Cancer is still a major of death cause worldwide and limited therapeutic options, severe therapy side effects, and poor clinical outcomes clearly depict the urgent need to develop novel strategies for this disease (1, 2). One important recently emerged hallmark of cancer is metabolic reprogramming. Otto Warburg and his co-workers showed early on that cancer cells produce lactic acid from glucose even under non-hypoxic conditions, since then known as the Warburg effect (3). Nowadays it is evident, that cancer cells undergo extensive metabolic reprogramming to meet their increased energetic demands. They alter cellular pathways to generate energy, produce co-factors for anabolic reactions, and build metabolic precursors, while several generated metabolites also exert a signaling function promoting tumor growth and progression. This extends not only to glycolysis, but also to glutamine, serine, methionine, arginine, and lipid metabolism (4).

A challenge when developing strategies is to decide which metabolic adaptation is best to address. A promising target in that regard is glutamine metabolism, or ‘glutaminolysis’. “Glutamine addiction,” i.e., excessive consumption of glutamine from human plasma or cell culture media in amounts beyond those necessary for protein synthesis, is exceedingly common across different tumor entities (5). Glutamine is imported into the cells by specialized amino acid transporters, such as SLC38A1 and SLC1A5 and transported into mitochondria mainly for energy generation. Glutamine is hydrolyzed to glutamate by glutaminases and can be further metabolized by glutamine dehydrogenase (GLUD) or glutamic-oxaloacetic transaminase (GOT) to α-ketoglutarate (αKG), which feeds into the TCA cycle to provide energy and building blocks (6). The metabolic pathways glutamine feeds into are versatile and at the same time essential for tumor growth and survival. Therefore, targeting glutamine metabolism represents a promising anti-cancer approach (7).

In recent research, the lysosomes turned out as master regulators of cellular nutrient sensing and adaptation to metabolic stress. Cellular glucose and amino acid levels are sensed in lysosomes by the LYNUS (lysosomal nutrient sensing) machinery. This multi-protein complex resides on the lysosomal surface and includes the V-ATPase, a lysosomal proton pump responsible for luminal acidification, and mTORC1, a master regulator of metabolic adaptation processes, which is tethered to the V-ATPase by several adapter proteins (8). Upon starvation, mTORC1 is inactivated, leading to the induction of lysosomal biogenesis and autophagy to maintain cellular energy generation (9). Recently, the V-ATPase has also been implicated in cellular metabolism and adaptation to starvation conditions by mTORC1 and AMPK signaling (10). There is also initial evidence that targeting V-ATPase influences glutamine dependence of cancer cells (11). Additionally, luminal acidification of lysosomes, facilitated by the V-ATPase, is a decisive prerequisite for fusion with autophagosomes and subsequent cargo degradation. V-ATPase inhibition is hence an effective way to inhibit autophagy (12). Given the role of V-ATPase nutrient sensing and autophagy, our working hypothesis was that inhibition of autophagy by V-ATPase inhibition, concomitantly with glutamine-deprivation, leads to increased growth-inhibition and apoptosis, especially in glutamine-dependent cancer cells. This strategy may be a novel therapeutic option and thus underlying cellular signaling processes warrant thorough examination.

To investigate the potential of the V-ATPase inhibition combined with glutamine deprivation in cancer cells, we first determined glutamine dependency in different cancer cell lines, originating form colorectal carcinoma (HCT-15, HT-29), pancreatic cancer (BxPC3, Panc03.27, Panc10.05), breast cancer (MCF-7), hepatocellular carcinoma (HUH-7), bladder cancer (T24), and cervical cancer (HeLa). Therefore, we analyzed proliferation capacity of these cells in the presence or absence of glutamine. Glutamine deprivation led to a reduction in proliferative capacity in all tested cells, however, to different extents (Figure 1A). As a study model, we focused on colorectal cancer and pancreatic cancer lines for two main reasons: (a) glutamine dependency has been reported in these tumor entities also in clinical settings (18, 19) and (b) they showed medium (about 40-60%) reduction in proliferation, which is a feasible amount for implementation of a potential combination therapy strategy. Since lysosomes are the cellular centers for nutrient sensing, we first assessed the impact of glutamine deprivation on lysosomal characteristics. While lysosomal volume was unchanged as indicated by LysoTracker™ Green staining (Supplementary Figure 1A), lysosomes were more acidic upon glutamine-deprivation indicated by semi-quantitative analysis of LysoSensor™ Green fluorescence (Figure 1B). This was further confirmed by pH-sensitive quenching of FITC-dextran fluorescence (Supplementary Figure 1B). This phenomenon could be reversed by simultaneous treatment with the V-ATPase inhibitor archazolid (Figure 1B and Supplementary Figure 1B).

