Combined inhibition of RAD51 and CHK1 causes synergistic toxicity in cisplatin resistant cancer cells by triggering replication fork collapse

Abstract

The therapeutic efficacy of the anticancer drug cisplatin is limited by acquired drug resistance. Cisplatin forms DNA crosslinks, that, if not removed, lead to replication stress. Due to this, the DNA damage response (DDR) gets activated regulating cell cycle arrest, DNA repair, cell death or survival. This makes DDR components promising targets for the development of new therapeutic approaches aiming to overcome acquired drug resistance. To this end, cisplatin‐resistant bladder cancer cells were analyzed regarding their sensitivity to combination treatments with selected pharmacological DDR inhibitors. Synergistic cytolethal effects were achieved after combined treatment with low to moderate doses of the non‐genotoxic RAD51‐inhibitor (RAD51_i) B02 and CHK1‐inhibitor (CHK1_i) PF477736. This effect was also found in cisplatin resistant tumor cells of other origin as well as with other RAD51_i and CHK1_i. Combined treatments promoted decelerated replication, S‐phase blockage, accumulation of DNA strand breaks, DDR activation and stimulation of apoptotic cell death as compared to mono‐treatment, which is independent of the expression of RAD51, CHK1, and PrimPol. Based on these data, we suggest combined inhibition of RAD51 and CHK1 to overcome acquired cisplatin resistance of malignant cells. We propose that the molecular mechanism of this synergistic toxicity relies on a simultaneous inactivation of two key DNA damage tolerance pathways regulating replication fork restart, thereby circumventing the activation of alternative compensatory mechanisms and, in consequence, eventually effectively triggering apoptotic cell death by replication fork collapse.

What's new?

Cisplatin induces replication stress by forming DNA crosslinks,  which activate the DNA damage response that may contribute to acquired drug resistance. This study presents novel evidence that simultaneous targeting of two different DNA damage response‐related components of the replication machinery, that is, checkpoint kinase 1 and RAD51, is a useful strategy to overcome acquired cisplatin resistance. Combined treatment promoted decelerated replication, S‐phase blockage, accumulation of DNA strand breaks, DNA damage response, and apoptotic cell death. The combination treatment may inhibit replication fork restart, leading to synergistic cytotoxicity due to the induction of apoptosis of cisplatin‐resistant malignant cells during the S‐phase.

1. INTRODUCTION

Acquired drug resistance is a major obstacle in the therapy of malignant diseases, including bladder cancer, which is a frequent type of cancer with urothelial carcinoma (UC) being the prevailing type. ¹ Cisplatin (CisPt)‐based therapeutic regimens are commonly used to treat muscle‐invasive as well as relapsed or metastatic UC. ² CisPt enters cells by passive diffusion and via transporters. ³ , ⁴ Covalent DNA adducts are formed when its chloride ligands are replaced by water inside the cell. ⁵ DNA intrastrand crosslinks (GpG and ApG) account for the majority (60%–80%) of CisPt‐generated DNA adducts. ⁶ , ⁷ They cause distortion of the DNA double helix, thereby causing replication blockage. ⁸ , ⁹ Stalled replication forks and sustained replicative stress may lead to DNA double‐strand breaks (DSBs), ¹⁰ which are a potent trigger of cell death if remaining unrepaired. ¹¹ The antitumor efficacy of platinum‐based anticancer therapy is limited by intrinsic or acquired resistance to the drug. ¹² Factors relevant to the development of CisPt resistance are poorly characterized for UC and have been classified according to their site of action as pre‐, on‐, post‐, and off‐target. ¹³ Representative of these different mechanisms are drug transport, DNA repair, apoptosis, and signal transduction, respectively.

A relevant target for CisPt and other approved conventional (i.e., genotoxic) anticancer therapeutics (CATs) is the genomic DNA of tumor cells. In consequence of CAT‐induced DNA damage and enhanced replication stress of rapidly dividing tumor cells, the DNA damage response (DDR) gets activated. Key regulators of the DDR are the PI3‐like kinase Ataxia telangiectasia mutated (ATM) and the ATM and Rad3‐related kinase (ATR). ¹⁴ , ¹⁵ , ¹⁶ They coordinate stress responses resulting from the generation of DSBs and replication blocking DNA lesions, respectively. ¹⁷ , ¹⁸ , ¹⁹ The DDR stringently controls cell cycle progression through activation of cell cycle checkpoints and, moreover, regulates DNA repair and cell death‐related pathways, ¹⁴ , ²⁰ thereby defining the balance between survival and death. ²¹ Activation of the DDR is a known precursor of genomic instability in bladder cancer. ²² Moreover, error prone repair of DSBs further favors genomic instability and, hence, progression of bladder carcinomas. ²³

Having in mind the pivotal role of DDR and DNA repair for the maintenance of genomic stability and survival, compounds targeting those mechanisms, such as inhibitors of the DNA repair‐associated factor PARP or inhibitors of DSB‐ and replicative stress‐related kinases ATM, ATR as well as checkpoint kinase, are promising drug candidates to achieve an improvement of anticancer therapy. ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ Moreover, targeting these factors can exploit genetic alterations of individual tumors as Achilles heels, thereby favoring personalized treatment strategies and synthetic lethality. ²⁹ , ³⁰ , ³¹ , ³² Here, we assess the usefulness of combination treatments with pre‐selected DDR inhibitors to evoke cell death of bladder cancer cell variants that acquired resistance to CisPt following a clinically oriented selection scheme. ³³

