Autophagy-Dependent Metabolic Reprogramming Sensitizes TSC2-Deficient Cells to the Antimetabolite 6-Aminonicotinamide

Autophagy is a catabolic process leading to the degradation of cytoplasmic content in lysosomal compartments, thereby providing an endogenous source of nutrients for energy production under stress conditions and allowing dysfunctional organelles, including mitochondria, to be “recycled” (1). Tumor cells often upregulate autophagy to promote survival and drug resistance (2). The role of autophagy in cancer therapy is an area of active clinical investigation.

Tuberous sclerosis complex (TSC) is an autosomal dominant syndrome caused by germline mutations of the TSC1 and TSC2 tumor–suppressor genes (3). Patients with TSC develop histologically benign but life-threatening proliferative lesions in multiple organs, including the brain, heart, lung, and kidney. The TSC1 and TSC2 gene products form a complex that suppresses mammalian target of rapamycin complex 1 (mTORC1). mTORC1 is a major regulator of protein translation, autophagy, and metabolism. Hyperactive mTORC1 occurs in the tumors that develop in patients with TSC and in many human cancers, leading to differential metabolic requirements in the setting of low autophagy levels (4, 5).

Autophagy inhibition has been shown to suppress the growth of tumor cells (6). We previously found that constitutive activation of mTORC1 in TSC2-null cells leads to decreased levels of basal and stress-induced autophagy, and increases their vulnerability to further autophagy inhibition (7), yet the metabolic consequences of autophagy inhibition in the context of hyperactive mTORC1 are not fully understood. Moreover, the metabolic pathways used by tumor cells to maintain growth and survival when autophagy is inhibited are poorly characterized. We report here that the metabolic reprogramming induced by autophagy inhibition leads to pentose phosphate pathway (PPP) “addiction” selectively in Tsc2-deficient cells. We also provide proof-of-concept that the metabolic consequences of mTORC1 hyperactivation can be therapeutically targeted without inhibiting mTORC1 itself, thereby revealing novel therapeutic strategies for TSC.

Tsc2^+/+p53^−/− and Tsc2^−/−p53^−/− mouse embryonic fibroblasts (MEF) were the gift of D.J. Kwiatkowski (Harvard Medical School, Boston, MA) and were confirmed to exhibit Tsc2 deficiency and constitutive activation of mTORC1 by immunoblotting of tuberin. Tsc2 deficiency and constitutive activation of mTORC1 were confirmed by immunoblotting of tuberin and phospho-S6, respectively. The novel Tsc2-null cystadenoma cell line, 105K, was derived from a Tsc2^+/− C57Bl/6 mouse renal tumor and isolated in the laboratory of Dr. Elizabeth Henske. These cells were confirmed to have loss of the second allele of Tsc2 by PCR, loss of tuberin expression, and increased phospho-S6 levels by immunoblotting. All cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% FBS, 100 μg/mL of penicillin, and 100 μg/mL of streptomycin.

Chloroquine diphosphate salt, bafilomycin A, and 6-aminonicotinamide (6-AN) were obtained from Sigma-Aldrich. Spautin-1 was obtained from Cellagen Technology. Short hairpin RNAs (shRNA) against Atg5 (TRCN0000099434 and TRCN0000099431), LAMP2a (TRCN0000086974), Beclin1 (TRCN0000087290), or control (shGFP) were obtained from the RNAi Consortium (TRC).

Tsc2^+/− mice were studied in the A/J genetic background. The severity of renal lesions in the Tsc2^+/− mice was scored using a quantitative index as previously described (8), incorporating standardized gross inspection and microscopic histologic parameters. The animal studies were approved by the Children's Hospital Boston Animal Care and Use Committee.

