Although platinum compounds are the first-line treatment for ovarian cancer, the majority of patients relapse and develop resistance to treatment. However, the mechanism underlying resistance is unclear. The goal of our study is to decipher the mechanism by which a metabolic kinase, diacylglycerol kinase alpha (DGKA), confers platinum resistance in ovarian cancer.
Metabolic kinase RNAi synthetic lethal screening was used to identify a cisplatin resistance driver in ovarian cancer. DGKA variants were used to demonstrate the need for DGKA activity in cisplatin resistance. Phospho-proteomic and genomic screens were performed to identify downstream effectors of DGKA. Therapeutic efficacy of targeting DGKA was confirmed and clinical relevance of DGKA signaling was validated using ovarian cancer patient-derived tumors that had different responses to platinum-based therapy.
We found that platinum resistance was mediated by DGKA and its product, phosphatidic acid (PA), in ovarian cancer. Proteomic and genomic screens revealed that DGKA activates the transcription factor c-JUN and consequently enhances expression of a cell-cycle regulator, WEE1. Mechanistically, PA facilitates c-JUN N-terminal kinase recruitment to c-JUN and its nuclear localization, leading to c-JUN activation upon cisplatin exposure. Pharmacologic inhibition of DGKA sensitized ovarian cancer cells to cisplatin treatment and DGKA–c-JUN–WEE1 signaling positively correlated with platinum resistance in tumors derived from patients with ovarian cancer.
Our study demonstrates how the DGKA-derived lipid messenger, PA, contributes to cisplatin resistance by intertwining with kinase and transcription networks, and provides preclinical evidence for targeting DGKA as a new strategy in ovarian cancer treatment to battle cisplatin resistance.
Platinum-based therapy is a first-line treatment option for ovarian cancer, but resistance often leads to therapeutic failure. Dysregulated kinase signaling and altered metabolism are implicated in human cancers, but how these pathways intertwine and contribute to chemotherapy resistance is largely unknown. Here we report that loss of DGKA selectively rescues cisplatin sensitivity in cisplatin-resistant ovarian cancer in a kinase-dependent manner through its metabolic product, phosphatidic acid (PA). Mechanistically, PA activates a transcription factor c-JUN by binding and translocating it into the nucleus to promote WEE1 expression. DGKA inhibition significantly sensitizes cancer cells to cisplatin in a patient-derived xenograft model. Moreover, the clinical relevance of DGKA signaling in platinum resistance was confirmed in patient tumors. Our findings not only suggest a mechanism by which DGKA provides cisplatin resistance through c-JUN–WEE1 signaling, but also implicate DGKA as a potent therapeutic target to overcome platinum resistance in ovarian cancer.
Platinum-based compounds have been the most actively used front-line clinical drug for the treatment of ovarian cancer for decades. The prognosis of patients with ovarian cancer can be defined by the response of the tumor to cisplatin; patients whose tumors are intrinsically resistant to platinum at the time of initial treatment or acquire resistance during treatment have very poor prognosis (1–3). Therefore, understanding of the crucial factors driving platinum resistance is needed to enhance the therapeutic outcomes of platinum-based chemotherapy. Although efforts have been made to identify platinum resistance factors including HOXD8 (4), ATP11B (5), Foxo3a (6), and Xiap (7) in ovarian cancer, most studies lack a clinical correlation approach or a detailed explanation for how these factors intertwine with signaling networks in ovarian cancer cells to control platinum-resistant survival.
Emerging evidence suggests that metabolite levels are altered in cancer cells and control of these changes may provide metabolic vulnerabilities that can be exploited for cancer therapy (8). The altered metabolites are reported to function as critical signaling molecules that contribute to cellular actions needed for proliferative and metastatic potential in cancer cells. For instance, ribulose-5-phosphate, an intermediate in the oxidative pentose phosphate pathway, inhibits AMPK signaling by disrupting the interaction of AMPK with its activating kinase LKB1 and eventually promotes lipogenesis by activating a downstream effector acetyl-CoA carboxylase 1 (9). We found fumarate, an intermediate metabolite produced by glutamine metabolism, binds to and activates a reactive oxygen species (ROS) scavenging enzyme glutamate peroxidase 1 to promote redox balance in cancer cells (10). In line with these findings, recent studies demonstrate that the response to first-line chemotherapy is influenced by the metabolic state of cancer cells (11). Cisplatin-resistant ovarian cancer cells are reported to have higher levels of reducing equivalents such as glutathione than parental cells and are vulnerable to ROS inducing agents (12, 13). The Bcl-2 inhibitor ABT-737 is known to reverse cisplatin resistance in ovarian cancer cells by targeting glucose metabolism (14). Cisplatin-resistant ovarian cancer cells were found to have higher rates of glycolysis and lower mitochondrial activity compared with the parental cells. In contrast, lung cancer cells have lower levels of hexokinase and glycolytic rates but instead rely on oxidative phosphorylation and have higher dependence on glutamine metabolism. Therefore, inhibition of glutaminase sensitized cisplatin-resistant lung cancer cells to cisplatin-based chemotherapy, which reflects the tumor-specific heterogeneity of metabolic alterations (15, 16). The current findings focus on an observation of metabolic alterations and related therapeutic strategies. However, comprehensive mechanisms by which cellular metabolites are intertwined with cellular signaling networks to confer platinum resistance specifically in ovarian cancer and whether these signaling pathways are virtually relevant in patients are not explored. Here we identify that phosphatidic acid (PA), a lipid metabolite produced by the diacylglycerol kinase alpha (DGKA) enzyme reaction, acts as a critical signaling effector in cisplatin resistance of ovarian cancer.
