RAD51-Mediated DNA Homologous Recombination Is Independent of PTEN Mutational Status

Simple Summary

PTEN is an important tumor suppressor that is frequently mutated in malignancy. PTEN mutational loss has been associated with reduced RAD51 expression and homologous recombination deficiency (HRD), however; recent studies have failed to recapitulate these findings. Here, we show that RAD51 expression, foci formation and homologous recombination repair activity are unaltered in normal and tumorigenic PTEN-deficient cells and patient samples. Furthermore, we show that PTEN-deficient tumor cell lines do not synergize with the clinical PARP inhibitor olaparib, underscoring a need to discontinue its use in treating patients with PTEN-deficient tumors that do not otherwise exhibit HRD.

Abstract

PTEN mutation occurs in a variety of aggressive cancers and is associated with poor patient outcomes. Recent studies have linked mutational loss of PTEN to reduced RAD51 expression and function, a key factor involved in the homologous recombination (HR) pathway. However, these studies remain controversial, as they fail to establish a definitive causal link to RAD51 expression that is PTEN-dependent, while other studies have not been able to recapitulate the relationship between the PTEN expression and the RAD51/HR function. Resolution of this apparent conundrum is essential due to the clinically-significant implication that PTEN-deficient tumors may be sensitive to poly (ADP-ribose) polymerase (PARP) inhibitors (PARPi) commonly used in the clinical management of BRCA-mutated and other HR-deficient (HRD) tumors. Methods: Primary Pten-deficient (and corresponding wild-type) mouse embryonic fibroblasts (MEFs) and astrocytes and PTEN-null human tumor cell lines and primary cells were assessed for RAD51 expression (via the Western blot analysis) and DNA damage repair analyses (via alkali comet and γH2AX foci assays). RAD51 foci analysis was used to measure HR-dependent DNA repair. Xrcc2-deficient MEFs served as an HR-deficient control, while the stable knockdown of RAD51 (shRAD51) served to control for the relative RAD51/HR-mediated repair and the phospho-53BP1 foci analysis served to confirm and measure non-homologous end joining (NHEJ) activity in PTEN-deficient and shRAD51-expressing (HRD) lines. Cell proliferation studies were used to measure any potential added sensitivity of PTEN-null cells to the clinically-relevant PARPi, olaparib. RAD51 levels and DNA damage response signaling were assessed in PTEN-mutant brain tumor initiating cells (BTICs) derived from primary and recurrent glioblastoma multiforme (GBM) patients, while expression of RAD51 and its paralogs were examined as a function of the PTEN status in the RNA expression datasets isolated from primary GBM tumor specimens and BTICs. Results: Pten knockout primary murine cells display unaltered RAD51 expression, endogenous and DNA strand break-induced RAD51 foci and robust DNA repair activity. Defective HR was only observed in the cells lacking Xrcc2. Likewise, human glioblastoma multiforme (GBM) cell lines with known PTEN deficiency (U87, PTEN-mutated; U251 and U373, PTEN-null) show apparent expression of RAD51 and display efficient DNA repair activity. Only GBM cells stably expressing shRNAs against RAD51 (shRAD51) display dysfunctional DNA repair activity and reduced proliferative capacity, which is exacerbated by PARPi treatment. Furthermore, GBM patient-derived BTICs displayed robust RAD51 expression and intact DNA damage response signaling in spite of PTEN-inactivating mutations. RNA expression analysis of primary GBM tissue specimens and BTICs demonstrate stable levels of RAD51 and its paralogs (RAD51B, RAD51C, RAD51D, XRCC2, XRCC3, and DMC1), regardless of the PTEN mutational status. Conclusions: Our findings demonstrate definitively that PTEN loss does not alter the RAD51 expression, its paralogs, or the HR activity. Furthermore, deficiency in PTEN alone is not sufficient to impart enhanced sensitivity to PARPi associated with HRD. This study is the first to unequivocally demonstrate that PTEN deficiency is not linked to the RAD51 expression or the HR activity amongst primary neural and non-neural Pten-null cells, PTEN-deficient tumor cell lines, and primary PTEN-mutant GBM patient-derived tissue specimens and BTICs.

1. Introduction

PTEN encodes a phosphatidylinositol-3,4,5-trisphosphate 3-phosphatase containing a tensin-like domain and a catalytic domain typical of those of the dual-specificity protein tyrosine phosphatases [1,2]. Unlike most protein tyrosine phosphatases, PTEN preferentially dephosphorylates phosphoinositide substrates. PTEN is involved in diverse biological processes, functioning as an important negative regulator of the phosphatidylinositol-3,4,5-trisphosphate kinase-signaling cascade in several downstream cellular processes, including cell regulation and apoptosis. Functioning as a tumor suppressor, PTEN is one of the most commonly-mutated (inactivated) genes in human cancer, including glioblastoma multiforme (GBM), breast, prostate, endometrium, ovary, colon cancers, melanoma, and lymphoma [3,4]. Tumors featuring PTEN mutations are characterized by pronounced genomic instability and chromosomal defects [5].

