|Home | About | Journals | Submit | Contact Us | Français|
Ionizing radiation is toxic to ovarian follicles and can cause infertility. Generation of reactive oxygen species (ROS) has been implicated in the toxicity of ionizing radiation in several cell types. We have shown that depletion of the antioxidant glutathione (GSH) sensitizes follicles and granulosa cells to toxicant-induced apoptosis and that supplementation of GSH is protective. The rate-limiting reaction in GSH biosynthesis is catalysed by glutamate–cysteine ligase (GCL), which consists of a catalytic subunit (GCLC) and a regulatory subunit (GCLM). We hypothesized that overexpression of Gclc or Gclm to increase GSH synthesis would protect granulosa cells against oxidant- and radiation-induced cell death. The COV434 line of human granulosa tumour cells was stably transfected with vectors designed for the constitutive expression of Gclc, Gclm, both Gclc and Gclm or empty vector. GCL protein and enzymatic activity and total GSH levels were significantly increased in the GCL subunit-transfected cells. GCL-transfected cells were resistant to cell killing by treatment with hydrogen peroxide compared to control cells. Cell viability declined less in all the GCL subunit-transfected cell lines 1–8 h after 0.5 mM hydrogen peroxide treatment than in control cells. We next examined the effects of GCL overexpression on responses to ionizing radiation. ROS were measured using a redox-sensitive fluorogenic dye in cells irradiated with 0, 1 or 5 Gy of γ-rays. There was a dose-dependent increase in ROS within 30 min in all cell lines, an effect that was significantly attenuated in Gcl-transfected cells. Apoptosis, assessed by terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick-end labelling and activated caspase-3 immunoblotting, was significantly decreased in irradiated Gclc-transfected cells compared to irradiated control cells. Suppression of GSH synthesis in Gclc-transfected cells reversed resistance to radiation. These findings show that overexpression of GCL in granulosa cells can augment GSH synthesis and ameliorate various sequelae associated with exposure to oxidative stress and irradiation.
Radiotherapy is based on the concept that exposure to sufficient doses of ionizing radiation will destroy target cells. Inherent to these treatments is the fact that non-target cells (i.e. normal tissue) will also be destroyed, and under certain clinical situations, damage to the limited pool of ovarian follicles may be unavoidable. Significant exposure to the ovaries occurs during radiotherapy of abdominal and pelvic cancers or when the pelvic lymph nodes are irradiated for haematological diseases such as Hodgkin’s lymphoma (1). The ovaries also incur significant exposure when total body irradiation is used to treat various forms of leukaemia or when it is used to condition patients prior to bone marrow transplants (1). Together, chemotherapy and radiotherapy have dramatically increased the long-term survival of young cancer patients, but there are major side effects that can be directly linked to the damage of normal tissue surrounding the treatment area that adversely impact the quality of life of the patient (1,2). These include temporary cessation of ovarian function and premature ovarian failure (1,3–7).
It is generally accepted that mammalian females are born with a finite stock of ovarian follicles and that the survival and maturation of these follicles are essential for the preservation of fertility (1,8,9). Follicles consist of the oocyte surrounded by layers of somatic cells. The granulosa cell layers immediately surround the oocyte, and granulosa cells form direct intercellular connections with the oocyte. Surrounding the granulosa cells and separated from them by a basement membrane are the theca cells. Ovarian failure and permanent sterility are a direct result of complete follicular depletion, which occurs through the normal, age-associated processes of ovulation and follicular atresia and through exogenously induced follicular apoptosis. Atresia is a controlled apoptotic process of follicular degeneration (8–11). Ionizing radiation destroys the least mature primordial follicles as well as the maturing antral and secondary follicles (1–3,12). Destruction of maturing follicles appears to be driven by induction of apoptosis of the granulosa cells, which secondarily leads to death of the oocyte (11). Premature ovarian failure can occur as a result of the depletion of primordial follicle reserves by ionizing radiation (1–3,12). Temporary loss of fertility can occur when more mature follicles are destroyed, but some primordial follicles remain. The nearly complete destruction of primordial follicles in young adult mouse ovaries occurs 2 weeks after a single exposure to 0.1 Gy of ionizing radiation (13). Mathematical models have predicted that the estimated dose of abdominal, pelvic or total body irradiation required to destroy half of all the primordial follicles in young humans is <2 Gy (5,6). The extent of ovarian damage caused by radiotherapy is dependent upon three critical factors: the dose, the irradiation field and the patient’s age, with older women having a greater risk of becoming infertile (1,2).
The molecular mechanisms underlying follicle destruction by radiation exposure are not completely understood. About one-third of DNA damage from irradiation is caused by direct ionizations to the DNA target molecule, while the remaining two-thirds of the net DNA damage are due to the reactive species derived from the indirect ionization of water (14–16). The most reactive of these species is the hydroxyl radical, which generates clustered DNA damage when energy deposition events yielding several ion pairs are in close proximity to the DNA helix. Hydroxyl radicals and other reactive oxygen species (ROS) have the potential to damage critical cellular components. These include protein oxidation, lipid peroxidation and DNA damage including DNA–protein cross-linking and double-stranded DNA breaks (17–19). Despite the transience of radiation-induced free radicals, irradiation has been shown to elicit more persistent increases in oxidative stress. Increased ROS persisting hours, days and even weeks post-irradiation likely reflect perturbations to the redox homoeostasis of cells (20) and have been linked to functional changes at the mitochondrial level as well as other ROS generators such as the nicotinamide adenine dinucleotide phosphate (NADPH) oxidases (21). Radiation-induced oxidative stress can lead to elevated superoxide anion radical (SO) and hydrogen peroxide (H2O2), whose relatively longer half-lives permit more extensive diffusion throughout the cell, thereby amplifying the impact of past irradiation events (16). Therefore, scavenging free radicals through the use of radioprotectors (agents present during irradiation) and/or mitigators (agents present after irradiation) have the potential to reduce radiation- and metabolic-derived reactive species and provide a tractable strategy for attenuating radiation-induced cytotoxicity.
Glutathione (GSH) is the most abundant low-molecular weight, non-protein thiol in eukaryotic cells. It is maintained at millimolar concentrations in most cells (22,23). Despite its multiple and diverse functions, GSH is most often recognized as one of the cell’s primary defences against exogenous xenobiotics and oxidative stress (22–24). GSH detoxifies ROS by spontaneous reduction, as in the case of hydroxyl radical, or by GSH peroxidase-mediated reduction, as in the case of H2O2 (22–24). GSH is synthesized by two sequential adenosine triphosphate (ATP)-dependent enzymatic reactions. The first and rate-limiting step is catalysed by glutamate–cysteine ligase (GCL), yielding γ-glutamylcysteine. The final step is catalysed by GSH synthetase and adds a glycine residue to yield γ-glutamylcysteinylglycine or GSH (22,23). GCL is a heterodimer composed of a catalytically active subunit (GCLC) that contains all substrate-binding sites and a regulatory subunit (GCLM) that modulates GCLC affinity for substrates and inhibitors (22,23,25).
