V-ATPase inhibition reduces GBM growth in vivo and in vitro

We previously showed that different conformations of the V-ATPase pump are expressed in low-grade glioma respect to GBM and that the pump conformation affects glioma growth in vivo. [25]. High expression of the catalytic V-ATPase G1 subunit is significantly associated with most aggressive gliomas. [19]. Therefore, we started this study inhibiting V-ATPase activity using Bafilomycin A1 (BafA1), a macrolide molecule that binds to the V0 domain of the pump thus inhibiting its rotation and proton translocation [26].

At the cytostatic concentration of 5 nM BafA1 (Suppl. Figure 1a,b), the pump activity in lysosomes was completely inhibited (Suppl. Figure 1c,d) and cell cycle progression was arrested in S phase (Suppl. Figure 1e,f). In the orthotopic mouse model of GBM, [25], treatment of GSCs with BafA1 was sufficient to slow tumor growth (Fig. 1a,b), and to decrease tumor cell invasion (STEM121), proliferation (Ki67) and growth (p-S6) (Fig. 1c).

Fig. 1

V-ATPase inhibition in GSCs decreases glioma growth in vivo. a, b Luciferase-transduced GSCs treated with vehicle or 5 nM BafA1 for 24 h were intracranially injected in nude mice and glioma growth in vivo was evaluated by luciferase emission up to 144 days (BL Intensity; a,b). At sacrifice brains were collected and formalin-fixed and paraffine-embedded for morpho-histological examination. Then, a section was stained with STEM121, phosphorylated S6 at Serine 240/244 (pS6) and ki67 antibodies (brown color; c) to visualize respectively glioma cell infiltration in the brain parenchyma, cell growth and proliferation. STEM121, pS6 and ki67 staining was quantified using Image Scope (Leica) (j). Scale bar 1 mm left panel, 100 μm zoomed inset. Representative images are shown for A and C panels. **, p = 0.003 by Mann–Whitney U test

BafA1 is a well-known inhibitor of the last stages of autophagy [24, 27], a pathway deeply connected with lysosomal function that could promote cell death [28]. Therefore, we analyzed whether autophagy was affected after V-ATPase block. In BafA1-treated GSCs the autophagic markers LC3B-II and p62 showed only a marginal accumulation over time (Suppl. Figure 2a,b), while no difference in autophagosome could be detected in BafA1-treated GSC compared to controls (Suppl. Figure 2c), supporting the previous observations about autophagy impairment upstream of autophagosome formation in GSC with elevated expression of V-ATPase [24].

In keeping with this, the lysosomal compartment was not affected by the V-ATPase block, given that Lamp1 expression and the number of lysosomes was unaffected by BafA1 (Suppl. Figure 2b,e respectively), and the major transcription regulator or the autophagy-lysosomal pathway TFEB was not detected in GSC nuclei (Suppl. Figure 2f). Therefore, this information supports the idea that autophagy is not a pathway key for GSC bioenergetic requirements.

On the contrary, the master energy sensor AMP-activated protein kinase (AMPK) was activated after BafA1 treatment (Suppl. Figure 3a-e) while the activity of the kinase p70S6K (Suppl. Figure 3b,c), the ribosomal protein S6 (Suppl. Figure 3f,g) and of eukaryotic initiation factor 2 (eI2F) alpha subunit (Suppl. Figure 3d,e) was inhibited suggesting that the mTORC1 signaling was inactivated by BafA1 treatment and global translation was compromised (Suppl. Figure 3a).

These data indicate that the inhibition of V-ATPase activity induces a state of energy deficit in primary GSCs without the involvement of autophagy. This metabolic change is crucial for GSCs growth in vivo.