Homologous recombination (HR) is a critical ATP-dependent DNA double-strand break repair (DSBR) pathway, particularly active in the G1-S phase of the cell cycle [6,7,8,9], wherein a template strand invades base-paired strands of homologous DNA molecules to guide repair of damaged DNA bases [10]. The RAD51 recombinase protein forms a tripartite complex with XRCC2 and BRCA2 that associates with additional co-factors and RAD51 paralog complexes to mediate this form of homologous sister-chromatid guided DSBR. Pathogenic germline and acquired somatic mutation, promoter hypermethylation, or other as yet to be identified mechanisms [11,12,13,14] can result in a complex loss of function leading to HR deficiency (HRD). HRD reduces overall DNA repair fidelity thus impacting cellular survival amongst dividing cell populations and can promote early tumorigenic events implicated in a number of malignancies, including breast, ovarian, brain, and endometrial cancers. Cellular survival in the context of HR inactivation is thought to be mediated via a combination of classical non-homologous end joining (c-NHEJ)-mediated DSBR, alternative end joining (alt-EJ), and/or base excision repair (BER) [15]. Nevertheless, tumors featuring bi-allelic HR-associated mutations offer a unique therapeutic opportunity that has heralded the use of inhibitors that target poly (ADP-ribose) polymerase (PARP) [16,17,18], a critical enzyme that functions as a DNA strand break sensor and activator of the BER response and other end joining repair pathways [19,20,21]. This “synthetic lethal” therapeutic strategy successfully exploits the tumor’s HRD and accompanying proliferation/replication stress whereby PARP inhibition/inhibitors (PARPi) target the cell’s outstanding break repair capacity. This approach has been successful in the treatment of breast, ovarian, and endometrial cancer patients that feature bi-allelic BRCA1/2 mutations/deletion [17,22,23,24] and recently expanded clinical usage in lung, prostate, pancreatic, colon cancer and other cancer patients featuring a variety of tumors featuring HRD [18,25,26,27,28].

A number of studies have implicated PTEN in the cellular expression of RAD51 (and/or its paralogs), linking PTEN to cellular HR and DSBR [29,30,31]. Likewise, studies have attributed tumors with PTEN-deficiency as having enhanced sensitivity to PARP inhibition, purportedly due to PTEN-associated loss of RAD51 function and HRD, similar to BRCA loss [32,33]. However, these findings have been met with controversy, as more recent studies have shown opposite conclusions: that PTEN-deficient cells do not display nor attenuated RAD51 expression nor HRD [34,35,36,37]. The continued uncertainty surrounding the PTEN-RAD51 connection adversely affects scientific research, forums, and academia as differing viewpoints yield conflicting inferences, assumptions and factor increasingly prominently in translational research, drug development, and clinical care.

To delineate any potential role for PTEN in the regulation of the RAD51/HR-dependent activity, we used a systematic approach to evaluate RAD51 expression, DNA repair activity, and PARPi responses in diverse PTEN-deficient primary and cancer cell models. Here, we demonstrate that PTEN-deficient cells retain RAD51 expression and activity, while HR and DNA repair activity in general remain robust independent of the PTEN status. Consequently, we show that PTEN-null/mutant cells show no overt HR-dependent sensitivity to PARP1 inhibition, indicating that any putative PARP-dependent tumor ablation in PTEN-null cells likely involves alternative mechanisms. These studies demonstrate unequivocally that PTEN does not impact RAD51 expression or affect cellular HR/DNA damage repair activity.

2. Materials and Methods

2.1. Generation of Primary Cells from Pten-Deficient Mice

Mouse experiments were approved by the Institutional Animal Care and Use Committee (protocol number 278). Primary astrocytes were prepared [38] from cortices of postnatal day 2 (P2) brains derived from controls (Pten^+/+ and Pten^LoxP/+) and conditional Pten (Pten^LoxP/LoxP) [39]. Cortices were dissociated by passage through a 5 mL pipette and cells were resuspended in Dulbecco’s modified Eagle medium and Ham’s nutrient mixture F-12 (1:1 DMEM/F12, Gibco-BRL) supplemented with 10% fetal bovine serum (v/v), 1x glutamax, 100 U/mL penicillin, 100 mg/mL streptomycin, and 20 ng/mL epidermal growth factor (EGF; Millipore). Primary astrocytes were established in Primeria T-25 tissue culture flasks (Falcon) at 37 °C in a humidified CO₂-regulated (5%) incubator. Following cell establishment, astrocytes at 25% confluency were transduced with retroviral Cre recombinase (MSCV-creGFP). DNA from green fluorescent protein (GFP)-positive cells was isolated and underwent PCR analysis to validate efficient Cre-mediated deletion of floxed sites within the Pten locus [39,40]. Immunoblotting analysis using a rabbit anti-PTEN monoclonal antibody (9188, Cell Signaling Technologies) was also used to confirm concordant PTEN expression status.

Primary mouse embryonic fibroblasts (MEFs) were prepared from an E13.5 embryonic mesenchyme [41]. Tissue was minced using dissection scissors, trypsinized, and resuspended in a Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (v/v), 1x glutamax, 100 U/mL penicillin and 100 μg/mL streptomycin (1× Pen/Strep), and β-mercaptoethanol and established in T-25 tissue culture flasks (Falcon) at 37 °C in a humidified CO₂-regulated (5%) incubator. Similar to the astrocyte preparation, deletion of Pten within primary MEFs occurred via expression of retroviral MSCV-creGFP, followed by PCR validation and Western analysis as outlined above.

2.2. PTEN-Deficient Cancer Cell Lines and RNA Interference

PTEN-mutant (U87MG) and PTEN-deleted (U251MG and U373MG) GBM cell lines were originally obtained from American Type Culture Collection (ATCC) (Manassas, VA, USA) and maintained in a DMEM supplemented with 5% fetal bovine serum (v/v), 1× GlutaMAX, 100 U/mL penicillin, 100 μg/mL streptomycin. The Burkitt lymphoma cell line, BJAB, was grown in RPMI 1640 supplemented with 1× Pen/Strep and 1× GlutaMAX and 10% fetal bovine serum. The use of all cell lines was restricted to passages (P) 5 to P25. Lentiviral particles were generated from MISSION^® shRNA plasmid DNA constructs (Millipore-SIGMA, St. Louis, MO, USA) to deliver and stably express shRNAs in GBM cell lines to diminish human RAD51 expression (Figure S1). Lentivirus was made via co-transfection of pLKO.1-puro plasmids with MISSION^® lentiviral packaging DNA (Millipore-Sigma, SHP001) into sub-confluent 293LTV cells (Cell Biolabs Inc, San Diego, CA, USA) using XtremeGene HD (Millipore-SIGMA). Lentivirus was collected 48–96 h later and filtered (0.45µm pore size). GBM cell lines were transduced with lentiviral particles for 48 h at 37 °C followed by puromycin (1 μM) selection for 72 h. Western analysis was used to confirm shRNA-mediated knockdown of RAD51 expression. Following transduction of 3 independent shRAD51 constructs into GBM cells and the subsequent Western analysis, TRCN0000329686 was selected for further knockdown studies due to its highest reduction in RAD51 protein expression (Figure S1). Non-target scrambled (SCM) shRNA and lentiviral TurboGFP (Millipore-SIGMA) expression cDNAs were used as negative and positive transduction controls, respectively.