Several previous studies provide support for the idea that GSH may be involved in radioprotection. Chemo- and radio-resistant cells have increased GSH levels (26–29). Depletion of GSH with the GCL-specific inhibitor L-buthionine sulfoximine (BSO) caused significant radiosensitization of A549 human lung cells (30,31). N-acetyl-cysteine (NAC), a GSH precursor, has been used as a radioprotective agent to attenuate acute liver injury, characterized by increased lipid peroxidation, GSH depletion, DNA single-strand breaks and hepatocyte apoptosis caused by X-rays in mice (32) and by total body γ-radiation in rats (33). These findings suggest that augmenting intracellular GSH levels can confer radiotolerance. Consequently, we set forth to genetically manipulate GSH concentrations of cells derived from a human granulosa cell tumour by overexpressing one or both subunits of GCL, the rate-limiting enzyme in GSH biosynthesis. We challenged these cells with various doses of pro-oxidants and γ-radiation to test the hypothesis that overexpression of GCL protects granulosa cells against oxidative and radiation-induced cell death.
All chemicals were purchased from Sigma–Aldrich (St Louis, MO) or Fisher Scientific (Pittsburgh, PA), unless otherwise noted. Tissue culture reagents including all culture media, foetal bovine serum (FBS), 0.25% trypsin–ethylenediaminetetraacetic acid (EDTA), trypan blue, Lipofectamine 2000 Transfection Reagent and Geneticin (G418) were purchased from Invitrogen (Carlsbad, CA). All tissue culture vessels were obtained from Corning Costar (Lowell, MA).
The COV434 line of immortalized granulosa cells was derived from a primary human granulosa cell tumour and was kindly provided by Dr Peter I. Schrier (Department of Clinical Oncology, University Hospital, Leiden, The Netherlands) (34). The COV434 cells are diploid and, like normal granulosa cells, grow in small follicle-like aggregates, form intercellular junctions with neighbouring cells, form intercellular gap junctions with native cumulus cells when co-cultured with a cumulus–oocyte complex and increase cyclic adenosine 3′,5′-monophosphate and 17β-oestradiol in response to follicle-stimulating hormone and androstenedione co-stimulation (34–36).
COV434 cells were maintained in Dulbecco’s modified Eagle’s medium–F12 (DMEM–F12) with GlutaMAX™ supplemented with 9% FBS, henceforth referred to as complete media, at 37°C in a humidified atmosphere of 95% air and 5% CO2. Cells from passages 41–77 were used for the experiments reported herein. Cells were routinely harvested by trypsinization with 0.25% trypsin–EDTA, counted using a haemacytometer and 0.4% trypan blue and plated in complete media or complete media supplemented with 1 mg/ml G418. The doubling time of these cells is ~24 h. Doubling times of the stably transfected cell lines were also ~24 h.
A full-length mouse Gclc cDNA cloned in the vector pCR2.1 (GCLc-pCR2.1) was a gift of Dr Julie K. Andersen (Buck Institute for Age Research, Novato, CA) (37). The Gclc cDNA was excised out of GCLc-pCR2.1 with the restriction endonucleases, BamHI and SalI, following the manufacturer’s recommendations (New England BioLabs, Ipswich, MA). The pCMV-Script vector (Stratagene, La Jolla, CA) was cleaved using the same restriction endonucleases. Digested DNA fragments were isolated using the QIAquick PCR Purification Kit (Qiagen, Valencia, CA); purified fragments were electrophoresed through 2% agarose gels and isolated using the DNA Gel Extraction Kit as per the manufacturer’s recommendations (Millipore, Billerica, MA). Cleaved vector was treated with calf intestinal alkaline phosphatase (New England BioLabs). Gclc cDNA was directionally subcloned downstream of the cytomegalovirus (CMV) promoter in the pCMV-Script vector using T4 DNA ligase (New England BioLabs). The resulting recombinant vectors, pCMVGCLC, were propagated in One Shot MAX Efficiency DH5α-T1® competent cells (Invitrogen) and plasmid preparations were performed using the QIAfilter Midi Plasmid Kit (Qiagen).
Dr Terrance J. Kavanagh (University of Washington, Seattle, WA) kindly provided us with a full-length mouse Gclm cDNA directionally cloned in pCR3.1-Uni, also a pCMV-directed mammalian expression vector that allows for the selection of stably transfected mammalian cells in media containing G418 antibiotic. This vector, GCLM-pCR3.1-Uni, will henceforth be referred to as pCMVGCLM vector (38). The Gclm-containing recombinant vector was propagated in One Shot MAX Efficiency DH5α-T1® competent cells and plasmid preparations were performed using the QIAfilter Midi Plasmid Kit.
All purified plasmid preparations were subjected to analytical restriction digests to identify correctly subcloned constructs. Vectors displaying the correct restriction map were subjected to CMVGCLC transgene-specific or CMVGCLM transgene-specific polymerase chain reaction (PCR) and/or automated sequence analysis at the UCI DNA Core Facility (University of California, Irvine, CA). Briefly, purified pCMVGCLC-Script vectors were amplified using the primer 5′-AATTAACCCTCACTAAAGGG-3′, which anneals to the 5′-flanking region just upstream of the ligation site beginning at nucleotide −109,where nucleotide +1 is the first nucleotide of the Gclc cDNA, and the primer 5′-GTAATAC-GACTCACTATAGGGC-3′, which anneals to the 3′ untranslated region of Gclc ending at nucleotide +2331, yielding an amplified product that is exactly 2.44 kb and spans part of the CMVGCLC transgene construct and the entire Gclc cDNA sequence. DNA sequencing has repeatedly revealed 100% homology between pCMVGCLC-Script vector sequence and published pCMV-Script and Gclc cDNA sequences (37,38).
Purified pCMVGCLM vectors were amplified using the primer 5′-TAATACGACTCACTATAGGG-3′, which anneals to the 5′-flanking region just upstream of the ligation site beginning at nucleotide −60,where nucleotide +1 is the first nucleotide of the Gclm cDNA, and the primer 5′-TAGAAGGCACAGTCGAGG-3′, which anneals to the 3′ untranslated region ending at nucleotide +1052, yielding an amplified product that is exactly 1.112 kb and spans part of the CMVGCLM transgene construct and the entire Gclm cDNA sequence. DNA sequencing has repeatedly revealed 99.8% homology between pCMVGCLM vector sequence and published Gclm cDNA sequences, including one alanine to proline substitution at amino acid 7 and one silent mutation at amino acid 33 (codon CGA to codon CGG, both encoding for the amino acid arginine). DNA sequencing was performed at least two separate times and in duplicate for each cell line described in this report.
Cells from passage 44, for Gclc and empty vector control transfections, and cells from passage 36, for Gclm transfections and Gclc/Gclm co-transfections, were trypsinized and counted using trypan blue exclusion. Cells were plated at a density of 1.4×106 per well of a six-well tissue culture plate and were incubated for 24 h prior to transfection.
