. Tumors were typically established 1 to 2 weeks after implantation. Tumor burden was measured by bioluminescence imaging (BLI) using a Xenogen IVIS-200 system (Xenogen). Mice were randomized into treatment groups and disseminated disease progression in the peritoneal cavity was measured serially once per week by BLI.
Prexasertib (mesylate monohydrate salt of LY2606368; Eli Lilly) was formulated in 20% Captisol (CyDex Inc.) pH 4.0 and administered subcutaneously at 8 mg/kg twice daily for 3 days, followed by 4 days of rest and repeated for 2 additional cycles. Olaparib (ChemExpress) was formulated in PBS containing 10% DMSO and 10% (w/v) 2-hydroxy-propyl-β-cyclodextrin (Sigma-Aldrich) and administered orally at 100 mg/kg daily for 3 weeks. Tumor-bearing mice were monitored for survival and tumor burden for approximately 100 to 150 days. For pharmacodynamic studies, mice were euthanized at specific time points and tumor cells were harvested from ascites.
Cell lines
OVCAR3, UWB1.289, UWB1.289+ BRCA1, ES2, OV90, SKOV3, PA-1, SW-626, SKOV3, and CAOV3 cell lines were purchased from ATCC. KURAMOCHI, OVSAHO, and OVTOKO cell lines were purchased from JCRB. TOV21G, TOV21G+FANCF, PEO1, PEO4, COV362, A2780, JHOS2, JHOS4, and CAOV4 cell lines were obtained from other laboratories. Cell line identities were confirmed by STIR profiling. The generation of UWB1.289-SYR12 and UWB1.289-SYR14 olaparib-resistant cell lines has been described previously (32). UWB1.289 cells were grown in RPMI (ATCC-R-30-2001) and MEGM Bullet Kit (Lonza CC-3150) at a 1:1 ratio, supplemented with 3% FBS and 1% penicillin/streptomycin (Lonza 17-602E). All other cell lines were cultured in RPMI (GIBCO 11875) supplemented with 10% FBS (GIBCO 7471) and 1% penicillin/streptomycin (Lonza 17-602E).
siRNAs
The siRNA sequence targeting 53BP1 was sense 5′-AGAACGAGGAGACGGUAAUAGUGGG-3′, antisense 5′-CCCACUAUUACCGUCUCCUCGUUCU-3′ (33). The control nontargeting siRNA was purchased from Dharmacon.
Western blotting and antibodies
Cells were lysed in a buffer containing 20 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 1 mmol/L Na2EDTA, 1 mmol/L EGTA, 1% Triton, 2.5 mmol/L sodium pyrophosphate, 1 mmol/L β-glycerophosphate, 1 mmol/L sodium vanadate, and 1 μg/mL leupeptin (Cell Signaling Technology, 9803S) with protease inhibitor PMSF (Cell Signaling Technology, 8553S). Western blots were performed with the following antibodies: CHK1 (G-4; Santa Cruz Biotechnology, SC-8408), phospho-CHK1 [Ser345] (Cell Signaling Technology, 2348S), phospho-histone H2AX [S139] (EMD, 05-636), phospho-RPA32 [S4/S8] (Bethyl, A300-245A), phospho-KAP1 [S824] (Abcam, ab84077), vinculin (Santa Cruz Biotechnology, SC-25336), RAD51 (EMD PC-130), RAD51C (Novus), and Cyclin E (Santa Cruz Biotechnology, SC-247).
Drug sensitivity assays
Stock solutions of olaparib (Selleck S1060; 100 mmol/L) and prexasertib (40 mmol/L) were prepared in dimethyl sulfoxide (DMSO) and stored in aliquots at −80°C. Appropriate dilutions were prepared in culture medium. For viability assays, cells were seeded in 96-well plates (2,000–4,000 cells/well) and exposed to DMSO or graded concentrations of prexasertib, olaparib, or a combination of both drugs for 3 to 6 days. Cell survival was determined by CellTiter-Glo (Promega, G7571). Cell viability was calculated relative to DMSO-treated samples and dose–response curves or IC50 plots were generated using GraphPad Prism. Synergy between prexasertib and olaparib was calculated using Combenefit software as described previously (34). For colony formation assays, 500 to 2,000 cells/well were plated in 6-well plates along with graded concentrations of drugs and cultured for 10–14 days for colony growth. Colonies were then fixed with methanol for 10 minutes at −20°C and stained with crystal violet. Plates were imaged with ImageQuant Las4000 (GE Healthcare) and quantification of the cell growth area was performed with ImageJ.
Immunohistochemistry
Ascites-derived tumor cells were harvested from euthanized mice, washed with saline, and transferred to DMEM-F12 supplemented with 10% FBS. Samples were divided into 2 aliquots, 1 of which was subjected to 10 Gy γ-irradiation (IR). Both unirradiated and irradiated samples were incubated for 4 hours at 37°C and fixed in 4% neutral buffered formalin for 10 minutes. Postfixation, cells were washed with PBS and encapsulated in Histogel (Richard Alen Scientific) embedded in paraffin and sectioned for IHC with antibodies to RAD51 (35), 53BP1 (Millipore, MAB3802), or phospho-histone H2AX (MiliporeSigma, 05-636).
Genomic analysis
Genomic DNA was isolated from ascites-derived tumor cells from PDX models and the Molecular Inversion Probe (MIP) assay (OncoSan3 platform) was performed. Resulting data were processed using Chromosome Analysis Suite (ChAS). Allelic copy number analysis was performed using Nexus Express (Affymetrix). Whole-exome sequencing of genomic DNA from ascites-derived tumor cells was performed using the Illumina HiSeq 4000 and analyzed by members of the DFCI Center for Cancer Computational Biology.
For bi-sulfite sequencing, genomic DNA was treated with bi-sulfite using EZ DNA Methylation-Direct Kit (ZYMO Research) and then the promoter region of RAD51C was amplified using specific primers (Forward primer: ATGGTGTATAAGTGTGAAAATTTATAAGA; reverse primer: CCTCTAAAAATTCCTCAACAATCTAAA). The PCR product was sequenced by Sanger sequencing.
To confirm the 14 bp deletion at the exon–intron boundary and the mutation identified at the TP53BP1 locus in the DF59 PDX model, PCR was performed on genomic DNA using primers flanking the deletion site (primer sequence 53BP1 exon12-intron12 boundary: GCCCATGCCTCACAAAGCTT, AAGCATGCAGTTCTCGCCAA). To verify the presence of the mutation, mutation specific primers were used, including primer sequence 53BP1 exon 26 T>C (p.L1806P) specific forward primer: AACACAGCTTACCAGTGTCC; and consensus reverse primer: ACGGTAGTTCTGGAGCTGGT. Real-time PCR analysis was performed using total RNA isolated from the DF83, DF20 and DF181 PDX models. Primer sequences for RAD51 expression were: TGCGCCCACAACCCATTTCAC, TCAATGTACAACCCATTTCCTTCAC. Primer sequences for RAD51C expression were: AAGACGTTCCGCTTTGAA, GCATTGCCAGAGGAAAAC. CCNE1 expression in PDX tumor cells was determined by qRT-PCR using a TaqMan Kit (Applies Biosystems, 4331182).