Cells were harvested in quadruplicate, and intracellular metabolites extracted using 4 mL of cold (−80°C) 80% volume for volume (v/v) aqueous methanol. Cells and the metabolite-containing supernatants were collected into conical tubes. Insoluble material in lysates was centrifuged at 2,000× g for 15 minutes, and the resulting supernatant was evaporated using a refrigerated speed vac. Samples were resuspended using 20 μL high-performance liquid chromatography (HPLC) grade water for mass spectrometry. Then, 10 μL was injected and analyzed using a 5500 QTRAP triple quadrupole mass spectrometer (AB/SCIEX) coupled to a Prominence UFLC HPLC system (Shimadzu) via selected reaction monitoring (SRM) of a total of 254 endogenous water-soluble metabolites for steady-state analyses of samples. Some metabolites were targeted in both positive and negative ion mode for a total of 287 SRM transitions using pos/neg polarity switching. Electrospray (ESI) voltage was +4,900 V in positive ion mode and −4,500 V in negative ion mode. The dwell time was 3 ms per SRM transition and the total cycle time was 1.55 seconds. Approximately 10 to 14 data points were acquired per detected metabolite. Samples were delivered to the Mass Spectrometry (MS) via normal phase chromatography using a 4.6 mm i.d. × 10 cm Amide Xbridge HILIC column (Waters Corp.) at 350 μL/min. Gradients were run starting from 85% buffer B (HPLC grade acetonitrile) to 42% B from 0 to 5 minutes; 42% B to 0% B from 5 to 16 minutes; 0% B was held from 16 to 24 minutes; 0% B to 85% B from 24 to 25 minutes; 85% B was held for 7 minutes to reequilibrate the column. Buffer A was comprised of 20 mmol/L of ammonium hydroxide/20 mmol/L of ammonium acetate (pH = 9.0) in a water:acetonitrile ratio of 95:5. Peak areas from the total ion current for each metabolite SRM transition were integrated using the MultiQuant v2.0 software (AB/SCIEX).

The resulting raw data from the MultiQuant software were normalized by total proteins and then uploaded in metaboanalyst (http://www.metaboanalyst.ca/MetaboAnalyst/) for subsequent data processing and analyses. Raw and normalized/processed data are reported in Supplementary Table S1. Metabolites that were not detected in at least one sample were excluded from the analysis. Data were normalized to the median (per sample) and processed through log transformation. Heat map and hierarchical clustering was generated using Pearson correlations and the Ward method. Metabolite Set Enrichment Analysis (MSEA) was performed using Metaboanalyst that exploits the Kyoto Encyclopedia of Genes and Genome (KEGG) (http://www.genome.jp/kegg/pathway.html) pathway database. Fifty-eight metabolites significantly increased in Tsc2^−/− MEF with chloroquine treatment compared with vehicle (P < 0.05, calculated by the Student t test with univariate analysis and equal group variation) were used for MSEA analysis. For reference metabolites, a total of 235 metabolites were used. Metabolite sets containing at least 5 compounds were used in the analysis. MSEA calculates hypergeometrics test score based on cumulative binominal distribution. Only 3 metabolites were found increased in Tsc2^+/+ MEFs upon chloroquine treatment.

Metabolite levels in the medium were measured using the Yellow Springs Instruments (YSI) 7100 as previously described (9).

Cells were seeded in 12-well plates, treated for 24 hours, and incubated for 3 hours at 37°C with either 1 μCi/mL of [14C(U)]-labeled D-glucose or [14C(U)]-labeled palmitate (American Radiolabelled Chemicals Inc.). Then, 3 mol/L of perchloric acid was added to the culture media and the dishes were sealed with a phenylethylamine (Sigma-Aldrich)-saturated Whatman filter paper to capture ¹⁴C-CO₂, as previously described (10). Following 3-hour incubation at room temperature on a gentle shaker, the filter paper was removed, placed into Ultima Gold F Scintillation Fluid (PerkinElmer Inc.), and radioactivity counts were read in a 1600 liquid scintillation analyzer (Packard).