Four micrograms of plasmid DNA (pCMVGCLC-Script, pCMVGCLM, a 1:1 molar ratio of pCMVGCLC-Script to pCMVGCLM or empty pCMV-Script vectors) was diluted in 250 μl Opti-MEM I reduced serum medium. To form DNA–liposome complexes, 260 μl diluted Lipofectamine 2000 reagent was added to each diluted vector preparation and incubated for 20 min at room temperature. DNA–liposome complexes were added to each well containing COV434 cells, and cells were incubated for 6 h at 37°C. Transfection media were then replaced with complete media, and cells were incubated for 42 h. Cells were trypsinized and subcultured at 1:12 in 10-cm tissue culture plates using complete media freshly supplemented with 1 mg/ml G418 and ensuring that adherent cells covered no more than 50% of the growth surface area. Cells were grown for 10–14 days, feeding them selective complete media every 4 days.
Once there were visible colonies growing in the selective tissue culture plates, stably transfected cells were harvested using trypsinization within cloning cylinders adhered around a single colony using sterile vacuum grease. Cells from a single colony were transferred to a 24-well tissue culture plate in freshly prepared selective complete media. Cultures were grown to 80% confluency. Confluent cells were subcultured in progressively larger tissue culture vessels for a minimum of five passages prior to beginning experiments.
Confluent cells were harvested by trypsinization and counted using trypan blue exclusion. In total, 4×106 cells were collected and pelleted by centrifugation at low speeds. Cells were washed with phosphate-buffered saline (PBS) and resuspended in PBS. Genomic DNA was extracted using the DNeasy Blood and Tissue Kit as per the manufacturer’s instructions (Qiagen). One microgram genomic DNA was amplified using the CMVGCLC- or CMVGCLM-PCR using the primers described above.
For the CMVGCLC-PCR, DNA was denatured for 30 sec at 94°C, primers were annealed for 30 sec at 50°C and new DNA was polymerized for 2 min at 72°C for 32 cycles, yielding a 2.44-kb amplified product. For the CMVGCLM touchdown PCR, DNA was denatured for 30 sec at 94°C, primers were annealed for 30 sec at 65°C and new DNA was polymerized for 1 min at 72°C for 36 cycles. The annealing temperature was decreased in increments of 1°C in each subsequent set of cycles until 45°C was reached (cycle 21) at which time all annealing temperatures were held constant. This PCR yielded an amplified product that was exactly 1.112 kb. The genotyping of all the cell lines described in this report was performed at least three separate times throughout the course of these experiments.
Total intracellular GSH levels were measured using a modified form of an enzymatic recycling assay developed by Griffith (39,40). Cells were plated at 2×106 per 25-cm2 flask using selective complete media and allowed to grow for 48 h. Cells were harvested by trypsinization and centrifuged at 300× g for 10 min. Cell pellets were lysed in ice-cold TES-SB buffer (20 mM Tris, 1 mM EDTA, 250 mM sucrose, 2 mM L-serine, 20 mM boric acid, supplemented with protease inhibitors, 100 μg/ml Pefabloc, 10 μg/ml n-alpha-p-tosyl-L-lysine chloromethyl ketone, 1 μg/ml pepstatin A, 1 μg/ml aprotinin and 1 μg/ml leupeptin) at a ratio of 50 μl buffer for every 2×106 cells. A small volume of cell homogenate was removed for subsequent protein assay with the Pierce bicinchoninic acid (BCA) Assay Kit (Pierce, Rockford, IL), and proteins were precipitated by acidifying the remaining cell homogenate with four parts homogenate to one part 5% sulfosalicylic acid (SSA). Homogenates were centrifuged at 15800× g for 10 min at 4°C. Sample aliquots were frozen at −80°C until use.
Standards containing known concentrations of GSH in TES-SB buffer with SSA or unknown samples were aliquoted in triplicate into the wells of a 96-well microplate, diluted with water and incubated at 30°C for 10 min. GSH reaction mixture [0.26 mM reduced β-NADPH, 0.74 mM 5,5-dithiobis(2-nitrobenzoic acid) (DTNB), 0.62 U/ml GSH reductase, 143 mM Na2HPO4 and 6.3 mM EDTA] was then added to each well. Absorbance was immediately monitored at 412 nm every 10 sec for 3.5 min using a VersaMax Tunable Microplate Reader with SOFTmax Pro version 3.1 Software (Molecular Devices, Sunnyvale, CA). DTNB is reduced to the yellow product 5-thio-2-nitrobenzoic acid with GSH acting as an electron donor. Oxidized glutathione is cycled back to its reduced form by GSH reductase and NADPH. The resulting slope of the absorbance versus time plot is proportional to the GSH concentration. The threshold of sensitivity for this assay is 0.04 nmol in our hands. Samples were analysed from four separate experiments.
Cells were plated at 2×106 per 25-cm2 flask in selective complete media and allowed to grow for 24 or 48 h. Cells were then processed as for GSH assays, except the acidification step was omitted. Forty micrograms of protein samples were diluted in sample buffer containing sodium dodecyl sulphate and β-mercaptoethanol. Proteins were electrophoresed through 12% Tris–HCl polyacrylamide gels (Bio-Rad Laboratories, Hercules, CA) and then transferred to Immobilon-P transfer membranes (Millipore).
Protein blots from three separate experiments were probed and exposed to film simultaneously for GCLC and GCLM immunodetection. For activated caspase 3 immunodetection, cells from two separate experiments were analysed simultaneously. Blots were blocked at 4°C overnight using 5% non-fat dry milk in PBS supplemented with 0.1% Tween (PBST). Blots were then washed and incubated for 2 h at room temperature with a mixture of 1:40000 rabbit anti-human GCLC and 1:20000 rabbit anti-human GCLM (kind gift of Dr Terrance J. Kavanagh, University of Washington) (41) diluted in blocking solution of 3% non-fat dry milk, 3% bovine serum albumin (BSA), 1% ovalbumin, 1% normal goat serum and 0.1% sodium azide in PBST. Activated caspase 3 antibody (Asp 175, Cell Signaling, Beverly, MA) was used at a dilution of 1:100 in the same blocking solution. Blots were then washed and incubated for 1 h at room temperature with 1:10000 horseradish peroxidase-conjugated donkey anti-rabbit IgG (Amersham Biosciences, Piscataway, NJ). The blots were washed and chemiluminescent detection was performed using the ECL Detection Kit and ECL Hyperfilm (both from Amersham Biosciences). Films were subjected to densitometric analyses to ascertain relative band intensities. The same blots were then probed for β-actin using 1:160000 mouse anti-human β-actin (Sigma–Aldrich), followed by 1:10000 horseradish peroxidase-conjugated anti-mouse IgG (Amersham Biosciences). GCLC and GCLM intensities were normalized to β-actin band intensities for the same sample, and results were expressed as fold the normalized CMV9b control levels for the same blot.