2. MATERIALS AND METHODS

2.1. Materials

J82 (RRID:CVCL_0359) UC cells, A549 (RRID:CVCL_0023) lung tumor cells, and SH‐SY5Y (RRID:CVCL_0019) neuroblastoma cells originate from the German Collection of Microorganisms and Cell Culture (DSMZ) (Braunschweig, Germany). A2780 (RRID:CVCL_0134) and CisPt‐resistant A2780^CisPt (RRID:CVCL_H745) ovary tumor cells originate from Sigma (Steinheim, Germany). Foreskin‐4 human induced pluripotent stem cells (hiPSCs) (RRID:CVCL_2337) originate from WiCell Stem Cell Bank (Madison, WI, USA). J82^CisPt, A549^CisPt, and SH‐SY5Y^CisPt cell lines were created in our laboratory. Cisplatin originates from TEVA (Ulm, Germany). Cytostatics, DDR modifiers and other compounds were obtained from the following suppliers: Doxorubicin (Cellpharm [Bad Vilbel, Germany]), OH‐Urea (Sigma [Steinheim, Germany]), 5‐Fluorouracil (Medac [Wedel, Germany]), Carboplatin (TEVA [Ulm, Germany]), Oxaliplatin (Accord Healthcare [Munich, Germany]), hydrogen peroxide (H₂O₂) (Carl Roth GmbH [Karlsruhe, Germany]), checkpoint kinase 1 (CHK1) inhibitor PF477736 and LY2603618 (Sigma [Steinheim, Germany]), RAD51 inhibitor B02 and RI(dl)2 (Tocris Bioscience [Bristol, UK]), MRE11 inhibitor Mirin (Abcam [Cambridge, UK]), PARP inhibitor Niraparib (MedChemExpress [Monmouth Junction, NJ, USA]), HDAC inhibitors Entinostat (Selleckchem [Munich, Germany]) and Vorinostat (Sigma [Steinheim, Germany]).

The following antibodies were used: antibodies detecting Ser139 phosphorylated histone H2AX (γH2AX) (Millipore (Billerica, MA, USA) or Abcam [Cambridge, UK]), α‐tubulin, β‐actin (Santa Cruz Biotechnology [Santa Cruz, CA, USA]), Ser345 phosphorylated CHK1, CHK1, H2AX, cleaved caspase 7, caspase 7, PARP, Ser15 phosphorylated p53 (Cell Signaling [Danvers, MA, USA]), Ser4/8 phosphorylated RPA32, Ser824 phosphorylated KAP1 (Bethyl Laboratories [Montgomery, AL, USA]), Ser33 phosphorylated RPA32, RAD51, Pericentrin (Abcam [Cambridge, UK]), RPA32 (Millipore [Billerica, MA, USA]), rat anti‐BrdU (Abcam [Cambridge, UK]), mouse anti‐BrdU (BD [Franklin Lakes, NJ, USA]) and Ser10 phosphorylated histone 3 (pH 3) (Thermo Fisher [Waltham, MA, USA]). The fluorescent antibodies Alexa Flour 488 and 555 were obtained from Life Technologies (Carlsbad, CA, USA). Horseradish peroxidase‐conjugated secondary antibodies were purchased from Rockland (Gilbertsville, PA, USA). The thymidine analogues 5‐Chloro‐2′‐deoxyuridine (CldU) and 5‐Iodo‐2′‐deoxyuridine (IdU) were bought from Sigma (Steinheim, Germany).

2.2. Cell culture

J82^WT, thereof derived J82^CisPt cells, ³³ SH‐SY5Y^WT, SH‐SY5Y^CisPt, A549^WT, and A549^CisPt were cultured in DMEM (Sigma [Steinheim, Germany]) containing 10% of fetal calf serum (FCS) (PAA Laboratories [Cölbe, Germany]) and 1% penicillin/streptomycin (Sigma [Steinheim, Germany]) at 37°C in an atmosphere containing 5% CO₂. A2780^WT and the cisplatin‐resistant variant A2780^CisPt were grown in RPMI‐1640 (Sigma [Steinheim, Germany]) with 10% FCS, 1% penicillin/streptomycin, and 1% glutamine (Sigma [Steinheim, Germany]). Foreskin‐4 hiPSCs were cultivated on plates coated with human embryonic stem cell qualified Matrigel (Corning, Inc. [Corning, NY, USA]) in mTeSR1 medium (StemCell Technologies [Vancouver, Canada]) containing 10 mM Y‐27632 dihydrochloride (Sigma [Steinheim, Germany]). ³⁴ The used cell lines were authenticated by STR profiling during the last 3 years. All experiments were performed with mycoplasma free cells. Unless otherwise noted, logarithmically growing cells were treated 24 h after seeding.

2.3. Determination of cell viability

Cell viability was determined using the AlamarBlue Assay, ³⁵ which measures the metabolism of the non‐fluorescent dye resazurin (Sigma [Steinheim, Germany]) to fluorescent resorufin. Cells were incubated with resazurin solution (40 μM) for 1.5–3 h before fluorescence was measured (excitation: 535 nm, emission: 590 nm, 5 flashes, integration time: 20 μs). Additionally, cytotoxicity was examined with the Neutral red assay, which measures lysosomal membrane integrity. Cells were incubated with Neutral red solution (0.01% Neutral red/0.1 M HEPES) for 90 min before cell fixation (1% formaldehyde/1% CaCl₂). After extraction of neutral red absorbance was measured (540 nm). For both assays, relative viability in the untreated controls was set to 100%. Unless otherwise stated, data are presented as the mean ± standard deviation (SD) of three independent experiments, each performed in quadruplicate.

2.4. Combination index

To determine synergistic, additive or antagonistic drug effects on viability, the combination index (CI) was calculated (AlamarBlue Assay‐derived data) using the Compusyn software version 1.0 (ComboSyn, Inc. [Paramus, NJ, USA]) based on the Chou–Talalay method. ³⁶ CI < 0.9 are considered as synergistic, 0.9 ≤ CI ≤ 1.2 as additive and CI > 1.2 as antagonistic effects.