As glutamine starvation has been reported to induce autophagy as a survival mechanism (20) and that V-ATPase inhibition blocks autophagy (16), we analyzed autophagy. In our model, glutamine deprivation led to a decrease in LC3-II protein levels, indicating autophagy induction. As expected, V-ATPase inhibition could reverse this effect and block autophagic flux (Figure 1C). Of note, p62 levels remained unaffected (Supplementary Figure 1C). Analysis of phosphorylation of its downstream target p70S6K revealed decreased mTORC1 activity, evident by a reduction in p70S6K phosphorylation, which was also not further affected by archazolid co-treatment (Figure 1D). Furthermore, glutamine deprivation led to increased phosphorylation of the cellular energy sensor AMPK, which was not affected by V-ATPase inhibition (Figure 1E). Taken together, glutamine deprivation apparently induces changes in lysosomal properties and autophagic flux of cancer cells, which can be reversed by V-ATPase inhibition. These data support our hypothesis, that inhibiting V-ATPase in glutamine-deprived conditions could block cellular survival mechanisms and might be beneficial for cancer therapy.

To test the therapeutic benefit, we analyzed proliferation and apoptosis upon glutamine deprivation by starvation or pharmacological glutaminase inhibition with BPTES or CB-839 together with V-ATPase inhibition. Using crystal violet staining, we found no difference in dose-sensitivity to archazolid in full-medium versus glutamine-deprived conditions. The same phenotype was observed upon glutaminase inhibition with CB-839 or BPTES (Figure 2A). Additionally, we did observe a moderate increase in apoptosis upon the combination, yet no synergistic effect as hypothesized (Figure 2B). Of note, we could observe a significant effect on proliferation and on apoptosis for a combination of BPTES with archazolid in HCT-15 cells (Figures 2A, B). However, BPTES is an inhibitor of the multidrug efflux transporter P-gp (Supplementary Figure 2), of which archazolid is a known substrate. Hence the additive effect might result from an increased intracellular drug concentration, rather than from glutaminase inhibition. Taken together, these findings indicate that contrary to our hypothesis, V-ATPase inhibition has no therapeutic benefit in glutamine-deprived cancer cells despite reversing lysosomal and autophagic effects of glutamine deprivation (Figure 1).

We were then curious to understand why there was no beneficial therapeutic effect of V-ATPase inhibition. We hypothesized that the cancer cells might be able to circumvent proliferation inhibition or apoptosis, by metabolic reprogramming. Therefore, we analyzed solute carrier family transporters (SLC) which have broad substrate specificity, including all neutral amino acids, like glutamine and alanine. Most SLCs were transcriptionally upregulated in HCT-15 cells and downregulated in BxPC-3 cells upon glutamine deprivation, while V-ATPase inhibition had no additional effect (Figure 3A and Supplementary Table 1). Ubiquitously expressed transporters SLC38A1 and SLC1A5 (21) were modestly upregulated in HCT-15 cells on mRNA and protein level, but not in BxPC-3 cells (Figures 3A, B). Confocal microscopy showed, that glutamine deprivation has no effect on transporter localization, while archazolid treatment leads to an accumulation of transporters in intracellular vesicles (Figure 3C). To assess transporter activity, we pre-starved the cells and then monitored L-[³H]-glutamine uptake, which was unchanged in BxPC-3 cells but increased in Panc03.27 cells. V-ATPase inhibition again had no effect (Figure 3D). Employing the amino acid analogue N-methylaminoisobutyric acid (MeAIB), and L-threonine, which are known substrates of SLC38A1 and SLC1A5, respectively, we identified SLC1A5 as main import transporter in our cells. While competition of L-[³H]-glutamine with MeAIB did not affect the uptake of glutamine, competition with L-threonine almost completely blocked uptake (Figure 3D). Taken together, we identified SLC1A5 as the main amino acid importer in our cells, which is upregulated upon starvation, supposedly to increase nutrient uptake and maintain intracellular energy generation constant.

There is evidence in the literature, that apart from glutamine uptake via SLC transporters, glutamine supply can also be sustained break-down or proteins to amino acids within the lysosome. These amino acids can be exported into the cytosol by SLC38A7 (17) and subsequently utilized to maintain energy generation. Analyzing intracellular fluorescence of a transferrin-AlexaFluor™488 conjugate as a model substrate, reveals an increased internalization upon glutamine starvation (Figure 3E). Additionally, we found an increase in SLC38A7 mRNA (Figure 3A), which might hint to a SLC38A7-mediated compensation mechanism upon glutamine deprivation. However, siRNA-mediated knock-down of SLC38A7 did not increase the effect of concomitant glutamine starvation and V-ATPase inhibition (Figure 3F, Supplementary Figure 3, and Supplementary Table 2). Further analysis of glutamine metabolism, i.e., expression levels of glutaminolysis-associated enzymes, revealed no differences (Figure 3G) that could account for the missing therapeutic benefit of combining glutamine deprivation with V-ATPase inhibition.