GCL enzymatic activity was measured as first described by White et al. (42). Cells were plated at 2×106 per 25-cm2 flask, cultured for 48 h and harvested and processed as for immunoblotting. Fifty microlitres of sample (in duplicate) or GSH standards in TES-SB were added in 20-sec intervals to prewarmed microcentrifuge tubes containing 50 μl of GCL reaction cocktail (400 mM Tris base, 40 mM ATP, 20 mM L-glutamic acid, 2 mM EDTA, 20 mM boric acid, 2 mM L-serine and 40 mM magnesium chloride hexahydrate). The GCL reaction was initiated by adding 50 μl of 2 mM L-cysteine to one of each of the sample tubes (GCL tube). After incubating for 20 min at 37°C, the GCL reaction was stopped by adding 50 μl of 200 mM SSA to all tubes. Immediately after the addition of SSA, 50 μl of 2 mM L-cysteine was added to the second tube of each sample (the baseline GSH tube) and to the GSH standard tubes. After the 20-min incubation on ice, all tubes were centrifuged for 5 min at 2000× g at 4°C. Twenty microlitres of supernatant was aliquoted in triplicate into the wells of a 96-well flat-bottomed, black-sided microplate. One hundred and eighty microlitres of 1.1 mM 2,3-naphthalenedicarboxaldehyde solution was then added to each well. The plate was incubated at room temperature for 30 min and read at an excitation wavelength of 485 nm and an emission wavelength of 535 nm using a GENios Plus Multi-Detection Microplate Reader with Enhanced Fluorescence (Tecan Group, Zurich, Switzerland). Results were expressed as the nanomoles of GSH synthesized above the baseline GSH level per minute per milligram protein, as determined using the Pierce BCA Protein Assay. There were three separate experiments performed for a total number of at least four samples per cell genotype.
Cells were irradiated in a J. L. Shepherd and Associates Mark I self-shielded 137Cs irradiator at a dose rate of 1.61 Gy/min. For preliminary experiments using parental COV434 cells, trypsinized cells were suspended in complete media in 15-ml conical tubes, irradiated and then plated for subsequent assays. In subsequent experiments, cells were plated and irradiated 24 h or more later. For water-soluble tetrazolium (WST)-1 assays, plating media were removed from the wells of a 96-well microplate and replaced with 30 μl DMEM–F12; after irradiation, media volumes were completed to 100 μl. For terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick-end-labelling (TUNEL) assay, plating media were removed from the chambers of Nunc Lab-Tek II CC2 8-Chambered Slide System and replaced with 90 μl DMEM–F12; after irradiation, media volumes were completed to 200 μl. For caspase 3 immunoblotting, plating media were removed from 25-cm2 flasks and replaced with 5 ml DMEM–F12 prior to irradiation. For the experiment in which GSH was suppressed with BSO, cells were irradiated in six-well plates after culture for 24 h with medium containing 0, 1, 5 or 100 μM BSO without removing the medium. These doses were chosen to cause varying degrees of GSH suppression from <50% to >85% suppression based on preliminary experiments in COV434 cells.
Cell metabolic activity, an indicator of cell viability, was measured based on the cleavage of a WST-1 salt to formazan by mitochondrial dehydrogenases of viable granulosa cells. For WST assays, 1×104 live cells were plated per well of a 96-well tissue culture-treated microplate. For the experiments in Figures 5 and and8,8, cells were treated with H2O2 or gamma irradiation directly in the plate. For the experiments in Figure 6, cells were trypsinized and suspended in medium, irradiated while in suspension and then plated for the assay. At the end of the treatment periods, 10 μl of WST-1 reagent was pipetted into each well, and the plates were incubated for 2 h at 37°C. Blank wells contained culture medium and WST-1 reagent with no cells. The absorbance at 440 nm was then read using a VersaMax Tunable Microplate Reader (Molecular Devices). WST-1 cleavage to a formazan results in a colour change. The absorbance at 440 nm is therefore proportional to the number of viable cells. The average of the blank absorbance values was subtracted from the absorbance reading for each well and the resulting values were used for data analysis.
ROS were measured using the cell permeable compound, 2′,7′-dichlorofluorescin diacetate (H2DCF-DA; Molecular Probes, Eugene, OR). H2DCF-DA is taken up by cells, where it is cleaved by cytosolic esterases to the non-fluorescent compound dichlorofluorescin, which in the presence of various ROS and nitrogen species is oxidized to the fluorescent compound, dichlorofluorescein (DCF) (43,44).
In preliminary experiments, suspensions of the parental COV434 cells were irradiated with 0, 1 or 5 Gy and then plated in six-well tissue culture plates. After 6, 24 or 48 h, conditioned media were removed, and 10 μM H2DCF-DA in fresh media was added. Cells were incubated for 30 min, washed, trypsinized and resuspended in the conditioned media. DCF fluorescence was then immediately measured using fluorescence-activated cell sorting (45).
For subsequent experiments, a previously described H2DCF-DA microplate assay was used (35,46). In total, 2×104 live cells per well of a 96-well tissue culture-treated, black-sided microplate were incubated for 24 h. Cells were then irradiated with 0, 1 or 5 Gy. After irradiation and just prior to the addition of H2DCF-DA, positive control COV434 cells were treated for 30 min with 0.5 mM H2O2. At 0.5, 2 or 6 h after irradiation, media were removed and replaced with media containing 100 μM H2DCF-DA. Negative control wells received media without H2DCF-DA. Cells were incubated for 30 min. The H2DCF-DA-containing medium was then removed; cells were washed with PBS and maintained in PBS while fluorescence intensity was read using a GENios Plus Multi-Detection Microplate Reader with Enhanced Fluorescence (Tecan Group) with excitation wavelength set at 485 nm and emission wavelength set at 535 nm. Three separate experiments were performed.
Apoptosis was detected in irradiated cells using the In Situ Cell Death Detection Kit, POD (Roche Molecular Biochemicals, Mannheim, Germany), as previously described (35,47). Cells were cultured at 5×104 per chamber of a Nunc Lab-Tek II CC2 8-Chambered Slide for 24 h, irradiated with 0, 1 or 5 Gy and returned to the incubator. After 2 or 6 h, cells were washed with ice-cold PBS and fixed with 4% paraformaldehyde for 30 min on ice. Slides were washed in PBS, treated with 3% hydrogen peroxide in methanol for 10 min, washed in PBS, permeabilized with 0.1% Triton X-100 in 0.1% sodium citrate for 2 min on ice, washed, blocked for 30 min at room temperature with 3% BSA in PBS, washed, incubated at 37°C for 1 h with terminal deoxynucleotidyl transferase/label solution (containing fluorescein-labelled deoxyuridine triphosphate) and washed again. Positive control slides were treated with DNAse I (500 U) prior to labelling, and the terminal transferase was omitted for negative control slides. Slides were then blocked with 3% BSA in PBS, washed, incubated for 30 min at 37°C with peroxidase converter solution (anti-fluorescein antibody, Fab fragment from sheep, conjugated with horseradish peroxidase) and diluted 1.5:1 in TUNEL dilution buffer. Slides were then washed and incubated for 5 min with diaminobenzidine substrate solution (Roche Molecular Biochemicals), washed in PBS and counterstained with haematoxylin. All washes and incubations were at room temperature unless otherwise indicated. Data from two separate experiments were pooled for analysis. The numbers of TUNEL-positive and TUNEL-negative cells in three ×400 microscopic fields per well were counted by an observer blind to treatment and genotype. The average of the counts was used for data presentation and statistical analyses.