2.5. Flow cytometry‐based analysis of cell cycle distribution, cell death, and cell cycle‐dependent DNA damage formation

Cell cycle distribution was analyzed by flow cytometry. Adherent cells were detached by trypsinization and combined with floating cells. After centrifugation (300 × g, 5 min, 4°C), cells were resuspended in hypotonic solution containing 50 μg/mL propidium iodide (Sigma [Steinheim, Germany]). After incubation (20 min, RT), cells were flow cytometrically analyzed (Becton Dickinson [Heidelberg, Germany]). The SubG1 fraction is indicative of apoptotic cells. To identify in which cell cycle phase DNA damage occurs, PI/γH2AX co‐staining was performed. To this end, cells were fixed with 1% formaldehyde in phosphate‐buffered saline (PBS) (15 min, 4°C), permeabilized in ice‐cold 80% ethanol (2 h, −20°C) and blocked with 1% BSA in PBS/0.3% Triton X‐100 (5 min, RT). After incubation with γH2AX antibody (1:100, overnight, 4°C), secondary antibody was added (1:200, 1 h, RT). For PI staining, cells were incubated with PBS containing 10 μg/mL PI and 100 μg/mL DNase‐free RNase (1 h, RT) and analyzed by flow cytometry (Becton Dickinson [Heidelberg, Germany]).

2.6. Analysis of DNA repair and replicative stress response by foci analyses

The formation of nuclear foci of factors involved in DSB repair by homologous recombination (HR) (RAD51) and replicative stress responses (RPA32) were monitored by immunocytochemistry. To this end, cells were fixed with ice‐cold methanol:acetone (7:3) (10 min, −20°C). After blocking (3 × 10 min, RT; blocking solution: 1% goat serum in phosphate‐buffered saline [PBS]), incubation with RPA2 antibody and RAD51 antibody was performed (1:500, overnight, 4°C), followed by washing with goat serum/PBS and addition of the secondary fluorescence‐labeled antibodies (1:500, 2 h, RT). Cells were mounted in Vectashield (Vector Laboratories [Burlingame, CA, USA]) containing DAPI and nuclear foci were counted by microscopic analysis (Olympus BX43 fluorescence microscope [100× objective]).

For RPA/γH2AX co‐staining, cells were fixed with 4% formaldehyde in PBS (15 min, RT) and incubated with ice‐cold ethanol (20 min, −20°C). After blocking (1 h, RT; blocking solution: 5% BSA in PBS/0.3% Triton X‐100), primary antibodies were added (RPA2 1:500, γH2AX 1:1000, overnight, 4°C), followed by washing (PBS/0.3% Triton X‐100) and addition of secondary fluorescence‐labeled antibodies (1:500, 2 h, RT). Afterwards, cells were mounted in Vectashield (Vector Laboratories [Burlingame, CA, USA]) containing DAPI and nuclei were analyzed microscopically as described above.

2.7. Western blot analysis

The activation status of the DNA damage response (DDR) was examined by Western blot analysis using a panel of phospho‐specific antibodies. Total protein cell extracts were obtained by sonication in RIPA buffer (EpiShear™ Probe sonicator, Active Motif [La Hulpe, Belgium]). Protein concentrations were determined using the DC Protein Assay (BioRad [Munich, Germany]). Samples were mixed with Roti®‐Load buffer (Carl Roth GmbH [Karlsruhe, Germany]) and separated by SDS‐PAGE. After transfer to nitrocellulose membrane (GE Healthcare [Little Chalfont, UK]) and blocking (5% non‐fat milk in TBS/0.1% Tween 20) (MERCK [Darmstadt, Germany]) (2 h, RT), the membrane was incubated with the respective primary antibody overnight (4°C) (pKAP1 1:5000, pChk1 1:500, Chk1 1:1000, pRPA32 1:2000, pP53 1:1000, γH2AX 1:1000, cleaved caspase 7 1:1000, caspase 7 1:1000, PARP 1:1000, β‐actin 1:50,000). Upon washing with TBS/0.1% Tween 20, the secondary (peroxidase‐conjugated) antibody was added (1:2000, 2.5 h, RT). For visualization, the ChemiDoc imaging system (BioRad [Munich, Germany]) was used.

2.8. EdU incorporation assay

For analysis of S‐phase activity, EdU incorporation was measured using the EdU Cell Proliferation Assay (baseclick GmbH [Neuried, Germany]). After EdU pulse‐labeling (10 μM EdU, 2 h), cells were fixed, permeabilized and incubated with the reaction cocktail (30 min, RT, in the dark) according to the manufacturers protocol. Ultimately, the cells were mounted in Vectashield (Vector Laboratories [Burlingame, CA, USA]) containing DAPI. EdU positive nuclei were counted by microscopic analysis (Olympus BX43 fluorescence microscope [20× objective]).

2.9. Analysis of mitotic index

The phosphorylation of histone 3 at serine 10 (pH 3) was analyzed by immunocytochemical staining. After fixation (4% formaldehyde in PBS, 15 min, RT), incubation with ice‐cold ethanol (20 min, −20°C) and blocking (1 h, RT; 5% BSA in PBS/0.3% Triton X‐100), incubation with the pH 3 antibody was performed (1:500, overnight, 4°C), followed by addition of the secondary fluorescence‐labeled antibody (1:500, 2 h, RT, in the dark). After washing, the cells were mounted in DAPI‐containing Vectashield (Vector Laboratories [Burlingame, CA, USA]) and the number of pH 3 positive nuclei was assessed by microscopic analysis (Olympus BX43 fluorescence microscope [20× objective]).

2.10. DNA fiber spreading assay

For analysis of DNA replication dynamics, the DNA Fiber Spreading Assay was applied as described before. ³⁷ , ³⁸ Briefly, cells were incubated with 20 μM CldU (20 min, 37°C, 5% CO₂) and right afterwards with 200 μM IdU (20 min, 37°C, 5% CO₂). After harvesting, cell suspensions were transferred onto glass slides and mixed with lysis buffer (0.5% SDS, 200 mM Tris–HCl, 50 mM EDTA). After 6 min incubation (RT), slides were tilted upwards to stretch the fibers, dried for 6 min lying horizontally flat, fixed (5 min, RT, fixing solution: methanol:acetic acid 3:1), again dried (7 min lying horizontally flat) and stored overnight (4°C, 70% ethanol). The next day slides were incubated with 100% methanol (5 min, RT), denatured with 2.5 M HCl (1 h, RT) and blocked (1 h, 37°C, 5% BSA in PBS). The fibers were stained with rat anti‐BrdU (1:140) for CldU detection and mouse anti‐BrdU (1:70) for IdU detection (1 h, RT, antibodies in 0.5% BSA‐PBS). After washing (0.05% Tween 20‐PBS and PBS) incubation with secondary antibodies anti‐rat 488 (1:400) and anti‐mouse 555 (1:250) was performed (1 h, RT, 0.5% BSA‐PBS). Following washing with 0.05% Tween 20‐PBS and PBS, the slides were mounted with Fluoroshield (Sigma [Steinheim, Germany]). Fibers were analyzed microscopically (Olympus BX43 fluorescence microscope [40× objective]).