Immunofluorescence for analysis of RAD51 foci
Cells were plated on autoclaved coverslips and allowed to grow for 24 hours before exposure to drugs or IR, followed by fixation in 4% paraformaldehyde for 15 minutes at room temperature. The cells were then washed with PBS and permeabilized using cold methanol for 2 minutes. After blocking with 10% normal goat serum containing 0.3% Triton X-100 for 1 hour, cells were stained with primary anti-RAD51 antibody (EMD, PC130; diluted 1:1,000 in PBS containing 1% BSA and 0.3% triton X-100) followed by staining with goat anti-rabbit IgG (H+L) Alexa Fluor-488 labeled secondary antibody (Thermo Fisher Scientific, A11034; diluted 1:1,500 in PBS containing 1% BSA and 0.3% Triton X-100). Coverslips were then mounted with a solution containing DAPI (Thermo Fisher Scientific, p36935) and cells were visualized under the fluorescence microscope.
DNA fiber assay
Replication fork stability of UWB1.289 cells was determined by the DNA fiber assay as described previously (36) using the FiberComb Molecular Combing System (Genomic Vision). A total of 200,000 cells/well were plated in 6-well plates and cultured for 48 hours before treatment with 5-Iodo-2′-deoxyuridine (IdU; Sigma, I7125) for 45 minutes, 5-Chloro-2′-deoxyuridine (CldU; Sigma, C6891) for 45 minutes, followed by hydroxyurea (HU) with or without prexasertb for 4.5 hours, with 3 PBS washes between each treatment. Cells were then embedded in low melting point agarose plugs, followed by proteinase K treatment and agarose digestion, after which DNA samples were combed onto silanized coverslips. The labeling of CldU and IdU was detected by staining with rat anti-BrdUrd antibody (clone BU1/75-ICR) specific to CldU (Abcam, ab6326) and mouse anti-BrdUrd antibody specific to IdU (BD Biosciences, 347580), followed by staining with chicken anti-mouse IgG Alexa Fluor-594 (Red; Life Technology, A-11031) and goat anti-rat IgG Alexa Fluor-488 (green; Life Technology, A-11006) secondary antibodies. DNA fibers were visualized by fluorescence microscopy and images were taken of at least 100 fibers per condition, quantified with ImageJ and graphed.
Statistical analysis
For monotherapy efficacy studies in vivo, a group size of n = 5 was utilized in order to provide >95% power to detect a difference of at least 50% in tumor growth inhibition between vehicle-treated mice and drug-treated mice using the Mann–Whitney–Wilcoxon test. For combinatorial efficacy studies, a group size of n = 9 or 10 was utilized in order to provide >95% power to detect a difference of at least 30% in tumor growth inhibition between combination and monotherapy-treated mice or drug- and vehicle-treated mice using the Mann–Whitney–Wilcoxon test. Survival estimates were computed using the Kaplan–Meier method and survival curves were compared between monotherapy and combination treated groups. GraphPad Prism was used to graph the data and carry out statistical analyses. A P value < 0.05 was considered statistically significant.
Results
PARP inhibitor–resistant BRCA1-mutant PDX models of HGSOCs exhibit restoration of HR repair
To determine the preclinical activity of prexasertib in ovarian cancer, we used PDX models of HGSOC. Fourteen ascites-derived PDX models from patients who had received platinum chemotherapy have been clinically-annotated, luciferized, characterized by IHC, and analyzed for BRCA status and other genomic changes (Fig. 1A; Supplementary Table S1; ref. 30). These models retain genetic alterations observed in the corresponding original tumors. One BRCA wild-type model was sensitive to the PARP inhibitor olaparib. However, the remaining 13 models, including 4 derived from BRCA1 germline carriers (DF59, DF68, DF86, and DF101), were resistant to olaparib (Fig. 1A; Supplementary Fig. S1A).
PARP inhibitor–resistant BRCA1-deficient ovarian cancer PDX models exhibit restoration of HR. A, BRCA status and PARP inhibitor sensitivity of 14 HGSOC PDX models. Four of the 14 models are BRCA1-mutated, and 13 of the models are olaparib resistant. B, RAD51 IHC staining in irradiated tumor cells from PDX models showing presence of RAD51 foci in 4 BRCA1-mutated PDX models and absence of RAD51 foci in the BRCA-WT DF83 model. FFPE tissue sections of the 14 PDX models were stained using an anti-RAD51 antibody, and representative images (60×) are shown. C, Bisulfite sequencing of tumor cells from PDX models showing hypermethylation of the RAD51C locus in the DF83 model. TSS indicates the transcription start site. Human genome build (hg18) was used as a control. D, qRT-PCR analysis showing absence of RAD51C mRNA expression (right) and reduced RAD51 mRNA expression (left) in the DF83 PDX model. ****, P < 0.0001; *, P = 0.0210 using the 1-way ANOVA (n = 4). E, Western blots of tumor lysates from the DF59, DF83, and DF20 PDX models. F, 53BP1 IHC staining of irradiated tumor cells from BRCA1-mutant PDX models showing absence of 53BP1 protein levels in the DF59 PDX model. FFPE tissue sections of the 4 BRCA1-mutant models were stained using an anti-53BP1 antibody. Representative images (60×) are shown. Subnuclear 53BP1 foci are present in all models except DF59. G, MIP-SNP array (OncoScan3 platform) analysis showing allelic copy number loss of the 15q segment harboring the TP53BP1 locus in ascites-derived tumor cells from the DF59 PDX model. Exome sequencing identified 2 mutations in the ascites-derived DF59 cells; 53% of the reads showed a 14 bp deletion at the end of exon 12, and 32.5% of the reads carried a missense mutation L1806P. A schematic representation of these events is shown. Results were verified using genomic PCR (H). The 14-bp deletion in ∼50% of the genomic DNA was observed as a lower migrating band in the PCR (top). A mutation-specific primer was used to verify the presence of the T>C mutation in the DF59 model (middle). DF181 was used as the control in the PCR analyses, and GAPDH was used as the loading control.