PA is a vital cellular phospholipid that is typically metabolized from a membrane lipid, diacylglycerol (DAG), by DGKs (17, 18). DGKs are a large family of conserved membrane lipid kinases (19). Among 10 known DGK enzymes, DGKA has been implicated in human cancers (20, 21). DGKA and its product PA are involved in ovarian cancer cell invasion that is mediated through α5β1 integrin trafficking. DGKA is involved in NF-κB or RAS–Raf–MEK activation, which suppresses TNFα-mediated apoptosis in melanoma or enhances hepatocellular carcinoma progression, respectively (22, 23). In addition, DGKA has been linked to activation of HIF-1α, c-MET, ALK, and VEGF (20, 24–27). DGKs are also appreciated as physiologic regulators of immune cell function through PA serving as second messengers in T-cell receptor signaling (28). DGKA is known to be overexpressed in CD8+ tumor-infiltrating T cells. Treatment with R59949 and R59022, small-molecule compounds that inhibit DGKA by binding to its catalytic domain, was reported to recruit CD8+ T cells to the antitumor response and may prevent inactivation of T cells (29, 30).
In this study, we report a novel role of DGKA and its product PA, connecting metabolic and cellular kinase signaling to gene regulatory machinery that provides resistance to platinum-based chemotherapy uniquely in ovarian cancer.
Materials and Methods
Lentiviral plasmids containing short hairpin RNA (shRNA) for DGKA, LPAAT2, PLD2, c-JUN, WEE1, ATX, PLA1A, PLA2G4A, DGKB, DGKI, and Human Kinase shRNA Gene Family Library were purchased from Dharmacon. The shRNA sense strand sequences were GCTCTGGAAGTTCCAGTATAT (clone #1), and GCTAAATATGTCCAAGGAGAT (clone #2) for DGKA, CGGACCTTATGGCTACAGTAA (clone #1) and CGCAAACCTCAGCAACTTCAA (clone #2) for c-JUN, CCACCCAGAGTAATAGAACAT (clone #1) and CTAGAAAGAGTGCAGAACAAT (clone #2) for WEE1, CGTGCATCATACATGAAGAAA for ATX, GCCAGATAAACCAAGTGAAAT for PLA1A, CCGACTTATTTGGAAGCAAAT for PLA2G4A, CCTGAACTGCATGATTTATTA for DGKB, GCACCCTAATTGACTTGGAAA for DGKI, CGAGGGTACTCGCAACGACAA for LPAAT2, and CCAAGAAGAAATACCGTCATT for PLD2. Image clones for human DGKA, c-JUN, and WEE1 were obtained from DNASU. Primers for DGKA shRNA-resistant silent mutant, DGKA kinase-dead mutant G434A, and qRT-PCR were obtained from Integrated DNA Technologies. The mutant constructs were generated using site-directed mutagenesis kit from Agilent Technologies. 1,2-dioctanoyl-sn-glycero-3-phosphate, the PA, was obtained from Avanti Polar Lipids. 1-oleoyl-2-acetyl-sn-glycerol, an analog of the protein kinase C (PKC)-activating second messenger DAG as a DGKA substrate, lysophosphatidic acid (LPA), and DGK Inhibitor II (R59949) were obtained from Sigma-Aldrich. Gemcitabine, paclitaxel, bortezomib, rapamycin, and MK-1775 were obtained from Selleckchem. Recombinant human proteins c-JUN and JNK1 were obtained from Abcam and Invitrogen, respectively. NE-PER Nuclear and Cytoplasmic Extraction Reagent was obtained from Thermo Fisher Scientific. Fluorometric PicoProbe Phosphatidic Acid Assay Kit was obtained from BioVision. LPA Bioassay ELISA Kit was purchased from US Biological Life Sciences. Chromatin Immunoprecipitation Assay Kit was purchased from Millipore. Proteome Profiler Human Phospho-Kinase and Human Cell Stress Array Kits were purchased from R&D Systems.