2.11. Analysis of mitotic progression

The formation of a functional spindle apparatus is essential for mitosis. Co‐staining of spindle (α‐tubulin) and centrosome (pericentrin) markers was applied to identify defects in mitotic progression. ³⁹ After fixation (4% formaldehyde in PBS [20 min, RT]), permeabilization (0.25% Triton X‐100 in PBS [5 min, RT]) and blockage (3% BSA in PBS [2 h, RT]) incubation with the α‐tubulin antibody (1:500) was done (4°C in 1% BSA/PBS, overnight). Next day, the samples were incubated with the pericentrin antibody (1:1000, 2.5 h, RT) followed by incubation with secondary antibodies (1:1000, 1% BSA/PBS, 2 h, RT) and coverslips were mounted with Prolong Gold containing DAPI (Invitrogen [Waltham, MA, USA]). Microscopic analysis was performed with an Olympus BX43 fluorescence microscope (100× objective; z‐stack images). Chromatin structure, number and location of centrosomes, and the arrangement of spindles were considered to classify different mitotic phases and abnormalities.

2.12. Alkaline comet assay

The formation of DNA strand breaks and apurinic/apyrimidinic sites was monitored via the alkaline comet assay. ⁴⁰ Directly after harvesting, cells were mixed with 0.5% low melting agarose and transferred onto glass slides coated with agarose (1.5%). After lysis in alkaline buffer (pH 10, 1 h, 4°C, protected from light) and denaturation of the DNA in precooled electrophoresis buffer (pH >13, 25 min, 4°C, protected from light), electrophoresis was performed (300 mA, 25 V). After neutralization and staining with propidium iodide solution (50 μg/mL), comets were evaluated by fluorescence microscopy. Quantification of migrated DNA was performed with the TriTek Comet ScoreTM software (version 1.5), evaluating 50 cells per experimental condition.

2.13. Analysis of germ cell apoptosis in C. elegans

Apoptotic germ cells corpses were analyzed 24 h after CisPt, B02 and/or PF477736 treatment using reporter strain MD701 (bcIs39 [lim‐7p::ced‐1::GFP + lin‐15 (+)]) as described. ⁴¹

2.14. Statistical analysis

For statistical analysis the unpaired two‐tailed Student's t‐test or one‐way ANOVA were applied using GraphPad Prism 6 software. p‐values ≤.05 were considered as significant and were marked with an asterisk, plus or hashtag. Paired t‐test was used for statistical analysis of C. elegans‐based data.

3. RESULTS AND DISCUSSION

3.1. Analyses of the cross‐sensitivity of CisPt resistant bladder carcinoma cells

An established cisplatin‐resistant bladder cancer cell variant (J82^CisPt) was used as in vitro model of acquired CisPt resistance. ³³ J82^CisPt are characterized by ~four‐times higher IC₅₀ compared to parental J82^WT cells (Figure 1A). In an initial screening approach, J82^CisPt cells were analyzed regarding their cross‐sensitivity to various CAT and pharmacological inhibitors of the DDR and DNA repair to (i) get hints towards the putative molecular mechanism(s) contributing to acquired CisPt‐resistance and (ii) identify DDR‐modulatory compounds that could effectively re‐sensitize J82^CisPt cells to CisPt or (iii) induce cell death in the CisPt resistant variants in a different way. J82^CisPt showed to be cross‐resistant to other platinum compounds (i.e., carboplatin and oxaliplatin) while revealing similar sensitivity to the topoisomerase II poison doxorubicin and ionizing radiation, both being prototypical inducers of DSBs (Figure 1B and Supplementary Table 1). J82^CisPt were found to be more sensitive than J82^WT to the ribonucleotide reductase inhibitor OH‐urea and the antimetabolite 5‐fluorouracil (5‐FU) (Figure 1C,D and Supplementary Table 1), which both are well known inducers of replication stress. ⁴² , ⁴³ Regarding pharmacological inhibitors of DNA repair and DDR, we observed cross‐resistance of J82^CisPt cells to the RAD51 inhibitor B02 (Figure 1E, Supplementary Figure 1A and Supplementary Table 1). This finding was confirmed by the NeutralRed Assay (Figure 1F). In contrast, J82^CisPt cells revealed enhanced sensitivity to the pan‐HDAC inhibitor vorinostat and class I inhibitor entinostat as well as to the CHK‐inhibitors LY2603618 and AZD‐7762, ³³ but not to the CHK1 inhibitor PF477736, the PARP inhibitor niraparib and the MRE11 inhibitor mirin (Figure 1G,H and Supplementary Table 1).

FIGURE 1.

Cisplatin resistant bladder carcinoma cells (J82^CisPt) react sensitive to replication stress‐inducing compounds and are cross‐resistant to the RAD51 inhibitor B02. Cells of bladder carcinoma cell line J82^WT and its cisplatin resistant cell variant J82^CisPt were treated with different concentrations of Cisplatin (A), Doxorubicin (B), Hydroxy‐Urea (C), 5‐Fluorouracil (D), B02 (E, F), Niraparib (G), or PF477736 (H) for 72 h. After the incubation period, cell viability was measured using the AlamarBlue Assay (A–E, G, H) or the neutral red assay (F). Data shown are the mean ± SD from three independent experiments each performed in quadruplicate. *Statistical significance of J82^WT cells vs. J82^CisPt cells. ***p ≤ .001; **p ≤ .01; *p ≤ .05.