PARP inhibitor–resistant BRCA1-deficient ovarian cancer PDX models exhibit restoration of HR. A, BRCA status and PARP inhibitor sensitivity of 14 HGSOC PDX models. Four of the 14 models are BRCA1-mutated, and 13 of the models are olaparib resistant. B, RAD51 IHC staining in irradiated tumor cells from PDX models showing presence of RAD51 foci in 4 BRCA1-mutated PDX models and absence of RAD51 foci in the BRCA-WT DF83 model. FFPE tissue sections of the 14 PDX models were stained using an anti-RAD51 antibody, and representative images (60×) are shown. C, Bisulfite sequencing of tumor cells from PDX models showing hypermethylation of the RAD51C locus in the DF83 model. TSS indicates the transcription start site. Human genome build (hg18) was used as a control. D, qRT-PCR analysis showing absence of RAD51C mRNA expression (right) and reduced RAD51 mRNA expression (left) in the DF83 PDX model. ****, P < 0.0001; *, P = 0.0210 using the 1-way ANOVA (n = 4). E, Western blots of tumor lysates from the DF59, DF83, and DF20 PDX models. F, 53BP1 IHC staining of irradiated tumor cells from BRCA1-mutant PDX models showing absence of 53BP1 protein levels in the DF59 PDX model. FFPE tissue sections of the 4 BRCA1-mutant models were stained using an anti-53BP1 antibody. Representative images (60×) are shown. Subnuclear 53BP1 foci are present in all models except DF59. G, MIP-SNP array (OncoScan3 platform) analysis showing allelic copy number loss of the 15q segment harboring the TP53BP1 locus in ascites-derived tumor cells from the DF59 PDX model. Exome sequencing identified 2 mutations in the ascites-derived DF59 cells; 53% of the reads showed a 14 bp deletion at the end of exon 12, and 32.5% of the reads carried a missense mutation L1806P. A schematic representation of these events is shown. Results were verified using genomic PCR (H). The 14-bp deletion in ∼50% of the genomic DNA was observed as a lower migrating band in the PCR (top). A mutation-specific primer was used to verify the presence of the T>C mutation in the DF59 model (middle). DF181 was used as the control in the PCR analyses, and GAPDH was used as the loading control.
To assess the HR-mediated DNA repair status in these models, we developed a RAD51-based IHC assay. Freshly derived tumor cells from PDX-bearing mice were exposed to radiation and stained with an anti-RAD51 antibody. Consistent with the olaparib response, the 13 olaparib-resistant models exhibited DNA-damage induced RAD51 foci (Fig. 1B). The presence of RAD51 foci in the BRCA1-mutant models suggested that restoration of HR had occurred. In contrast, the olaparib-sensitive PDX model DF83 showed absence of RAD51 foci (Fig. 1B). Bisulfite sequencing demonstrated hypermethylation of the RAD51C locus (Fig. 1C) along with low expression of both RAD51C and RAD51 mRNAs (Fig. 1D) and their encoded proteins (Fig. 1E), indicating that RAD51C, a known FA gene (FANCO) and regulator of HR repair (37, 38), is silenced in this model, accounting for HR deficiency and PARP inhibitor sensitivity.
Because loss of 53BP1 is known to restore HR (9, 39) and confer PARP inhibitor resistance in BRCA1-mutant cancer cells (9), we assessed the levels of 53BP1 in the 4 BRCA1-mutant, olaparib-resistant models. DF59 showed low expression of 53BP1 staining by IHC (Fig. 1F). CGH and MIP-SNP array analysis showed allelic deletion in the TP53BP1 locus (30). Whole- exome sequencing of this tumor identified 2 mutations in the TP53BP1 locus. A 14-base pair deletion at the end of exon 12 was present in 53% and the missense mutation L1806P was present in 32.5% of the reads, respectively (Fig. 1G), confirmed by PCR (Fig. 1H). These results confirm that BRCA1-mutant HGSOCs can restore HR and become PARP inhibitor resistant by mutation of TP53BP1.
Prexasertib exhibits monotherapy activity in PARP inhibitor–resistant HGSOC PDX models
We next sought to determine whether prexasertib has activity in the PARP inhibitor–resistant PDX models. Prexasertib monotherapy demonstrated significant antitumor activity across all of the models (Supplementary Fig. S1A), including the 4 that are BRCA1-mutated, PARP inhibitor–resistant, as well as the remaining BRCA wild-type models, several of which demonstrate moderately elevated CCNE1 mRNA expression and one with extremely high expression (DF172) (Supplementary Table S1; Supplementary Fig. S1B). Of note, prexasertib monotherapy activity was not restricted to the models with the highest levels of CCNE1 expression.
For 3 models, the activity was confirmed in larger cohorts of mice, including 2 BRCA1-mutated models (DF101 and DF86) and 1 BRCA wild-type model with high MYC expression (DF149; Fig. 2; Supplementary Table S1). These 3 models were resistant to olaparib as expected, but sensitive to prexasertib, with profound and durable tumor regression. Prexasertib was overall well tolerated with 5% to 15% loss of body weight during the treatment period (Supplementary Fig. S2A). Collectively, these results demonstrate that prexasertib has monotherapy activity in PARP inhibitor–resistant HGSOC PDX models, including BRCA-mutant tumors and tumors with genomic features indicative of replication stress.
Prexasertib exhibits monotherapy activity in PARP inhibitor–resistant HGSOC PDX models. NSG mice bearing luciferized PDXs were treated with vehicle, olaparib (100 mg/kg, daily), or prexasertib (8 mg/kg, BIDx3, rest 4 days; n = 5 mice per group for each model) for 3 weeks. Tumor growth was monitored by weekly bioluminescence imaging of the mice for approximately 125 days. Tumor measurements are represented as mean ± SE. P values were determined using the Mann–Whitney test. A, Left. Representative images of mice bearing DF101 tumors. Right. Tumor growth in mice bearing DF101 xenografts. **, P = 0.0079, for vehicle vs. prexasertib at day 22. B, Tumor growth in mice bearing DF86 xenografts. **, P = 0.0079, for vehicle vs. prexasertib at day 36. C, Tumor growth in mice bearing DF149 xenografts. **, P = 0.0079, for vehicle vs. prexasertib at day 22.
Prexasertib exhibits monotherapy activity in PARP inhibitor–resistant HGSOC PDX models. NSG mice bearing luciferized PDXs were treated with vehicle, olaparib (100 mg/kg, daily), or prexasertib (8 mg/kg, BIDx3, rest 4 days; n = 5 mice per group for each model) for 3 weeks. Tumor growth was monitored by weekly bioluminescence imaging of the mice for approximately 125 days. Tumor measurements are represented as mean ± SE. P values were determined using the Mann–Whitney test. A, Left. Representative images of mice bearing DF101 tumors. Right. Tumor growth in mice bearing DF101 xenografts. **, P = 0.0079, for vehicle vs. prexasertib at day 22. B, Tumor growth in mice bearing DF86 xenografts. **, P = 0.0079, for vehicle vs. prexasertib at day 36. C, Tumor growth in mice bearing DF149 xenografts. **, P = 0.0079, for vehicle vs. prexasertib at day 22.