Oxygen consumption was measured under basal conditions or in the presence of the Trifluorocarbonylcyanide phenylhydrazone (FCCP) using the Seahorse Bioscience XF24 analyzer as previously described (7). Levels of oxygen consumption were normalized to cell number.

Mitochondrial reactive oxygen species (ROS) were measured by using MitoSOX (Invitrogen) staining (5 mmol/L for 15 minutes at 37°C). Cells were washed with PBS, trypsinized and resuspended in PBS containing 1% FBS. Data were acquired with a BD FACSCanto II flow cytometer (BD Biosciences) and analyzed with FlowJo analytical software (Treestar; ref. 11).

Cells were plated into 96-well plates (1,000 cells/well). After treatment, cells were fixed with 10% formalin for 5 minutes, stained with 0.05% crystal violet in distilled water for 30 minutes, washed 2 times with tap water, and drained. Crystal violet was solubilized with 100 μL of methanol and the plate was read with a BioTek plate reader [optical density (OD) 540].

Phospho-S6 (Ser235/236), phospho-S6K (Thr 389), total S6, total S6K, phospho-4E-BP (Thr 37/46), total p65, β-actin, and phospho-p65 antibodies were obtained from Cell Signaling Technology, tuberin antibody from Abcam, Beclin-1 and caspase-1 antibodies from Santa Cruz, LC3 antibody from Novus Biologicals, Atg5 antibody from Sigma, and LAMP2A antibody from Invitrogen. For immunoblot analyses, cells were washed with PBS and harvested in a lysis buffer containing Nonidet P-40. Whole-cell lysates were resolved by electrophoresis, and proteins were transferred onto polyvinylidene difluoride membrane (Immobilon P; Millipore), blocked in Tris-buffered saline Tween-20 buffer (Cell Signaling Technology), and probed with the indicated antibodies in this buffer.

IL-6 levels were measured in the conditioned media using a Luminex LX100 instrument.

One- and two-sample t tests and the Mann–Whitney test were used for in vitro and in vivo studies (GraphPad Prism version 5.04 for Windows; GraphPad Software, www.graphpad.com). Significance was defined as P < 0.05. Data are shown as means ± SD.

To investigate the impact of autophagy suppression on the proliferation and metabolism of cells with hyperactive mTORC1 signaling, we used chloroquine, which raises lysosomal pH and inhibits the fusion between autophagosomes and lysosomes, to treat Tsc2-deficient MEFs in monolayer culture. Chloroquine partially suppressed cell proliferation (38% after 96 hours of treatment, P < 0.05; Fig. 1A and Supplementary Fig. S1A) in a TSC2-dependent manner (Fig. 1A). Next, we treated Tsc2^+/− mice, which develop spontaneous renal cystadenomas, with chloroquine (50 mg/kg/day, intraperitoneally, 5 days/week) for 4 months. Chloroquine decreased the number of macroscopic and microscopic renal lesions by approximately 50% (Fig. 1B). Chloroquine treatment did not affect body weight (Supplementary Fig. S1B).

To identify compensatory metabolic mechanisms activated upon suppression of autophagy, metabolite profiling of Tsc2-null MEFs treated with chloroquine (5 μmol/L) was performed. Chloroquine induced extensive alterations, including changes in the glucose, amino acid, and nucleotide utilization/biosynthesis pathways (Fig. 1C and Supplementary Table S1). Importantly, mTORC1 activity was not affected by chloroquine as assessed by the phosphorylation of ribosomal protein S6 (Fig. 1C). MSEA applied to metabolites whose levels increased in chloroquine-treated Tsc2-null cells identified the PPP as the most enriched metabolic pathway (Fig. 1C and D). In contrast, metabolites of the PPP were either unchanged or decreased upon treatment with chloroquine in Tsc2^+/+ cells (Supplementary Fig. S2 and Supplementary Table S1).