2.3. Isolation and Propagation of GBM Patient Brain Tumor Initiating Cells (BTICs)

Human GBM samples from primary (treatment-naïve patients 428 and 935) and recurrent (treatment-resistant patients 241 and 972) tumors were obtained from consenting patients as approved by the Hamilton Health Sciences/McMaster Health Sciences Research Ethics Board (REB# 07-366). Brain tumor samples were dissociated in the PBS (Thermo Fisher Scientific, cat# 10010049, Waltham, MA, USA) containing 0.2 Wünsch unit/mL Liberase Blendzyme 3 (Millipore Sigma, cat# 5401119001) and incubated in a shaker at 37 °C for 15 min. The dissociated tissue was filtered through a 70 μm cell strainer (Falcon, cat# 08-771-2) and collected by centrifugation (1500 rpm, 3 min). Red blood cells were lysed using an ammonium chloride solution (STEMCELL Technologies, cat# 07850, Vancouver, BC, CA). GBM cells were resuspended in the Neurocult complete (NCC) media, a chemically defined serum-free neural stem cell medium (STEMCELL Technologies, cat# 05751), supplemented with the human recombinant epidermal growth factor (20 ng/mL; STEMCELL Technologies, cat# 78006), basic fibroblast growth factor (20 ng/mL; STEMCELL Technologies cat# 78006), heparin (2 μg/mL 0.2% heparin sodium salt in the PBS; STEMCELL technologies, cat# 07980), antibiotic-antimycotic (1×; Wisent, cat# 450-115-EL, Montreal, QC, Canada), and plated on ultra-low attachment plates (Corning, cat# 431110, Corning, NY, USA) and cultured as neurospheres and used with minimal culturing (<20 passages).

Neurospheres were plated on polyornithine-laminin coated plates for adherent growth. Adherent cells were replated in low-binding plates and cultured as tumorspheres, which were maintained as spheres upon serial passaging in vitro. Compared to commercially available GBM cell lines which are cultured in mono-layers, patient-derived 3D cultures represent the variety of heterogeneous clones present within patient samples [42].

2.4. Western Analysis

Protein extracts were prepared using a lysis buffer (50 mM Tris–HCl, 200 mM NaCl, 0.2% NP-40, 1% Tween-20 (v/v)) supplemented with 1 mM NaF, 1 mM sodium vanadate, 50 mM β-glycerophosphate, 2 mM phenylmethylsulfonyl fluoride (PMSF), and Complete protease inhibitor (Roche, Basil, Switzerland). Following quantification using the Bradford assay (Bio-Rad, Hercules, CA, USA), proteins (20 µg per lane) were separated through a Novex 4–12% (w/v) Bis-Tris Sodium Dodecyl Sulfate (SDS) polyacrylamide gel (Life Technologies) and transferred onto a nitrocellulose membrane (Bio-Rad). Blots were sequentially immunostained with either anti-RAD51 monoclonal antibody (mAb) (Cell Signaling Technology, cat# 8875, 1:1000, Danvers, MA, USA), anti-PTEN mAb (Cell Signaling Technology, cat# 9188, 1:1000), mouse anti-PAR polyclonal antibody (pAb) (Trevigen, cat# 4335-MC-100, 1:1000, Gaithersburg, MD, USA), mouse anti-PARP1 mAb (Millipore-Sigma, cat# AMAb90959, 1:1000), rabbit anti-p-KAP1 mAb (Cell Signaling Technology, cat# 4127, 1:1000), rabbit anti-KAP1 mAb (Cell Signaling Technology, cat# 4124, 1:1000), rabbit anti-p-ATM^S1981 mAb (Cell Signaling Technology, cat# 13050, 1:1000), rabbit anti-ATM mAb (Cell Signaling Technology, cat# 2873, 1:1000), rabbit anti-p-53BP1^S1778 pAb (Cell Signaling Technology, cat# 2675, 1:1000), or rabbit anti-53BP1 pAb (Cell Signaling Technology, cat# 4937, 1:1000), followed by a horseradish peroxidase–conjugated goat anti-mouse (Bio-Rad, cat# 1706516, 1:2000) or a goat anti-rabbit antibody (Bio-Rad, cat# 1706515, 1:2000) and detected using the Clarity chemiluminescence reagent (Bio-Rad) and imaged using an ImageQuant LAS 500 automated chemiluminescence imager. Rabbit anti-actin pAb (Millipore-Sigma, cat# A2066, 1:2000) or rabbit anti-Scythe ab [43] (1:5000) were used as protein-loading controls. Cell treatments for Western analysis are in each respective figure legend. Blots were visualized using a LAS 500 Imager (GE Healthcare). Densitometric analysis of individual protein bands was performed via ImageJ (National Institute of Health, Bethesday, MD, USA). Protein levels were first normalized to a loading control and any concomitant phospho-protein levels were normalized to their non-phosphorylated counterparts.

Control (Ctrl; GM06889) and A–T (ataxia–telangectasia) (GM01526) human lymphoblastoid cell lines (LCL) were obtained from the Coriell Institute for Medical Research (Camden, NJ, USA). LCLs were maintained in the RPMI media supplemented with 15% fetal bovine serum (v/v), 1× glutamax, 100 U/mL penicillin, 100 μg/mL streptomycin. Cells underwent 8 Gy irradiation using an RS-2000 (Rad Source Technologies, Inc., Buford, GA, USA), followed by 60 min incubation (37 °C) to mediate cellular DNA damage-mediated signaling/phosphorylation events. Protein extracts were prepared, run, and analyzed as above.

