|Home | About | Journals | Submit | Contact Us | Français|
Cisplatin-based regimens are the standard of care for epithelial carcinoma of the ovary. Because cisplatin is known to increase intracellular levels of toxic reactive oxygen species (ROS), an increase in cisplatin toxicity selectively in cancer cells could result from further increasing the cisplatin-elevated ROS levels by targeting antioxidant genes upregulated in ovarian cancers. The serine/threonine kinase Mirk/dyrk1B is a transcriptional coactivator that increased the expression of the antioxidant genes superoxide dismutase 2 and ferroxidase in ovarian cancer cells. As a result, depletion of Mirk increased cellular ROS levels in each of 4 ovarian cancer cell lines. Mirk depletion averaged only about 4-fold, yet combined with cisplatin treatment, enabled low levels of drug to increase ROS to toxic levels in both SKOV3 and TOV21G ovarian cancer cells. Lowering ROS levels by treatment with N-acetyl cysteine limited cisplatin toxicity, resulting in higher cell numbers and decreased cleavage of the apoptotic proteins PARP and caspase 3. Mirk has also been shown to block cells in G1 by inducing proteolysis of cyclin D1. Mirk depletion increased cyclin D1 levels in 3 of 4 ovarian cancer cell lines, implying that some Mirk-depleted cells could more readily enter cycle, potentially increasing their sensitivity to cisplatin. Because Mirk is upregulated in a large subset of human ovarian cancers, but is expressed at low levels in most normal tissues, and embryonic knockout of Mirk results in viable and fertile mice, targeting Mirk may sensitize ovarian cancers to lower levels of cisplatin while sparing normal tissues.
Ovarian cancer is a frequent cause of cancer death in women. Most patients with ovarian cancer have widespread disease at presentation, with yearly mortality approximately 65% of the incidence rate. The high death rate is due to widely disseminated disease at the time of diagnosis. Ovarian cancer often spreads by local shedding into the peritoneal cavity.1 The implantation of exfoliated ovarian cancer cells on the peritoneum leads to ascites formation and local invasion of the bowel and bladder. Gynecologic Oncology Group trials support treatment with cisplatin, carboplatin, and paclitaxel. Cisplatin increases reactive oxygen species (ROS) levels in many types of cancer cells, and recently, the serine/threonine kinase Mirk was found to mediate the survival of pancreatic cancer cells by reducing ROS levels, but the effect of combining Mirk depletion with a drug known to increase ROS was not tested.2 Mirk/dyrk1B is a member of the Minibrain/dyrk family of kinases,3-5 which aid differentiation in certain normal tissues: skeletal muscle (Mirk/dyrk1B),6 neuronal cells (Dyrk1A),3,7 erythropoietic cells (Dyrk3),8,9 and sperm (Dyrk4).10 Mirk is not an essential gene because embryonic knockout of Mirk/dyrk1B caused no evident phenotype in mice,11 and normal diploid fibroblasts exhibited no alteration in survival after 20-fold depletion of Mirk.12 Thus, inhibition of Mirk might cause few deleterious effects in normal tissue. Mirk is widely expressed in both ovarian cancers and cancer cell lines and is upregulated in a large subset of cancers compared with normal ovarian tissue (Hu and Friedman, in preparation). Mirk/dyrk1B is one of 16 genes within a consistently amplified 660-kb subregion of the 19q13 amplicon found in about 10% of pancreatic cancers,13 but more frequently in ovarian cancers, about one third,14 suggesting that Mirk’s possible function in ovarian cancers would merit further investigation. Furthermore, if Mirk depletion increased ROS levels in ovarian cancer cells, Mirk depletion might sensitize ovarian cancer cells to cisplatin, a drug that was reported to increase ROS levels in other cancers. Targeted therapies for ovarian cancer that improve response to low cisplatin levels may improve patient survival.
