Mitochondrial superoxide dismutase has a pro-tumorigenic role in ovarian clear cell carcinoma

Abstract

Epithelial ovarian cancer (EOC) is the fourth leading cause of death due to cancer in women and comprises distinct histological subtypes, which vary widely in their genetic profiles and tissues of origin. It is therefore imperative to understand the etiology of these distinct diseases. Ovarian clear cell carcinoma (OCCC), a very aggressive subtype, comprises >10% of EOCs. In the present study we show that mitochondrial superoxide dismutase (Sod2) is highly expressed in OCCC compared to other EOC subtypes. Sod2 is an antioxidant enzyme that converts highly reactive superoxide (O₂^•−) to hydrogen peroxide (H₂O₂) and oxygen (O₂), and our data demonstrate that Sod2 is pro-tumorigenic and -metastatic in OCCC. Inhibiting Sod2 expression reduces OCCC ES-2 cell tumor growth and metastasis in a chorioallantoic membrane (CAM) model. Similarly, cell proliferation, migration, spheroid attachment and outgrowth on collagen, and Akt-phosphorylation are significantly decreased with reduced expression of Sod2. Mechanistically, we show that Sod2 has a dual function in supporting OCCC tumorigenicity and metastatic spread. First, Sod2 maintains highly functional mitochondria, by scavenging O₂^•−, to support the high metabolic activity of OCCC. Secondly, Sod2 alters the steady-state ROS balance to drive H₂O₂-mediated migration. While this higher steady-state H₂O₂ drives pro-metastatic behavior it also presents a doubled-edged sword for OCCC, as it pushed the intracellular H₂O₂ threshold to enable more rapid killing by exogenous sources of H₂O₂. Understanding the complex interaction of antioxidants and ROS may provide novel therapeutic strategies to pursue for the treatment of this histological EOC subtype.

Introduction

Ovarian clear cell carcinomas (OCCC) represent approximately 10%–25 of all epithelial ovarian cancer (EOC), depending on ethnic background (1). It is now evident that OCCC differs widely from the more common high-grade serous adenocarcinoma. While the primary tumor mass of OCCC is found on the ovary, its origin is not thought to be the ovary or fallopian tube, but rather stem from endometrioid tissue and endometriosis. Due to the ROS stress associated with endometrioisis OCCC has been characterized as a stress-responsive cancer (2,3). Gene expression studies have shown increased expression of a number of stress-related and metabolic genes, in particular those related to hypoxic insult, glycolysis and antioxidant defense mechanisms (4). The Nrf2 stress response pathway has been implicated in driving some of these changes, including the enhanced expression of mitochondrial manganese-containing superoxide dismutase (Sod2) (5).

Sod2 is a nuclear-expressed mitochondria-targeted antioxidant enzyme, which catalyzes the conversion of two molecules of superoxide anion (O₂^•−) to hydrogen peroxide (H₂O₂) and oxygen (O₂). Since a small amount of O₂^•− leakage occurs during normal oxidative phosphorylation from the mitochondrial electron transport chain, Sod2 is of importance in preventing redox-mediated damage of mitochondrial proteins and preserving mitochondrial function. Enhanced O₂^•− production can occur in response to stress, such as hypoxia, and has been linked to a number of cancer types. For this reason, Sod2 was initially characterized as a tumor suppressor gene (6). However, recent data now point to a dichotomous role of Sod2 during tumor progression. Evidence suggests that Sod2 expression is often increased during metastatic progression, potentially as an adaptation to enhanced levels of intra- and extracellular ROS (7). While Sod2 may initially prevent reactive oxygen species (ROS)-mediated DNA damage to facilitate tumor initiation, increased Sod2 expression appears to conversely contribute to the metastatic phenotype by altering redox-signaling pathways (8). We and others have observed that increased Sod2 expression correlates with a shift to higher intracellular H₂O₂ levels, contributing to pro-metastatic behavior (7,9,10).

In the present study we set out to interrogate the role of Sod2 in OCCC and found a dual function for this mitochondrial antioxidant in OCCC tumorigenicity. Sod2 not only provides a protective role in scavenging mitochondrial O₂^•−, thereby maintaining high mitochondrial function and proliferation, but also alters the steady-state ROS balance to drive H₂O₂-mediated migration and metastasis of OCCC cells.

