Coordinated AMPK activity and LC3-II processing during spheroid formation
We demonstrated previously that AMPK is activated in EOC spheroids to promote cytostasis[8]. In an independent study, our group generated evidence that autophagy is rapidly induced in EOC spheroids also, and autophagy is required to maintain cell viability in these structures. Thus, we now seek to connect the kinetics and requirement for AMPK signaling with autophagy activation in our in vitro spheroid model of EOC metastasis. Assessment of autophagic flux can be initially performed by measuring protein expression of both microtubule-associated protein 1A/1B-light chain (LC3) and p62 (sequestosome1). Being recruited to autophagosomal membranes, LC3 is proteolytically cleaved at its C-terminus followed by lipidation to generate LC3-II, making it an excellent marker for monitoring the progression of autophagy [12]. Due to its ubiquitin-binding domain, p62 is known to function as a mediator protein, targeting ubiquitinated proteins to the autophagosomal membrane. Accumulation of p62 protein levels is indicative of reduced autophagic flux, whereas its decrease over time indicates sustained autophagy induction [12].
To address AMPK phosphorylation kinetics and its relation to autophagy induction, iOvCa147-MA cells were seeded in ULA conditions and protein was isolated at various time points during spheroid formation. Following immunoblot analysis, we identified increased levels of LC3-II processing and increased phosphorylation of AMPK at T172 between 24 and 72 hours relative to adherent cells (Figure 1a&b). The highest levels of both p-AMPK and LC3-II was observed at 48 hours; therefore, subsequent spheroid culture experiments were taken to the 48-hour time point. Time course experiments conducted using OVCAR8 spheroids further confirmed the 48-hour time point as optimal for evaluating AMPK activity and autophagy markers (Figure 1c&d). Different levels of basal autophagy were observed in standard adherent conditions between the iOvCa147-MA and OVCAR8 cell lines, as we have seen previously among several ovarian cancer cell lines[11].
AMPK knockdown inhibits autophagic flux in EOC spheroids but does not alter p62 or LC3 processing
To elucidate the requirement of AMPK signaling regulation of autophagy in spheroids, we performed siRNA-mediated knockdown of the AMPK α1 and α2 catalytic subunits in iOvCa147-MA and OVCAR8 cells. AMPK exists as a heterotrimeric protein consisting of one catalytic α-subunit and two regulatory β- and γ-subunits. Although up to 12 different isomeric configurations are possible, there are only two known catalytic subunits encoded by the genes PRKAA1 and PRKAA2 [9]. Combined knockdown of PRKAA1 and PRKAA2 allowed us to control for variations in catalytic subunit expression and potential compensatory mechanisms, and to maximize AMPK attenuation. Following transfection in adherent conditions, cells were trypsinized and seeded into ULA conditions for 48 hours, at which point protein was collected for immunoblot analysis. To our surprise, PRKAA1/2 knockdown in iOvCa147-MA or OVCAR8 spheroids did not significantly alter LC3-II or p62 relative to siNT-transfected control spheroids (Figure 2a&b). This was intriguing since AMPK has been implicated in several models as a canonical activator of autophagy, with its loss typically inhibiting autophagic flux [14, 19, 20]. No significant difference in spheroid cell viability was observed between the PRKAA1/2 knockdown and siNT controls (data not shown), which corroborates the results from our previous study [8].
To further investigate the effect of PRKAA1/2 knockdown on autophagic flux in EOC spheroids, we used OVCAR8 cells stably-transfected with an eGFP-LC3B reporter construct [10]. Following PRKAA1/2knockdown, OVCAR8-eGFP-LC3B cells were seeded as spheroids and assessed using live-cell fluorescence microscopy. We observed a notable increase in green fluorescence in spheroids following PRKAA1/2 knockdown indicating a block in autophagic flux (Figure S1). However, it is difficult to draw this conclusion, as well as adequately monitor autophagic progression from early-to-late stages, with a single fluorescence reporter construct. To address this issue, we stably transfected OVCAR8 cells with the dual fluorescence mCherry-eGFP-LC3B reporter [21]. Following autophagosome fusion with the acidic lysosome, the pH-sensitive eGFP signal is quenched, whereas the mCherry signal remains unaffected. Highly autophagic cells will exhibit predominantly red fluorescent punctae indicative of increased autophagic flux. Conversely, inhibiting autophagy induces an increase in green fluorescence due to reduced autophagosome fusion with lysosomes. Although this reporter has been used in adherent culture systems[21, 22], it can also be applied to spheroid models [23]. By placing OVCAR8 mCherry-eGFP-LC3B cells into ULA conditions and assessing overall fluorescence colour shift rather than individual autophagic punctae, we can characterize general autophagic flux within spheroids in a rapid manner.
PRKAA1/2 knockdown in OVCAR8 mCherry-eGFP-LC3B spheroids resulted in a dramatic increase in green and red fluorescence relative to siNT-transfected control spheroids, which had predominantly low levels of fluorescence signal (Figure 2c&d). To confirm our interpretation of a block in autophagic flux, we treated spheroids with chloroquine (CQ), a well-characterized lysosomotropic agent that inhibits lysosomal fusion to the autophagosome[12], and which we have demonstrated previously inhibits autophagy in EOC cells and spheroids [10, 11]. Treatment of OVCAR8 mCherry-eGFP-LC3B spheroids with 50 μM CQ for 4 hours resulted in similar accumulation of green fluorescence as we observed with the PRKAA1/2 knockdown (data not shown). Thus, PRKAA1/2 knockdown can reduce autophagic flux in EOC spheroids; however, based on our immunoblot data, this observed AMPK-mediated regulation of autophagy may occur in an LC3- and p62-independent manner.