Anti-DGKA antibody for immunoblotting and IHC staining was purchased from ProteinTech. Antibodies against PLD2 (13904), LPAAT2 (14937) phospho-c-JUN Ser63 (9261 for immunoblotting and 2361 for IHC), c-JUN (9165/60A8), phospho-JNK (c-JUN N-terminal kinase) Thr183/Tyr185 (9255/G9), JNK (9252), His-tag (2366/27E8), myc-tag (2278/71D10), PARP (5625/D64E10), WEE1 (13084/D10D2), phospho-S6 Ser235/236 (4858/D57.2.2E), S6 (2217/5G10), phospho-cdc2 Tyr15 (4539/10A11), and cdc2 (9112) were obtained from Cell Signaling Technology. Anti–β-actin antibody (A1978/AC-15), anti-GST antibody (G1160/GST-2), and anti-flag antibody (F7425) were obtained from Sigma-Aldrich. Antibody against Ki-67 (ab92742/EPR3610) and phospho-Serine/Threonine (ab17464) were from Abcam.
RNAi, phospho-proteomic, and genomic screens
Primary synthetic lethal RNAi screen was performed using the TRC Human Kinase shRNA Gene Family Library (31). Among the 100 top ranking candidates from the primary screen, 16 kinases that use small molecular metabolites as substrates were selected out and further investigated for synthetic lethality with cisplatin in cisplatin-resistant ovarian cancer cell lines. In brief, A2780cisR and SK-OV-3cisR cells were infected with lentivirus pools targeting individual genes and treated with sublethal doses that resulted in 15% cell death for each cell line (SK-OV-3cisR: 2 μg/mL, A2780cisR: 5 μg/mL) of cisplatin. For phospho-proteomic screen, A2780cisR derivatives were treated with PBS or cisplatin for 48 hours. Cell lysates were obtained and applied to the Proteome Profiler Human Phospho-Kinase Array and the results were quantified by ImageJ software. For targeted genomic screen, probes for Encyclopedia of DNA Elements (ENCODE)-based 44 potential DGKA–c-JUN target genes were designed and the gene expression profile for each gene was assessed by qRT-PCR using A2780cisR cells with DGKA or c-JUN knockdown.
Cell culture and generation of cell lines
A2780cisR cells were cultured in RPMI1640 medium with 10% FBS. 293T cells were cultured in DMEM with 10% FBS. SK-OV-3cisR cells were cultured in McCoy 5A medium with 10% FBS. A2780, A2780cisR, COV318, COV504, and PEO1 cells were obtained from Sigma-Aldrich. BG-1 and 1A9 cells were as described previously (31). 293T, CAOV3, SW626, OV90, and SK-OV-3 cells were from the ATCC. SK-OV-3cisR cells were generated by constantly exposing SK-OV-3 cells to increasing concentrations (3–10 μmol/L) of cisplatin. A2780cisR and SK-OV-3cisR cells are 6.8- and 3.5-fold more resistant to cisplatin, respectively, than their parental cell lines in terms of cisplatin IC50. Lentivirus and retrovirus production, RNAi, and protein overexpression in human cancer cells were as described previously (10, 32). In brief, shRNA lentivirus was generated in 293T cells using pLKO.1 vector encoding shRNA, psPAX2, and pMD2.G. A2780cisR cells were infected with lentivirus for 2 days and selected using 2 μg/mL of puromycin. Human DGKA, JNK, c-JUN, and WEE1 were flag or myc tagged by PCR and cloned into pLHCX or pDEST-27 for overexpression. Selection of cells with stable DGKA expression was carried out using 50 μg/mL of hygromycin.
DGKA kinase activity assay
GST-fused DGKA wild-type or kinase-dead mutant proteins enriched from 293T cells were mixed with 100 μg/mL of 1-oleoyl-2-acetyl-sn-glycerol (DAG) in lipid dilution buffer [25 mmol/L HEPES (pH 7.5), 0.5 mmol/L EGTA]. The kinase activity of DGKA was assessed by ADP-Glo Kinase Assay (Promega). GST protein was used as a negative control for normalization.
Cell viability assay and colony formation assay
Cells were seeded at 7,000 cells per well in 96-well plates and compounds were added at the indicated concentrations for 48 hours. Cell viability was determined using CellTiter-Glo Luminescent Viability Assay (Promega). For colony formation assay, 500 cells were seeded in 35 mm dishes and treated with sublethal doses of cisplatin for 24 hours. The cells were cultured in complete media for 10 additional days before the colonies were stained and counted by ImageJ software.
