Mitochondrial membrane depolarization enhances TRAIL-induced cell death in adult human granulosa tumor cells, KGN, through inhibition of BIRC5

Among the subtypes of ovarian cancer, granulosa cell tumors (GCTs) of the ovary are relatively rare in occurrence, representing only approximately 2–5% of all ovarian neoplasms, yet represent the most common sex cord tumors [1]. This heterogeneous classification includes tumors derived from the stroma, rete ovarii, and developing ovarian follicles [1]. GCTs can be categorized as either adult (> 95%) or juvenile (< 5%) types, with adult GCTs (aGCTs) harboring a somatic missense mutation in the FOXL2 gene, (FOXL2^C134W), which is not present in the rarer juvenile subtype [2]. Arguably pathogenic, the mechanisms by which FOXL2^C134W drive tumorigenesis are not well characterized, but set this tumor type apart from not only the juvenile subtype, but ovarian cancers of epithelial origin as well [2‐5]. Due to the relative paucity of information regarding the etiology, progression, and molecular properties of aGCTs given their comparatively infrequent manifestation, the prevailing therapeutic strategies used to treat aGCTs are derived from those used in treatment of more common epithelial ovarian cancers (EOCs), despite the distinct phenotype of aGCTs [6, 7]. Among the stark differences that distinguish GCTs from EOCs is the capacity of aGCTs to exhibit steroidogenesis associated with gonadotropin-responsiveness, categorizing aGCTs as endocrine tumors [4]. Consequently, unlike EOCs which can remain asymptomatic and evade detection until late-stage, the elevated estrogen synthesized by aGCTs initiates symptoms characteristic of estrogen excess, including menorrhagia and endometrial hyperplasia [8], resulting in earlier clinical detection than other ovarian cancers. However, recurrence is common and often delayed, with relapses occurring between 6– and 30–years from the initial diagnosis [9]. Accordingly, additional characterization of the aGCT response, as compared to EOCs, is an important consideration for the treatment of aGCTs.

Recent evidence from a number of cancer types highlights the cellular metabolic properties of tumor cells as a potential mediator of cancer growth and progression, metastasis, and evasion of chemotherapeutic-induced cell death. In cancer cells, a distinctive and near-universal trait is a metabolic switch in carbohydrate metabolism from oxidative phosphorylation (OXPHOS) to glycolysis [10]. While most normal cells generate ATP primarily from OXPHOS in aerobic conditions, tumor cells are more likely to synthesize ATP via conversion of glucose to lactate, and consequently exhibit a reduction in OXPHOS activity [11]. Although this metabolic shift to aerobic glycolysis at first glance appears to decrease energetic efficiency, it is believed the upregulation of glycolysis is a key step in carcinogenesis, providing a growth advantage to malignant cells capable of surviving a hypoxic microenvironment [12]. Malignant transformation is known to be associated with other alterations in mitochondrial function as well. For example, in normal cells a high mitochondrial membrane potential (Δψ_m) reflects mitochondrial respiration and generation of ATP, whereas mitochondria in cancer cells demonstrate an aberrant Δψ_m [13]. These types of alterations in mitochondrial function specifically in cancer cells have raised prospects for therapeutic targeting of mitochondria in cancer cells. Notably, metabolic profiles defining OXPHOS vs glycolytic status for aGCT cells, as well as EOCs, have yet to be well- reported [14‐17].

We have previously reported on the efficacy of the naturally occurring cytokine, tumor necrosis factor-related apoptosis-inducing ligand (TRAIL), to promote cell death in aGCTs under specific conditions [18, 19]. This is of particular interest, as evidence from primary cultured ovarian granulosa cells have demonstrated that under normal conditions, healthy granulosa cells are inherently resistant to TRAIL-induced apoptosis [20]. However, while treatment of aGCTs with TRAIL in combination with therapeutic agents such as cisplatin or proteasome inhibitors enhances TRAIL-induced cell death, this action is mediated largely through upregulation of death receptors (e.g., DR4 or DR5), and downstream effectors, such as p53, are not impacted [18, 19]. DR4 and DR5 are members of the tumor necrosis factor receptor (TNFR) superfamily [21], and are characterized by multiple cysteine-rich domains present within the extracellular domain, in addition to an intracellular death domain (DD) [22]. Following activation through TRAIL-binding, the DD links with Fas-associated death domain (FADD) protein, resulting in the formation of the apoptosis-inducing signaling complex (DISC) [23], which subsequently leads to activation of caspase-8, which cleaves bid and triggers the release of cytochrome c from the mitochondria into the cytosol [24], triggering the caspase cascade that ultimately results in cellular apoptosis. However, the ability of TRAIL to readily induce apoptosis is diminished, as the TRAIL-induced caspase cascade is opposed by pro-survival proteins, namely the inhibitor of apoptosis (IAP) family, which includes Survivin, also called baculoviral inhibitor of apoptosis repeat-containing 5 (BIRC5). Notably, BIRC5 is a potent inhibitor of caspase activation, and when present, actively promotes cell survival through inhibition of caspase-induced apoptosis. Moreover, and perhaps not surprisingly, BIRC5 levels are elevated in tumors as compared to normal tissues [25]. Additionally, BIRC5 is thought to play an important role in mediating aerobic glycolysis in cancer cells through suppression of OXPHOS, which occurs, at least in part, through inhibition of respiratory complex-I [26].