2.5. Alkali Comet Assay

Quiescent (non-proliferating) primary astrocytes and MEFs were treated with γ–irradiation (IR, 20 Gy). Following IR treatments, cells were incubated for various times in a drug-free medium at 37 °C (recovery). Cells were trypsinized and suspended in a pre-chilled 1× PBS mixed with an equal volume of 1.2% UltraPure low-melting point agarose (Invitrogen) maintained at 42 °C and immediately layered onto frosted glass slides (Fisher) pre-coated with 0.6% agarose and maintained in the dark at 4 °C for all further steps. Slides were immersed in a pre-chilled lysis buffer (2.5 M NaCl, 10 mM Tris–HCl, 100 mM Ethylenediaminetetraacetic acid (EDTA) (pH 8.0), 1% (v/v) Triton X-100, 3% (v/v) DMSO, pH10) for 1 h, washed with pre-chilled distilled water (twice for 10 min each), and placed into a pre-chilled alkaline electrophoresis buffer (50 mM NaOH, 1 mM EDTA, 1% DMSO) for 45 min. Electrophoresis was carried out at 95 mA (astrocytes) or 190 mA (MEFs) for 25 min, followed by neutralization in 0.4 M Tris–HCl (pH 7.0). Comets were stained with SYBR Green (1:10,000 in 1× PBS, Sigma) for 10 min. A minimum of 100 comet tail moments were measured using the Comet Assay IV system (Perceptive Instruments) coupled to an Axioskop2 plus microscope (Zeiss) at 200× magnification as previously described [38,41]. All data reflect experiments performed in triplicate (n = 3, N ≥ 300).

Human glioblastoma cell lines, U87MG, U251MG, and U373MG, were transduced with shRNAs to silence RAD51 gene expression (or with Scm as control) and underwent treatment with the radiomimetic Neocarzinostatin (NCS; Sigma cat# N9162-100UG, St. Louis, MO, USA) at 340 ng/mL for 1 h followed by alkaline comet analysis. Briefly, cells were resuspended in 1% UltraPure low melting point agarose and plated on a 96-well CometSlide (Trevigen; cat# 4253-096-03) and allowed to dry. Slides were incubated in a lysis buffer for 1 h at 4 °C and then placed into an alkaline unwinding buffer (200 mM NaOH, 1 mM EDTA; pH > 13) protected from the light at room-temperature (RT) for 20 min. Slides were then placed into a cold electrophoresis buffer (50 mM NaOH, 1 mM EDTA, 1% DMSO) within an electrophoresis manifold (Trevigen; #4250-050-ES), followed by electrophoresis at 21 V for 40 min. Slides underwent two successive washes with double-distilled water, followed by a 70% ethanol wash and air drying. DNA was stained with SYBR Green (Sigma-Aldrich; cat# S9430; 1:10000 in 1× PBS) for 10 min. A minimum of 300 comet tail moments per experiment per cell line/treatment were measured using a Cytation 5 (Biotek Instruments, Winooski, VT, USA) multi-mode reader with a 2.5× objective capturing a fixed area encompassing ≥ 100 cells per well, with 3 wells being measured for each treatment group in each experimental set (N ≥ 300). Gen5 analysis software (Biotek Instruments, Winooski, VT, USA) was used to determine the average length of a comet tail moment. For U87 and U373, n = 5 and N ≥ 1500; for U251, n = 4 and N ≥ 1200.

2.6. Quantification of DNA Damage-Induced Nuclear Foci

For the fluorescent labeling of quiescent primary murine cortical astrocytes and MEFs, cells were grown to confluence on glass coverslips, underwent irradiation (+/− recovery at 37 °C) and then fixed with 4% paraformaldehyde (PFA) in PBS for 10 min, followed by three successive washes with 1× PBS, and then permeabilized for 5 min in 0.5% Triton X-100/PBS. Following three successive washes with 1× PBS, cells were immunostained with antibodies diluted in 3% bovine serum albumin (BSA) + 1× PBS; anti-γH2AX (Ser139, mouse, 1:500; Millipore, cat# JBW301) and anti-RAD51 (rabbit, 1:500; Calbiochem), followed by Alexa Fluor 488/555-conjugated secondary antibodies (1:500, Life Technologies, cat# A-21206 and A-31570, respectively), and counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Experiments were performed in triplicate (n = 3) with 30 cells measured per experiment (N = 90). Xrcc2/p53-double-deficient primary MEFs served as an HRD control and were previously described [44].

GBM cells were seeded at a density of 8000 cells per well in a 96-well black clear flat-bottomed plate (Corning, cat# 07-200-588). The cells were treated with neocarzinostatin (NCS) (140 ng/mL) or media as the vehicle control for 1 h. Cells were then fixed with 4% PFA (in 1× PBS) for 10 min at room temperature followed by a washing step with PBS (2×) and permeabilization with 0.5% Triton X-100 in the PBS for 5 min at room temperature. Following an additional washing step with the PBS (2×), primary antibodies against phospho-53BP1 (S1778) (rabbit, Cell Signaling Technology, cat# 2675S, 1:1000) and RAD51 (mouse, Abcam, cat# ab213, 1:500) were prepared in a 3% BSA (in 1× PBS) and added to the cells at a volume of 50 μL/well for overnight incubation at 4 °C in the dark. Cells were then washed with 1× PBS (2×) and immunostained with Alexa Fluor-488 donkey anti-rabbit and Alexa Fluor-555 donkey anti-mouse dyes (1:500, Life Technologies, cat# A-21206 and A-31570, respectively) prepared in 3% BSA (1× PBS) at a volume of 50 μL/well for 2 h at room temperature in the dark. The cells were then washed with the PBS (1×), fixed with 1% PFA (in 1× PBS) for 10 min, and permeabilized with 0.5% Triton X-100 for 5 min at room temperature, taking care not to expose the assay plate to light. Following another wash step with 1× PBS, the fixed cells were immunostained with an anti-phospho-H2AX (Ser139) antibody pre-conjugated to Alexa Fluor-647 (Biolegend; cat# 93144; 1:500 dilution) for 1 h at room temperature. The cells were washed in 1× PBS (2×) then counterstained with Hoechst 33342 (0.2 μg/mL in the PBS; Life Technologies, cat# H1399) to visualize cell nuclei. Image acquisition and analysis were carried out using a Cytation 5 cell imaging multi-modal reader and the Gen5™ Imaging Software (Biotek Instruments). Experiments were performed in triplicate (n = 3) with 100 cells measured per experiment (N = 300).