Data from two or three replicate experiments were pooled for statistical analyses unless otherwise noted. Homogeneity of variances was assessed by Levene’s test. Data expressed as proportions (percentages of TUNEL-positive cells) were subjected to arcsine square root transformations prior to analysis (48). The differences among groups were analysed by analysis of variance (ANOVA) followed by Fisher’s least significant difference (LSD) test. When variances were non-homogeneous, the Kruskal–Wallis test for non-parametric data was used. If the Kruskal–Wallis test was significant at P<0.05, the Mann–Whitney test was then used for intergroup comparisons. Statistical analyses were carried out using SPSS 16.0 for Mac OS X.
Genomic DNA from all stably transfected cell lines was genotyped using CMVGCLC-PCR and/or CMVGCLM-PCR to confirm that the cells had stably integrated the correct transgenes. Figure 1 shows examples of cell lines that had and had not stably integrated the correct transgenes, as dictated by the vectors used during the transfection process (pCMV-Script control vector, pCMVGCLC-Script vector, pCMVGCLM vector or both the pCMVGCLC-Script and the pCMVGCLM vectors). In this and subsequent sections, cell lines transfected with both pCMVGCLC-Script vector and pCMVGCLM vector are designated ‘CM’ followed by a number and lower case letter, cell lines transfected with pCMVGCLC-Script vector only are designated ‘C’ followed by a number and lower case letter and cell lines transfected with pCMVGCLM vector only are designated ‘M’ followed by a number and letter. The empty vector-transfected control cells are designated CMV9b. Cell lines CM4b and CM3a did not appear to integrate the CMVGCLC transgene, but both did have very light bands corresponding to the CMVGCLM transgene. The remaining cell lines integrated the transgenes with which they were transfected.
Figure 2 shows the total GSH concentrations in the parental and stably transfected cell lines. GSH levels varied significantly by cell genotype whether expressed as nanomole per 106 cells or nanomole per milligram protein (P<0.001 by Kruskal–Wallis test). When expressed as nanomole GSH per 1×106 cells, parental COV434 cells displayed virtually the same GSH levels as control cells transfected with empty pCMV-Script vector, the CMV9b cells (7.2±0.3 and 6.6±0.5 nmol GSH/1×106 cells, respectively). The cell lines CM2a, CM4b, C5a, C6a, C6b and M5a displayed statistically significantly higher total GSH levels relative to COV434 parental cells and CMV9b control cells (P<0.04 and P<0.03, respectively, by Mann–Whitney tests; Figure 2A). GSH concentrations were 1.8-fold higher in CM2a cells, 1.3-fold higher in CM4b cells, 1.6-fold higher in C5a cells, 2.4-fold higher in C6a cells, 1.9-fold higher in C6b cells and 1.7-fold higher in M5a cells than in control CMV9b cells when expressed as nanomole GSH per 1×106 cells. The cell lines C3b, C6c, M7a and M8c did not display statistically significant differences in total GSH concentrations relative to either COV434 parental cells or CMV9b control cells (Figure 2A). The cell line CM3a displayed a statistically significant decrease of 33% in total GSH levels relative to parental COV434 cells (P=0.002 by Mann–Whitney test), but not CMV9b cells (Figure 2A). When total intracellular GSH levels were expressed as nanomole GSH per milligram protein, the results followed the same trends as the data described above, although the results of statistical analyses varied somewhat (Figure 2B).
Because CM2a, C6a, C6b and M5a cells displayed statistically significant increases in total GSH relative to control CMV9b cells and had stably integrated the transgenes by PCR, these cells were selected for use in subsequent experiments in which cells were challenged with H2O2 or ionizing radiation.
GCLC and GCLM protein levels varied in a statistically significant manner with genotype (P=0.001 and P<0.001 by Kruskal–Wallis tests, respectively, Figure 3). Control CMV9b cells had virtually the same GCLC and GCLM protein levels as parental COV434 cells (1.3- and 0.9-fold COV434 controls, respectively). CM2a cells co-transfected with Gclc and Gclm displayed a statistically significant increase in GCLC protein expression levels of 1.5-fold relative to parental COV434 controls (P=0.040 by Mann–Whitney test) and a small, statistically non-significant increase relative to CMV9b controls. They also had small, non-statistically significant increases in GCLM levels relative to both COV434 and CMV9b controls (Figure 3). C6a cells transfected with Gclc had statistically significant increases in GCLC protein expression of 14.9-fold relative to COV434 control cells (P<0.001 by non-parametric Mann–Whitney test) and 11.5-fold relative to CMV9b controls (P<0.001). C6b cells transfected with Gclc had 8-fold increases in GCLC in an initial experiment, but did not show any increase in GCLC levels in a subsequent experiment, suggesting that copies of the transgene were lost over time. As a result, the pooled data for C6b cells from all experiments showed a non-significant increase in GCLC protein levels of 4.6-fold over COV434 control cells. In contrast, C6b cells consistently showed a reproducible and statistically significant decrease in GCLM protein expression of 0.3-fold relative to COV434 control cells (P=0.017 by Mann–Whitney test). M5a cells transfected with Gclm had statistically significant increases in GCLM protein expression of 3.5-fold relative to COV434 controls (P=0.008 by Mann–Whitney test) and 4-fold relative to CMV9b controls (P=0.001).
GCL enzymatic activity varied in a statistically significant manner with genotype (P<0.001 by one-way ANOVA; Figure 4). Parental COV434 cells and control CMV9b cells had virtually the same GCL activity levels (0.134±0.011 and 0.142±0.011 nmol GSH/mg protein/min, respectively). GCL enzymatic activity increased by 2.0-fold in CM2a cells transfected with both Gclc and Gclm relative to CMV9b cells (P<0.001 by LSD test). C6a cells transfected with Gclc also displayed a statistically significant increase in GCL enzymatic activity of 2.8-fold when compared to CMV9b cells (P < 0.001 by LSD test). C6b and M5a cells displayed small, statistically insignificant increases in GCL enzymatic activity when compared to either COV434 or CMV9b cells.
Overall, the changes in GCLC and GCLM protein expression, GCL enzymatic activity and GSH concentrations in the stably transfected cell lines are consistent with one another, with C6a cells transfected with Gclc having the greatest fold increases in GCLC protein levels, GCL enzymatic activity and GSH levels.
Cells were treated with various concentrations of H2O2 for 30 min to induce oxidative stress and then assayed for metabolic activity based on the cleavage of WST-1. For all cell genotypes, metabolic activity decreased in a concentration-dependent manner with H2O2 treatment (Figure 5A). The effects of H2O2 concentration and genotype were statistically significant by two-way ANOVA (P<0.001 and P=0.003, respectively); however, because variances were not homogeneous, non-parametric tests were used for subsequent comparisons. The effect of cell genotype on metabolic activity was statistically significant at every concentration, except 1 mM H2O2 (P≤0.001 by Kruskal–Wallis tests at each concentration; Figure 5A). At all concentrations, excluding 1 mM H2O2, C6a and C6b Gclc-transfected cells had significantly higher WST-1 cleavage than COV434 and CMV9b control cells (Figure 5A). At all H2O2 concentrations, the differences in metabolic activity between parental COV434 and control CMV9b cells were negligible and not statistically significant (Figure 5A).