For anabolic biosynthesis analyses, cells were spiked with 4 μCi/mL of D-[U-14C]-glucose for 2 hours. Lipids were extracted using hexane and isopropanol (3:2 vol/vol), dried, and resuspended in chloroform. The amount of labeled lipid was quantified by liquid scintillation counting and normalized to the total protein amount. For RNA synthesis assay, total RNA was isolated from the spiked cells. The labeled RNA was quantified by liquid scintillation counting and normalized to the total RNA content. Intracellular ATP levels were determined using ATP Bioluminescent Somatic Cell Assay (Sigma-Aldrich). The total ATP levels were normalized to cell number. Intracellular ROS levels were determined by staining cells with CM-H2DCFDA.
Apoptosis assay and cell-cycle assay
A total of 3 × 105 cells were treated with cisplatin for 48 hours and stained with FITC-conjugated annexin V labeling reagent (BD Pharmingen) and propidium iodide followed by FACS analysis for apoptotic cell population analysis. Cell cycle was assessed by propidium iodide flow cytometric assay and analyzed with FlowJo Software V10.
Cellular thermal shift assay
Cellular thermal shift assay was performed as described previously with a modification (33, 34). In brief, 293T cells were transfected with c-JUN or JNK expressing vectors for 24 hours. Cell lysates were treated with or without 10 μmol/L of PA, equally aliquoted, and heated at 55, 57, 60, 63, 66, 69, 72, and 75 degrees for 3 minutes. Cells were lysed and c-JUN or JNK in the soluble fraction was quantified by Western blot analyses.
Surface plasmon resonance
Surface plasmon resonance (SPR) assays were performed using a Biacore ×100. A series of increasing concentrations of recombinant JNK in running buffer [0.01 M HEPES (pH 7.4), 0.15 mol/L NaCl, 0.005% vol/vol surfactant P20] were injected over immobilized c-JUN in the presence of 10 μmol/L PA, LPA, or vehicle at 30 μL/minute for 180 seconds at 20°C. Multiple-cycle kinetics analyses were used to quantify c-JUN and JNK interaction. Sensorgrams were obtained by subtracting values of vehicle alone and dissociation constants (Kd) were calculated using BIAevaluation Software.
WEE1 promoter reporter assay
Cells with knockdown or overexpression of DGKA or c-JUN were transfected with WEE1 promoter reporter construct followed by cisplatin treatment. Activity of WEE1 promoter was assessed using luciferase assay kit. Cypridina luciferase reporter was used for normalization.
A2780cisR cells were seeded on coverslips and fixed in PHEMO buffer (68 mmol/L PIPES, 25 mmol/L HEPES, 3 mmol/L MgCl2, 15 mmol/L EGTA, 0.05% glutaraldehyde, 3.7% formaldehyde, 0.5% Triton X-100). Samples were blocked in 10% goat serum and stained with anti-c-JUN antibody followed by Alexa 488–conjugated anti-rabbit IgG antibody. The coverslips were mounted and imaged on a Zeiss LSM 510 META confocal microscope.
The total RNA of A2780cisR and SK-OV-3cisR were isolated using RNeasy Mini Kit (Qiagen). To detect the expression levels of target genes, reverse transcription of RNA was performed using High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific) followed by qPCR using Universal SYBR Green Supermix (Bio-Rad). The relative expression levels of the target genes were computed by the 2−ΔΔCt method. The primers for qRT-PCR were obtained from Integrated DNA Technologies.
Approval to use human-derived tissue specimens was given by the Institutional Review Board of Emory University (Atlanta, GA). Clinical samples were collected with informed consent under Health Insurance Portability and Accountability Act approved protocols. Formalin-fixed and paraffin-embedded tissues derived from patients with ovarian cancer receiving platinum-based chemotherapy were kindly provided by Champions Oncology.
Formalin-fixed and paraffin-embedded tissues derived from patients with ovarian cancer receiving platinum-based chemotherapy were kindly provided by Champions Oncology. Tumors collected from cell line–based or patient-derived xenograft (PDX) mice were stained with anti-DGKA (1:500), anti–phospho-c-JUN Ser63 (1:100), anti-WEE1 (1:100), or anti–Ki-67 (1:1,000) antibodies. Staining intensity of DGKA, phospho-c-JUN Ser63, WEE1, and Ki-67 in the tumors was scored as 0 to +3.
Animal studies were performed according to protocols approved by the Institutional Animal Care and Use Committee of Emory University (Atlanta, GA). Nude mice (Hsd:Athymic Nude-Foxn1nu, female, 6-week-old; Envigo) were injected with 1.2 × 106 A2780cisR cells and cisplatin (5 mg/kg) was administered by intraperitoneal injection twice a week when tumor sizes reached 100–150 mm3. Ovarian cancer PDX tumors (Jacksons Laboratory) were implanted in the flank of 6-week-old nude mice. The mice were randomly divided into four groups when the tumor size reached 100–150 mm3. Cisplatin (5 mg/kg, twice a week) and R59949 (10 mg/kg, every 2 days) were administered by intraperitoneal and subcutaneous injection, respectively. Tumors were measured blindly, and volumes were calculated as 4π/3 × (width/2)2 × (length/2). Tumor proliferation was assessed by Ki-67 staining.