Cisplatin increases ROS levels in various cancers, including bladder cancer,15 MCF-7 breast cancer cells,16 and Hep2-K1 human laryngeal carcinoma cells.17 Cisplatin increased ROS levels in a dose-dependent manner in SKOV3, TOV21G, OV90, and OVCAR4 ovarian cancer cell lines, as expected (data not shown). Because Mirk increased expression of a cohort of antioxidant genes in pancreatic cancer cells through its transcriptional coactivator function,2 we speculated that depletion of Mirk might increase ROS in ovarian cancers and thus increase the toxicity of cisplatin. To test this hypothesis, cisplatin was added to TOV21G cells, and the more resistant SKOV3 cells after Mirk were depleted by RNA interference (Fig. 1). Mirk depletion by either of 2 RNAi duplexes was incomplete, averaging about 3-fold in SKOV3 cells (Fig. 2A). After incubation with trypan blue dye, viable, dye-excluding SKOV3 cells were counted by microscopy. Over twice as many Mirk-depleted cells than mock-depleted cells were killed by cisplatin over the 4-day period (P = 0.0278) (Fig. 1A). Viable cell number increased 6-fold in the mock-depleted cultures and continued to increase until the end of the experiment at 96 hours, reflecting the low toxicity of the cisplatin concentration used. In contrast, the number of viable Mirk-depleted cells increased half as much, reaching a plateau after 48 hours of cisplatin treatment. A parallel study using metabolism of MTT to reflect cell number yielded similar results (Fig. 1B). The numbers of Mirk-depleted SKOV3 cells reached a plateau after 48 hours, while the mock-depleted cells continued to proliferate for 96 hours and increased almost 4-fold even in the presence of cisplatin (Fig. 1B). Thus, even partial Mirk depletion increased the toxicity of a low concentration of cisplatin.
Mirk was also a factor mediating cisplatin resistance in the TOV21G ovarian cancer cell line. Mirk was depleted 5-fold by RNA interference from TOV21G cells (Fig. 2B), which were then treated with the same concentration of cisplatin (3.3 µM) for up to 120 hours. Mirk depletion led to a statistically significant 2-fold increase in cell kill by cisplatin (Fig. 1C), with viability analyzed by dye exclusion. A time-course study using metabolism of MTT to reflect cell number confirmed that mock-depleted cells were more resistant to cisplatin than Mirk-depleted TOV21G cells, achieving on average a 40% higher total cell number (Fig. 1D). The viable cell number by the dye exclusion assay leveled off after 24 hours in cisplatin in the Mirk-depleted cells (Fig. 2C), as did the total cell number measured by the MTT assay (Fig. 1D). Thus, Mirk was one factor in mediating resistance to cisplatin in both SKOV3 and TOV21G ovarian cancer cells.
The 2- to 3-fold increase in cell kill after depleting Mirk 3- to 5-fold in SKOV3 and TOV21G cells is probably an underestimation of the potential biological effect of targeting Mirk kinase. RNA interference still leaves Mirk protein at 20% to 30% of original levels in part because Mirk expression is elevated in these ovarian cancer cells, and the kinase is relatively stable in vivo. Signaling by the Raf kinase within melanomas had to be reduced by at least 90% by the small molecule inhibitor PLX4032 to block Raf’s biological effects,18,19 suggesting that the 20% to 30% of Mirk protein left after RNA interference may maintain many Mirk functions. Inhibiting Mirk kinase activity by a selective small molecule inhibitor would be a preferable method, but no such inhibitor is commercially available.
Cancer cells often exhibit high ROS levels because of their increased metabolism or because of expression of mutant ras genes. Such cancer cells cannot tolerate much increase in their already elevated ROS levels before cell death is induced.20 For example, immortalized, nontumorigenic T72 normal ovarian epithelial cells exhibited a 2-fold increase in ROS levels when stably transfected with oncogenic H-rasV12.20 In earlier studies, Mirk depletion in SU86.86 pancreatic cancer cells led to, at most, a 2-fold increase in ROS levels, with cell death occurring within 24 hours of this rise.2 Mirk depletion induced a similar, at most, 2-fold increase in ROS in the ovarian cancer lines (Fig. 2). Mirk was depleted using either of 2 RNAi duplexes targeting different Mirk exons (Fig. 2, insets). Partial depletion of Mirk by either siD or siC led to increases in ROS levels (P < 0.001) for SKOV3 cells (Fig. 2A), for TOV21G cells (P < 0.001) (Fig. 2B), for OV90 cells (P = 0.0418) (Fig. 2C), and for OVCAR4 cells (P < 0.0001 [siD], P = 0.026 [siC]) (Fig. 2D). As a control to confirm that the increase in ROS levels following Mirk depletion was due to a loss in antioxidant activity, the general antioxidant N-acetyl-cysteine (NAC) was added to parallel cultures 6 hours before analysis (Fig. 2). Although this short treatment was insufficient to eliminate all ROS, NAC reduced both the baseline ROS levels and the increase in ROS caused by Mirk depletion.