Prexasertib induces DNA damage and replication stress in HGSOC PDX models
To determine whether prexasertib causes DNA damage and replication stress in HGSOC PDX models, we assessed pharmacodynamic markers of prexasertib activity, using DF59, the PARP inhibitor–resistant model with absent TP53BP1 expression (Fig. 3A). Mice bearing DF59 xenografts were exposed to vehicle, prexasertib, olaparib, or the combination. Lysates from tumors harvested at variable timepoints were subjected to Western blotting for p-CHK1[S345], a pharmacodynamic marker of CHK1 inhibition, markers of DNA damage, including pKAP1 and γ-H2AX, as well as a marker of replication stress, pRPA32. Prexasertib-mediated CHK1 inhibition is expected to result in increased phosphorylation at the S345 ATR phosphorylation site (28, 29). As expected, olaparib alone did not cause significant DNA damage in this model, although small increases in RAD51 and γ-H2AX foci were observed by IHC (Supplementary Fig. S3). In contrast, prexasertib, either alone or in combination with olaparib, caused DNA damage as early as 6 hours, with continued DNA damage and replication stress evident after 52 hours of exposure (Fig. 3A; Supplementary Fig. S3).
Prexasertib causes DNA damage and synergizes with PARP inhibition in ovarian cancer HGSOC PDX models. A, Prexasertib causes DNA damage in PDX tumors. NSG mice bearing luciferized PDX model DF59 were treated with vehicle (n = 2 mice), olaparib (100 mg/kg; n = 3 mice), prexasertib (8 mg/kg, twice a day; n = 3 mice), or olaparib (100 mg/kg) + prexasertib (8 mg/kg, twice a day; n = 3 mice). Lysates from individual tumors harvested at 6 hours (left) and 52 hours (right) after dosing were subjected to Western blotting with the indicated antibodies. In the left panel, data are from a single mouse in each group. B and C, Tumor growth (left) and survival (right) in mice bearing DF59 or DF83 PDX tumors. NSG mice bearing luciferized PDXs, namely DF59 (n = 10 mice/group) or DF83 (n = 9 mice/group) were treated with vehicle, olaparib (100 mg/kg, daily), prexasertib (8 mg/kg, twice day ×3, rest 4 days), or olaparib (100 mg/kg, daily) + prexasertib (8 mg/kg, twice day ×3, rest 4 days) for 3 weeks. Tumor growth by weekly bioluminescence imaging and survival of tumor-bearing mice were monitored. Tumor measurements are represented as mean ± SE. In B and C, P values for tumor growth curves were determined using the Mann–Whitney test. In the DF59 model (B), **P = 0.002 for prexasertib vs. prexasertib+olaparib at day 35; ****, P < 0.0001 for prexasertib vs. vehicle at day 21. In the DF83 model (C), ****P < 0.0001 for prexasertib vs. prexasertib+olaparib at day 70 and for vehicle vs. olaparib, as well as vehicle vs. prexasertib group at day 42. In the right-hand graphs, log-rank survival plots were generated using Prism GraphPad. In the DF59 model (B), P = 0.0005 for vehicle vs. olaparib; P < 0.0001 for vehicle vs. prexasertib; P = 0.0045 for prexasertib vs. prexasertib+olaparib; and P = 0.0010 for olaparib vs. prexasertib+olaparib. In the DF83 model (C), P < 0.0001 between all groups except olaparib vs. prexasertib.
Prexasertib causes DNA damage and synergizes with PARP inhibition in ovarian cancer HGSOC PDX models. A, Prexasertib causes DNA damage in PDX tumors. NSG mice bearing luciferized PDX model DF59 were treated with vehicle (n = 2 mice), olaparib (100 mg/kg; n = 3 mice), prexasertib (8 mg/kg, twice a day; n = 3 mice), or olaparib (100 mg/kg) + prexasertib (8 mg/kg, twice a day; n = 3 mice). Lysates from individual tumors harvested at 6 hours (left) and 52 hours (right) after dosing were subjected to Western blotting with the indicated antibodies. In the left panel, data are from a single mouse in each group. B and C, Tumor growth (left) and survival (right) in mice bearing DF59 or DF83 PDX tumors. NSG mice bearing luciferized PDXs, namely DF59 (n = 10 mice/group) or DF83 (n = 9 mice/group) were treated with vehicle, olaparib (100 mg/kg, daily), prexasertib (8 mg/kg, twice day ×3, rest 4 days), or olaparib (100 mg/kg, daily) + prexasertib (8 mg/kg, twice day ×3, rest 4 days) for 3 weeks. Tumor growth by weekly bioluminescence imaging and survival of tumor-bearing mice were monitored. Tumor measurements are represented as mean ± SE. In B and C, P values for tumor growth curves were determined using the Mann–Whitney test. In the DF59 model (B), **P = 0.002 for prexasertib vs. prexasertib+olaparib at day 35; ****, P < 0.0001 for prexasertib vs. vehicle at day 21. In the DF83 model (C), ****P < 0.0001 for prexasertib vs. prexasertib+olaparib at day 70 and for vehicle vs. olaparib, as well as vehicle vs. prexasertib group at day 42. In the right-hand graphs, log-rank survival plots were generated using Prism GraphPad. In the DF59 model (B), P = 0.0005 for vehicle vs. olaparib; P < 0.0001 for vehicle vs. prexasertib; P = 0.0045 for prexasertib vs. prexasertib+olaparib; and P = 0.0010 for olaparib vs. prexasertib+olaparib. In the DF83 model (C), P < 0.0001 between all groups except olaparib vs. prexasertib.
Combination of prexasertib and olaparib is synergistic in a subset of HGSOC PDX models
We next tested the efficacy of the combination of prexasertib and olaparib in the DF59 model (Fig. 3B). Olaparib alone had no activity; however, the combination of prexasertib and olaparib demonstrated substantial tumor growth inhibition, greater than that observed with prexasertib alone, along with prolonged survival, consistent with CHK1 inhibitor-mediated sensitization to PARP inhibition. Combination treatment was well tolerated with approximately 10% loss of body weight during the time of drug exposure (Supplementary Fig. S2B). We also assessed the combination in the HR-deficient olaparib-sensitive PDX model, DF83 (Fig. 3C; Supplementary Fig. S2B). Here, the combination augmented the degree and duration of response compared with either of the monotherapies, again translating to improved survival. Collectively, these results suggest that the combination of prexasertib and olaparib may be relevant in both PARP inhibitor–resistant and sensitive backgrounds.
