Dysregulation of autophagy has long been implicated in numerous pathologies [29]. In the context of metastatic ovarian cancer, it appears that autophagy serves a tumour protective role. We have demonstrated previously that EOC spheroids display increased levels of autophagy and that its inhibition can reduce overall EOC spheroid viability [10, 11]. Independently, we demonstrated that p-AMPK is increased in both patient-derived EOC spheroids as well as those generated in vitro, relative to proliferating adherent cells [8]. In this study, we sought to bridge these two phenomena and demonstrate that AMPK activity is required for autophagy induction in EOC spheroids to maintain cell viability.
In several biological contexts, AMPK activity on its own can lead directly to autophagy induction. However, treatment of adherent OVCAR8 cells with AMPK activators did not significantly change LC3-II or p62 expression. Metformin and oligomycin treatments increased p-AMPK levels, but these drugs function indirectly to activate AMPK by inhibiting mitochondrial respiration [30, 31]. Thus, caution must be taken when using these methods for AMPK activation and correlating results with autophagy induction. However, we consider it unlikely that AMPK activation is sufficient on its own to induce autophagy in proliferating adherent EOC cells.
We show that RNAi-mediated AMPK inhibition strongly inhibits autophagic flux as visualized by fluorescence microscopy. Interestingly, this phenotype may occur in an LC3- and p62-independent manner. The discrepancy in our results between fluorescence reporter and immunoblot assays raise certain questions as to what specific players mediate autophagy induction in ovarian cancer. Initially described in yeast as ATG8, several orthologs of the ubiquitin-like LC3 protein have been identified in mammals, although most work focuses on LC3B. More recently, studies have implicated LC3B-independent forms of autophagy. One candidate LC3 ortholog is gamma-aminobutyric acid receptor-associated protein (GABARAP). GABARAP has been shown to possess separate functions from LC3, as it is involved in late-stage autophagosome maturation [32]. More recently, LC3-independent autophagy in rat hepatocytes was shown to be regulated primarily through the GABARAP complex in the autophagosome [33]. Since we did not observe major effects on LC3 processing, we are currently investigating whether AMPK inhibition affects GABARAP expression and function in autophagic flux in EOC spheroids.
Although Compound C did not attenuate p-AMPK levels nearly to the same extent as either RNA interference or STO-609 treatment, it clearly inhibited autophagic flux in EOC spheroids. Compound C may affect autophagy as a combination of AMPK inhibition as well as with other potential targets of this agent. Previous literature identified multiple intersecting pathways that are potently affected by Compound C that are independent of AMPK (Harhaji-Trajkovic et al., 2010; Zhao et al., 2018). In fact, Compound C is known to affect BMP and mTOR signaling[23, 34]; we have demonstrated that both of these signaling pathways impinge upon the EOC spheroid phenotype [7, 10, 35]. As such, the combined action of Compound C on multiple different kinases could lead to our observed LC3 reporter results in EOC spheroids. In fact, we observed poor p-AMPK attenuation using Compound C, and it has conflicting roles as either an activator or inhibitor of autophagy[34]. Thus, use of this agent alone poses a limitation for analysis of AMPK regulation of autophagy in our system.
To address this, we present new findings regarding the requirement of CAMKKβ-mediated AMPK signaling in modulating autophagy in EOC. Treatment of EOC spheroids with the CAMKKβ inhibitor, STO-609, supports our PRKAA1/2 knockdown data, thus strengthening the notion that AMPK is required for autophagy induction in EOC cells under spheroid conditions. This phenotype holds true not only for EOC cell lines, but also in non-malignant fallopian tube epithelial cells. Perhaps the autophagic stress response mediated by AMPK is conserved in secretory epithelial cells, as well as the high-grade serous EOC cells from which they arise.
Furthermore, work in our laboratory identified recently that LKB1-deficient EOC spheroids still retain the capacity to induce p-AMPK through CAMKKβ activity [26]. This finding together with our results herein suggest a crucial role for CAMKKβ in regulating p-AMPK levels in this disease. It has been previously reported that a rise in cytosolic calcium can induce autophagy through CAMKKβ in both MCF-7 and HeLa cell lines, highlighting an ATP-independent mechanism for autophagy induction [36]. More recently, cellular matrix deprivation has been identified as an inducer of intracellular calcium spikes, which in turn can activate AMPK through CAMKKβ signaling [28]. Examination of the calcium-oxidant signaling network in EOC spheroids might highlight a unique characteristic of these cancer cells that would lend itself to therapeutic inhibition. As such, it would be prudent to further characterize both the AMPK-dependent and -independent roles of CAMKKβ in the context of ovarian cancer. Our encouraging results of CAMKKb-AMPK inhibition using the STO-609 and its negative impact on EOC spheroid cell viability lends even more support for such an intervention.
Overall, it appears that AMPK is required in part to induce autophagy in EOC spheroids, although this may occur in an LC3-and p62-independent manner. We also show AMPK phosphorylation is regulated by CAMKKβ activity in EOC spheroids to promote autophagic flux in these structures. These findings have contributed to our understanding of signaling axes regulating autophagy induction in EOC cells, and may represent novel therapeutic targets for this critical stress response in the setting of metastatic disease.