Materials and Methods

Oncomine Data and Ovarian cancer cell line microarray data

Oncomine.org was used to screen Sod2 expression in ovarian cancer histological subtypes (Supplementary Fig. S1). Two representative data sets are shown in Fig. 1A & B (GEO Accession no. GSE2109 & GSE6008). Microarray data of the following ovarian cancer cell lines was obtained using the GeneChip Human Genome U133A 2.0 Array (affymetrix; GEO Accession no.: GSE25428) (4,11). Data represents expression of Sod2 probe 215223_s_at (log2 RMA normalized). OCCC: JHOC-5, JHOC-7, JHOC-8, JHOC-9, KOC-5C, KOC-7C, OVISE, OVTOKO, RMG-1, RMG-2, RMG-5, TAYA, TOV-21-G. Serous adenocarcinoma: CAOV3, Fuov1, HEY, Hey-A8, Hey-Ce, JHOS-2, JHOS-3, JHOS-4, M41, M41-cisR, OV90, OVARY1847, OVCA420, OVCA429, OVCA432, OVCAR3, PEO1, PEO4, SKOV3. Mucinous: JHOM-1, JHOM-2B, MCAS, OMC-3. Endometrioid: OVK-18, TOV-112D. Adenocarcinoma: A2780 (A2780-J), A2780J-cisR, DOV13, OVCAR2, OVCAR5, OVCAR8. Teratocarcinoma: CH1, PA1. Undifferentiated: TYK-nu, TYK-nu cisR. Prior to microarray analysis, cell lines were authenticated by STR analysis at the Fragment Analysis Facility, Johns Hopkins University (PowerPlex 1.2 System; Promega) or at the University of Colorado Cancer Center (AmpFℓSTR Identifier Plus PCR Kit, Applied Biosystems) (11).

Figure 1.

Increased Sod2 mRNA expression is observed in ovarian clear cell carcinomas (OCCC) compared to other ovarian carcinoma histological subtypes. Two representative data sets from ovarian cancer micro-array studies are displayed as box and whisker plots (data obtained through oncomine.org). A. Clear cell (n=16), Serous (n=24), Mucinous (n=8), Endometrioid (n=29), Serous Surface Papillary (n=107); One-way ANOVA (p<0.0002), Tukey’s post-test **** p<0.0001, ***p<0.001, **p<0.01 (GSE2109). B. Clear cell (n=8), Serous (n=41), Mucinous (n=13), Endometrioid (n=37),; One-way ANOVA (p<0.0001), Tukey’s post-test ***p<0.001 (GSE6008). C, Sod2 mRNA expression was significantly increased in a panel of OCCC lines, listed in Methods section, compared to serous adenocarcinoma cell lines (Affymetrix array; ANOVA Tukey’s post-test *p<0.05)

D, Semi-quantitative real-time RT-PCR was performed to assess Sod2 mRNA expression in serous ovarian cancer cell lines OVCA433, OVCA429 and OVCAR3 and OCCC cell lines ES-2 and TOV-21-G (data expressed relative to OVCA433, which displayed lowest Sod2 expression). E, Immunoblot analysis and densitometric quantification of Sod2 protein expression in ovarian cancer cell lines (For D & E: Mean ± SEM, n=3, ANOVA, Tukey’s post-test **** p<0.0001, ***p<0.001, **p<0.01, * p<0.05).

Cell lines and cell culture conditions

At commencement of this study ES-2 and TOV-21-G cells were newly obtained from American Type Culture Collection (ATCC, Manassas, VA). Authenticity was verified by ATCC using STR analysis. ES-2 cells were maintained in McCoy’s 5A media + 10% FBS and TOV-21-G cells in 40% Media199/40% MCBD supplemented with 20% FBS and sodium bicarbonate. Cells were maintained at 37°C with 5% CO₂.

Sod2 knockdown using RNA interference

Scramble non-targeting control and Sod2-specific siRNA oligonuecliotides were synthesized by Life Technologies/Dharmacon. 5’- CAACAGGCCUUAUUCCACU-3’ and 5’- AAGUAAACCACGAUCGUUA-3’ sequences were used as siSod2_#1 and siSod2_#2 respectively (Supplementary Fig.S2) and 10pmol transfected into cells using lipofectamine RNAiMax (Invitrogen). Short hairpin RNA (shRNA) with non-targeting scramble sequence or targeting Sod2 (shSod2_#1: 5’-CTGACGGCTGCATCTGTTGGTGTCCAAGG-3’, and shSod2_#2: 5’-ACCTGAACGTCACCGAGGAGAAGTACCAG-3’) in pGFP-V-RS vector (Origene; TG309190) were used to stably transfect ES-2 cells (Fig. 2; Supplementary Fig.S3). The clone expressing shSod2_#1 was used in Figures 2–7.

Figure 2.

Inhibition of Sod2 expression abrogates OCCC tumorigenicity A, Sod2 expression and activity were reduced using RNA interference in OCCC cell lines ES-2 (shSod2_#1) and TOV-21-G (siRNASod2_#1). Sod2 expression was analyzed using western blotting (upper panels) and activity determined using in-gel zymography (lower panel). B, Cell proliferation rate was decreased with reduced Sod2 expression (shSod2_#1), analyzed by cell counting using trypan blue exclusion assay (mean ± SEM. ANOVA Tukey’s post-test; ****p<0.0001). C, Sod2 RNA interference decreases single cell survival of ES-2 cells in clonogenicity assays. D, Reduced Sod2 levels significantly inhibited the tumor size and weight of ES-2 tumors grown in the ex ovo CAM model. Fertilized chicken eggs were incubated at 37 °C for 7 days after inoculating the cells (shSod2_#1 or scramble transfected ES-2) onto the chorioallantoic membrane, after which tumors were collected and measured (control, n=12; shSod2, n=10; mean ± SD. Student t-test; **p < 0.01).

Figure 7.