Prexasertib reduces viability across a panel of HGSOC cell lines
We next evaluated prexasertib activity in a panel of ovarian cancer cell lines including BRCA wild-type lines, BRCA-mutant lines, and cell lines with high cyclin E expression suggestive of replication stress. Prexasertib was cytotoxic in most of the ovarian cancer cell lines with IC50 values of 1 to 10 nmol/L in a Cell Titer-Glo viability assay (Fig. 4A), regardless of BRCA status or cyclin E expression (Supplementary Fig. S4). The activity of prexasertib was lowest in the BRCA1-mutant, TP53-mutant JHOS2 cell line, which demonstrated an IC50 of 8.4 μmol/L (Fig. 4A). Prexasertib exposure resulted in increased levels of pKAP1, pRPA32, and γ-H2AX in sensitive cell lines, whereas these events were absent in resistant JHOS2 cells (Fig. 4B), suggesting that monotherapy activity may be mediated by the induction of DNA damage and replication stress.
Ovarian cancer cells are highly sensitive to prexasertib. A, IC50 of prexasertib in ovarian cancer cell lines. Cells were grown for 24 hours before exposure to graded concentrations of prexasertib. Viability was assessed after 3 days of treatment using Cell Titer-Glo. IC50 values for growth inhibition were calculated by Prism GraphPad and mean ± SE (n = 4) are shown. The experiments were done at least 2 times for each cell line, and the data from a representative experiment are shown. B, Western blots of lysates from prexasertib-sensitive ovarian cancer cells (ES2, KURAMOCHI, and TOV112D) and prexasertib-resistant ovarian cancer cells (JHOS2) treated with DMSO or prexasertib (40 nmol/L) for 18 hours.
Ovarian cancer cells are highly sensitive to prexasertib. A, IC50 of prexasertib in ovarian cancer cell lines. Cells were grown for 24 hours before exposure to graded concentrations of prexasertib. Viability was assessed after 3 days of treatment using Cell Titer-Glo. IC50 values for growth inhibition were calculated by Prism GraphPad and mean ± SE (n = 4) are shown. The experiments were done at least 2 times for each cell line, and the data from a representative experiment are shown. B, Western blots of lysates from prexasertib-sensitive ovarian cancer cells (ES2, KURAMOCHI, and TOV112D) and prexasertib-resistant ovarian cancer cells (JHOS2) treated with DMSO or prexasertib (40 nmol/L) for 18 hours.
Prexasertib and olaparib are synergistic in PARP inhibitor–resistant HGSOC cell lines
Several of the cell lines were relatively PARP inhibitor–resistant, with olaparib IC50 values in the 4 to 10 μmol/L range, including BRCA wild-type TOV112D and ES2 (Fig. 5A and B). The combination of prexasertib and olaparib resulted in synergistic cytotoxicity in these and other cell lines (Fig. 5A and B; Supplementary Fig. S5).
Prexasertib increases sensitivity of ovarian cancer cells to olaparib and disrupts homologous recombination. A and B, Synergy between prexasertib and PARP inhibition in TOV112D (A) and ES2 (B) ovarian cancer cells. Cells were grown for 24 hours in triplicate before exposure to graded concentrations of prexasertib and olaparib. Viability was assessed at 6 days using CellTiter-Glo. Synergy/antagonism between the drugs was determined using Combenefit software. (Left) Survival plots. (Middle) Bliss synergy/antagonism levels in a matrix format. (Right) Bliss synergy/antagonism levels on the experimental combination dose–response surface. Bliss scores greater than zero (green/blue shading) indicate synergy between prexasertib and olaparib. C, Prexasertib inhibits homologous recombination activity in a DR-GFP reporter assay in U2OS cells. Twenty-four hours after transfection with a plasmid encoding SceI endonuclease, U2OS-DR-GFP cells were exposed to graded concentrations of prexasertib for 48 hours, and GFP expression analyzed by flow cytometry. The relative GFP-positive cells normalized to vehicle-treated cells are shown. ****, P < 0.0001 using the 1-way ANOVA (n = 3). D, Prexasertib inhibits radiation-induced RAD51 foci in U2OS cells. Cells were pretreated with prexasertib for 16 hours before exposing them to radiation. Six hours after radiation, cells were analyzed for RAD51 foci by immunofluorescence. Representative images at 63× (left) and quantitation (right) of RAD51 foci are shown. At least 100 cells with foci were scored. ****, P < 0.0001 using the 1-way ANOVA. E, Prexasertib inhibits radiation-induced RAD51 foci in ES2 ovarian cancer cells. Cells were pretreated with prexasertib for 1 hour before exposing them to radiation. Six hours after radiation, cells were analyzed for RAD51 foci by immunofluorescence as in D. ****, P < 0.0001 using the 1-way ANOVA (n = 3). F, (left) Western blots of the lysates from BRCA1-mutated COV362 cells after siRNA-mediated knockdown of 53BP1. (Middle and right) Survival of COV362 cells with siRNA-mediated knockdown of 53BP1 after olaparib or prexasertib treatment, showing resistance to olaparib. Forty-eight hours after siRNA transfection, cells were exposed to graded concentrations of olaparib or prexasertib in 96-well plates. Viability was assessed at 6 days by CellTiter-Glo, and IC50 values were calculated by Prism GraphPad. IC50 values for olaparib in siControl and si53BP1 cells were 2.7 and 14.3 μmol/L, respectively. IC50 values for prexasertib in siControl and si53BP1 cells were 3.2 and 5.1 nmol/L, respectively (n = 4). G, Synergy between prexasertib and PARP inhibition in COV362 ovarian cancer cells after siRNA-mediated knockdown of 53BP1. Forty-eight hours after siRNA transfection, cells were exposed to graded concentrations of prexasertib and olaparib in triplicates in 96-well plates. Viability was assessed at 6 days using CellTiter-Glo. Survival and synergy were determined as described in A and B. H, Olaparib-induced RAD51 foci in COV362 cells are increased after depletion of 53BP1. Forty-eight hours after siRNA transfection, cells were pretreated with vehicle or olaparib for 24 hours before exposing them to vehicle or prexasertib. Six hours after the second exposure, cells were analyzed for RAD51 foci by immunofluorescence. Quantitation of RAD51 foci is shown. At least 100 cells with foci were scored. ****, P < 0.0001 using the 1-way ANOVA (n = 3). The experiments in all panels except panel D were done 2 or 3 times, and the data from a representative experiment are shown.