3.2. Sensitivity of J82^CisPt to replication stress is related to mitotic defects and chromosomal instability (CIN)

Since cisplatin induces replication stress by forming DNA crosslinks, the high susceptibility of J82^CisPt to replication stress‐inducing substances 5‐FU and OH‐Urea was unexpected. Aneuploidy is not only very common in cancer cells, ⁴⁴ but is also associated with both chemotherapy resistance and cellular stress. ⁴⁵ , ⁴⁶ This basal intrinsic stress could be a consequence of accelerated depletion of the nucleotide pool because of the high proliferation activity of malignant cells. ⁴⁷ By that, under‐replicated DNA is presumably formed leading to chromatin bridges that can be detected in J82^CisPt cells (Supplementary Figure 2A). In line with that, data obtained from analyzing the mitotic spindle apparatus, J82^CisPt showed a higher percentage of cells with abnormalities occurring during mitosis as compared to J82^WT cells (Supplementary Figure 2B), which can lead to chromosome mis‐segregation and numerical alterations. Alternatively, mitotic abnormalities of J82^CisPt cells might be the consequence of aneuploidy‐based replication stress.

3.3. Combination treatment with RAD51_i  + CHK1_i causes synergistic cytotoxicity in J82^CisPt cells

Replication fork stability is maintained by a complex and fine‐tuned system involving RAD51, ATR/CHK1, and PARP. ⁴⁸ Based on the findings in J82^CisPt cells, we speculated that replication fork protecting proteins are a promising target for disrupting the replication machinery in CisPt resistant cells thereby triggering cell death. Since many different factors can contribute to acquired CisPt resistance, ¹³ we hypothesize targeting superordinate characteristics of malignant cells, such as rapid proliferation, as a more promising strategy to overcome acquired drug resistance than targeting individual resistance factors of a heterogenous CisPt resistant cell population. Therefore, J82^CisPt cells were exposed to multiple combination treatments using various pharmacological DDR modifiers ± CisPt or combination of two DDR modifiers, followed by measurement of the cell viability via the AlamarBlue Assay and calculation of the corresponding combination indices (CI). Combination of the RAD51 inhibitor B02 with CisPt caused additive toxicity (Supplementary Figure 3). Notably, using low to moderate toxic doses of the CHK1 inhibitor PF477736 (CHK1_i) in combination with the RAD51 inhibitor B02 (RAD51_i) caused additive to synergistic cytotoxicity in J82^CisPt (Figure 2A). This synergism was also detectable in Western Blot analyses for the apoptotic markers cleaved caspase 7 and cleaved PARP (Figure 2B). This finding shows that the synergistic cell death evoked by combined treatment with RAD51_i plus CHK1_i is at least partly dependent on apoptosis. In concordance with that, the pan‐caspase inhibitor QVD reduced SubG1 fraction after B02 plus PF477736 co‐treatment (Figure 2C). Additionally, we investigated the CI for additional combinations of DDR inhibitors and chemotherapeutics. Synergistic toxicity was also found upon co‐treatment with CHK1_i and OH‐Urea or CHK1_i and 5‐FU (Supplementary Table 2).

FIGURE 2.

Synergistic toxicity evoked by combined treatment with the CHK1 inhibitor PF477736 (PF) and the RAD51 inhibitor B02 in J82^CisPt. (A) The viability of J82^CisPt cells was analyzed after treatment with differently combined low and moderate toxic doses of CHK1_i PF477736 and RAD51_i B02 as indicated. Cell viability was measured after a 72 h treatment period using the AlamarBlue Assay. Based on the data obtained from three independent experiments each performed in quadruplicate, the combination indices (CIs) were calculated using CompuSyn software (CI < 0.9 indicating synergistic effects, CI ≈1 additive effects and CI > 1.2 antagonistic effects). (B) Protein expression and activation of different apoptosis‐related factors were examined via Western Blot analyses using protein extracts of J82^CisPt cells treated for 6 or 24 h with 10 μM B02, 1 μM PF477736 or both substances. (C) 10 μM B02 and 1 μM PF477736 ± the pan caspase inhibitor QVD were added 24 h after seeding and after 24, 48 and 72h treatment period the induction of SubG1 fraction was analyzed by propidium iodide staining followed by detection by flow cytometry. Data presented are the mean + SD from three independent experiments. ***p ≤ .001; *p ≤ .05.

3.4. Combination treatment with RAD51_i  + CHK1_i disrupts cell cycle progression in J82^CisPt cells

To elucidate the molecular mechanisms of synergistic toxicity caused by combined treatment with CHK1_i and RAD51_i, proliferation‐ and cell cycle‐related endpoints were examined. Cell cycle analyses showed that both mono‐treatment with PF477736 and co‐treatment (B02 + PF477736) induced an S‐phase arrest accompanied by an increased frequency of apoptotic cells (Figure 3A,B). While the S‐phase arrest after 24 h was not significantly different between those two groups, the combined treatment led to a higher number of cells in subG1 fraction at both early (24 h) and late (72 h) time point. This finding indicates that under conditions of CHK1‐inhibition alone, the cells are still able to recover from the S‐phase arrest, whereas they are more likely to undergo cell death under conditions of simultaneous inhibition of CHK1 plus RAD51.

FIGURE 3.

S‐phase arrest in J82^CisPt cells following treatment with B02 and PF477736. (A, B) Inhibitors (10 μM B02 ± 1 μM PF477736) were added 24 h after seeding and cell cycle distribution was analyzed after a treatment period of 24 h (A) and 72 h (B) employing propidium iodide staining and flow cytometric analysis. Data are presented as mean + SD from n = 3 independent experiments. (C) The EdU incorporation of J82^CisPt cells was analyzed after treatment with 10 μM B02 or/and 1 μM PF477736. EdU incorporation was analyzed after 24 h treatment period with an EdU pulse of 2 h. The graph shows the mean + SD of n = 3 independent experiments (1000–2000 nuclei analyzed per sample). The scale bars in the representative pictures correspond to 50 μm. ***p ≤ .001; **p ≤ .01; *p ≤ .05; significant compared to control (*), to B02 mono‐treatment (#) and to PF477736 mono‐treatment (+).