Because elevated ROS levels led to cell death, we measured ROS levels after a short, 2-day cisplatin treatment of Mirk-depleted TOV21G and SKOV3 cells (Fig. 3 A and andC),C), then measured cell death by counting cells after 72 and 96 hours (Fig. 3 B and andD).D). Every concentration of cisplatin tested, from 2 to 10 µM, caused a greater increase in ROS in the Mirk-depleted TOV21G cells than in the mock-depleted TOV21G cells (Fig. 3A). After Mirk depletion, 3.3 µM cisplatin induced similar ROS levels to those induced by 10 µM cisplatin in cells without Mirk depletion. Thus, less cisplatin was needed to achieve high ROS levels in Mirk-depleted TOV21G cells.
To confirm that elevated ROS levels played a role in the increased cell death seen after coupling Mirk depletion with cisplatin treatment, a similar experiment was performed, but the antioxidant NAC was added to reduce ROS levels in TOV21G cells, with fresh drugs and media added daily and cultures followed for 3 and 4 days to allow the elevated ROS levels to kill cells. NAC treatment did prevent some cell kill by 3 or 4 days of cisplatin treatment as more TOV21G cells remained viable when NAC was added to Mirk-depleted cells (Fig. 3B). Thus, NAC treatment, by blocking the increase in ROS caused by Mirk depletion, reduced the amount of cell death caused by cisplatin. Also, NAC partially reversed cell loss induced by cisplatin in mock-depleted cells after both 3 and 4 days of cisplatin treatment (Ct si lanes with and without NAC) (Fig. 3B). Thus, both Mirk depletion and cisplatin kill TOV21G cells by increasing ROS levels, allowing low levels of cisplatin to induce more tumor cell death when Mirk is depleted.
Similar experiments were performed with SKOV3 cells. Cisplatin at 20 µM increased ROS levels 60% higher in Mirk-depleted cultures compared to mock-depleted cultures (Fig. 3C). At the higher level of 33 µM cisplatin, ROS levels were only about 20% higher in the Mirk-depleted cultures because the Mirk-depleted cells with the most elevated ROS levels began to die, lyzing and releasing their contents so their ROS could not be measured, resulting in a 60% decrease in cell number after another day of exposure to cisplatin (data not shown). Thus, Mirk depletion increased the capacity of cisplatin to induce cell death by further increasing ROS to toxic levels.
To confirm that elevated ROS levels played a role in the increased cell death induced by Mirk depletion in cisplatin-treated SKOV3 cultures, the antioxidant NAC was added to reduce ROS levels, and cell numbers were measured after 3 days (Fig. 3D) and showed that reduction of ROS levels by NAC led to more survival of Mirk-depleted cells. ROS levels in SKOV3 cells were raised by either Mirk depletion or by cisplatin treatment, enabling cisplatin to cause more tumor cell death when Mirk was depleted. Thus, lower cisplatin levels were needed to raise cellular ROS to toxic levels in both SKOV3 cells and TOV21G cells when Mirk was depleted.
Mirk induced transcription of a series of antioxidant genes in pancreatic cancer cells, including superoxide dismutase 2 (SOD2), SOD3, and the ferroxidase ceruloplasmin, which blocks formation of the hydroxyl ion.2 Induced depletion of Mirk by addition of doxycycline to SKOV3 cells bearing a short hairpin (shRNA) to Mirk mRNA led to decreased expression of Mirk and ferroxidase after 1 day and reduced expression of SOD2 after 2 days (Fig. 4A). Depletion of Mirk by each of 2 RNAi duplexes in OV90, OVCAR4, and OVCAR3 ovarian cancer cells also led to decreased levels of ferroxidase. Plotting the protein levels normalized to actin showed a linear relationship between protein levels of Mirk and ferroxidase (one of 2 similar experiments with r2 values >0.89) (Fig. 4B), indicating that Mirk increased expression of this antioxidant gene in ovarian cancer cells, possibly by Mirk’s known capacity as a transcriptional coactivator. Thus, Mirk controlled ROS levels in ovarian cancer cells at least in part by upregulating expression of ferroxidase, SOD2, and possibly other antioxidant genes.