2.7. Cell Viability Assays

Cell viability was determined by counting the cells stained with the Hoechst 33342 nucleic acid stain solution at 24 h increments over a 96 h period. Briefly, glioblastoma cell lines U87MG, U251MG, and U373MG transfected with shRNAs to RAD51 or Scm control, were seeded at a density of 1000 cells/well in a 96-well plate. Wells were treated with either DMSO (control) or the PARPi olaparib (AZD2281, Ku-0059436, Apexbio, cat# A4154) at a concentration of 1 μM and the cells were incubated at 37 °C (with 5% CO₂) for 96 h. At each 24 h interval, Hoechst 33342 (Life Technologies, cat# H1399) at a concentration of 0.2 μg/mL was applied to a subset of wells and the subsequently-labeled cell nuclei were counted and analyzed via Cytation 5 and Gen5. Experimental assays were repeated in triplicate with 3 wells per cell treatment/time-point per experiment (n = 3, N = 9). Cell growth rate was determined relative to the baseline cell count performed upon cell seeding (Day 1). Statistical analysis was performed using the 2-way ANOVA using the Tukey’s post hoc test with no correction.

2.8. Gene Expression Analyses According to the PTEN Status (TCGA and SKS Datasets)

Mutational (whole exome sequencing) and transcriptomic (mRNA expression, RSEM values) data of 145 glioblastoma tissue specimens were gathered from The Cancer Genome Atlas (TCGA) PanCancer Atlas study [45] using cBioPortal [46,47]. Samples with one or more amino acid changes (missense or frameshift mutant) in the PTEN coding sequence were categorized as mutant/deletion (mut/del), whereas all remaining samples were considered wild-type (WT). Transcriptomic RSEM values for genes of interest (PTEN, RAD51, RAD51B, RAD51C, RAD51D, XRCC2, XRCC3 and DMC1) were matched with the PTEN status and compared between WT and mut/del.

Patient-derived primary glioblastoma specimens were obtained, processed, cultured, and enriched for glioblastoma brain tumor initiating cells (BTICs) as previously described [48]. RNA was extracted from each BTIC line using the RNeasy Kit (Qiagen) or Total RNA Purification Kit (Norgen). Samples were submitted for mRNA-Seq (SKS dataset) at the Donnelly Sequencing Centre at the University of Toronto (http://ccbr.utoronto.ca/donnelly-sequencing-centre) or at MedGenome Inc. (https://diagnostics.medgenome.com/research-services). Further details about sequencing and data analysis can be requested from Dr. Sheila K. Singh ([email protected]). Variant calls for PTEN were obtained using the Genome Analysis Toolkit (GATK) best practices pipeline [49,50]. Samples with one or more amino acid changes (missense or frameshift mutant) in the PTEN coding sequence were categorized as mutant/deletion (mut/del), whereas all remaining samples were considered wild-type (WT). Normalized transcript read count values for genes of interest (PTEN, RAD51, RAD51B, RAD51C, RAD51D, XRCC2, XRCC3 and DMC1) were matched with the PTEN status and compared between WT and mut/del.

3. Results

3.1. RAD51 Expression and HR Activity Are Not Impaired in PTEN-Deficient Primary Cells

To investigate the link between PTEN and RAD51, multiple approaches were undertaken. Primary Pten-null (and corresponding wild-type) mouse embryonic fibroblasts (MEFs) and astrocytes were established (Figure 1A). Western analysis confirmed the Pten expression status, which was consistent with corresponding Pten genotypes. Subsequent immunostaining revealed Rad51 expression (Figure 1B) regardless of the Pten genotype, indicating that loss of Pten does not affect Rad51 expression. To determine if a functional relationship exists between the PTEN expression and DNA repair activity, cells were irradiated and analyzed via the alkaline comet analysis. In both primary fibroblasts and astrocytes, cells incurred similar levels of DNA damage following irradiation, irrespective of Pten expression (Figure 2). One hour following damage, both the Pten^-/- and Pten^+/+ (wild-type) cell populations show complete DNA repair, indicating that DNA repair activity is equivalent in wild-type and Pten-null cells. In comparison, comet analysis of Xrcc2-null cells, which normally express the Rad51 protein [51], display higher comet tail moment values than wild-type or Pten^-/- cells one hour following irradiation (Figure 2A,B), confirming the deficiency in homologous recombination (HR) [44]. Complementary DNA repair assays examining γH2AX and RAD51 foci, indicators of DNA double-stranded breaks and HR activity, respectively, yielded similar results. In both fibroblasts and astrocytes, wild-type and Pten^-/- cells accumulated equivalent numbers of γH2AX foci following irradiation (Figure 2C,D). Moreover, all cells displayed pervasive RAD51 immunostaining accompanied by formation of RAD51 immunopositive foci, indicating active HR activity. In contrast, Xrcc2-null cells are RAD51-immunopositive but fail to form RAD51 foci in spite of the presence of endogenous and radiation-induced γH2AX foci. Combined, these DNA damage repair data indicate that PTEN-deficiency does not impact RAD51 expression or disrupt RAD51/HR-dependent DNA damage repair.