We then posed the hypothesis, that cells undergo a metabolic shift from glutaminolysis to glycolysis when challenged with V-ATPase inhibition under glutamine-deprived conditions. A Seahorse™ glycolytic stress test revealed an increase in the extracellular acidification rate (ECAR) in all cell lines, when challenged with glutamine deprivation alone and together with archazolid treatment (Figure 4A). Subsequent addition of glucose, to induce glycolysis, oligomycin, to block mitochondrial respiration and therefore forcing the cells to fully rely on glycolysis, and 2-DG, to completely shut down glycolysis, allows assessment of glycolytic parameters. Upon glutamine deprivation in combination with V-ATPase inhibition, cells have an increased glycolytic capacity and glycolytic reserve (Supplementary Figure 4). By using CYRIS^® flox, a novel real-time platform to analyze cellular metabolism, we could verify our observation that cells have an increased glycolytic capacity and glycolytic reserve upon glutamine deprivation, which were both even enhanced upon V-ATPase inhibition (Supplementary Figures 5A, B). To confirm the metabolic shift toward glycolysis, we assessed glucose uptake and lactate production in our cells. Analyzing the uptake of fluorescent glucose analogue 2-NBDG by flow cytometry revealed an increased substrate uptake upon glutamine deprivation (Figure 4B). Along the line, lactate production increased likewise (Figure 4C). We additionally analyzed several glycolytic key enzymes but found no significant change in expression (Supplementary Figure 5C and Supplementary Table 3). As enzyme activity does not necessarily correlate with expression, we assessed the combined activity of all hexokinase isoforms upon glutamine deprivation in combination with archazolid treatment. Hexokinases catalyze the phosphorylation of glucose to glucose-6-phosphate, which represents the first, obligatory, and rate-limiting step of glycolysis. Upon glutamine deprivation or V-ATPase inhibition, Hexokinase activity did not significantly change, however, upon a combination of both, enzyme activity greatly increases (Figure 4D). This further strengthens the finding that glycolysis is induced to maintain cellular metabolism when challenged with glutamine deprivation and inhibition of autophagy.

Metabolic reprogramming, like increased reliance on glutamine metabolism, is an established hallmark of cancer cells, which is an intensively investigated anti-cancer treatment strategy (22, 23). In this study, we aimed to enhance the therapeutic effectiveness of glutamine deprivation by co-inhibiting the V-ATPase. Glutamine starvation induced pro-survival autophagy, which could be blocked by V-ATPase inhibition. However, we could not observe a synergistic anti-cancer effect. Mechanistically, cells upregulate glycolysis which we conclude contributes to the maintenance of survival.

Cellular nutrient sensing is centered at lysosomes, which therefore regulate metabolic adaptations in accordance with nutrient availability, by catabolic autophagy pathways (11, 24). However, the exact role of autophagy upon starvation is still debated and differs with the diminished nutrients. Concerning glutamine deprivation, there is accumulating evidence, that lysosome-dependent pro-survival autophagy is induced. For instance, Yu et al. (25) show that deprivation of glutamine and serum leads to a rapid fusion of lysosomes with autophagosomes to form large autolysosomes in a variety of cell lines across different species, indicating that degradation of lysosomes is an evolutionarily conserved response to starvation. In line with that, Mukha et al. (26) report that ATG5-dependent autophagy is a pro-survival response upon glutamine deprivation of prostate cancer cells. Additionally, others have shown that also glucose starvation induces ATG5 and LC3-dependent microautophagy in cancer (27). With our study, we can add evidence that glutamine starvation leads to significant acidification of lysosomes accompanied by autophagy induction, as evident by a decrease in LC3-II level, which goes in hand with a decreased mTORC1 activity (Figure 1F). In contrast to these findings, an elevation of lysosomal pH following serum starvation was reported, which did not impact autophagy (28). In addition, Wilden et al. (29) found that starvation with Hank’s Balanced Salt Solution (HBSS) does not affect lysosomes or autophagy induction. Yet, when they simultaneously disrupted lysosomal function with chloroquine, they observed large vacuoles, which could be dissipated by the re-addition of glutamine. Interestingly, Chiodi et al. (30) further report autophagy induction by glucose starvation, but not serum or glutamine starvation, in human fibroblasts. Taken together, our results support evidence in the literature, that glutamine starvation induces LC3-dependent autophagy in cancer cells. However, whether autophagy induction is a conserved response to any lack of nutrients is still a matter that requires further investigation.