Cisplatin can induce apoptosis. Mirk was depleted by either of 2 RNAi duplexes in SKOV3 cells (Fig. 5 A and andB)B) or in TOV21G cells (Fig. 5C). Cells were then treated with cisplatin for up to 4 days, and the amount of the apoptotic proteins cleaved PARP and cleaved caspase 3 were measured in cell lysates by Western blotting. Depletion of Mirk before addition of cisplatin enhanced the cleavage of apoptotic marker proteins 2- to 5-fold both in SKOV3 and TOV21G cells (Fig. 5). In addition, apoptosis occurred with a more rapid onset after Mirk depletion. PARP cleavage and caspase 3 cleavage were initiated in control-depleted SKOV3 cells after 2.5 days but could be detected 12 hours earlier in Mirk-depleted cells (Fig. 5B). Thus, measurements of cell viability by dye exclusion (Fig. 1), cell number by the MTT assay (Fig. 1), and the appearance of apoptotic proteins seen in Western blotting (Fig. 5) confirmed that Mirk depletion increased the toxicity of cisplatin to each of 2 ovarian cancer cell lines. ROS played a role in the induction of apoptosis. Reducing ROS levels by daily addition of the antioxidant N-acetylcysteine led to a reduction in the amount of cleaved PARP induced by Mirk depletion in cisplatin-treated TOV21G cells (Fig. 5C). Also, NAC reduced the basal level of cleaved PARP caused by cisplatin alone. Similarly, NAC decreased the amount of cleaved caspase 3 in both Mirk-depleted and mock-depleted cisplatin-treated TOV21G cells.
It was possible that Mirk mediated cisplatin resistance by other mechanisms in addition to lowering ROS levels. Mirk blocks the cycling of both nontransformed cells and cancer cells by destabilizing the cyclin D family of G1 cyclins by phosphorylation at a conserved ubiquitination site.21,22 Mirk also increases the length of G1 by stabilizing p27kip1 by phosphorylation at S10.23,24 In 5-fluorouracil–treated colon cancer cells, Mirk phosphorylated cyclin D1 at a conserved ubiquitination site to mediate a G1 checkpoint. This G1 checkpoint was reduced by overexpression of cyclin D1 mutated at the Mirk phosphorylation site (T288A) but not by wild-type cyclin D1, indicating that the checkpoint was mediated by destabilization of cyclin D1 following phosphorylation by Mirk.25 In SKOV3 and TOV21G, depletion of Mirk increased the amount of cyclin D1 2- to 3-fold, with a smaller effect on OVCAR4 cells and little effect on OV90 cells (Fig. 5D). Thus, in some, but not all, ovarian cancer cells, Mirk depletion would enable more cells to enter cycle by increasing cyclin D1 levels, potentially increasing the fraction of cells in S phase sensitive to the killing effect of cisplatin.
In the current study, the kinase Mirk has been shown to reduce the toxicity of cisplatin by 2 mechanisms, reducing ROS levels in ovarian cancer cells through increasing levels of ferroxidase and SOD2 and, in some cancer cell lines, by decreasing cyclin D1 levels to reduce cell cycling. These functions may provide a selective pressure to maintain the upregulation of Mirk seen in ovarian cancers and the 19q13 amplicon containing the Mirk gene found in about one third of ovarian cancers.14 Significantly, Mirk depletion increased the toxicity of low levels of cisplatin to SKOV3 and TOV21G cells (Figs. 1, ,5),5), at least in part by increasing ROS levels in both lines. Cisplatin-treated cells underwent more apoptosis if Mirk was depleted, leading to a loss of viable cells. Knockout of Mirk/dyrk1B caused no evident phenotype in mice,11 suggesting that Mirk is not an essential gene. Supporting this interpretation, normal diploid fibroblasts exhibited no alteration in survival after 20-fold depletion of Mirk.12 Thus, Mirk appears not to be an essential gene for normal cells but is upregulated in ovarian cancer cells where Mirk mediates survival by reducing ROS. The efficacy of targeting Mirk kinase to improve cisplatin-initiated ovarian cancer cell kill was suggested by the results of this study. A modest 2- to 3-fold increase in cell kill was seen possibly because the RNA interference method only achieved a mean 4-fold depletion of Mirk. A more complete inhibition of Mirk’s kinase activity, for example, by a small molecule inhibitor might further sensitize ovarian cancer cells to cisplatin.