3.2. PTEN-Deficient Cancer Lines Express RAD51 and Exhibit Robust DNA Repair Activity

Although PTEN loss is a frequent genetic driver during tumorigenesis, a vast majority of tumors may accumulate mutation(s) to one or more DNA repair genes resulting in aberrant DNA repair activity [52,53]. We examined whether PTEN-deficient human GBM cell lines U87MG, U251MG, and U373MG display dysfunctional DNA double-strand break repair, particularly, RAD51-dependent HR activity. These cells harbor previously described PTEN mutations resulting in either exon deletion (U87MG: PTEN-mut) or premature termination (U251MG and U373MG: PTEN-null), thus inactivating PTEN function [54]. To quantify relative RAD51 expression and HR/DNA repair activity in these PTEN-deficient lines, GBM cells underwent lentiviral transduction and puromycin selection in order to generate stable cell lines expressing shRNAs targeting RAD51 or scrambled control (SCM) (Figure S1 and Figure 3A). While RAD51 protein is highly expressed in all three PTEN-deficient shSCM-GBM lines (control), ≥83% of the RAD51 protein is reduced in shRAD51-GBM (RAD51-deficient) lines. Likewise, DNA damage analysis of all stable shRAD51-expressing GBM lines demonstrate up to a 4-fold increase in endogenously occurring γH2AX foci compared to shSCM-GBM control lines (Figure 3B). Furthermore, induction of DNA double-stranded breaks with the NCS treatment (140 ng/mL for 1 h at 37 °C) results in a ~2 fold increase in γH2AX and 53BP1 foci in shRAD51-expressing cells compared to the shSCM-GBM counterparts (Figure 3C,D). However, RAD51 foci are only present in shSCM-GBM cells and are concomitantly increased with the NCS treatment but are absent in shRAD51-GBM cells regardless of DNA damage levels (Figure 3C,D). Similarly, the alkali comet analysis shows a significant accumulation of DNA damage in all NCS-treated shRAD51-GBM (RAD51-deficient) lines compared to NCS-treated shSCM-GBM control lines (p ≤ 0.05) (Figure 3E). Analysis of an unrelated malignant PTEN-deficient human B-cell line, BJAB, shows robust RAD51 expression and intact ataxia telangiectasia mutated (ATM)- and KRAB-associated protein 1 (KAP1)-dependent activation of DNA damage responses (Figure S2) [55]. Combined, these data indicate that PTEN-deficient GBM cells retain both RAD51/HR- and 53BP1/NHEJ-dependent DNA DSBR activities, while induced HRD results in enhanced accumulation of DNA damage and concomitant compensatory NHEJ activity.

Recent studies suggest that PTEN-deficient tumors are responsive to PARPi due to purported RAD51/HR loss similar to PARPi-induced synthetic lethality in tumors that feature BRCA/HR-deficiency [32]. Although we find that RAD51 expression remains intact in PTEN-deficient cells, we examined whether PTEN-deficiency could result in enhanced DNA damage and sensitivity to PARPi. PTEN-deficient GBM cells underwent treatment with olaparib (AZD-2281, Ku-0059436), a clinically-relevant PARPi regularly used in the treatment of BRCA-deficient breast, ovarian, and endometrial cancer [27]. Despite effective inhibition of Poly ADP-Ribosylation (PARylation) (Figure 4A), olaparib treatment failed to induce a discernable DNA damage response in GBM cells, whereas mono- or co-treatment with topotecan (TPT), a DNA strand break-inducing Topoisomerase-1 inhibitor, stimulated ATM-dependent KAP-1 phosphorylation, indicating DNA damage signaling activation (Figure 4A) [55].

All three GBM cell lines underwent cellular viability analysis to determine whether PTEN-deficient cancer cells have enhanced sensitivity to PARPi (Figure 4B,C). Olaparib treatment of both shSCM- and shRAD51-GBM lines was used to compare the specific requirement of RAD51 expression in PARPi-mediated growth suppression. The three GBM cell lines showed varying sensitivity to olaparib, which was not related to endogenous RAD51 levels. Interestingly, expression of shRAD51 and the resultant loss of the RAD51 protein (Figure S1) reduces cellular growth independently of PARPi treatment (Figure 4B,C). However, in U373 cells where RAD51 depletion causes a modest decrease in growth, olaparib co-treatment completely ablates cell growth in the other shRAD51 lines, while only moderately suppressing cell growth in all shSCM control lines (Figure 4C). The apparent difference in response with the shRAD51-U373 line compared to the shRAD51-U87 and shRAD51-U251 lines is likely due to the previously described reduced proliferative rate and colony-forming capacity of U373 compared to U251 and U87 [56]. As HR is primarily associated with proliferation, the impact of shRAD51-induced HRD in U373 would be expected to be muted compared to those of U87 and U251. Combined, these data demonstrate that PTEN deficiency has no role in mediating RAD51-dependent sensitivity to olaparib or in disrupting HR-dependent DNA damage signaling and repair responses which would otherwise synergize with PARPi.

To determine whether expression of RAD51 or its paralogs is impacted by PTEN-dysfunction in primary tumors, we performed an RNA expression analysis amongst PTEN-mutant/deleted tumors derived from glioblastoma patients. Analysis of The Cancer Genome Atlas (TCGA) dataset (Figure 5A) of glioblastoma tissue specimens and the SKS dataset (Figure 5B) of patient-derived glioblastoma brain tumor initiating cells (BTICs) revealed no notable differential expression of RAD51 or its paralogs in PTEN-mutated or wild-type samples. A minor decrease in RAD51 expression was observed in PTEN-mutant samples from the TCGA dataset as compared to the wild type (p = 0.012), although this difference was nominal (<7%). Furthermore, we performed protein expression and DNA damage response signaling analyses on 2 primary (treatment-naïve) and 2 recurring (treatment-resistant) GBM patient-derived BTIC lines (Figure 5C). These cells recapitulate the key morphological, architectural, and expression features that are present in the initial clinically presented GBM tumor while retaining their self-renewal potential and capability to form tumors in vivo [57]. Each patient-derived BTIC line contains an independent PTEN mutation predicted to abrogate the PTEN protein function and/or stability [49,50,58]. In all cases, RAD51 expression was apparent and topotecan-induced DNA damage resulted in robust ATM-dependent activation of KAP1 (S824) and phosphorylation of 53BP1 (S1778). Combined, these patient tumor data indicate that primary GBM tumors do not incur PTEN-dependent expression changes neither in RAD51 nor its paralogs and retain functional DNA damage signaling, including HR and NHEJ activity.