Prexasertib increases sensitivity of ovarian cancer cells to olaparib and disrupts homologous recombination. A and B, Synergy between prexasertib and PARP inhibition in TOV112D (A) and ES2 (B) ovarian cancer cells. Cells were grown for 24 hours in triplicate before exposure to graded concentrations of prexasertib and olaparib. Viability was assessed at 6 days using CellTiter-Glo. Synergy/antagonism between the drugs was determined using Combenefit software. (Left) Survival plots. (Middle) Bliss synergy/antagonism levels in a matrix format. (Right) Bliss synergy/antagonism levels on the experimental combination dose–response surface. Bliss scores greater than zero (green/blue shading) indicate synergy between prexasertib and olaparib. C, Prexasertib inhibits homologous recombination activity in a DR-GFP reporter assay in U2OS cells. Twenty-four hours after transfection with a plasmid encoding SceI endonuclease, U2OS-DR-GFP cells were exposed to graded concentrations of prexasertib for 48 hours, and GFP expression analyzed by flow cytometry. The relative GFP-positive cells normalized to vehicle-treated cells are shown. ****, P < 0.0001 using the 1-way ANOVA (n = 3). D, Prexasertib inhibits radiation-induced RAD51 foci in U2OS cells. Cells were pretreated with prexasertib for 16 hours before exposing them to radiation. Six hours after radiation, cells were analyzed for RAD51 foci by immunofluorescence. Representative images at 63× (left) and quantitation (right) of RAD51 foci are shown. At least 100 cells with foci were scored. ****, P < 0.0001 using the 1-way ANOVA. E, Prexasertib inhibits radiation-induced RAD51 foci in ES2 ovarian cancer cells. Cells were pretreated with prexasertib for 1 hour before exposing them to radiation. Six hours after radiation, cells were analyzed for RAD51 foci by immunofluorescence as in D. ****, P < 0.0001 using the 1-way ANOVA (n = 3). F, (left) Western blots of the lysates from BRCA1-mutated COV362 cells after siRNA-mediated knockdown of 53BP1. (Middle and right) Survival of COV362 cells with siRNA-mediated knockdown of 53BP1 after olaparib or prexasertib treatment, showing resistance to olaparib. Forty-eight hours after siRNA transfection, cells were exposed to graded concentrations of olaparib or prexasertib in 96-well plates. Viability was assessed at 6 days by CellTiter-Glo, and IC50 values were calculated by Prism GraphPad. IC50 values for olaparib in siControl and si53BP1 cells were 2.7 and 14.3 μmol/L, respectively. IC50 values for prexasertib in siControl and si53BP1 cells were 3.2 and 5.1 nmol/L, respectively (n = 4). G, Synergy between prexasertib and PARP inhibition in COV362 ovarian cancer cells after siRNA-mediated knockdown of 53BP1. Forty-eight hours after siRNA transfection, cells were exposed to graded concentrations of prexasertib and olaparib in triplicates in 96-well plates. Viability was assessed at 6 days using CellTiter-Glo. Survival and synergy were determined as described in A and B. H, Olaparib-induced RAD51 foci in COV362 cells are increased after depletion of 53BP1. Forty-eight hours after siRNA transfection, cells were pretreated with vehicle or olaparib for 24 hours before exposing them to vehicle or prexasertib. Six hours after the second exposure, cells were analyzed for RAD51 foci by immunofluorescence. Quantitation of RAD51 foci is shown. At least 100 cells with foci were scored. ****, P < 0.0001 using the 1-way ANOVA (n = 3). The experiments in all panels except panel D were done 2 or 3 times, and the data from a representative experiment are shown.
To demonstrate that prexasertib directly compromises HR repair as a mechanism of PARP inhibitor sensitization, we utilized U2OS osteosarcoma cells engineered to express GFP during HR repair in response to a restriction enzyme-induced double-strand break. Disruption of HR activity by prexasertib was demonstrated in this assay (Fig. 5C). Consistent with this finding, prexasertib exposure reduced γ-irradiation–induced RAD51 foci in U2OS (Fig. 5D) and ES2 ovarian cancer (Fig. 5E) cells in a concentration-dependent manner. Similarly, we studied BRCA1-mutant COV362 cells rendered olaparib-resistant by siRNA-mediated depletion of TP53BP1. The PARP inhibitor–resistant derivative retained sensitivity to prexasertib (Fig. 5F). Prexasertib was synergistic with olaparib in these cells and compromised induction of RAD51 foci after olaparib exposure (Fig. 5G and H). Taken together, these results suggest that prexasertib compromises HR repair, sensitizing resistant HGSOC cells to PARP inhibition. Of note, BRCA1−/− retinal pigmented epithelial (RPE) cells (36) rendered resistant to PARP inhibition via HR restoration by TP53PB1 depletion also retained sensitivity to prexasertib (Supplementary Fig. S6A–S6D).
Because prexasertib disrupts HR, we asked whether brief exposure of cells to prexasertib prior to olaparib would further augment cytotoxicity. As shown in Supplementary Fig. S6E, there was no improvement over the degree of synergism achieved with concomitant exposure to both drugs at the outset of the experiment.
Prexasertib reverses stabilization of replication forks in PARP inhibitor–resistant HGSOC cells
UWB1.289 BRCA1-mutated HGSOC ovarian cancer cells rendered PARP inhibitor–resistant after prolonged exposure to drug have been demonstrated to simultaneously exhibit multiple mechanisms of resistance, including restoration of HR, as well as replication fork stabilization (32). Similar to the engineered COV362 cells (Fig. 5F), the UWB1.289 olaparib-resistant derivatives SYR12 and SYR14 (32) retained sensitivity to prexasertib, comparable to that of parental cells (Fig. 6A and B; Supplementary Fig. S7A). These subclones demonstrated partial restoration of HR, assessed by RAD51 focus assembly in response to olaparib compared with parental cells complemented with wild-type BRCA1 cDNA. As expected, prexasertib compromised the induction of RAD51 in response to olaparib (Fig. 6C).
Prexasertib has activity in BRCA1-mutated ovarian cancer cells that have restored fork stability and acquired PARP inhibitor resistance. A and B, UWB1.289 SYR12 and UWB1.289 SYR14 ovarian cancer cells with acquired PARP inhibitor resistance are sensitive to prexasertib. BRCA1-mutated parental UWB1.289 cells, resistant derivatives SYR12 and SYR14, and BRCA1 add-back cells were exposed to olaparib or prexasertib in 6-well plates for 12 days and stained with crystal violet. Representative colony formation assays (n = 3) after olaparib or prexasertib exposure are shown in A and B, respectively. C, Prexasertib blocks restored HR in PARP inhibitor–resistant UWB1.289 SYR12 and UWB1.289 SYR14 ovarian cancer cells. The indicated cell lines were pretreated with olaparib (5 μmol/L) for 24 hours before exposing to prexasertib (60 nmol/L) for 6 hours and analyzed for RAD51 foci by immunofluorescence. Representative images (top) and quantitation (bottom) of RAD51 foci are shown. At least 100 cells were scored. ****, P < 0.0001 using the 1-way ANOVA (n = 3). (D) Prexasertib reactivates degradation of stalled replication forks in PARP inhibitor–resistant cells as determined by DNA fiber assays. Schematic of IdU, CldU, hydroxyurea (HU), and prexasertib treatment (top), representative images of normal and stalled fibers (middle), and graphical quantitation of DNA fiber analysis of stalled forks (degraded forks) with each fiber (bottom) are shown. Fiber assays were repeated twice, and data from a representative experiment are shown. For fiber assays, newly synthesized DNA was sequentially labeled with IdU for 45 minutes and CldU for 45 minutes. Cells were then treated with DMSO, replication stalling agent HU (2 mmol/L), or HU (2 mmol/L) + prexasertib (100 nmol/L) for 4.5 hours. At least 100 DNA fibers were analyzed per condition. IdU labeling was detected in red and CldU labeling was detected in green, and fork degradation was assessed by measuring the ratio of CldU to IdU labeling, with a ration <1 indicating fork degradation. ****, P < 0.0001; **, P = 0.0023; *, P < 0.05 using the Mann–Whitney test. The experiments in A–D were done at least 2 times, and the data from a representative experiment are shown.