For a detailed analysis of S‐phase‐related effects, EdU incorporation was monitored. J82^CisPt cells treated with the combination of RAD51_i and CHK1_i showed a significantly reduced EdU incorporation after 24 h treatment, whereas the corresponding mono‐treatments did not reveal any significant changes in EdU incorporation compared to the untreated control (Figure 3C). These data, together with the detected S‐phase arrest, lead to the assumption that DNA replication is effectively hampered in B02/PF477736 co‐treated J82^CisPt cells.

3.5. A small number of cells still enter mitosis with under‐replicated DNA

Errors occurring during S‐phase can be passed on into mitosis and impair separation of cells. Indeed, combination treatment of cancer cells with the CHK1 inhibitor prexasertib and B02 causes premature entry into mitosis with under‐replicated DNA. ⁴⁹ This under‐replicated DNA requires to be completed via mitotic DNA synthesis. However, since RAD51 is an important factor in this pathway, ⁵⁰ inhibition of RAD51 is anticipated to promote mitotic catastrophe. Flow cytometry‐based cell cycle analyses showed a reduction in G2/M‐phase cells following B02 plus PF477736 co‐treatment (Figure 3A). Moreover, a reduction of the percentage of phospho‐histone 3 (pH 3)‐positive mitotic cells was observed after combination treatment and PF4777736 mono‐treatment (Supplementary Figure 4A). Indeed, double‐staining with pH 3 antibody and propidium iodide showed some pH 3 positive cells with a DNA content of S‐phase cells (2.8%), pointing to a premature entry of the co‐treated J82^CisPt cells into mitosis (Supplementary Figure 4B). Also, when examining DAPI stained nuclei by fluorescence microscopy, some of the co‐treated cells' nuclei showed morphological features of a mitotic catastrophe (Supplementary Figure 4C). ⁵¹ Still, this phenomenon is considered a minor finding in J82^CisPt cells, since it cannot explain the severe cytotoxic and pro‐apoptotic effect caused by the combined treatment (Figure 2A,B). Assuming that mitotic effects are likely not of major relevance for the synergistic toxicity of a RAD51_i + CHK1_i co‐treatment, further investigations were focused on S‐phase directly. We speculate that treatment with CHK1_i + RAD51_i represents a “final hit,” so that J82^CisPt cells do not enter mitosis anymore but rather die during S‐phase.

3.6. Decelerated S‐phase progression is due to hampered replication

To analyze the impact of B02/PF477736 co‐treatment on the replication fork level, a DNA fiber spreading assay was performed. Cells co‐treated for 6 h with CHK1_i + RAD51_i revealed shorter chlorodeoxyuridine (CldU) (data not shown) and iododeoxyuridine (IdU) labeled DNA fibers as compared to the untreated control and the mono‐treatments, indicating a decelerated replication fork progression (Figure 4A, upper panel). The shorter DNA tracks are likely not a consequence of excessive origin firing, since the percentages of origins did not show major differences between the groups, in particular between PF477736 treated cells with 20% origins and B02 + PF77736 treated cells showing 18% origins (Figure 4A, middle panel). However, evaluation of asymmetries at tri‐colored replication origins revealed a statistically significant rise after co‐treatment as compared to all other groups suggesting severe replication fork stalling (Figure 4A, lower panel). In conclusion, DNA replication dynamics in J82^CisPt cells treated with B02 + PF77736 are severely perturbed by the stalling of replication forks, which is not seen mono‐treatments. Notably, this did not result in transcriptional upregulation of different replicative and non‐replicative polymerases trying to compensate for stalled replication by subsequent increase in DNA synthesis or bypassing of replication impairments (data not shown).

FIGURE 4.

Hampered replication occurring after combined treatment of J82^CisPt with PF477736 and B02. (A) 24 h after seeding, J82^CisPt cells were treated with either 10 μM B02, 1 μM PF477736 or both substances for 6 h. After the treatment, cells were incubated for 20 min with CldU, followed by 20 min incubation with IdU. The BrdU analogs were labeled by immunofluorescence, staining was analyzed microscopically and fiber lengths were measured using ImageJ. Data presented are from two independent experiments, whereby 200 fibers were measured for each sample. Each dot represents one analyzed fiber and the black lines show the mean ± SEM. The mean value is also given above the graphs. Upper panel, nascent DNA elongation, graphically displayed as IdU track lengths of bi‐colored DNA fibers. Middle panel, table summarizing the evaluation of proportions of origins and terminations in the total fiber population (ns, not significant). Lower panel, as measure of DNA replication fork stalling fork asymmetry was determined from three‐colored replication origins as the ratio of the longer red IdU fiber track length versus the shorter red IdU fiber track length departing from the same green CldU track. (B) Formation of RPA foci in the nuclei of J82^CisPt cells was analyzed after 6 and 24 h treatment with 10 μM B02 or/and 1 μM PF477736 via immunocytochemical staining. Data are shown with each dot representing one analyzed nucleus and the black lines showing the mean ± SEM from three independent experiments, where in each case 50 nuclei were counted. The scale bars in the representative pictures correspond to 10 μm. ***p ≤ .001; **p ≤ .01; *p ≤ .05; significant compared to control (*), to B02 mono‐treatment (#) and to PF477736 mono‐treatment (+).