In addition to PRKAA1/2 knockdown, we sought to examine the effect of a pharmacological inhibitor of AMPK on EOC spheroids. Currently, Compound C (also known as dorsomorphin) is the only known selective inhibitor of AMPK [24]. Treatment of both iOvCa147-MA and OVCAR8 cells with Compound C resulted in modest reduction of p-AMPK at 10 mM (Figure 3a), yet significant increases in LC3 processing and a slight increase in p62 levels were observed (Figure 3a&b). OVCAR8 mCherrry-eGFP-LC3B spheroids treated with 10 μM Compound C for 24 hours exhibited a detectable increase in green fluorescence relative to their DMSO-treated controls (Figure 3c).
AMPK activation alone is insufficient to induce autophagy
We have shown previously that proliferating adherent EOC cells have relatively low levels of both autophagy and p-AMPK yet are robustly induced upon spheroid formation [8, 11]. As such, we deemed it important to test whether AMPK activity on its own is sufficient to induce autophagy in EOC cells. To achieve this, we activated AMPK by using the mitochondrial inhibitors oligomycin (100 nM) and metformin (2 mM), since both drugs are known to increase p-AMPK and its activity [25]. Oligomycin and metformin treatment for 24 hours led to increased p-AMPK in adherent OVCAR8 cells, however no significant changes were observed in LC3-II processing or p62 levels (Figure S2). Taken together, our results suggest that AMPK activation is required in part for autophagic flux in EOC spheroids, yet on its own is insufficient to induce autophagy in adherent EOC cells.
Pharmacologic inhibition of CAMKKβ reduces AMPK phosphorylation and inhibits autophagic flux
Due to the limited availability of small molecule inhibitors of AMPK, we sought to attenuate AMPK phosphorylation by targeting upstream kinases that lead to AMPK activation. Liver Kinase B1 (LKB1) encoded by STK11, is the best-characterized upstream kinase of AMPK. LKB1 is a highly-conserved serine-threonine kinase that typically functions as a regulator of cellular metabolism within the AMPK signaling axis [26]. Surprisingly, recent work from our laboratory identified that EOC spheroids lacking LKB1 expression by CRISPR-mediated STK11 knockout sustain elevated p-AMPK suggesting alternative kinase(s) target AMPK in our system [27].
Previous literature has implicated calcium/calmodulin-dependent protein kinase beta (CAMKKβ) as an alternative AMPK activating kinase [28]. For example, cellular matrix deprivation leads to CAMKKβ-mediated AMPK phosphorylation in breast cancer cell lines [29]. As such, we decided to use a selective CAMKKβ inhibitor, STO-609, as another method to attenuate AMPK phosphorylation. We included additional cell lines, the high-grade serous cancer COV318 cells, and immortalized human fallopian tube secretory epithelial FT190 cells. Treatment of non-malignant FT190 and EOC cell line spheroids with 10 μM ST0-609 resulted in significant reduction in p-AMPK (Figure 4a). In addition, we observed a significant increase in p62, but no change in LC3-II levels (Figure 4b). ST0-609 treated spheroids exhibited increased green fluorescence relative to their DMSO control indicating robust autophagic flux inhibition (Figure 4c). Enhanced green fluorescence was observed in FT190 spheroids, suggesting that CAMKKβ-mediated regulation of AMPK and autophagic flux occur in both EOC cells and their potential premalignant precursor cells, too.
We have demonstrated previously that autophagy is critical to maintain EOC cell viability in spheroids[10, 11], thus we postulate that potent inhibition of AMPK activity using STO-609 over an extended period would negatively impact EOC spheroid cell viability. First, we treated OVCAR8 mCherrry-eGFP-LC3B spheroids with 10 mM STO-609 or DMSO for up to 12 days to visualize extent of autophagic flux inhibition and spheroid integrity. Autophagic flux was potently blocked by STO-609 over the complete time period as evidenced by increased green fluorescence; we also observed a small decrease in overall spheroid size due to STO-609 treatment (Figure 5a&b). Subsequently, we evaluated the effects of CAMKKb-AMPK signaling inhibition by treating nine different high-grade serous EOC spheroids, and FT190 spheroid controls, with STO-609 for 3 and 7 days prior to quantifying viable cell number. After 3 days of STO-609 treatment, we observed a significant reduction in cell viability in 6 out of 9 EOC cell line spheroids (Figure 6a); extending this treatment to 7 days resulted in an additional cell line (OVCAR4) sensitive to CAMKKβ-AMPK inhibition (Figure 6b). Cell viability for two EOC cell lines, CaOV3 and COV318, as well as normal FT190 spheroids, was unaffected by STO-609 treatment at both time points. In summary, our results implicate CAMKKb-mediated activation of AMPK is required for autophagy induction and resultant cell survival in EOC spheroids.