Prexasertib has activity in BRCA1-mutated ovarian cancer cells that have restored fork stability and acquired PARP inhibitor resistance. A and B, UWB1.289 SYR12 and UWB1.289 SYR14 ovarian cancer cells with acquired PARP inhibitor resistance are sensitive to prexasertib. BRCA1-mutated parental UWB1.289 cells, resistant derivatives SYR12 and SYR14, and BRCA1 add-back cells were exposed to olaparib or prexasertib in 6-well plates for 12 days and stained with crystal violet. Representative colony formation assays (n = 3) after olaparib or prexasertib exposure are shown in A and B, respectively. C, Prexasertib blocks restored HR in PARP inhibitor–resistant UWB1.289 SYR12 and UWB1.289 SYR14 ovarian cancer cells. The indicated cell lines were pretreated with olaparib (5 μmol/L) for 24 hours before exposing to prexasertib (60 nmol/L) for 6 hours and analyzed for RAD51 foci by immunofluorescence. Representative images (top) and quantitation (bottom) of RAD51 foci are shown. At least 100 cells were scored. ****, P < 0.0001 using the 1-way ANOVA (n = 3). (D) Prexasertib reactivates degradation of stalled replication forks in PARP inhibitor–resistant cells as determined by DNA fiber assays. Schematic of IdU, CldU, hydroxyurea (HU), and prexasertib treatment (top), representative images of normal and stalled fibers (middle), and graphical quantitation of DNA fiber analysis of stalled forks (degraded forks) with each fiber (bottom) are shown. Fiber assays were repeated twice, and data from a representative experiment are shown. For fiber assays, newly synthesized DNA was sequentially labeled with IdU for 45 minutes and CldU for 45 minutes. Cells were then treated with DMSO, replication stalling agent HU (2 mmol/L), or HU (2 mmol/L) + prexasertib (100 nmol/L) for 4.5 hours. At least 100 DNA fibers were analyzed per condition. IdU labeling was detected in red and CldU labeling was detected in green, and fork degradation was assessed by measuring the ratio of CldU to IdU labeling, with a ration <1 indicating fork degradation. ****, P < 0.0001; **, P = 0.0023; *, P < 0.05 using the Mann–Whitney test. The experiments in A–D were done at least 2 times, and the data from a representative experiment are shown.
As evidenced by DNA fiber assays, the SYR12 and SYR14 derivatives also demonstrated stabilization of replication forks after hydroxyurea-induced stalling. Prexasertib also reversed fork stabilization in these cells (Fig. 6D). Consequently, prexasertib resensitized these resistant derivatives to olaparib (Supplementary Fig. S7B). Taken together, these results indicate that prexasertib can induce cytotoxicity in BRCA1-mutated PARP inhibitor–resistant HGSOC cells and reverse the two major mechanisms of acquired PARP inhibitor resistance.
Discussion
In the current report, we demonstrate that the CHK1 inhibitor prexasertib has activity as monotherapy and in combination with PARP inhibition in preclinical EOC in vivo models. Prexasertib has activity in both HR repair-deficient and -proficient models and is effective in killing tumor cells with de novo or acquired PARP inhibitor resistance.
As a monotherapy, prexasertib-mediated CHK1 inhibition compromises checkpoint control and repair in response to endogenous DNA damage, as well as increases replication stress in tumor cells. Cells with TP53 deficiency, which occurs nearly universally in HGSOC, coupled with HR deficiency (3, 4), may be highly dependent on the ATR–CHK1 pathway for maintenance of genomic stability, as evidenced by the response to prexasertib of the RAD51C-deficient PDX model DF83. Additionally, cells with unstable replication forks, as can occur with BRCA-deficiency, may require the ATR–CHK1 pathway to prevent replication fork collapse. In our previous work in HGSOC organoid cultures, those with baseline unstable replication forks were highly sensitive to CHK1 inhibition (40). Alternatively, in cells with a high baseline level of replication stress, including those driven by CCNE1 or MYC copy number gain, amplification or overexpression of the corresponding proteins, further exacerbation of replication stress mediated by CHK1 inhibition may be lethal, accounting for the sensitivity of such cells both in preclinical models tested here and in clinical trials (26, 27).
Importantly, prexasertib also demonstrates activity in PARP inhibitor–resistant BRCA1-mutant cells both as monotherapy and in combination with olaparib. The four BRCA1-mutant PDX models have clearly demonstrated restored HR, as evidenced by formation of DNA damage-induced RAD51 foci; in the case of one model, restored HR is the result of TP53BP1 mutation and reduced expression, which is known to facilitate BRCA1-independent DNA end resection (9). A variety of other mechanisms may restore HR in PARP inhibitor–resistant cells (41), including BRCA reversion mutation (42), stabilization of mutant BRCA proteins (43), or depletion of proteins such REV7 (10) and other components of the Shieldin Complex (44) and DYNLL1 (45). Irrespective of the mechanism of HR restoration, in HGSOC cell lines, prexasertib compromises the formation of RAD51 foci in response to DNA damage. It has been previously shown that CHK1 phosphorylates RAD51 (23), an event required for filament formation that is blocked by CHK1 inhibition.
As shown by UWB1.289 BRCA1-mutant HGSOC cancer cells rendered PARP inhibitor–resistant, multiple mechanisms of resistance may occur simultaneously, so that restored HR may be accompanied by replication fork stabilization (32). The latter event typically results from the exclusion of nucleases, as occurs with reduced expression of PTIP, compromising the recruitment of the MRE11 nuclease (11). Similar to previously reported results with ATR inhibition (32), CHK1 inhibition compromises replication fork stability in PARP inhibitor–resistant cells with stabilized replication forks, likely by preventing RAD51 accumulation at sites of replication fork stalling. In this regard, prexasertib reverses a second major mechanism of PARP inhibitor resistance and sensitizes cells with acquired resistance to PARP inhibition.