Exogenous H₂O₂ treatment significantly decreases cell viability of cells with higher steady state H₂O₂ levels A, ES-2 scramble control cells show significantly lower cell viability than shSod2 ES-2 cells using crystal violet assays, in response to increasing concentrations of H₂O₂ for 72 h. Each data point represents an average of 6 replicates ± SEM. Tukey’s post-test *p<0.05. B, Proposed mechanisms of Sod2-mediated OCCC tumorigenesis and metastasis. OCCC has high Sod2 expression, which provides efficient superoxide scavenging and maintenance of high mitochondrial function to drive increased cell proliferation. In addition, Sod2 shifts the intracellular ROS balance from O₂^·− to H₂O₂, which drives tumor cell migration and metastasis. This concomitantly enhances the intracellular H₂O₂ threshold enabling more rapid killing by exogenous H₂O₂.

Immunoblotting

Protein expression was analyzed by standard western blotting using antibodies Antibodies were from Cell Signaling Technology (Boston, MA; pAkt-s473, Akt, pFAK-Y397, FAK, p-p130cas-Y165, p130cas) or Abcam (Cambridge, MA; Sod2). Primary Antibodies were diluted in blocking solution (5% nonfat milk in TBS with 0.1% tween 20, 1:1000), and incubated overnight at 4 °C. Blots were visualized using Femto and Pico ECL chemiluminescence substrate (Thermo scientific, Rockford, IL) and imaged using a ChemiDoc MP system (BioRad). Densitometric analysis was performed using ImageJ software (NIH). Each protein band was normalized to the respective GAPDH or β-Actin loading control band.

Sod2 Zymography

Sod2 activity was analyzed using Sod2 in-gel zymography as previously described (12). Briefly, cell lysates were loaded on non-denaturing acrylamide gels, followed by electrophoresis. Sod2 activity is visualized by the inhibition of nitroblue tetrazolium reduction.

Clonogenicity & cell viability

Single cell survival clonogenicity assays were performed as previously described (13). Briefly, 100 cells were plated in each well of a 6 well plate colonies visualized after 10 days using crystal violet. Viability was assessed by cell counting using trypan blue (1%) staining or crystal violet uptake assays (13).

Chorioallantoic Membrane Assay (CAM)

Each CAM was inoculated with 5× 10⁵ ES-2 cells stably expressing either scramble-shRNA-GFP or Sod2-shRNA_#1-GFP that were suspended in 50 µl PBS (with 1 mM MgCl₂, 0.5 mM CaCl₂, 100 U/mL penicillin, and 100 µg/mL streptomycin), essentially as previously described (14). Tumors were allowed to form for 7 days prior to termination of the experiments by sacrificing the chick embryo. Tumors on the CAM were removed and measured. Chorioallantoic membrane and chick embryo organs (liver and lung) were collected for tumor metastasis analysis by surveying for GFP-labeled cells.

Seahorse XF24 extracellular flux analysis

Oxygen consumption rate (OCR), extracellular acidification rate (ECAR) and mitochondria stress tests were measured using the Seahorse XF24₃ Extracellular Flux Analyzer (Seahorse Bioscience; Billerica, MA), as described previously (13). Cells were plated at a density of 40,000 cells/well and media replaced with XF media the following day 1hr prior to the assay. Three measurements of OCR and ECAR were taken at baseline and after each injection of the following mitochondrial stress test compounds: oligomycin (1µM; complex V inhibitor); FCCP (0.75µM; proton gradient uncoupler); antimycin A (1µM; complex III inhibitor). Basal and maximal respiration were normalized by subtracting non-mitochondrial OCR (i.e. after Antimycin A addition). Respiratory reserve capacity was calculated as the difference between maximal and basal OCR. ATP-linked OCR was derived as the difference between basal and Oligomycin A inhibited OCR. Data was normalized to total protein content in each well.

Wound healing assay

Cell migration was assessed in serum free media by wound healing assays using Ibidi inserts (Martisried, Germany) and quantified after 72hrs. Ibdi inserts were removed from a monolayer of GFP-labeled cells to expose the cell-free wound area. Fluorescence images were taken after 72 hrs of migration and overlayed with corresponding images at time 0hr. Pixels representing GFP-labeled cells were quantified within the wound area using Image J and corrected by subtracting any GFP-detected cells in the same area at time 0.

Spheroid attachment assay

Cells were plated at a density of 1000 cells per well in ultra-low attachment 96 well plates (Corning) and incubated for 5 days. Spheroids were transferred to 24 well plates with or without Collagen I coating. Percentage outgrowth was calculated by subtracting the area covered by migrating cells onto the collagen matrix from the area of the spheroid at time 0h.

Live cell imaging of MitoSox oxidation

As an indicator of mitochondrial O₂^•− the oxidation and fluorescence of the redox-sensitive MitoSox Red dye (Life Technologies) was monitored by live cell imaging according to manufacture’s instructions. Cells were imaged to detect oxidation and fluorescence of MitoSox using a Leica SP5 II AOBS confocal microscope following incubation with dye in HBSS for 30 min at 37°C. The Manganese Porphyrin O₂^•− scavenger ortho tetrakis(N-n-butoxyethylpyridinium-2-yl) porphyrin (MnTnBuOE-2-PyP5+) was generously provided by Dr. Ines Batinic Haberle (Duke University).