4. Discussion

“Synthetic lethality” (and the related “synthetic sickness”) in anti-cancer therapy arose from success in killing cancer cells by specifically targeting compensatory mechanisms, pathways, or activities that usually allow for tumor cell survival and growth despite the loss of the functionally important parallel cell growth/survival pathways [59,60]. Perhaps the best known example of such phenomenon involves mutations and/or low expression of BRCA1/2, whereby HR deficiency sensitizes tumor cells to inhibit PARP1, a critical mediator of the corresponding DNA single-strand break repair pathway [17,61,62,63]. This finding enabled the use of the first clinically relevant chemical inhibitor of PARP, olaparib, which spares normal HR-proficient cells while specifically targeting HRD tumor cells to undergo enhanced DNA damage-induced cell death [64,65]. As a corollary, a functional loss of other proteins involved in HR fidelity and DNA DSBR is thought to confer tumor cells a “BRCAness” phenotype [66,67,68,69,70]. Clinical olaparib use in BRCA-deficient breast, ovarian, and endometrial cancers is well-documented and has spurred recent trials in prostate, lung, and colon cancers that feature HRD [27,37,69]. Furthermore, this clinical success is the basis of numerous biomedical studies that seek to link other known tumorigenic mutations to the DNA repair function, either directly or indirectly, to enable wider therapeutic utilization of current and next generation PARPis [23,71]. Amongst the more controversial is the finding that PTEN associates with E2F-1 to regulate the RAD51 transcription, a key component of the tripartite BRCA1/2 HR DNA repair complex [30]. Subsequent studies suggested that PTEN-deficient tumor lines acquire HRD and in turn a BRCA-like phenotype, thus sensitizing these cells to olaparib [32,33,72,73]. Considering that PTEN is one of the most frequently mutated genes in cancer [3,4], with 40–80% of most human tumors acquiring somatic PTEN mutation, deletion, and/or loss of heterozygosity [74,75], use of PARPi to treat these tumors would represent a profound therapeutic advancement. However, more recent studies, including our findings, demonstrate the opposite; that there is no functional link between PTEN and RAD51 co-expression nor any basis for the use of PARPi to ablate PTEN-deficient tumors as PTEN loss does not uniformly predict response to olaparib. In our study, we employ multiple types of PTEN-deficient cell models, both primary and tumorigenic, to demonstrate that: (i) expression of PTEN has no bearing on the expression of RAD51 or its paralogs, (ii) PTEN loss does not impart cellular HRD, (iii) PTEN expression has no effect on cellular DNA strand break repair rates, and (iv) PTEN loss does not synergize with PARPi. Consistent with our findings, studies have demonstrated that RAD51 expression and foci formation remain unchanged in other PTEN-deficient tumor cell lines (isogenic HCT116/PTEN^−/−, PC-3, and U251MG), primary astrocytes, RNAi studies (shPTEN-expressing H1299), comparative prostatic tumor xenograft analyses (PTEN^+/+ 22RV1; PTEN^+/− DU145; and PTEN^−/− PC3), and human tumor tissue microarrays [29,34,35,76]. Non-malignant primary Pten wild-type, heterozygote, and knockout astrocytes and MEF cells all display robust RAD51 expression and HR-dependent RAD51 foci formation in response to DNA damage regardless of the Pten status. Similarly, we find that PTEN-deficient glioblastoma lines show pronounced RAD51 expression and form HR-responsive RAD51-positive and NHEJ-responsive 53BP1-positive DNA damage-induced foci. Consistent with our data, Miyasaka and associates found that RAD51 levels were expressed broadly and IR-induced γH2AX and RAD51 foci formation was uniform amongst a set of 16 PTEN-proficient and deficient endometrial cell lines [77].

It has been suggested that the apparent discordance in co-expression amongst these findings and other reports may reflect cell type-specific differences involving PTEN/RAD51 [35]. Our study interrogates this assertion by comparing primary versus malignant and neural versus non-neural cell types with the identical conclusion: that RAD51 expression is independent from the PTEN status irrespective of the cell type. Furthermore, RAD51 expression is apparent in replicating (tumor lines) or quiescent (contact-inhibited primary astrocytes/MEFs) PTEN-deficient lines. McEllin et al. suggested that PTEN-associated HRD is not mediated by the loss of RAD51 expression, but rather via reduced expression of the RAD51 paralogs RAD51B, RAD51C, and RAD51D [29]. Although these analyses were strictly performed in primary murine Pten-null astrocytes, our RNA expression analysis in human GBM patient-derived primary tumors and outgrowing BTICs clearly demonstrates that the expression of RAD51 and its paralogs remain unchanged regardless of PTEN status.

Although two studies show that PTEN-deficient endometrial tumor cells are not sensitive to PARPi monotherapy [37,77], we observe that PTEN-deficient GBM cell lines are somewhat attenuated with olaparib treatment; observations more consistent with the studies performed on PTEN-deficient prostate cancer cell lines [35]. Furthermore, we show that cell sensitization to olaparib appears to be replication-dependent, as we find that non-replicating cells do not show olaparib sensitivity (data not shown), similar to Gupta et al. [36]. Importantly, only upon shRAD51 co-expression does olaparib treatment enhance DNA damage and completely ablate PTEN-deficient tumor cell survival. Similarly, Turchick et al. [78] showed that PTEN-deficient tumor cells could only be sensitized upon internalization of a cell-penetrating anti-RAD51 antibody. Combined, these data underscore the functional nature of RAD51-mediated HR activity in PTEN-deficient cells, which mediate cell survival upon inhibition of cellular PARylation.