Statistical parameters are indicated in the figure legends and figures. One representative experiment from multiple experiments is shown. Error bars represent mean ± SEM for tumor volume curves and SD for all others. Statistical significance is based on two-tailed Student t test for two-group comparisons and one-way or two-way ANOVA for multiple comparisons in experiments with more than two groups. P values of 0.05 or less were considered statistically significant. Sample size was not predetermined using statistical methods. For in vivo experiments, animals were randomly chosen and blinding outcome assessment and concealed allocation were used. The in vitro studies were not randomized and allocation and outcome assessment were not blinded. Statistical analyses and graphical presentation were performed using GraphPad Prism 8.
DGKA contributes to platinum resistance in ovarian cancer
To identify a unique metabolic kinase target that can overcome platinum resistance of human ovarian cancer, we performed targeted RNAi screening in cisplatin-resistant ovarian cancer cell lines with lentiviral shRNA of metabolic kinases selected among the top 100 kinase targets found to sensitize cisplatin response in a kinome-wide RNAi screen. Among 16 enzymes tested, DGKA was identified as the most effective target that commonly sensitizes ovarian cancer cells to cisplatin treatment (Fig. 1A). The RNAi screening result was further validated in vitro and in vivo using individual DGKA shRNA clones. DGKA knockdown using two different shRNA clones significantly attenuated colony formation efficiency, cell proliferation, and cisplatin resistance, and enhanced apoptotic cell death when cells were treated with cisplatin (Fig. 1B). Similar results were obtained when the DGKA knockdown cells were treated with another platinum-based compound, carboplatin, or with chemotherapy agents that impede cell division including gemcitabine and paclitaxel, but not with nonchemotherapy agents including the proteasome inhibitor bortezomib or molecularly targeted inhibitor rapamycin (Supplementary Fig. S1 and S2). Knockdown of DGKA significantly sensitized 10 ovarian cancer cell lines to cisplatin treatment regardless of subtype, suggesting that DGKA commonly provides cisplatin-resistant potential to ovarian cancer (Supplementary Fig. S3). Furthermore, tumor growth potential was dramatically decreased in cisplatin-treated xenograft mice bearing ovarian cancer cells with DGKA knockdown (Fig. 1C–E). These results suggest that DGKA contributes to chemotherapy resistance of human ovarian cancer and DGKA could be a potential target in ovarian cancer with treatment of chemotherapy agents such as cisplatin.
PA is a key factor for DGKA-mediated cisplatin resistance
We next examined whether enzyme activity and the product of DGKA are required for cisplatin resistance. Mutation at glycine 434 in the catalytic subunit to alanine in DGKA abrogated its enzyme activity that produces PA (Supplementary Fig. S4A; ref. 18). Rescue expression of DGKA wild-type but not the enzyme-dead mutant G434A (GA) fully restored the decreased level of intracellular PA, cell viability, and cisplatin resistance in cisplatin-resistant ovarian cancer cells with DGKA knockdown (Fig. 2A). We next examined whether the cisplatin resistance is mediated through PA or its downstream LPA signaling by target downregulating enzymes that produce LPA in cells, including phospholipase A1 (PLA1), phospholipase A2 (PLA2), and autotaxin (ATX; Supplementary Fig. S4B and S4C). Decrease in intracellular LPA level and LPA signaling, which was assessed by S6 phosphorylation in the mTOR pathway, did not alter cisplatin resistance, whereas accumulated PA levels resulted in enhanced cisplatin resistance (Supplementary Fig. S4D–S4L). In line with this observation, replenishment of the decreased intracellular PA by external PA significantly rescued the decreased cell viability and cisplatin resistance whereas LPA did not influence these phenotypes in DGKA knockdown cells (Fig. 2B and C). These data suggest that the effect of DGKA on cisplatin resistance is mediated through PA but not downstream metabolite LPA.
To identify an essential enzyme that manages PA levels in ovarian cancer cells, gene expression levels of PA producing enzymes including isoforms of DGK, LPAAT, and PLD were examined in ovarian cancer cells (35). Although DGKB and DGKI were additionally identified from the RNAi screen as potential targets to enhance cisplatin response, DGKA was more abundant than the other isoforms in ovarian cancer cells (Fig. 2D). Furthermore, target downregulation study indicated that DGKA is the critical enzyme responsible for PA production among the three PA regulating enzymes, DGKA, LPAAT2, and PLD2, in ovarian cancer cells (Fig. 2E). These data suggest that PA, mainly produced by DGKA in ovarian cancer cells, is the key factor that provides DGKA-mediated cisplatin resistance in ovarian cancer cells.