Measurement of cellular H₂O₂ using catalase activity assay

Concentration of cellular H₂O₂ was determined by measuring the rate of inactivation of catalase with amino1,2,4-triazol (ATZ), according to Yusa et al. (15). ATZ irreversibly inactivates catalase by covalently binding with intermediate compound I, formed following oxidation of catalase by H₂O₂. Briefly, cells were treated with 20 mM ATZ for different time intervals (15 and 30 min). Cells were washed with PBS and protein lysates were prepared in 50 mM phosphate buffer (pH=7.4) with protease inhibitors. Decomposition of H₂O₂ by catalase in protein lysates was analyzed using ultraviolet spectroscopy at 240nm wavelength. H₂O₂ concentration was determined using the equation [H₂O₂] = k/k1, where k is the empirically-determined pseudo first order rate constant of catalase inactivation in the cells by ATZ (Figure 6A), while k1 is the rate of compound I formation (1.7 × 10⁷ M⁻¹ s⁻¹).

Figure 6.

H₂O₂ contributes to OCCC migration. A, Steady-state H₂O₂ levels decrease with reduced Sod2 expression. Steady-state H₂O₂ levels were derived in ES-2 cells stably transfected with either scramble shRNA Control (closed circles) or shRNA targeting Sod2 (open circles), using aminotriazole inhibition of catalase kinetics assays, as described in methods. First order decay curves for catalase inhibition by aminotrizaole were used to determine rate constants of inactivation (k; one representative decay curve from 3 replicate experiments shown). H₂O₂ concentration were determined using the equation [H₂O₂] = k/k1, where k1 is the rate of catalase compound I formation (1.7 × 10⁷ M⁻¹ s⁻¹). H₂O₂ data represents an average of three experiments ± SD. B. There was no change in catalase activity and catalase protein expression between control and shRNA-Sod2 cells. C, H₂O₂ scavenging by catalase abrogates migration of ES-2 cells, while H₂O₂ can stimulate migration in cells with low Sod2 expression. Migration was analyzed by wound healing assays using ibidi inserts and the area covered by cells quantified following 72 hrs of migration (normalized to day 0). Scavenging of endogenous H₂O₂ by catalase significantly reduces the migration of ES-2 scramble-transfected control cells. Recombinant catalase protein (500 units/mL) was added in serum free media at time 0 and left on cells for 72hrs. Conversely, migration of shSod2 cells was significantly increased with treatment of 5 µM H₂O₂, while this level of H₂O₂ was toxic to control cells, and effects on migration could therefore not be determined (N/D). Data represents one of two independent cellular migration experiments (n=5, mean ± SEM. ANOVA Tukey’s post-test *p<0.05). D. H₂O₂ can stimulate Akt signaling in OCCC. Dose dependent increases in phospho-Akt S473 levels were observed by western blotting following 2hrs of H₂O₂ treatment with indicated doses in serum free media.

Statistical analysis

All data presented are representative of at least three independent experiments and expressed as mean ± SEM, unless otherwise stated. Statistical data analysis (ANOVA with Tukey’s post-test or student t-test) was performed using GraphPad Prism Software v6. p < 0.05 was considered to be significant.

Results

Sod2 expression is significantly increased in OCCC compared to other EOC histological subtypes

Mining publicly available expression data (Oncomine.org) revealed that Sod2 mRNA levels are elevated in OCCC compared to other ovarian cancer histological subtypes (Supplementary Fig. S1). Two representative data sets displayed in Figure 1A and B show that levels of this antioxidant enzyme are statistically higher than in any other ovarian cancer subtype. Similarly, significantly higher Sod2 mRNA expression was observed by microarray analysis in a panel of OCCC cell lines when compared to cell lines of high grade serous adenocaricnoma origin (Fig. 1C). This was further verified by assessing Sod2 expression in cultured ovarian cancer cell lines, using semi quantitative real-time RT-PCR and western blotting (Fig. 1D and E).

Reduced Sod2 levels significantly inhibit OCCC cell proliferation in vitro and tumor formation in the CAM model

To further investigate the role of Sod2 we utilized two OCCC cell lines, ES-2 and TOV-21-G. Sod2 expression was inhibited by shRNA and siRNA transfection, which was demonstrated to lead to a concomitant decrease in Sod2 enzyme activity (Fig. 2A, Supplementary Fig. S2 & S3A). Following Sod2 expression knockdown, ES-2 cell proliferation rate (Fig. 2B, Supplementary Fig. S3B) and clonogenicity (Fig. 2C, Supplemental Fig. 3SC) were significantly attenuated. This appeared to be Sod2 concentration dependent. A 30% reduction in Sod2 levels mediated by stable shRNA transfection reduced clonogenicity by approximately 50% (shSod2_#1), while a 70% Sod2 decrease almost completely abrogated the cells ability to survive in this assay (shSod2_#2; Supplemental Fig. S3C). Analysis of PARP cleavage and Annexin V staining suggested that the decrease in cell viability observed in cells with 30% Sod2 knock-down is not related to a significant increase in apoptosis (Supplemental Fig. S4). This cell line, referred to as shSod2 in subsequent Figures, was chosen for subsequent studies to achieve pathophysiologically relevant changes in Sod2 expression (rather than complete loss), which also closely reflects Sod2 levels observed in non-OCCC cell lines OVCA433 and OVCA429 (Fig. 1E). The CAM ex ovo model was used to further test the role of Sod2 on ES-2 tumorigenicity. A significant decrease in tumor size and weight was observed in tumors grown from ES-2 cells with reduced Sod2 expression (shSod2; Fig. 2D). In addition, the shRNA-Sod2 tumors exhibited less vascularization compared to the control groups (Fig. 2D), suggesting that Sod2 may contribute to both proliferation and the recruitment of blood vessels to the tumor.