It is therefore a reasonable therapeutic approach to block these adaptations simultaneously with glutamine deprivation. There are to date many pre-clinical and clinical studies available, that show a beneficial effect of combining glutaminase inhibitors with additional therapeutic strategies. In this regard, several studies in a broad variety of tumor entities have shown synergistic effects of glutaminase inhibitors with frequently used cytostatics (31–35). Other studies have even found that glutaminase inhibition can overcome therapy resistance for instance in prostate cancer cells, which can be re-sensitized to radiation therapy (26), or methotrexate. Another study reports that glutamine deprivation in the core regions of tumors contributes to the formation of cancer stemness, which can be inhibited by co-inhibition of mitochondrial fission (36). As our results demonstrate that glutamine deprivation induces autophagy (Figure 1), we investigated the therapeutical benefit of a combination of glutaminase inhibitors with the V-ATPase inhibitor archazolid, a known anti-cancer compound that inhibits autophagy. However, we found no beneficial effect.

In contrast to our observation, several reports in the literature describe a synergistic effect of glutaminase inhibition and inhibition of lysosomal function. This was shown in colorectal cancer cells for a combination of glutaminase inhibitor 968 with the lysosomotropic compound chloroquine (36–38). Similar results were described by Seo et al. (39) who could show that in pancreatic adenocarcinoma, glutamine, but not glucose, deprivation in combination with chloroquine resulted in a tremendous induction of cell death. In our study, however, glutamine deprivation in combination with autophagy inhibition did not lead to an augmentation in proliferation inhibition or apoptosis except for the effect of a BPTES and archazolid combination in HCT-15 cells (Figure 2). As this result does not fall in line with starvation or CB-839 treatment and data of the other cell lines, we figured an alternative mode of action could be underlying. Archazolid is a known substrate of multidrug resistance transporter P-glycoprotein (P-gp) (40), which is highly expressed in HCT-15 cells (41). Since we found that BPTES blocks P-gp activity, we conclude that the additive effect we observed by combining archazolid and BPTES rather results from increased cellular drug availability of archazolid than from glutaminase inhibition (Supplementary Figure 2).

Curious as to why we only detected moderate effects of glutamine deprivation and simultaneous autophagy inhibition in our model, we hypothesized, that when challenged with these conditions, cells can further adapt their metabolism to use other nutrients to ensure survival. There is evidence, that for instance exogenous asparagine or alanine might be used as alternative substrates in glutamine deprived conditions (37, 42, 43). Extracellular alanine, like glutamine, is taken up by SLC1A5 in exchange for export of arginine (44), hence SLC1A5 could play an important role in switching metabolism. Our results identify SLC1A5 as the main amino acid transporter in our model, which is activated upon starvation, however, it is also trapped in intracellular vesicles upon glutamine deprivation in combination with archazolid (Figure 3). This indicates, that increased uptake of exogenous amino acids might not be the primary metabolic escape of our cells. Interestingly, Liu et al. (45) discovered a close interaction between glutaminolysis and glycolysis in glioblastoma. In essence, they discovered a MTOR2/C-MYC/GFAT1 axis, which is responsible for a cross-talk between these pathways and is switched on and off to adapt to substrate availability. It is noteworthy, that the V-ATPase has been controversially implicated in glucose metabolism. While it was shown that the V-ATPase inhibitor archazolid leads to an upregulation of glucose uptake (46), others report that knock-down of the V-ATPase reduced glucose uptake and decreased extracellular acidification rate, indicating glycolysis inhibition (47). Our results on the other hand show, that V-ATPase inhibition alone does not yet influence extracellular acidification rate or glycolytic parameters in a glycolysis stress test significantly, but moderately induce glucose uptake and lactate production. Yet, when cells that are subjected to glutamine deprivation in combination with archazolid, they undergo a metabolic shift toward glycolysis, as evident by an increased glycolytic capacity and reserve, as well as strongly increased glutamine uptake an lactate production (Figure 4). Supporting this finding, we also detected an increase in activating phosphorylation of AMPK activity (Figure 1E) upon starvation, which is known to promote glycolysis (48). Along the line with our previous reports, archazolid did not further affect AMPK phosphorylation (10). We conclude that this metabolic adaptation to increased glucose reliance can at least partly restore cellular survival capacity, thereby preventing a beneficial effect of combining glutamine deprivation with archazolid. Furthermore, our report indicates an considerable metabolic plasticity, as we previously reported an increased glutamine dependency of cancer cells upon V-ATPase inhibition (49), which can, as we describe in this study, be reversed to increased reliance on glycolysis. Whether further metabolic routes contribute to metabolic plasticity in response to different treatment strategies is an interesting subject for future research.

The original contributions presented in this study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

FL and AG performed the experiments and analyzed data. FG and KB wrote the manuscript. FL, FG, AV, KS, and KB contributed to the conception and design of the study. KB supervised the project. All authors contributed to the manuscript revision, read, and approved the submitted version.