ROS plays a role in the evolution of ovarian cancers. During ovulations, normal ovarian surface epithelial cells experience oxidative stress, which increases the potential for DNA damage and subsequent transformation. Once malignancy is achieved, the elevated ROS levels resulting from oncogene action and from increased metabolism in cancer cells must be countered if cancer cells are to survive. The platinum drugs kill ovarian cancer cells, in part, by further increasing the already high ROS levels in cancer cells to toxic levels.20 Cancer cells cope with high ROS levels when drug treated by selecting for chemoresistant variants.26 This selection lowers the efficacy of cisplatin, carboplatin, and oxaliplatin in ovarian cancer patients.27
Cisplatin and related compounds are widely used for the treatment of ovarian cancers, but efficacy is limited by acquired or endogenous drug resistance, so strategies that improve tumor cell apoptosis are being sought. Major mechanisms of resistance include decreased membrane transport of the drug, increased cytoplasmic detoxification, and increased DNA repair.27,28 Upregulation of ROS by various mechanisms, such as depletion of Mirk, may induce tumor cell death and could function in either an additive or synergistic manner with cisplatin. For example, the synthetic retinoid fenretinide has antitumor activity in ovarian cancer cells because it induces ROS, which in turn leads to endoplasmic reticulum stress, JNK activation, and eventually apoptosis.29 ROS effects on tumor progression were also shown by tumor-associated mesenchymal cells. Oxidative stress in peritoneal mesothelial cells due to elevated ROS levels was found to promote ovarian cancer cell adhesion, facilitating binding and dissemination of ovarian cancer cells.30 Ovarian cancer cells also combat ROS by increasing intracellular levels of antioxidants, such as glutathione, the major intracellular thiol,31 so lowering levels of antioxidant proteins in ovarian cancers might be of therapeutic benefit.32
All ovarian adenocarcinoma cell lines were obtained from the ATCC in 2007 and maintained in DMEM supplemented with 7% heat-inactivated fetal bovine serum. All lines were screened for mycoplasma (Sigma-Aldrich, St. Louis, MO) in November 2009 and found to have remained negative. Affinity-purified polyclonal antibody to a region within the Mirk-unique C-terminus was described previously.34 Antibodies to ferroxidase/ceruloplasmin and to superoxide dismutase 2 were from Santa Cruz Biotechnology (Santa Cruz, CA). Cisplatin and N-acetylcysteine were from Sigma-Aldrich.
The following sequences in the Mirk mRNA were used as templates for synthetic RNAi duplexes: siA: GTGGTGAAAGCCTATGATCAT, siC: GCCTGGTATTTGAGCTGCTGTCCTA, siD: GAAGAAGGTCCTGAACCATGGTTAT. These and GC-matched control RNAi duplexes were transfected into cells with Lipofectamine 2000 (Invitrogen, Carlsbad, CA) in serum-containing growth medium as described.2 A lentiviral construct containing a doxycycline-inducible shRNA to the Mirk mRNA sequence starting at bp5302 was used to infect SKOV3 cells.
Cells were trypsinized and resuspended at 2 × 105/mL in 5 µM CM-H2DCFA, made from a fresh stock at 10 mM in dimethylformamide. After 60 minutes at 37°C, cells were resuspended in fresh DMEM, incubated 60 minutes at 37°C, and ROS activity levels measured in a Turner BioSystems modulus fluorometer (Sunnyvale, CA) with filters optimized to detect fluorescein. Data were corrected using cell-free DMEM.
This work was supported by the Jones-Rohner Endowment (E.F.) and the National Institutes of Health (grant number NCI RO1 CA67405 to E.F.).
The author(s) declared no potential conflicts of interest with respect to the authorship and/or publication of this article.