Cells were also treated with 0.5 mM H2O2 and then similarly assayed for metabolic activity at various times after treatment. The effect of genotype on metabolic activity was statistically significant at every time point (P<0.001 by Kruskal–Wallis tests; Figure 5B). For all cell genotypes, granulosa cell metabolic activity decreased with H2O2 treatment. Control cells showed a time-dependent decrease in metabolic activity until the 2-h time point after which metabolic activity reached a plateau. All GCL-transfected cells showed decreases in metabolic activity with H2O2 treatment, but after 1 h, the cells recovered and metabolic activity began to increase (Figure 5B). At all treatment times, the cell lines that had been stably transfected with Gclc and/or Gclm (C6a, C6b, CM2a and M5a) had statistically significantly smaller decreases in metabolic activity than did the CMV9b control cells. The greatest protective effect of overexpression of GCL subunits was observed in the C6a cell line, which had the highest intracellular GSH concentrations. At all treatment time points, the differences in WST-1 cleavage between parental COV434 and control CMV9b cells were smaller than differences between either control cell line and any of the GCL subunit-overexpressing cell lines.
Taken together, the H2O2 dose–response data and H2O2 time course data suggest that the Gclc- and Gclm-transfected granulosa cells, due to their increased ability to synthesize GSH, are able to respond to and overcome the oxidant insults. These results support the hypothesis that GSH is involved in protecting granulosa cells against ROS-mediated cell death.
In a preliminary experiment to select the optimal radiation doses for subsequent experiments, parental COV434 cells were irradiated with 0, 1, 2, 5 or 10 Gy and metabolic activity was measured at 48, 72 and 96 h after irradiation (Figure 6A). At all time points, there was a dose-dependent decrease in metabolic activity, with significant decreases in the number of viable cells observed at all doses compared to controls (P<0.001, effect of dose by Kruskal–Wallis test at all times). To begin to test whether ROS may play a role in radiation-induced cell death in granulosa cells, COV434 cells were irradiated with 0, 1 or 5 Gy, and ROS were measured as DCF fluorescence after 6, 24 and 48 h. DCF fluorescence increased in a dose-dependent manner at all time points (P<0.04 by Kruskal–Wallis test; Figure 6B). Because there was a significant increase in ROS generation observed after radiation treatment of the parental cells, we hypothesized that GCL-transfected cells would be protected when irradiated much like they were protected against H2O2 treatment. Because there was a significant increase in ROS generation already 6 h after radiation treatment, we chose to focus on the early, acute effects of radiation treatment in subsequent studies of ROS generation and apoptotic end points.
We tested the effects of GCL overexpression on the generation of ROS after gamma irradiation using DCF fluorescence (Figure 7). The effects of radiation dose, time since irradiation and genotype were statistically significant (all P<0.001 by three-way ANOVA). However, since the variances were not homogeneous, subsequent comparisons were made using non-parametric tests. When cells were irradiated with 1 or 5 Gy, all cell genotypes demonstrated increases in ROS relative to sham-irradiated controls of the same genotype after 0.5, 2 and 6 h (P<0.01, effect of dose for all genotypes and time points by Kruskal–Wallis test).
Irradiated cells displayed a genotype-specific response in ROS generation. ROS levels varied in a statistically significant manner by genotype for both the 1- and the 5-Gy doses at all time points (P≤0.01 by Kruskal–Wallis test). At all three time points after both doses of gamma irradiation, the CMV9b control cells had the highest relative increases in ROS levels, and one or more of the GCL subunit-overexpressing cell lines had statistically significantly lower ROS generation than the control CMV9b cells (Figure 7). The Gclc-overexpressing C6a cells, which had the highest GSH concentrations, also had the greatest attenuation of ROS generation relative to the CMV9b controls.
We tested the effects of overexpression of GCL subunits on radiation-induced cell death using the WST-1 metabolic activity assay. We observed significant effects of radiation dose, treatment time and cell genotype on metabolic activity (P<0.001 by three-way ANOVA for all comparisons). However, since the variances were not homogeneous, subsequent comparisons were made using non-parametric tests.
The fold increase in metabolic activity relative to 0 h varied with cell genotype at both the 72- and the 96-h time points and for all three doses of radiation (P<0.009 by Kruskal–Wallis test). All cell genotypes demonstrated cellular proliferation, despite having been treated with 1 Gy of γ-rays, as indicated by the increase in the number of viable cells over time (Figure 8). In contrast, only the C6a and C6b cells, which overexpressed Gclc, and the control CMV9b cells demonstrated any proliferation, albeit very reduced, after treatment with 5 Gy (Figure 8). After 96 h and despite being irradiated with 5 Gy, C6a and C6b cells displayed statistically significant, ~2-fold increases in the numbers of viable cells relative to their initial numbers at 0 h (P<0.001, effect of time by Kruskal–Wallis test; P<0.001, 96 versus 0 h by Mann–Whitney test). While the CMV9b controls demonstrated a small 1.3-fold increase over 0 h levels at 96 h (P=0.004 by Mann–Whitney test), the differences between the C6a and C6b lines and the CMV9b line at 96 h were statistically significant (P<0.001 by Mann–Whitney test). The 5-Gy-irradiated COV434 and CM2a cells, on the other hand, displayed significantly smaller fold increases than CMV9b controls at 96 h (P=0.005 and P=0.037 by Mann–Whitney tests, respectively).
Interestingly, the effect of cell genotype on cellular proliferation was also statistically significant in untreated (0 Gy) cells at both 72 and 96 h (P<0.008 by Kruskal–Wallis test). Post hoc analysis by Mann–Whitney test revealed that only the Gclc-overexpressing C6a cells had statistically significant fold increases in viable cells relative to the CMV9b controls at 72 and 96 h (P<0.01 by Mann–Whitney test). All other untreated cells displayed no statistically significant differences relative to CMV9b controls at either time point.
These results suggest that Gclc overexpression protects cells against radiation-induced cell death, as well as conferring a growth advantage to cells growing in culture in the absence of exposure to radiation. C6a cells, the stably transfected cell line with the greatest increase in intracellular GSH, were most protected against the deleterious effects of γ-rays. Together with the observation that C6a cells also demonstrated the greatest attenuation in ROS generation in response to irradiation, these results led us to select C6a cells for further studies. We have previously shown that treatment of parental COV434 cells with the alkylating agent cyclophosphamide resulted in ROS generation, followed by apoptosis. We hypothesized that the increase in ROS in response to γ-rays would similarly lead to the initiation of apoptosis and that C6a cells would be protected against this effect.