Sod2 maintains OCCC mitochondrial respiration

We previously demonstrated that OCCC cell lines are highly energetic, and depend on both oxidative phosphorylation and glycolysis for their energy needs (13). Given that Sod2 has a primary role in protecting mitochondria from excess O₂^•−, the effect of Sod2 knock-down on mitochondrial respiration was assessed. Oxygen consumption rate (OCR), representing mitochondrial oxidative phosphorylation, and extracellular acidification rate (ECAR), correlating with glycolytic activity, was measured using extracellular flux analysis in both ES-2 and TOV-21-G OCCC cell lines following Sod2 knock-down (13,16). Stable shRNA-Sod2 decreases in ES-2 and transient siRNA-mediated knock-down of Sod2 in TOV-21-G cells resulted in significant reduction in basal OCR compared to control scramble RNAi transfected cells (Fig. 3A–D, Supplementary Fig. S5A). Further, respiratory reserve capacity, a measure of the cells’ ability to enhance respiration in response to physiological cues and stress, was significantly inhibited with reduced Sod2 levels (Fig. 3B & D), suggesting that Sod2 plays a major role in maintaining mitochondrial health to support maximal respiration. Although slight, a consistent negative effect on OCR was observed in stable shRNA-Sod2 cells following FCCP treatment (Fig. 3B). A wide range of FCCP concentrations were tested on these cells, but none were able to enhance OCR with Sod2 loss. FCCP can be inhibitory at high concentrations and it has been speculated that this may be due to a loss in the ability of mitochondria to accumulate respiratory substrates (16). While not tested here, it is possible that sustained Sod2 expression decreases, and concomitant increases in mitochondrial O₂^•− levels, may exacerbate this FCCP-depended OCR inhibition, thereby influencing mitochondrial membrane integrity and substrate transport. No significant increases in ECAR were observed between control and sh- or si-RNA transfected cells, suggesting that decreases in Sod2 expression do not influence a compensatory shift towards glycolysis (Supplementary Fig. S5B–D). The above observations show that both si- and shRNA mediated decreases in Sod2 reduced basal OCR and respiratory reserve capacity, indicating that Sod2 is important in maintaining mitochondrial respiration in OCCC.

Figure 3.

Sod2 preserves OCCC mitochondrial respiration A, Reduced Sod2 levels significantly attenuated the mitochondrial oxidative phosphorylation in stably transfected ES-2 cells (shSod2_#1). Oxygen Consumption Rate (OCR) was measured using extracellular flux analysis and pharmacological manipulation of mitochondrial activity used to interrogate bioenergetics parameters. Oligomycin A (O) was used to derive ATP-linked OCR, FCCP (F) to stimulate maximal OCR, and Antimycin A (A) to inhibit all mitochondrial OCR. B, Basal oxygen consumption rate (OCR), ATP linked OCR, maximum OCR and respiratory reserve capacity (Max OCR- basal OCR), were significantly decreased with stable shRNA-mediated Sod2 knock down in ES-2 (one representative of three experiments is shown, mean ± SEM, control n=8, shSod2 n=9. ANOVA, Tukey’s post-test ***p<0.001, **** p<0.0001). C, Transiently transfected siRNA targeting Sod2 in TOV-21-G cells showed significantly reduced mitochondrial oxidative phosphorylation (siRNA construct #1). D, Basal OCR, maximum OCR and respiratory reserve capacity (Max OCR- basal OCR), were significantly decreased with Sod2 knock down in TOV-21-G cells. (one representative of three experiments is shown, mean ± SEM, control n=4, siSod2 n=5, ANOVA Tukey’s post-test *p<0.05). E, Superoxide-mediated oxidation of MitoSox redox-sensitive dye was measured in live cells. Increased MitoSox fluorescence was observed in ES-2 cells stably transfected with shRNA-Sod2_#1 compared to scramble-transected control cells, suggesting that the amount of mitochondrial superoxide accumulation is higher in cells with reduced Sod2 expression. MitoSox florescence was abrogated by addition of 10 µM of the porphyrin superoxide scavenger ortho tetrakis(N-n-butoxyethylpyridinium-2-yl) porphyrin (MnTnBuOE-2-PyP5+) (a representative of three experiments is shown, mean ± SEM, n=4, ANOVA, ANOVA Tukey’s post-test *p<0.05, **p < 0.01, ***p < 0.001).