Our study reinforces the apparently autonomous nature of RAD51 expression and functional HR activity within PTEN-deficient cells, which clearly diverges with the previous studies that originally suggested a PTEN/RAD51 relationship [30,32]. One reason may be the differing origins of Pten-deficient mouse lines [39,79,80] and methods used to develop Pten-null MEFs [40,41,79]. However, our data demonstrating that RAD51 expression and HR activity are similar amongst both primary Pten^-/- MEFs and astrocytes suggest that this divergence is likely due to the inherent differences amongst the two genetic Pten mouse models. It is noteworthy that cre recombinase expression in our conditional Pten (Pten^LoxP/LoxP) astrocyte and MEF cell lines results in deletion of exons 4 and 5, while other murine Pten-Rad51 analyses utilized a conditional Pten mouse line with loxP sites flanking only exon 5. Furthermore, McEllin and associates utilized a compound Pten-Ink4a/Arf double knockout line [29], which complicates these conclusions as this same group previously published that de novo DNA damage induced in Ink4a/Arf single knock-out mice promotes DNA double-strand break-mediated high-grade gliomas [81], which may suggest synergism between PTEN and INK4a/ARF in regulating RAD51 paralog-mediated expression and DNA repair. Future genetic and biochemical comparison of these Pten lines may be required to resolve these issues.

The conflicting PTEN-deficient tumor cell data in these studies may be due to the inherently complex genomic and biochemical landscape of tumors, which can accumulate multiple genetic lesions (in addition to PTEN loss), and heterogeneity, but also multiple genetic vulnerabilities, pleotropic roles, and microenvironmental changes throughout tumorigenesis, anti-tumor therapy, and recurrence [75,82]. The PTEN-deficient tumor cells originally used in the findings described by Mendes-Pereira et al. (PTEN^-/- HCT116: MLH1-deficiency) [32] and Shen et al. (PTEN-deficient PC3-AR: PARP1^S383Y) [30] have since been shown to contain additional genetic lesions resulting in RAD51-independent HRD and enhanced cell sensitivity [83,84,85]. Furthermore, veliparib treatment (another type of PARPi) of PTEN-null PC3-AR prostate cancer mouse xenografts has been shown to enhance RAD51 expression and HR activity under hypoxic conditions and reduce chemotherapy-mediated DNA damage and cell apoptosis [86]. However, further confounding this issue are the recent findings of Ma et al. who showed that in a singular PTEN-deficient cell line (U87MG), RAD51 is functionally expressed, yet these cells display pronounced DNA repair deficiency following radiation-induced DNA damage [87]. Further, they present data showing that radiation therapy can induce fibroblast growth factor receptor (FGFR) family-mediated PTEN^Tyr-240 phosphorylation and co-recruitment of Ki-67 and RAD51 to chromatin in PTEN-proficient GBM, thus enhancing DSBR activity and radioresistance in GBM patient-derived stem-like cells (GSCs). Although conflicting, their study suggests that contextual modifications of the PTEN protein in tumors may regulate RAD51-mediated HR activity, possibly via FGFR-TACC fusions that accumulate in subsets of PTEN-positive/WT tumors [88]. Our study is unique and definitive as we employ a multitude of PTEN-deficient normal and tumorigenic models to unequivocally demonstrate that RAD51 expression and HR DNA repair activity remain unaffected by PTEN dysfunction. These include PTEN-null/mutated primary murine cells (MEFs and astrocytes), human tumor cell lines (U87, U251, U373, and BJAB), and numerous patient-derived glioblastoma tumor specimens and BTICs.

Our findings are further reinforced by the striking lack of clinical evidence supporting therapeutic use of PARPis in the treatment of PTEN-deficient tumors despite the wide breadth of clinical trials [27,28]. In a phase 1 dose escalation trial examining BRCA mutations and sporadic prostate cancer patients treated with a next generation PARPi niraparib, stabilization of the disease occurred in 43% of patients; however, no correlation was observed between PTEN loss and anti-tumor activity [89]. Only a single case report has been published showing a complete remission of a PTEN-null sarcomatoid prostate cancer following olaparib treatment [90]. However, RB loss, TMPRSS2-ERG rearrangement, and TP53 alterations were also noted in this patient’s tumor. A more detailed delineation of additional genetic vulnerabilities amongst individual PARPi-responsive PTEN-deficient tumors may reveal additional mechanistic interactions with the PARP-mediated DNA repair [82], thus invoking additional targets for clinical use in combination therapy [91]. In fact, PTEN-deficient endometrial tumor cells are shown to only be sensitive to the PARPi-PI3Ki combination therapy, but not to the PARPi monotherapy [37], while co-inhibition of both RAD51 and ATR is required to synergize ablation of PTEN-deficient melanoma and glioma cells [78], further underscoring the independence of RAD51/HR from the PTEN status. In the context of our findings and the apparent lack of clinical evidence, future clinical use of PARPi in the treatment of tumors should be discontinued absent a clear indication of HRD independently of PTEN mutational status.

5. Conclusions

The identification of HRD mutations in tumors has heralded widespread use of PARPis in a targeted therapeutic approach. Recent findings indicating that PTEN-deficient tumors suppress RAD51 expression and HR activity have been met with controversy, as recent studies have failed to recapitulate these results. Resolving this conflict is paramount due to the ensuing impacts to clinical care and patient trials. Our study demonstrates that RAD51 expression, HR DNA repair activity, and RAD51 foci formation remain intact and unaffected in PTEN-deficient normal (both neural and non-neural) and tumorigenic cells. We further show that PTEN loss does not synergize with PARPi. Our findings underscore the functional independence of the PTEN mutational status in RAD51-mediated HR and advocate discontinuance of PARPi treatments in PTEN-deficient tumors that do not feature HRD.