V-ATPase modulates mitochondria homeostasis in GSCs

Next, we better characterize GSCs bioenergetics following V-ATPase inhibition, to elucidate the contribution of the proton pump to GSCs viability. BafA1 treatment in GSCs increased reactive oxygen species (ROS) levels (Fig. 2a,b), and induced mitochondria depolarization (Fig. 2c,d and Suppl. Figure 4a). ROS levels were not reverted to basal conditions after incubation with ROS inhibitor (Fig. 2a,b), suggesting that the inhibition of V-ATPase activity causes irreversible mitochondrial damage. To corroborate this finding, we performed ultrastructural analysis of GSCs at baseline or after BafA1 treatment. Despite the total number of mitochondria was not affected by V-ATPase activity, BafA1 treatment increased damaged organelles while diminishing the number of healthy ones (Suppl. Figure 4b-d). Using a live cell metabolic assay to dynamically measure mitochondrial function (Fig. 2e), we found that V-ATPase impairment decreases basal (Fig. 2f,g) and maximal (Fig. 2f,h) respiration in GSCs cultures. In contrast, the non-mitochondrial respiration was unaffected by BafA1 treatment (Suppl. Figure 4e).

Fig. 2

A mitochondrial pool of V-ATPase is present in GSC and V-ATPase targeting perturbs mitochondria homeostasis. a-d GSCs were incubated with vehicle (Ctrl) or 5 nM BafA1 for 48 h and then ROS production (a,b) or a mitochondrial membrane potential assay (TMRE; c,d) were performed. *, p = 0.03 by Mann–Whitney U test. e–h The mitochondrial function (e,f) was evaluated live in GSCs incubated with vehicle (Ctrl) or 5 nM BafA1 for 48 h as the oxygen consumption rate (OCR) using selective uncouplers of the electron transport chain (oligomycin, FCCP, and rotenone-antimycin A). The basal (g) and maximal (h) respiratory capacity was then measured in GSCs. **, p = 0.005; #, p = 0.008 by Mann–Whitney U test. i,j Mitochondrial proteins were analyzed by immunoblot (i) separately in cytoplasmic (Cyto) and mitochondrial (Mito) extracts and quantified by densitometric analysis (j) in GSCs treated as in a. β-tubulin was used to verify cytoplasmic contamination. k A proximity ligation assay (PLA) was performed with antibodies recognizing mitochondria (Tomm20) and V-ATPase G1 (V1G1), or lysosomes (Lamp1) and V-ATPase G1. Nuclei were stained with Hoechst 33,342. Left, representative images; scale bar, 20 µm. Right, quantification of PLA spots per nuclei. Double Tomm20-V1G1 and Lamp1-V1G1 staining were compared to single antibody staining. See Supplementary Fig. 3C for controls. l Colocalization of V-ATPase G1 (V1G1) with mitochondria (Tomm20) in GSCs was analyzed by confocal microscopy followed by 3D deconvolution. Left, representative image; scale bar, 20 µm. Right, quantification of colocalization was performed with Leica software. Bars, mean with SEM

Analysis in purified mitochondrial extracts of enzymes from the matrix and the outer, and inner membrane showed that their expression was reduced after BafA1 treatment (Fig. 2i,j), without modulation of mitophagy (Suppl. Figure 5a,b). Further, we observed V-ATPase G1 localization at the organelle membrane (Fig. 2 i,j).

While a connection between the proton pump and mitochondria, the cell powerhouse, has already been shown in C. Elegans [29] and in a zebrafish model of deafness [30], no evidence in mammalian systems has been reported before.

To validate our novel finding, we analyzed V-ATPase G1 subcellular localization in GSCs using a proximity ligation assay (PLA) and confocal microscopy followed by 3D deconvolution. Both PLA (Fig. 2k and Suppl. Figure 5c) and confocal microscopy (Fig. 2l) showed that the V-ATPase G1 subunit colocalizes with Tomm20, confirming its presence on GSCs mitochondria stained with mitochondrial marker Tomm20. We also analyzed and detected the V-ATPase subunit in lysosomes by PLA (Fig. 2k; Lamp1-positive spots) as expected [7], while no interaction was appreciated between mitochondria and lysosomes (Suppl. Figure 5d).

These data suggest the involvement of mitochondrial V-ATPase in regulating GSCs bioenergetics, resulting in changes in mitochondria homeostasis.