To assess whether a decrease in Sod2 expression result in compromised O₂^•− scavenging, which may be one of the causes of compromised mitochondrial function, the presence of mitochondrial O₂^•− was evaluated using the mitochondria-targeted redox-sensitive dye MitoSox. As expected, increased oxidation and consequential enhanced fluorescence of MitoSox were observed in the Sod2 knockdown cells compared to controls (Fig. 3E & F). Addition of the Sod2-mimetic porphyrin (MnTnBuOE-2-PyP5+), which acts as a O₂^•− scavenger, reduced MitoSox oxidation and in both control and shRNA-Sod2 groups (Fig. 3E & F).

Sod2 knockdown attenuates metastasis of cancer cells in the CAM model

We have previously demonstrated that enhanced Sod2 expression is implicated with metastatic progression (7,8,17). To investigate the role of Sod2 during OCCC metastasis, the appearance of metastatic lesions of GFP-labeled ES-2 cells were investigated in the CAM tumor model. Single cells and micrometer-sized cellular clusters were highly abundant throughout the membrane in the control group, which could be observed 2–3 cm from the tumor (Fig. 4A). In contrast, metastatic spread from shRNA-Sod2 knockdown tumors was limited to the appearance of single cells in the membrane confined to an approximate 1–1.5 cm radius from the tumor (Fig. 4A). Further, lung metastases in the chick embryo were observed in 11 of the 12 controls compared to only 5 of 10 embryo’s in the shRNA-Sod2 group (Fig. 4B). Further, 10 of 12 control tumors metastasized into the liver, whereas only 4 of10 liver metastases were observed in the shRNA-Sod2 group (Fig. 4B). While clusters of five or more cells were found in the lungs and livers of control groups, only single cells were detected in the Sod2 knockdown group (Fig. 4B).

Figure 4.

Metastatic spread of ES-2 cells is attenuated with reduced Sod2 expression in the ex ovo CAM tumor model. A, Representative images show the location of the primary tumors and micro-metastasis of GFP-labeled cells in the membrane (scale bar = 1mm, CAM tumor study carried out as in Fig. 2). B, Reduced Sod2 expression inhibited tumor metastasis into the chick embryo and CAM (shSod2_#1). Quantification of percentage metastasis of tumor cells into CAM, liver and lung of chick embryo. Representative images show the percent of tumors with metastatic cancer cells present in the chicken embryo liver and lung (control, n=12; shSod2, n=10).

Sod2 levels modulate cell migration and tumor spheroid outgrowth

Due to the significant abrogation of metastatic spread in response to Sod2 expression decreases, the role of Sod2 on cell migration was further investigated. Cell migration, assessed by wound healing assays, was significantly inhibited with reduced Sod2 expression in ES-2 cells (Fig. 5A; Supplementary Fig. S6). Further, the ability of cellular spheroid clusters to attach and cells to migrate from the spheroid onto collagen-I and un-coated surfaces was also compromised with reduced Sod2 expression (Fig 5B). Anchorage independent spheroid formation is a commonly observed phenotype of ovarian cancer cells metastasizing via the transcoelomic route through the IP cavity and these have been shown the ability to attach on the peritoneum to form metastatic lesions. These data suggest that Sod2 plays an important role in tumor spheroid metastasis (Fig. 5B). While spheroids of equal size were chosen for this assay, it should be noted that Sod2 knock-down also decreased ES-2 spheroid growth in anchorage independence (data not shown).

Figure 5.

A, Cell migration is significantly inhibited with Sod2 knockdown. A. Migration was analyzed by wound healing assays using ibidi inserts and the area covered by cells quantified following 72 hrs of migration (normalized to day 0). Data from one representative experiment is shown (n=6, mean ± SEM, t-test; **p < 0.01, scale bar = 500µm). B, Representative images of attachment and outgrowth of ES-2 spheroids on uncoated and collagen-I coated surfaces (scale bar = 1mm). Reduced Sod2 levels significantly inhibited the outgrowth of spheroids in both uncoated and collagen-I coated surfaces. Data from one representative experiment is shown (n=4, mean ± SEM, ANOVA, Tukey’s post-test; **p < 0.01). C. Akt-phosphorylation is dependent on Sod2 expression. Protein expression of phosphorylated and total levels of Akt, p130cas, and FAK were analyzed by western blotting. Data from one representative experiment is shown. Levels of phosphorylated proteins were quantified by densitometry and normalized to total levels (n=3, mean ± SEM).

To gain mechanistic insights into the signaling pathways that may be altered by Sod2-mediated metastasis, phosphorylation profiles of Akt, p130cas and focal adhesion Kinase (FAK) were investigated. These were chosen based on previous observations of their redox regulation and involvement in tumor cell migration (8,9,18,19). shRNA-Sod2 cells exhibited a 50% decrease in phospho-Akt levels compared to scramble control cells, while there was no appreciable change observed in phosphorylation of FAK or the focal adhesion adapter protein p130cas (Fig. 5C). The effects on Akt phosphorylation were also Sod2 concentration dependent, where cells with lower Sod2 expression demonstrated a more striking decrease in Akt phosphorylation (Supplementary Fig S6). These data suggest that Akt signaling may be important in driving Sod2-mediated tumrigenicity and metastasis of OCCC.