Excessive DNA unwinding takes place at stalled replication forks due to uncoupling of the helicase from the polymerase, resulting in a high amount of ssDNA. ⁵² Such ssDNA will be coated with RPA for protection from degradation. RPA‐coated ssDNA creates analyzable foci, which therefore serve as a surrogate marker of ssDNA formation. Indeed, the number of nuclear RPA foci was increased after CHK1_i‐treatment for 24 h but even more and earlier after combination treatment with B02 + PF477736 (Figure 4B). In addition, high protein levels of the Ser33‐phosphorylated RPA subunit RPA32, which is catalyzed by ATR, and of Ser4/Ser8‐phosphorylated RPA32, which is mediated by DNA‐PK and ATM, were detected 24 h after mono‐treatment with CHK1_i and already after 6 h of combination treatment with RAD51_i (Figure 5A). RPA phosphorylations are catalyzed by the mentioned ATM/ATR kinases in response to replication stress in order to recruit DNA repair proteins. ⁵³ In addition, increased levels of pKAP1 (Ser824), pCHK1 (Ser345), and pp53 (Ser15), which are other targets of ATM/ATR kinases were also detected both after CHK1_i mono‐treatment and upon co‐treatment with RAD51_i (Figure 5A). From this we conclude that CHK1_i‐treatment is sufficient to induce ATM/ATR signaling towards different phosphorylation substrates. However, rapid phosphorylation and chromatin association of RPA requires additional treatment with RAD51_i, underscoring the synergistic effects of CHK1 and RAD51i on DNA replication structures. Of note, synergism at an early timepoint is independent of the expression of the compounds target proteins, since CHK1 and RAD51 protein levels remained unaltered upon inhibitor treatment (Figure 5A).

FIGURE 5.

S‐phase‐dependent formation of DNA damage and activation of DDR‐ and DNA repair‐related mechanisms following treatment of J82^CisPt with PF477736 and B02. (A) Protein expression and activation of different replication stress‐ and DDR‐related factors was examined via Western Blot analyses with protein extracts of J82^CisPt cells treated for 6 h or 24 h with 10 μM B02, 1 μM PF477736 or both substances. (B) Formation of DNA strand breaks was analyzed via alkaline Comet Assay after 24 h mono‐ and combination‐treatment with 10 μM B02 and 1 μM PF477736. Tail intensity (% DNA in tail) is displayed as dot for every analyzed cell and the mean ± SEM calculated from n = 3; N = 50. ***p ≤ .001; **p ≤ .01; *p ≤ .05; significant compared to control (*), to B02 mono‐treatment (#) and to PF477736 mono‐treatment (+). (C) To analyze in which cell cycle phase the damage predominantly occurs, a double staining with γH2AX antibody and propidium iodide was applied and examined by flow cytometry after a treatment period of 6 and 24 h in J82^CisPt. Displayed representative images of the flow cytometrical analyses were generated using FlowJo software. (D) J82^CisPt cells were co‐treated with 10 μM B02 + 1 μM PF477736 for 24 h, afterwards immunocytochemical co‐staining of γH2AX and RPA was performed to analyze the correlation of both markers. For γH2AX, the mean fluorescence intensity of the nuclei was measured and the number of RPA foci per nucleus were counted. Data are shown with each dot representing one analyzed nucleus and the black lines showing the mean ± SEM from two independent experiments, where in each case 50 nuclei were measured. The scale bar in the representative picture corresponds to 20 μm. ***p ≤ .001; significant compared to nuclei with <10 RPA foci.

Protection of ssDNA at stalled replication forks by RPA is considered as a transient effect, and the function of RPA is subsequently taken over by RAD51 filaments. ⁴⁸ The RAD51 inhibitor B02 inhibits binding of RAD51 to ssDNA, ⁵⁴ as reflected by loss of RAD51 foci induction after IR (Supplementary Figure 1B). ⁵⁰ Thereby B02 impedes not only RAD51's ssDNA‐protecting function at stalled replication forks but also its role in DNA strand exchange during homologous recombination. Consequently, inhibition of DNA binding by B02 will abrogate the exchange of RPA by RAD51 and therefore also protection of nascent DNA. It has been demonstrated that DNA strand breaks are generated when the amount of ssDNA exceeds the corresponding level of RPA available in the cell. ⁵⁵ This threshold will most likely be reached earlier with RAD51_i treatment, since under these conditions RPA is sequestered on ssDNA instead of being exchanged by RAD51. In consequence, the free RPA pool is likely decreased. If the source for replication fork stalling is removed, replication fork restart can take place. Homologous recombination, and RAD51 as important protein in this pathway, is a crucial factor for fork restart. Hence, cells with impaired RAD51 cannot restart replication forks as efficiently as control cells and therefore more likely suffer from persistent fork stalling, resulting in replication fork collapse and one‐ended DSBs. ⁵⁶

3.7. Replication stress results in replication fork collapse, induction of DNA strand breaks and DDR signaling

Phosphorylated histone 2AX at Ser139 (γH2AX), known as a marker for DNA damage as well as replication fork collapse, ⁵⁷ was detectable already after 6 h of RAD51_i + CHK1_i co‐treatment, while RAD51_i or CHK1_i mono‐treatment did not yet induce such signal (Figure 5A). Synergistic induction of DNA strand break formation was confirmed using the alkaline comet assay. Accordingly, B02 + PF477736 co‐treated cells showed a significantly higher comet tail intensity than PF477736 mono‐treated cells (Figure 5B). Analysis of the number of γH2AX positive cells in the different cell cycle phases revealed that DNA damage is generated mainly in S‐phase cells (Figure 5C). Furthermore, immunocytochemical co‐staining suggests that the γH2AX signal predominantly originates from cells whose nuclei also harbor a high number of RPA foci (Figure 5D). A distinct increase in γH2AX positive S‐phase cells was already detected after 6 h of RAD51_i + CHK1_i incubation, which was not observed upon mono‐treatment. In accordance with the Western Blot analysis, the PF477736‐treated group showed a marked, but still less pronounced γH2AX signal than the co‐treated group after 24 h of incubation, whereas B02 mono‐treatment failed to increase the γH2AX signal (Figure 5A,C). Taken together, combined treatment synergistically induces S‐phase specific DNA damage, ssDNA and DNA strand‐breaks.