Mitochondrial metabolism is hijacked by inhibition of V-ATPase activity

To get further insights into the reliance of GSCs on V-ATPase activity to preserve viability and bioenergetics, we evaluated basal ATP production rates from mitochondrial respiration and glycolysis in living GSCs (Fig. 3a). Following BafA1 treatment, GSCs showed decrease in PDH protein expression (Fig. 2i,j) and mitochondrial ATP production (Fig. 3b,c) and significant increase in glycolysis-derived ATP production as well as lactate intra-cellular accumulation (Fig. 3b-d), suggesting a downregulation of TCA/ oxidative phosphorylation (OxPhos) metabolism and a compensatory enhancement of glycolytic metabolism. In keeping with this, we observed increase of GLUT1 levels, modest decrease of LDHB. No modulation of LDHA and MCT4 (Fig. 3e,f) were observed, indicating that the increased glycolytic rate/lactate accumulation is most likely sustained by increased glucose uptake from the extracellular space. Combination of BafA1 and UK5099, a selective inhibitor of pyruvate import into the mitochondria, induces a further decrease maximal respiration, confirming the impairment of using pyruvate as energetic source under BafA1 treatment (Fig. 3g,h).

Fig. 3

Inhibition of V-ATPase activity reprograms GSCs metabolism. a-c A live ATP-rate assay was performed to simultaneously detect energy production from glycolysis and mitochondria in GSCs treated with vehicle (Ctrl) or with 5 nM BafA1 for 48 h. The oxygen consumption rate (OCR; b) was measured after oligomycin and rotenone-antimycin A injection and ATP produced by mitochondrial respiration (mitoATP) or by glycolysis (glycoATP) was measured and expressed as percentage of the total ATP (c). §, p = 0.012 by Mann–Whitney U test. D) GSCs cultures were incubated with 13C-glucose and the treated as in A. Lactate production was assessed by a Surveyor high performance liquid chromatography system. *, p = 0.026 by Mann–Whitney U test. e,f GSCs were treated as in A, then the expression of the indicated metabolic enzymes was analyzed by immunoblot (e) and quantified by densitometric analysis (f). β-Actin was a loading control. g,h GSCs were treated for 48 h with vehicle (Ctrl), 5 nM BafA1, the pyruvate transporter inhibitor UK5099 or the combination of BafA1 and UK5099 (g). Then, mitochondria reliance on pyruvate was assessed measuring the OCR and the maximal respiration capacity was determined (h). i Targeted metabolic profiling was performed in Ctrl- or BafA1-GSCs cell extracts. The heatmap show significantly decreased levels of amino acids in BafA1-treated cultures (for complete list see Suppl. Table S3)

In line with this, increased pyruvate reduction to lactate and reduced pyruvate entry into mitochondria affects the levels of pyruvate-derived amino acids, including alanine, aspartate, threonine, and mostly glutamine/glutamate (Fig. 3i, Suppl. Table S3). Equally possible is that their reduction reflects also their use as compensatory energetic/biosynthetic sources in the TCA. Indeed, GSCs have been previously shown to rely on Gln for survival and for maintaining ATP production, cell growth and survival [31].

As lactate acts as signaling molecule to activate transcription of genes involved in mitochondria biosynthesis and pro-survival factors in normal astrocytes and in neuroblastoma cells, we assessed whether this might occur in our system as well [32, 33].

In contrast to previous reports, gene expression data showed that NFR1, a transcription factor involved in mitochondrial DNA replication, and several cell-cycle related genes were decreased in BafA1-treted GSCs (Suppl. Figure 6a,b), suggesting that increased intra-cellular lactate levels cannot compensate for BafA1-mediated mitochondrial metabolism dysregulation.