OCCC cell migration is H₂O₂ dependent

Sod2 is the primary enzyme involved in converting O₂^•− to H₂O₂ within the mitochondria. While it serves as a protective mechanism to maintain mitochondrial function (Fig. 3) by removal of damaging O₂^•−, a shift towards increasing levels of H₂O₂ has also been observed in response to enhanced Sod2 expression (8,10,18,20,21). Due to its relative stability and ease in traversing cellular membranes, H₂O₂ can mediate redox signaling, including events that drive migration (9). To test if Sod2 changes the steady-state H₂O₂ levels in OCCC, we assessed intracellular H₂O₂ status in control and shRNA-Sod2 ES-2 cells using a biochemical assay based on the irreversible inhibition of catalase by aminotriazole (15,17). Steady-state levels of H₂O₂ were reduced approximately 2-fold in Sod2 knockdown cells compared to controls (Fig. 6A), while baseline catalase activity and protein expression were comparable (Fig. 6B).

In order to test whether OCCC cell migration is H₂O₂ depended, wound-healing assays were carried in the presence of catalase, which catalyzes the conversion of H₂O₂ to H₂O and O₂. As previously demonstrated, exogenous application of recombinant catalase resulted in accumulation of catalase within ES-2 cells (Supplementary Figure S7) (9). Catalase significantly reduced the migration of both control and shRNA-Sod2 cells, suggesting that H₂O₂ is a promoter of ES-2 cell migration (Fig. 6C). Conversely, treatment with low levels of H₂O₂ (5 µM) significantly reversed the slow migration of shRNA-Sod2 cells (Fig. 6C). Further, 5 and 50 µM H₂O₂ treatment was able to increase Phospho-Akt levels (Fig. 6D), while catalase expression abrogated Akt phosphorylation (Supplementary Fig. S8), suggesting that this may be an important redox-dependent signaling pathway in OCCC.

It was noted that ES-2 control cells were not able to tolerate long-term exposure to low dose H₂O₂ during migration assays (Fig. 6C). To examine this further, cell viability was assessed in response to H₂O₂. A significant reduction in cell survival in response to H₂O₂ was observed in the control group compared to cells with decreased Sod2 expression (Fig. 7A). This suggests that a higher intracellular steady-state H₂O₂ milieu in OCCC predisposes cells to enhanced killing by additional exposure to low level exogenous H₂O₂. The above data imply that high Sod2 expression provides several advantages to OCCC, by protecting mitochondrial function through scavenging of O₂^•− and driving H₂O₂ dependent migration. While these attributes are advantageous for OCCC survival and metastatic progression, an enhanced intracellular steady-state H₂O₂ level presents a double-edged sword, as these cells are consequentially more susceptible to H₂O₂ toxicity (Fig. 7B).

Discussion

Although the five different EOC histological subtypes share the same primary tumor location on the ovaries, it is now evident that these are distinct diseases with vastly different tissue origins, and genetic and epigenetic profiles (2,11). In the present study, we show that Sod2 is highly expressed in OCCC compared to other EOC histological subtypes, and that this mitochondrial antioxidant plays a significant role in OCCC tumorigenicity and metastasis.

Intracellular ROS are maintained within a narrow range, tightly regulated by the balance of the rate of ROS production and ROS scavenging/detoxifying by antioxidant enzymes. This balance is often disrupted in the context of cancer, due to high ROS production as a consequence of changes in metabolism or the tumor environment (e.g. hypoxia), and the resulting changes in antioxidant expression. Since the mitochondrial respiratory chain is the major site of O₂^•− generation within cells, Sod2 plays an important role in maintaining cellular ROS balance. Based on the above findings Sod2 appears to play a dual role in enhancing OCCC tumorigenicity; first by protecting cells from mitochondrial O₂^•− damage; and second, by shifting the steady-state ROS balance towards H₂O₂.

We recently demonstrated that a distinguishing feature of OCCC is their unique metabolic phenotype. Compared to serous adenocarcinoma cells, OCCC cell lines were significantly more energetic, displaying both very high levels of mitochondrial oxidative phosphorylation and glycolytic flux (13). Our data suggest that Sod2 is intricately involved in maintaining this high rate of oxygen consumption, potentially by preserving mitochondrial function as a consequence of O₂^•− scavenging (Fig. 3). By preventing damage mediated by O₂^•− or secondary products of mitochondrial electron transport chain complexes, Sod2 likely supports the high rate of OCCC proliferation, clonogenicity and tumor growth. The present study suggests that inhibiting mitochondrial antioxidant defenses may provide and alternate strategy to therapeutically target OCCC.