The disruption of mitochondria-driven metabolism, as well as mitochondria-mediated pro-apoptotic cascades, has been well-characterized in many models of cancer, although not yet including aGCTs. A better understanding of the molecular characteristics of aGCT metabolism, as well as the response to traditional therapeutics, could serve to inform novel therapies which specifically target aGCTs. Herein, we evaluated the metabolic profile of three human EOC cell lines, SKOV3, OVCAR3, and Kuramochi, as well as KGN, a well-established human aGCT cell line heterozygous for the FOXL2^C134W mutation. Through the use of metabolic inhibitors, we investigated the role of KGN metabolism in survival following treatment with either TRAIL or cisplatin, and determined an essential role for Δψm in the opposition of pro-apoptotic signals.

Inherent in studies of ovarian cancer is the caveat of the disease’s high degree of heterogeneity in presentation. Because aGCTs represent a small proportion of clinical diagnoses under the umbrella term of “ovarian cancer” many clinical treatment paradigms are developed in context of epithelial adenocarcinomas [7], with a limited number of small clinical studies specifically targeting aGCTs [30]. While prognosis following stage I detection is excellent [7], high rates of recurrence of aGCTs necessitates a further evaluation of the molecular characterization of these tumors, which are likely distinct from epithelial carcinomas [7]. These data represent a characterization of the metabolic profile of KGN cells as a valuable model for aGCTs, with comparative assays run using epithelial-origin tumor derived cell lines instructive in demonstrating the heterogeneous response to treatments, even at the basic level of cells in vitro.

KGN cells were found to be primarily glycolytic, in both short-term responses to ATP production as well as over longer proliferation assays utilizing inhibitors against glycolysis, oxidative phosphorylation, or Δψ_m. Treatment with FCCP was surprisingly effective in inhibiting proliferation in otherwise glycolytic cells, implicating a role for Δψ_m outside of metabolism. In fact, pretreatment with FCCP, a proton ionophore which uncouples oxidation from phosphorylation within mitochondria via export of H⁺ ions [27], sensitized cells to TRAIL or cisplatin, more effectively than treatment with oligomycin A, which specifically inhibits oxidative phosphorylation by blocking the F0 subunit of ATP synthase [31]. We confirmed dual treatment was inducing apoptosis, rather than slowing proliferation rates, through Annexin V labeling, with significant increases in apoptosis occurring only in cells treated with both FCCP and TRAIL in combination. When evaluating cytochrome c as an upstream mediator of apoptotic induction, we found significantly higher rates of cytochrome c release in both the FCCP single treatment and dual treated cells, suggesting FCCP treatment is sensitizing cells to further induction via uncoupling Δψ_m. Interestingly, the Δψ_m probe JC-1 indicated that both FCCP and TRAIL single treatments were each sufficient to depolarize Δψ_m, which indicates a balance between extrinsic signaling, as TRAIL alone is insufficient to induce cell death, and intrinsic regulation, as FCCP alone is sufficient to decrease cell proliferation but not to induce apoptosis. Additionally, we found the IAP, BIRC5, to be essential in this balance, as treatment with TRAIL significantly increased BIRC5 expression, however this response was lost upon pretreatment with FCCP. A knockdown of BIRC5 prior to treatment further sensitized cells to apoptotic induction. These data indicate an essential role for both Δψ_m and BIRC5 in KGN cell response to TRAIL as summarized in Fig. 6. These data present the first molecular characterization of Δψ_m in KGN cells and its essential role in the inhibition of pro-apoptotic signaling, independent of a metabolic role. Such findings suggest the development of new therapeutic strategies targeting Δψ_m to more specifically treat aGCTs, in comparison to EOCs. Although there is limited clinical data focused solely on aGCTs, others have also described glycolytic signatures [32], with promising initial studies in mural models of GCTs [33], however the GCT field remains sparse, in comparison to EOCs. The availability of the KGN cell line to model aGCTs in vitro provides an important opportunity to characterize new basic research strategies, such as deciphering the role played by mitochondria as a hub of cellular signaling in response to cytotoxic treatment, independent of small clinical cohorts. These data indicate a promising new strategy for potential future aGCT-specific treatments, targeting mitochondrial signaling networks in otherwise glycolytic cells.