Bafilomycin A1-mediated metabolic reprogramming of GSCs

To further support results from metabolomics/metabolic assays, we analyzed the expression levels of metabolic-related genes in control and BafA1-treated GSCs (Suppl. Table S4). In line with results from metabolomics/metabolic assay, transcriptomics data showed that GSCs are characterized by proliferative signaling and OxPhos as main metabolic pathway (Fig. 4a and Suppl. Figure 7a). Treatment of GSCs with BafA1 shut-downs OxPhos while activates glucose metabolism and induces of amino-acids stress responses (mTORC1) (Fig. 4a and Suppl. Figure 7b). Hallmark analysis of K-means clusters (Suppl. Figure 8a-c and Supplementary Table S5) confirmed down regulation in BafA1-treated GSCs of key genes involved in oxidative phosphorylation, such as NADH dehydrogenase and Cytochrome C Oxidase subunits, and cell cycle progression, including CDC20, POLE, BUB1 as well as the Polo Like Kinase 1 (PLK1), which has been reported as crucial for GSCs stemness and viability (Fig. 4b) [34]. In contrast, BafA1-treated GSCs upregulated antioxidant genes (ROS hallmark), transcripts involved in fatty acid de novo synthesis (FASN, ACACA, SCD). In regards of the mTORC1 pathway, BafA1-treated GSCs up-regulated factors involved in amino acids and glucose transport (Solute Carrier Family members) (Suppl. Figure 8d), glucose metabolism (HK2, PFKL, GPI, GSPD), as well as anabolic enzymes (GOT1, ASNS) (Fig. 4b, Suppl. Figure 8a-c and Suppl. Table S6).

Fig. 4

Molecular insights into BafA1-mediated metabolic rewiring of GSCs. a A metabolic gene expression panel (n = 748 genes; see also Suppl. Table S4) was analyzed in GSCs treated with vehicle (Ctrl) or 5 nM Bafa1 for 48 h. List of genes whose expression was significantly up (adj p value < 0.05 and log2FC > 0.5) in Ctrl-GSCs or in BafA1-GSCs were separately imported in Reactome web-tool and the pathway analysis was performed with voronoi pathway visualization. The most relevant activated signaling in Ctrl- or BafA1-GSCs are shown. For the complete voronoi charts please refer to Suppl. Figure 7. b Transcripts belonging to the indicated hallmark and up (red) or down-modulated (green) in BafA1-treated GSCs are shown (see also Suppl. Table S5). c The expression of the activating transcription factor 4 (ATF4) was analyzed in Ctrl- and BafA1-treated GSCs. Data are showed as violin plots and each dot is a sample. **, p = 0.002 by Mann–Whitney U test. d Ingenuity pathway analysis was performed integrating gene expression (GEX) and metabolic data. Pathway predicted as activated or inhibited (positive and negative z-score, respectively) in BafA1-GSCs are shown (see also Suppl. Table S7). f Protein de novo synthesis was evaluated in GSCs incubated for 24 h with vehicle (Ctrl) or 5 nm BafA1 using fluorescence microscopy. The mean fluorescence intensity (MFI) was measured and quantified by Image J software (bottom panel). Scale bar, 20 µm. ***, p < 0.0001 by Mann–Whitney U test

These data depict a scenario whereby BafA1-GSCs as cells are facing hindered proliferation and an energetic/biosynthetic crisis, demanding sources from the extracellular milieu antioxidant enzymes to dampen increased ROS [31].

Finally, we measured the expression levels of the transcription factor ATF4, a nutrient sensor and master regulator of the amino acid deprivation response [35]. In BafA1 treated GSCs, the expression of ATF4 increased significantly (Fig. 4c). The combination of gene expression output and metabolomic analyses showed that BafA1 treatment promotes AMPK while inhibiting glutamate/glutamine metabolism (Fig. 4d and Suppl. Table S7). Accordingly, BafA1 treatment induces a significant impairment of protein biosynthesis in GSCs (Fig. 4e), in line with the reduced levels of amino acids observed in metabolomics (Fig. 5).

Fig. 5

Schematic summary of V-ATPase-directed bioenergetic states in GSCs. GSCc are characterized by a mitochondrial pool of V-ATPase, and the active form of the pump supports GSCs growth and proliferation through mitochondria-dependent energetics