DGKA signals through c-JUN to provide cisplatin resistance
Although DGKA contributes to cisplatin resistance through PA, a key intermediate in lipid metabolism, the knockdown of DGKA along with cisplatin treatment did not induce changes in lipid synthesis, nucleotide synthesis, bioenergetics, or redox status in ovarian cancer cells, suggesting DGKA-PA–induced cisplatin resistance does not occur through metabolic reprogramming (Supplementary Fig. S5). To investigate how DGKA regulates cellular signaling effectors to confer cisplatin resistance in cancer cells, we performed proteome profiling of human phospho-kinase and apoptosis signaling and monitored expression and phosphorylation of 71 factors related to cell survival and apoptosis in ovarian cancer cells with DGKA knockdown and cisplatin treatment. We observed that the phosphorylation of c-JUN at serine 63, which is known to enhance its transcription activity, dramatically decreased when A2780cisR cells lacked DGKA and were treated with cisplatin (Fig. 3A; Supplementary Fig. S6A and S6B). The array results were further confirmed in SK-OV-3cisR and A2780cisR cells by immunoblotting (Fig. 3B). The decrease in c-JUN phosphorylation in DGKA knockdown cells with cisplatin treatment was not mediated through a change in JNK, one of the MAPK family (Supplementary Fig. S6C). Target downregulation of c-JUN in these cells mimicked the DGKA knockdown effect, leading to attenuated cell viability and enhanced apoptosis induction when combined with cisplatin treatment, which provides a potential link between DGKA and c-JUN in cisplatin-resistant cell survival (Supplementary Fig. S6D).
We explored the molecular mechanism underlying DGKA and PA-mediated activation of c-JUN. DGKA did not directly phosphorylate c-JUN (Supplementary Fig. S6E). However, in vitro incubation with PA enhanced the complex formation of c-JUN and JNK in a dose-dependent manner (Fig. 3C). Furthermore, cellular thermal shift and SPR assays showed that PA binds to the c-JUN–JNK complex and PA but not LPA enhances the interaction between c-JUN and JNK (Fig. 3D and E). Nuclear accumulation of c-JUN is associated with transcriptional activation of c-JUN. Interestingly, we found that treatment with cisplatin enhances the nuclear localization of c-JUN and JNK. Loss of DGKA results in reduced nuclear import of c-JUN while treatment with PA restored the accumulation of c-JUN in the nucleus in cisplatin-treated ovarian cancer cells (Fig. 3F and G; Supplementary Fig. S6F and S6G). These results suggest that DGKA and its product PA promote c-JUN and JNK complex formation and their nuclear localization, which eventually leads to the transcriptional activation of c-JUN in ovarian cancer cells upon cisplatin exposure.
We next examined whether DGKA signals through c-JUN to mediate cisplatin-resistant cancer cell survival and proliferation. Overexpression of phospho-mimetic mutant form (S63D/S73D) but not phospho-deficient mutant form (S63A/S73A) of c-JUN fully restored the decreased cell viability and elevated apoptotic cell death in cisplatin-treated DGKA knockdown cells (Fig. 4A and B). Consistent with the observation in vitro, the phospho-mimetic active mutant form of c-JUN rescued the attenuated tumor growth potential in A2780cisR xenograft mice, suggesting that DGKA mediates cisplatin-resistant cell proliferation and tumor growth by signaling through the transcription factor c-JUN in ovarian cancer (Fig. 4C–E).
DGKA and c-JUN–mediated WEE1 expression promotes cisplatin-resistant cell survival
c-JUN is involved in various activities in cancer cells via direct regulation of target genes. To further delineate the downstream signaling pathways of DGKA and c-JUN that contribute to cisplatin resistance in ovarian cancer, we assessed gene expression of 44 potential c-JUN transcription targets, which were identified in human melanoma cell lines using ChIP-Seq of the ENCODE, in DGKA or c-JUN knockdown cells treated with cisplatin (36). Among 44 candidates tested, levels of WEE1, ITGAV, ITPKA, and HPSE were significantly decreased in both DGKA and c-JUN knockdown cells (Fig. 5A). Further determination of both gene and protein expression in multiple ovarian cancer cell lines revealed that the level of WEE1, a cell-cycle checkpoint regulator, is commonly decreased by DGKA knockdown and cisplatin treatment, suggesting c-JUN and WEE1 as potential downstream effectors of DGKA (Fig. 5B and C). Genetic or pharmacologic inhibition of WEE1 mimicked the DGKA or c-JUN knockdown effect, resulting in attenuated cell viability and enhanced apoptosis induction when combined with cisplatin treatment (Supplementary Fig. S6H and S6I). ChIP assay and WEE1 promoter reporter assay in ovarian cancer cells with c-JUN or DGKA modulation revealed that WEE1 is the direct transcription target of c-JUN, which is activated by DGKA, in ovarian cancer cells upon cisplatin exposure (Fig. 5D–F). Knockdown of DGKA led to aberrant cell-cycle progression and decreased inhibitory phosphorylation at Y15 of CDK1, which is essential for the premitotic checkpoint (Fig. 5G and H). Overexpression of WEE1 partially restored the accumulated 2n and 4n DNA content, CDK1 Y15 phosphorylation, enhanced cisplatin-induced apoptosis, and decreased cell viability upon cisplatin exposure in DGKA knockdown cells (Fig. 5I). These data suggest that c-JUN activated by DGKA promotes WEE1 expression, which protects cisplatin-exposed ovarian cancer cells from cell-cycle arrest leading to cisplatin-resistant cell survival and proliferation.