Recently, the preliminary activity of prexasertib monotherapy in BRCA wild-type HGSOC was reported (27). Most patients had platinum-resistant or refractory disease; among 24 evaluable patients, 8 achieved RECIST partial response. Four of the 8 patients with partial tumor responses had both CCNE1 amplification or copy number gain and CCNE1 mRNA upregulation, indicating that prexasertib may have potential activity in this population. Although PDX models with high CCNE1 expression and cell lines with high cyclin E levels used in our study were sensitive to prexasertib, activity was not restricted to this subset of models.
In the ovarian cancer study, prexasetib carried substantial hematologic toxicity resulting in transient grade 4 neutropenia and grade 3/4 thrombocytopenia in 79% and 25% of patients, respectively (27). Nonetheless, based on this study, a larger study of prexasertib in platinum-resistant and refractory disease is underway (NCT03414047). One cohort in this study will also evaluate patients with BRCA-mutant tumors with platinum-resistant disease, who have already received a PARP inhibitor. Results in the preclinical BRCA1-mutant models evaluated here suggest promise for this approach.
The substantial activity of prexasertib monotherapy in the PDX models made it difficult to demonstrate synergism with olaparib in all cases. Nonetheless, in the TP53BP1-deficient, HR-proficient, BRCA1-mutated model, combined prexasertib and olaparib was superior to monotherapy treatment. Additionally, several cell lines with either de novo or acquired PARP inhibitor resistance demonstrated synergism between prexasertib and olaparib, similar to previous reports (29, 46). This occurred in both BRCA wild-type cells, but also in BRCA1-mutated cells with acquired PARP inhibitor resistance. For this reason, a phase I trial combining prexasertib and olaparib has been initiated, focusing on BRCA-mutant, PARP inhibitor–resistant HGSOC (NCT03057145). Although overlapping hematological toxicity has not permitted administration of full doses of both agents, tolerable attenuated doses of the drugs used in combination concurrently have been achieved, with responses observed in this population (47). Nonetheless, ultimately, in the PARP inhibitor–resistant setting, the efficacy of prexasertib monotherapy and combined prexasertib/olaparib may need to be directly compared. Mechanistically, it is possible that combined prexasertib/olaparib may compromise stabilized replication forks to a greater degree than prexasertib alone. We have previously observed this phenomenon when prexasertib combined with a replication stress-inducing agent such as gemcitabine produced greater fork instability than prexasertib alone in HGSOC organoids with stabilized replication forks (40).
In addition to work in the PARP inhibitor–resistant setting, combined prexasertib and olaparib was superior to either monotherapy in the RAD51C-mutant, PARP inhibitor-sensitive model. These results suggest that comparison of the combination to olaparib monotherapy in this setting may also be warranted, and that prexasertib may play a role in delaying the development of resistance.
Furthermore, several cell lines examined in this work, including TOV112D and ES2, may not represent classic HGSOC (48), but rather may demonstrate characteristics of clear cell or endometrioid ovarian cancer (49, 50). These cancers do not typically respond well to standard treatments, suggesting that prexasertib as monotherapy or in combination with olaparib could be relevant therapeutic options in these subtypes as well.
In summary, prexasertib, either as monotherapy or in combination with olaparib, has extensive activity in in vivo and cell line models of HGSOC, in BRCA wild-type and mutant backgrounds and in PARP inhibitor-sensitive and resistant settings. These results are translating to ongoing clinical trials designed to fully assess the potential role of prexasertib in the HGSOC armamentarium.
Disclosure of Potential Conflicts of Interest
B.S. Kochupurakkal reports receiving commercial research grants and commercial research support from Eli Lilly and Company. K.T. Do is a consultant/advisory board member for Seattle Genetics. P.A. Konstantinopoulos is a consultant/advisory board member for AstraZeneca, Vertex, Tesaro, Merck, and Pfizer. U.A. Matulonis is a consultant/advisory board member for Eli Lilly. J.F. Liu is a consultant/advisory board member for Tesaro, Mersana, Clovis, AstraZeneca, and Merck. A.D. D'Andrea is a consultant/advisory board member for and reports receiving commercial research grants from Eli Lilly. G.I. Shapiro is a consultant/advisory board member for Lilly, Sierra Oncology, Merck-EMD Serono, Pfizer, Astex, Almac, Roche, Bicycle Therapeutics, Fusion Pharmaceuticals, G1 Therapeutics, Bayer, Ipsen, Cybrexa Therapeutics, Angiex, Daiichi Sankyo and Seattle Genetics, and reports receiving commercial research grants from Lilly, Sierra Oncology, Merck-EMD Serono and Merck & Co. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: K. Parmar, J.-B. Lazaro, S. Palakurthi, K.T. Do, A.D. D'Andrea, G.I. Shapiro
Development of methodology: B.S. Kochupurakkal, J.-B. Lazaro, Z.C. Wang, S. Palakurthi, H.D. Reavis, G.I. Shapiro
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Parmar, B.S. Kochupurakkal, S. Palakurthi, P.T. Kirschmeier, C. Yang, L.A. Sambel, A. Färkkilä, E. Reznichenko, H.D. Reavis, U.A. Matulonis, J.F. Liu, A.D. D'Andrea, G.I. Shapiro
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Parmar, B.S. Kochupurakkal, Z.C. Wang, S. Palakurthi, C. Yang, L.A. Sambel, A. Färkkilä, E. Reznichenko, H.D. Reavis, C.E. Dunn, P.A. Konstantinopoulos, U.A. Matulonis, A.D. D'Andrea, G.I. Shapiro
Writing, review, and/or revision of the manuscript: K. Parmar, B.S. Kochupurakkal, A. Färkkilä, K.T. Do, P.A. Konstantinopoulos, U.A. Matulonis, J.F. Liu, A.D. D'Andrea, G.I. Shapiro
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Parmar, P.T. Kirschmeier, C. Yang, C.E. Dunn, L. Zou, G.I. Shapiro
Study supervision: S. Palakurthi, P.T. Kirschmeier, G.I. Shapiro
Acknowledgments
This work was supported by Lilly-Dana-Farber Cancer Institute Program Grant A09640 (to A.D. D'Andrea, G.I. Shapiro), and Stand Up To Cancer-Ovarian Cancer Research Fund Alliance-National Ovarian Cancer Coalition Dream Team Translational Cancer Research Grant (SU2C-AACR-DT16-15). Stand Up To Cancer is a division of the Entertainment Industry Foundation. Research grants are administered by the American Association for Cancer Research, the Scientific Partner of SU2C. This work was also supported by the NIH grants R37 HL052725 (to A.D. D'Andrea), P01 HL048546 (to A.D. D'Andrea), P50 CA168504 (to A.D. D'Andrea, G.I. Shapiro), and R01 CA090687 (to G.I. Shapiro); U.S. Department of Defense grants BM110181 (to A.D. D'Andrea) and BC151331P1 (to A.D. D'Andrea); the Breast Cancer Research Foundation (to A.D. D'Andrea); and the Fanconi Anemia Research Fund (to A.D. D'Andrea).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.