We used TUNEL assay and activated caspase-3 immunodetection to test the hypothesis that overexpression of Gclc would protect human granulosa cells against the induction of apoptosis by ionizing radiation. The percentage of TUNEL-positive cells increased in a dose-dependent manner for the control CMV9b and COV434 cells, as well as for the Gclc-overexpressing C6a cells (Figure 9). The increase in TUNEL-positive cells was already evident at 2 h and became pronounced at 6 h after irradiation (Figure 9A). The effects of genotype, radiation dose and treatment time, as well as of the two-way and three-way interactions among those variables, on the arcsine square root-transformed fraction of TUNEL-positive cells were statistically significant by three-way ANOVA (P<0.001 for all effects). Because the variances were not homogeneous, non-parametric tests were used for subsequent comparisons. The effect of genotype was statistically significant for the 5-Gy dose at 2 h and for both the 1- and the 5-Gy doses at 6 h (P<0.02 by Kruskal–Wallis test). The Gclc-overexpressing C6a cells had significantly fewer TUNEL-positive cells 2 h after the 5-Gy dose and 6 h after both the 1- and the 5-Gy doses than either control CMV9b or COV434 cells (P<0.02 by Mann–Whitney test).
The increase in TUNEL-positive apoptotic cells was associated with activation of caspase 3, which is indicated by the appearance of cleaved 17- and 19-kDa activated forms of the enzyme (Figure 9B). Cleaved caspase 3 bands were not visible in control CMV9b or COV434 cells or in Gclc-overexpressing C6a cells at 2 h after irradiation (data not shown). By 6 h after irradiation, cleaved caspase 3 was clearly evident in the two control cell lines and was still absent in the C6a cells (Figure 9B).
To test whether the protective effect of stable transfection with Gclc is due to increased GSH synthesis, GSH synthesis was biochemically suppressed with BSO, an inhibitor of GCL activity. C6a cells were treated with 0, 1, 5 or 100 μM BSO for 24 h prior to 5-Gy irradiation or sham irradiation; control CMV9b cells were also irradiated but not treated with BSO. Three days after irradiation or sham irradiation, cells were trypsinized and live cells were counted using trypan blue exclusion. As shown in Figure 10, suppression of GSH synthesis in the Gclc-overexpressing C6a cells with increasing doses of BSO dose-dependently increased sensitivity to ionizing radiation. The C6a cells treated with 100 μM BSO were as sensitive to 5-Gy radiation as the control CMV9b cells (P=1.00; C6a, 100 μM BSO versus CMV9b, 0 μM BSO by Tukey’s honestly significant difference (HSD) post hoc test).
Radiation therapy can damage ovarian follicles and cause temporary or permanent infertility. Radiation can also indirectly damage cells by generating ROS that can persist for various times post-irradiation. The redox environment is critical to intracellular homoeostasis and has been shown to be important in the progression of programmed cell death. Herein, we showed for the first time that ionizing radiation treatment generates ROS and subsequently causes apoptotic death in cultured human granulosa cells. GSH is an important antioxidant, and its synthesis has been shown to be upregulated in radiation-resistant cancer cells. Our results show that stable overexpression of one or both subunits of GCL, the rate-limiting enzyme in GSH synthesis, in a human granulosa tumour cell line enhances GSH synthesis and protects the cells against hydrogen peroxide and ionizing radiation-induced cell death. Our novel findings demonstrate how strategies aimed at manipulating GSH levels can modulate radiation-induced granulosa cell apoptosis.
Radiation therapy, together with chemotherapy, has greatly enhanced the survival rates of young women afflicted with various forms of malignant cancers. Destruction of ovarian follicles, leading to infertility, is an unfortunate side effect of such aggressive treatments that can adversely affect quality of life. Ionizing radiation has been shown to destroy immature primordial follicles and maturing secondary and antral follicles (1–3,12). In the latter follicles, apoptotic death of cells in the granulosa cell layers surrounding the oocyte appears to occur first, followed by death of the oocyte. COV434 granulosa cells are derived from a human granulosa cell tumour; however, they possess many properties of normal granulosa cells that make them useful as models for understanding mechanisms by which ionizing radiation destroys growing follicles (36,49).
GCL is a heterodimer composed of GCLC and GCLM. Although GCLC possesses the catalytic activity, GCLM decreases the Km for glutamate and ATP and increases the Ki for GSH, thereby greatly increasing the catalytic efficiency of the enzyme (25). In the present experiments, we generated stably transfected COV434 human granulosa tumour cell lines that constitutively overexpressed Gclc, Gclm or both Gclc and Gclm. Total intracellular GSH levels increased significantly by 1.5- to 2.2-fold, depending on the cell line, as a result of increased GCLC and/or GCLM protein levels and increased GCL catalytic activity. Several studies have shown that overexpression of Gclc, Gclm or both Gclc and Gclm results in increased GCL protein expression and enzymatic activity as well as in higher total GSH levels in other cell types (50–57). The effect of GCL overexpression often varied in an unpredictable manner between different stably transfected clones. Indeed, the same cell line transfected with the same amount of GCL overexpression vector can result in various clones that have differing degrees of GCL protein expression, enzymatic activity and/or total GSH levels (50,51,56). In some instances, overexpression of Gclc alone maximally increased total GSH levels; in others overexpression of Gclm alone had the greatest effect and in yet others co-transfection yielded the highest GSH levels (50,53,54,58). When various molar ratios of Gclc to Gclm cDNA ranging from 4:1 to 1:4 were used to co-transfect COS-7 cells, there were no significant differences in GSH concentration among the co-transfected cells regardless of the molar ratio used. In the same study, when COS-7 cells were transfected with Gclc only, the effect on GSH levels was not significantly different from the changes seen in the various co-transfected clones (53). These results suggest that the copy number of integrated transgenes has the greatest effect on total GSH levels and not the absolute amount of Gclc and/or Gclm cDNA that is introduced into the transfected cells. Because transgene integration is a random event, one cannot predict the results of single-gene or double-gene transfection on biochemical end points. When Gclc and/or Gclm are overexpressed to various randomly generated degrees, the bioavailability of the opposite heterodimer may become limiting. Similarly, the bioavailability of L-cysteine, the rate-limiting substrate in GSH synthesis, may become limiting (59). Finally, the effect of GCL overexpression on total GSH levels also seems to be highly cell type specific (56).
The results of the present studies add to the growing body of evidence that GSH is involved in protecting cells against ROS-mediated apoptosis. In particular, this and our previous studies support an anti-apoptotic role of GSH in granulosa cells. Induction of granulosa cell apoptosis by hormone withdrawal and by the polycyclic aromatic hydrocarbon dimethylbenz[a]anthracene was preceded by an early rise in ROS in cultured antral follicles (60,61). Similarly, induction of apoptosis by the anticancer drug cyclophosphamide in COV434 human granulosa cells was preceded by an early increase in ROS (35). The induction of apoptosis by all these stimuli was enhanced by GSH depletion and attenuated by supplementation of GSH (35,60,61). Gclc overexpression in transgenic flies increased resistance to H2O2 treatment and increased mean fly lifespan by inhibiting age-associated ROS accumulation that culminates in neuronal apoptosis (54). Similarly, Gclc overexpression rescued primary rat cortical neurons from undergoing apoptosis induced by exogenous nitrosative stress (58). Rat Gclc overexpression in Chinese hamster ovary cells enhanced resistance to cytotoxic effects of tert-butyl hydroperoxide treatment (51), and overexpression of both Gclc and Gclm cDNAs in Hepa-1 cells conferred a protective effect against H2O2-induced single-strand DNA breaks (55). In the present study, GCL subunit-overexpressing cells were challenged with the pro-oxidant H2O2 and then assayed for cell viability. For all cell genotypes, granulosa cell viability decreased in a dose-dependent manner with H2O2 treatment; however, viability decreased significantly less in Gclc-transfected cells than in control cells, showing that these cells were better able to respond to the oxidative assault. These results support the hypothesis that GSH is involved in protecting granulosa cells against ROS-mediated cell death. It is important to note that in our experiments, the C6a cell line, which displayed the greatest increase in total GSH levels relative to parental controls, was most protected when challenged with H2O2. Three other GCL subunit-transfected cell lines, which had smaller increases in GSH concentrations relative to control cells, were also protected, but to a lesser extent than C6a cells. Moreover, the protective effect of stable transfection with Gclc in the C6a cell line was reversed by suppression of GSH synthesis with BSO, an inhibitor of GCL activity.