A DGK inhibitor sensitizes ovarian cancer cells to cisplatin treatment
We demonstrated that the knockdown of DGKA replenishes cisplatin sensitivity, implicating DGKA as a promising synthetic lethal target in combination with cisplatin in ovarian cancer. Thus, we examined the effect of targeting DGKA with a small-molecule inhibitor on cisplatin sensitization in ovarian cancer cells and PDX mouse models of ovarian cancer. Treatment with the DGK inhibitor, R59949, attenuated the activity of DGKA and intracellular PA levels (Supplementary Fig. S7A–S7C). R59949 enhanced cisplatin-induced apoptotic cell death, whereas cells with knockdown of DGKA but not DGKB or DGKI were resistant to cisplatin-induced apoptosis, indicating that the enhancement of cisplatin-induced cell death by R59949 occurs by targeting DGKA, but not other kinases (Fig. 6A; Supplementary Fig. S7D–S7F). In line with this observation, treatment with R59949 sensitized cisplatin-resistant ovarian cancer cells to cisplatin treatment (Fig. 6B). Furthermore, treatment with R59949 in combination with cisplatin dramatically attenuated tumor growth in ovarian PDX with no significant changes in body weight or histopathology (Fig. 6C–E; Supplementary Fig. S7G and S7H). These data collectively suggest that DGKA could be a potent therapeutic target together with cisplatin for ovarian cancer.
The DGKA–c-JUN–WEE1 signaling axis is associated with platinum resistance in ovarian cancer patient-derived tumors
To further validate the clinical relevance of the DGKA–c-JUN–WEE1 signaling axis, we examined levels of DGKA, phosphorylation of c-JUN, and WEE1 expression in tumors originally derived from patients with ovarian cancer who received cisplatin- or carboplatin-containing therapy before tumor excision. Patients with ovarian cancer were divided into two groups: patients who responded to cisplatin- or carboplatin-based therapy for a duration of 6 months and patients who lost response to the therapy within 6 months. Expression of DGKA, phosphorylation of c-JUN, and WEE1 expression were assessed in 26 patient-derived tumors that had different responses to platinum-based therapy (Fig. 6F; Supplementary Table S1). DGKA level was significantly higher in tumors collected from patients with ovarian cancer who did not respond to platinum-based therapy compared with patient-derived tumors that were sensitive to a platinum-based regimen (Fig. 6G). Primary tumor samples were collected before platinum treatment, suggesting that initial expression of DGKA in tumors contributes to platinum resistance. Moreover, DGKA expression positively correlated with phospho-c-JUN and WEE1 levels in these tumors with correlation coefficient r values of 0.876 and 0.741, respectively (Fig. 6H and I). The signaling axis was further confirmed by IHC staining (Fig. 6J). These data suggest that the DGKA–c-JUN–WEE1 signaling axis is connected with cisplatin resistance in patients with ovarian cancer.
Platinum-based chemotherapeutic regimens are the mainstay of treatment for advanced ovarian cancer. However, around 70% of patients with ovarian cancer relapse after cisplatin-containing first-line chemotherapy and understanding of the resistance mechanisms in a whole cellular context is needed (37). Our findings provide a comprehensive view of the interrelationship between metabolic pathways and cellular kinase signaling that confers cisplatin resistance in ovarian cancer. We present a distinct role of the key intermediate in lipid metabolism, PA, as a signaling molecule that activates the transcription factor c-JUN upon cisplatin exposure and consequently enhances transcription of the cell-cycle checkpoint regulator WEE1 gene. Mechanistically, we found that PA activates the c-JUN transcriptional machinery by recruiting c-JUN to upstream kinase JNK and translocating the complex to the nucleus for its function. Cisplatin treatment relocates c-JUN and JNK complex into the nucleus. While removal of DGKA impairs this translocation, refueling with PA restored the nuclear accumulation. Therefore, it is possible that PA binds within the c-JUN and JNK complex and induces the conformational change of the complex to facilitate nuclear translocation. Indeed, studies report PA as a cellular translocation mediator (38, 39). Further structural and mutational study is warranted to resolve the structure-based mechanism by which PA influences cellular location of c-JUN and JNK in ovarian cancer cells.