3.8. Analysis of the specificity of B02 + PF477736‐induced stress responses

LY2603618 and RI(dl)2 were used as additional inhibitors of CHK1 and RAD51, respectively, to confirm the synergistic cytotoxicity and molecular mechanism observed with B02 and PF477736. As anticipated, co‐treatment with B02 + LY2603618 also evoked additive to synergistic cytotoxicity (Supplementary Table 2) and induced S‐phase arrest as well as phosphorylation of RPA32 and H2AX (Figure 6A). Combining CHK1_i PF477736 with RAD51_i RI(dl)2, that specifically inhibits D‐loop formation, which is important for RAD51‐mediated replication fork restart, ⁵⁸ strong synergistic cytotoxicity, robust S‐phase arrest and induction of pRPA32 and γH2AX were observed (Supplementary Table 2 and Figure 6B). Combining CHK1_i PF477736 with PARP_i niraparib, that presumably promotes the restart of regressed forks, affected cytotoxicity and S‐phase arrest in antagonistic manner (data not shown). This data supports the idea that synergistic toxicity evoked by CHK1‐ + RAD51‐inhibitor is a general phenomenon due to impeded replication fork restart.

FIGURE 6.

Combining B02 or PF477736 with other CHK1‐ or RAD51‐inhibitors, respectively, likewise induces S‐phase arrest, replication stress, and DNA damage in J82^CisPt. J82^CisPt cells were co‐treated with 10 μM B02 + 1 μM LY2603618 (LY) (A) or 1 μM PF477736 (PF) + 30 μM RI(dl)2 (RI2) (B). Following 24 h treatment, propidium iodide‐based cell cycle analysis was performed by flow cytometry with emphasis on the proportion of cells in S‐phase. A total of 10,000 counts were measured for quantification. Induction of γH2AX and pRPA32 (S4, S8) was examined via Western Blot analyses with protein extracts of J82^CisPt cells treated for 6 or 24 h with the corresponding combination or mono‐treatments.

In addition, we investigated the outcome of a combination treatment with B02 and PF477736 on the viability of cisplatin‐resistant cell variants of other origin. Noteworthy, additive to synergistic cytotoxicity was also achieved in CisPt‐resistant neuroblastoma cells (SH‐SY5Y^CisPt), lung tumor cells (A549^CisPt) and ovary tumor cells (A2780^CisPt) (Supplementary Figure 5A). Moreover, SH‐SY5Y^CisPt cells also revealed a moderate S‐phase block and substantial increase in pRPA and γH2AX protein upon co‐treatment (Supplementary Figure 5B). Interestingly, A2780^CisPt cells, which only responded additively to B02 + PF477736, showed much weaker protein expression of CHK1 and RAD51 than the other CisPt resistant cells (Supplementary Figure 5C), which was not reflected on the mRNA levels. This suggests that the efficacy of the combined treatment correlates with the basal protein expression levels of RAD51 and CHK1. Protein expression of repriming polymerase PrimPol, that is frequently upregulated in cancer cells ⁵⁹ and is involved in the bypassing of replication obstacles, does not affect the outcome of RAD51 + CHK1 inhibitor treatment (Supplementary Figure 5C).

Furthermore, in vitro experiments employing normal cells were performed. Synergistic toxicity of the CHK1_i + RAD51_i combination treatment was found in human induced pluripotent stem cells (hiPSCs), representing highly proliferative non‐malignant cells such as stem/progenitor cells of the hematopoietic system. (Supplementary Figure 6A). Employing the nematode Caenorhabditis elegans as in vivo model, no apoptotic cell death in the posterior gonads was observed using 5‐fold higher concentrations of the inhibitors than in J82^CisPt, indicating an acceptable tolerability of CHK1_i + RAD51_i on the level of an organism (Supplementary Figure 6B).

4. CONCLUSION

Simultaneous targeting of two different DNA damage tolerance‐associated pathways by combined treatment with non‐ to moderate toxic doses of the CHK1_i PF477736 and the RAD51_i B02 is useful to overcome acquired CisPt resistance of bladder carcinoma cells and malignant cells of other origin. Synergistic toxicity evoked by the combination treatment is probably due to blockage of multiple mechanisms enabling the bypassing of DNA replication impediments. ⁶⁰ On the one hand, recombination‐mediated template switching enabling replication restart as well as fork protection regulated by RAD51 are prevented by the RAD51_i B02. At the same time, repriming is not efficient in the presence of the CHK1_i PF477736 because CHK1 regulates the DNA primase PrimPol. ⁶¹ Altogether, simultaneous impairment of both of these replication fork protecting factors leads to persistent replication fork stalling without adequate protection of accumulating ssDNA by RPA because lack of exchange through RAD51 eventually depletes the RPA pool. We hypothesize that, ultimately, this results in replication fork collapse and subsequent accumulation of DNA strand breaks triggering cell death in S‐phase.

AUTHOR CONTRIBUTIONS

Julia Mann: Conceptualization; formal analysis; investigation; methodology; visualization; writing – original draft. Kathrin Niedermayer: Methodology. Johannes Krautstrunk: Investigation. Lena Abbey: Investigation. Lisa Wiesmüller: Methodology; writing – original draft. Roland P. Piekorz: Methodology; writing – original draft. Gerhard Fritz: Conceptualization; funding acquisition; project administration; resources; supervision; writing – original draft.

FUNDING INFORMATION

This work was supported by the Düsseldorf School of Oncology of the Heinrich Heine University Düsseldorf, of which Julia Mann is a member, and the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation – 417677437/GRK2578) (Gerhard Fritz). Lisa Wiesmüller was supported by a grant from the German Cancer Aid, Priority Program “Translational Oncology: DETECT‐CTChigh” (Applying liquid biopsies to decipher therapy resistance mechanisms and develop adaptive treatment strategies for metastatic breast cancer – 70114705). Kathrin Niedermayer is a member of the International Graduate School in Molecular Medicine, Ulm University.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Data S1. Supporting Information.

Mann J, Niedermayer K, Krautstrunk J, et al. Combined inhibition of RAD51 and CHK1 causes synergistic toxicity in cisplatin resistant cancer cells by triggering replication fork collapse. Int J Cancer. 2025;156(2):389‐402. doi: 10.1002/ijc.35164

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon request.