In addition to scavenging O₂^•− and maintaining mitochondrial health for optimal cell proliferation, we believe that Sod2 has an additional role in promoting the aggressiveness of OCCC, by shifting steady-state H₂O₂ levels and driving pro-metastatic behavior (Fig. 4–6). It has been previously shown that high expression of Sod2 is associated with metastatic progression (8,10,17,20–23), and that dependence of cancer cell migration is related to cellular H₂O₂ production (9,19,24,25). For example, we have shown that steady-state increases in H₂O₂ can lead to induction of the focal adhesion kinase pathway and migration of metastatic bladder cancer cells and cells with enforced Sod2 expression (9,19). This was mediated by oxidation-dependent inhibition of the phosphatase PTPN12, leading to enhanced phosphorylation of p130cas and Rac1 activation. In addition, work from the Melendez’ group has shown that Sod2 expression significantly contributes to the expression of the matrix-degrading enzyme MMP-1 in a H₂O₂-dependent manner (21), and that the Sod2/H₂O₂-dependent inhibition of the dual lipid protein tyrosine phosphatase PTEN enhances Akt/GSK3β/VEGF-dependent angiogenesis (18), both processes contributing significantly to metastasis. Our present data suggest that Sod2 may similarly contribute to metastatic progression of OCCC by activating Akt signaling (Fig 5 and 6). In addition to its pro-survival function, Akt has been shown to influence metastasis and cell migration by regulating cytoskeletal rearrangement, pro-metastatic cell signaling and gene transcription (26). These results are of specific importance to OCCC, which, unlike other ovarian cancer histological subtypes, has been characterized by high frequency Akt pathway activation. 70% of early and 68% of late stage OCCC cases have been shown to display phospho-Akt (S473) staining (27). About 38% of OCCC cases show PTEN loss (28) and 40% of cases PI3K activating mutations (29). Our data imply that Sod2-dependent Akt phosphorylation may also contribute to high activation of Akt signaling in OCCC. Since Akt-phosphorylation was highly susceptible to H₂O₂ treatment, it suggests that this signaling pathway is redox regulated in OCCC, with a plausible mechanism for this being the oxidation of PTEN (18).

Although an increase in Sod should theoretically not result in higher levels of H₂O₂ production based on the kinetic properties of Sod2 (30), a number of studies have demonstrated increases in H₂O₂ levels that correlate with Sod2 expression (18,31–33). While the reason for this observation in OCCC has not been investigated, there are plausible explanations for this increase in steady-state H₂O₂ as a consequence of Sod expression. These primarily relate to changes in the reaction rates within the mitochondrial electron transport chain (ETC). For example, it has been proposed that Sod2 in the mitochondria may alter the flux of O₂^·− from some quinone/semiquinone/hydroquinone triads, such as coenzyme Q, thereby driving the reaction into the direction of O₂^·− production, potentially leading to enhanced localized dismutation to H₂O₂ by Sod2 (31,34). Alternatively, inhibition of cytochrome c oxidase by nitric oxide, arising as a consequence of Sod2 expression may influence the reduction state of the ETC and drive O₂^·− and H₂O₂ production (35). Our observation that Sod2 knock-down also decreases H₂O₂ levels, suggests that Sod2 is involved in regulating H₂O₂ balance within cells and this may contribute to H₂O₂-mediated redox signaling.

While Sod2 appears to contribute to H₂O₂-mediated metastatic progression, an enhanced steady-state H₂O₂ milieu may also present a disadvantage to OCCC cells. Our data suggest that cells with high Sod2 levels and concomitant increases in intracellular H₂O₂ are more susceptible to exogenous sources of redox stress (Fig. 7A). This likely puts cells closer to the cytotoxic threshold of H₂O₂, which is reached once cells are further challenged by exogenous ROS. Interestingly, OCCC do not appear to have enhanced expression of Catalase to provide additional scavenging of excess H₂O₂ (Fig. 6A). While sub-lethal levels of H₂O₂ have been shown to contribute to redox signaling, high levels of H₂O₂ can elicit tumor cell death by a number of pathways, including apoptosis, protein/DNA damage and mitochondrial dysfunction (36–38). Further, it was recently reported that H₂O₂ exposure of tumor cells with enforced Sod2 expression can result in Sod2 peroxidase activity, leading to mitochondrial damage and dysfunction (39). An increased H₂O₂ steady-state has been observed in a number of cancer cells (17,40–42) and lends credence to the idea that this higher H₂O₂ threshold may be exploited therapeutically. In this regard, the use of high-dose ascorbic acid, which is oxidized within tumor cells to produce H₂O₂, has recently been revisited for use in cancer treatment (43–45) and has shown promise in early clinical trials in advanced stage cancers (46,47). Ascorbate and concomitant H₂O₂ – mediated DNA damage and apoptosis were shown to enhance ovarian cancer cell death and increased chemosensitivity (48). While that study was not focused on OCCC, this type of treatment may be of particular benefit to this histological subtype given the high expression of Sod2. It is important to highlight that cancer cells with enhanced Sod2 expression may respond differently to ROS-producing agents, depending on both the type and cellular location of the ROS/reactive nitrogen species generated. For instance, Sod2 may enhance scavenging of O₂^•−, and therefore provide chemoresistance benefits to the tumor cells in response to these ROS. Conversely, while an increase in steady-state H₂O₂ facilitates redox-signaling beneficial to the cancer cells, this higher threshold may facilitate H₂O₂-mediated OCCC cell death in response to further insult by exogenous sources of H₂O₂. Understanding the complex interaction of antioxidants and ROS in OCCC is therefore of importance and may provide novel therapeutic avenues to pursue for this histological subtype of ovarian cancer.