Our data suggest that the role of DGKA as a cisplatin resistance driver is prominent in ovarian cancer, as the effects of targeting DGKA in other types of cancer including cervical cancer, head and neck cancer, and lung cancer were not as significant as in ovarian cancer in our series of synthetic lethal RNAi screens using a 781-kinome shRNA library in multiple types of cancer cells (31). In line with these results, targeting DGKA itself effectively attenuates cancer cell proliferation in other types of cancers including melanoma and hepatocellular carcinoma, which is different from observations in ovarian cancer cells (22, 23). This implies that metabolic and cellular needs of tumor cells may differ depending on cancer type and their disparate metabolic status.
We recently reported that microtubule-associated serine/threonine kinase 1 (MAST1) and inositol triphosphate kinase B (ITPKB) are commonly upregulated in various types of cisplatin-resistant tumors compared with cisplatin-sensitive tumors from patients with head and neck cancer, lung cancer, and ovarian cancer (31, 40, 41). MAST1 contributes to cisplatin resistance by rewiring the Raf–MEK pathway, whereas ITPKB balances redox status in cisplatin-treated cancer cells by inhibiting a ROS producing enzyme, NADPH oxidase 4. It is worthwhile to evaluate whether these factors operate in entirely separate signaling axes or are cross-linked with one another to overcome cisplatin resistance in human cancers. If these signaling factors work in parallel by separately contributing to kinase signaling, metabolic balance, and gene regulation, a combinatorial targeting approach of these factors may provide synergistic or additive effects on sensitizing cisplatin response in cancer cells.
Previous studies suggest that JNK-mediated c-JUN activation can subsequently induce both transcription of proapoptotic and prosurvival factors. Therefore, c-JUN signaling could be a double-edged sword in cisplatin treatment, simultaneously not only being a significant factor in cell death but also being associated with increased resistance to cisplatin-based chemotherapy through inducing cell survival (42). In the context of ovarian cancer, c-JUN mainly induces prosurvival signaling by regulating the cell-cycle progression regulator WEE1 upon cisplatin exposure. WEE1 regulates the G2 checkpoint and prevents entry into mitosis in response to DNA damage (43, 44). Therefore, targeting DGKA leads to inhibition of c-JUN and WEE1 and this may result in unrepaired DNA entering mitosis, leading to mitotic catastrophe and consequently to apoptotic cell death in response to cisplatin. In contrast to c-JUN, the transcription factor KLF2 is reported to specifically downregulate WEE1 expression in ovarian cancer (45). It is possible that c-JUN may be a predominant transcriptional regulator of WEE1 induced by DGKA when ovarian cancer cells are exposed to cisplatin.
We also demonstrated the clinical relevance of the DGKA–c-JUN–WEE1 signaling axis in cisplatin resistance using patient-derived tumors from patients with ovarian cancer. Our findings support that DGKA signaling could be used as a promising predictive marker of cisplatin-based chemotherapy response. Finally, genetic and pharmacologic inhibition of DGKA effectively enhanced cisplatin response in ovarian cancer, which provides evidence that the strategy of targeting DGKA signaling has potential efficacy in the treatment of ovarian cancer in combination with platinum-based chemotherapy. Further clinical trials and treatment optimization are warranted to target the DGKA signaling axis to overcome cisplatin resistance in ovarian cancer.
Disclosure of Potential Conflicts of Interest
S. Lonial is a paid advisory board member for Bristol-Myers Squibb, Amgen, GlaxoSmithKline, Takeda, and Janssen. No potential conflicts of interest were disclosed by the other authors.
Conception and design: S. Kang
Development of methodology: L. Jin
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Li, C. Pan, A.C. Boese, J. Kang, K.R. Magliocca, W. Yang, Y. Zhang
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Li, K.R. Magliocca, S. Lonial, S. Kang
Writing, review, and/or revision of the manuscript: J. Li, A.D. Umano, K.R. Magliocca, S. Lonial, S. Kang
Study supervision: S. Kang
Other (assisted with in vitro studies): A.D. Umano
We acknowledge Dr. Anthea Hammond for critical reading and editing of the manuscript. We thank Champions Oncology team for providing ovarian cancer patient tumor specimens. This study was supported in part by NIH/NCI (R01CA175316, R01CA207768, DRP of P50CA217691, to S. Kang), Department of Defense (W81XWH-17-1-0186, to S. Kang), American Cancer Society (Winship IRG-17-181-04, to L. Jin), Elsa U. Pardee Foundation (research grant, to L. Jin), University of Florida (UF Health Cancer Center Pilot Grant to L. Jin), and the Emory University Integrated Cellular Imaging Microscopy Core of the Winship Cancer Institute comprehensive cancer center grant, P30CA138292. S. Kang is a Georgia Cancer Coalition Scholar, a Robbins Scholar, and an American Cancer Society Basic Research Scholar.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.