When we irradiated parental COV434 cells, we noted a dose-dependent increase in ROS generation within 30 min of irradiation that persisted at 2, 6, 24 and 48 h. GCL-overexpressing human granulosa cells also demonstrated rapid, dose-dependent increases in ROS generation after irradiation; however, radiation-induced ROS generation was significantly attenuated in cells overexpressing GCL. Cells that produced the highest levels of GSH (i.e. C6a) were most protected against radiation-induced ROS. Several studies have also demonstrated that ROS increase after radiation exposure in other cell types. Immediately after exposing U937 human premonocytic leukaemic lymphoma cells to 2 Gy of γ-irradiation, levels of intracellular ROS and specifically H2O2 increased 2-fold. This increase in ROS generation was accompanied by a smaller, but significant, increase in lipid peroxidation and protein oxidation (62). In vivo radiation exposure has also been shown to increase generation of hepatic nitrosative species after 24 h (33). Primary rat neural precursor cells exhibited a dose-dependent increase in ROS over the first 12 h after exposure to X-rays, with maximal ROS levels being attained at 12 h (45). Post-irradiation ROS levels remained persistently elevated over a 3- to 4-week interval and then returned to background levels (45). When exposed to 250 MeV protons, neural precursor cells exhibited significant and dose-dependent increases in ROS, with maximal ROS levels at 6 h (63). Different radiation sources and regimes and different cell types, therefore, seem to elicit distinct temporal changes in ROS generation. Moreover, when primary neural precursor cells were pretreated with the pro-oxidant H2O2, they became sensitized to both proton and γ-rays (64).
In the present studies, caspase 3 activation and DNA fragmentation measured by TUNEL were minimally increased at 2 h after irradiation and were prominently increased by 6 h in both parental and empty vector-transfected control cell lines. The Gclc-transfected cell line C6a was protected against radiation-induced apoptosis. It is well established that radiation induces acute apoptotic cell death in various cell types including primary neural precursor cells (45), U937 cells (62), mouse hepatic tissue (32) and rat hepatic tissue (33). Others have established that pharmacologic replenishment of GSH or other thiol-containing reducing agents can protect cells against radiation-induced apoptosis or cell death. The GSH precursor, NAC, dose-dependently attenuated hepatic toxicity induced by ionizing radiation, as assessed by increased DNA damage, lipid peroxidation and apoptosis in mice (32) and rats (33). In both instances, pretreatment with NAC was shown to significantly and dose dependently increase GSH levels, suggesting that GSH, not NAC, mediates the protective effect against radiation-induced cell death. To our knowledge, the present study is the first to demonstrate a protective effect of genetic enhancement of GSH synthesis against radiation-induced cell death.
It is important to identify the key elements involved in follicular apoptosis in an effort to develop appropriate therapies for the protection and survival of these limited ovarian structures. We studied granulosa cells because it has been noted that follicular apoptosis often begins in the granulosa cells. Targeted GSH supplementation, for example using techniques such as biodegradable nanoparticle-controlled or magnetically controlled systems to deliver Gclc expression vectors to the ovary, could in the future serve as methods for protecting follicles in young women being treated with radiation therapy (65,66). Currently, the only methods for conserving these follicles are invasive and include ovarian transposition to reposition the ovaries outside of the irradiation field, sex hormone replacement or collection and cryopreservation of embryos, oocytes and ovarian tissue (1,2).
Along the same lines, it is imperative to understand the mechanisms by which cancerous cells circumvent the effects of conventional cancer treatments. Although COV434 cells have many characteristics of normal human granulosa cells and can serve as a model to understand normal granulosa cells (35,36), one cannot ignore the fact that they were derived from a granulosa cell tumour. These cells, therefore, inherently possess many characteristics of both primary granulosa cells and malignant cancerous cells. Our study demonstrates that increased GSH synthesis is radioprotective in tumour cells. Other researchers have shown that GSH depletion may serve as a potential therapy to radiosensitize tumour cells prior to the onset of radiotherapy (28–33).
Our studies suggest that polymorphisms of the Gclc and Gclm genes found in the human population may be responsible, at least in part, for interindividual variation in sensitivity to radiation-induced ovarian damage. In African American, Caucasian, Asian and Hispanic populations, the Gclc gene contains a GAG trinucleotide repeat in the 5′ untranslated region. There are five reported alleles with 4, 7, 8, 9 or 10 GAG repeats (67,68). The genetic polymorphisms that cause a deficit in Gclc expression, GCL activity and GSH biosynthesis have been associated with schizophrenia—a disease whose pathogenesis is associated with oxidative stress and GSH deficiency (69). There have also been two polymorphisms identified in an Asian population in the 5′-flanking region of the Gclm gene, each with two distinct allele variations. The −588T allele variant of the −588C/T polymorphism is positively correlated with decreased Gclm expression, decreased plasma GSH levels and increased risk of oxidative stress-induced vascular injury (70,71). By identifying correlations between allele frequencies and radiosensitivity, it may be possible to predict the probability of infertility resulting from radiotherapy.
In conclusion, we have shown that exposure of human granulosa cells to therapeutic doses of ionizing radiation rapidly increases intracellular ROS, followed by induction of apoptosis. Further, we have shown that granulosa cells with increased capacity to synthesize GSH, due to stable overexpression of GCL subunits, are protected against oxidant- and radiation-induced apoptosis. These results provide support for the hypothesis that the rise in ROS mediates the induction of apoptosis in granulosa cells by ionizing radiation and contribute to our understanding of the effects of radiation on normal ovarian follicles as well as ovarian tumours.
National Institutes of Health (ES10963 to U.L., T32 NS07444 to M.M.C.-W.); American Cancer Society (RSG-00-036-04-CNE to C.L.L); National Science Foundation (Graduate Teaching Fellowship DGE-0638751 to M.M.C.-W.); Howard Hughes Medical Institute University of California, Irvine, Teaching Fellows Program (to M.M.C.-W.); University of California Irvine Office of Research and Center for Occupational and Environmental Health (to U.L.).
We thank Dr Terrance J. Kavanagh, Dr Julie K. Andersen and Dr Peter I. Schrier for gifts of reagents. Portions of this work were presented in abstract form at the annual meetings of the Society for the Study of Reproduction in 2007 and the Society of Toxicology in 2008.
Conflict of interest statement: None declared.