We next measured uptake of glucose and secretion of lactate in Tsc2^−/− and Tsc2^+/+ MEFs treated for 24 hours with chloroquine. Interestingly, glucose uptake was significantly increased by chloroquine in Tsc2-null cells (P < 0.05; Fig. 2A) but not Tsc2^+/+ cells. The amount of lactate in the media was unchanged, indicating that the chloroquine-induced glucose uptake in Tsc2^−/− cells is not associated with increased aerobic glycolysis. Tsc2^−/− MEFs in which the autophagy gene Atg5 or Beclin 1 (Atg6) was inhibited by shRNA also exhibited significantly higher glucose consumption relative to control shRNA, with no increase in lactate secretion (Fig. 2A and Supplementary Fig. S3), consistent with the chloroquine results.

To define the metabolic routes to which glucose is directed when autophagy is inhibited, we assayed oxidative metabolism. First, we measured the oxygen consumption rate of cells treated with chloroquine. After 24 hours treatment with chloroquine, Tsc2^−/− MEFs displayed a significant suppression of both basal and maximal (FCCP-induced) oxygen consumption rate (Fig. 2B), consistent with our previous results in Tsc2-null rat leiomyoma ELT3 cells (7). A concomitant increase in mitochondrial ROS was observed in Tsc2^−/− MEFs treated with chloroquine for 24 hours (Supplementary Fig. S4), suggesting progressive accumulation of damaged mitochondria.

Next, we dissected changes in oxidative utilization of individual nutrients by measuring ¹⁴C-CO₂ release from cells labeled for 3 hours with D[U-¹⁴C]glucose or [U-¹⁴C]palmitate. Consistent with the metabolomic results, chloroquine induced an increase in glucose oxidation selectively in Tsc2-null cells (P < 0.05; Fig. 3A and D). In addition, fatty acid oxidation decreased in chloroquine-treated Tsc2-null cells (P < 0.05; Fig. 3B and D), consistent with the reduction in oxygen consumption rate. Treatment with rapamycin markedly reduced glucose oxidation even in the presence of chloroquine (P < 0.05; Fig. 3C), suggesting mTORC1 dependence.

Together, these results suggest that autophagy inhibition in Tsc2-null cells leads to increased glucose utilization via the PPP, coupled with decreased oxygen consumption, in an mTORC1-dependent manner.

On the basis of the above results, we hypothesized that Tsc2-null cells become dependent on the PPP under autophagy inhibition. To test this hypothesis, we used the antimetabolite 6-AN, a competitive inhibitor of glucose-6-phosphate dehydrogenase (G6PD) and 6-phosphogluconate dehydrogenase (6PGD), the two key enzymes of the oxidative branch of the PPP. G6PD catalyzes the conversion of glucose-6-phosphate into 6-phosphogluconolactone, which is the rate-limiting step of the pentose phosphate shunt, and 6PGD the subsequent oxidation of 6-phosphogluconolactone. Cells were treated with chloroquine, 6-AN, or both drugs for 96 hours. Strikingly, the chloroquine–6-AN combination selectively inhibited the proliferation of Tsc2^−/− cells (Fig. 4A and B). No consistent effect on mTORC1 signaling was noted (Supplementary Fig. S5A). To validate these results, we used a mechanistically distinct inhibitor of autophagy, spautin-1, which promotes the proteasomal degradation of Beclin 1 (12). Consistent with the chloroquine results, treatment with the combination of spautin-1 and 6-AN selectively inhibited the proliferation of Tsc2^−/− cells (Fig. 4C). In addition, 6-AN treatment of cells infected with either Atg5 or LAMP2a (a chaperone-mediated autophagy gene) shRNA decreased proliferation of Tsc2^−/− cells compared with control shRNA (Fig. 5). The addition of rapamycin did not further affect proliferation in the presence of chloroquine and 6-AN (Supplementary Fig. S5B). The PPP generates ribose-5-phosphate and NADPH. Interestingly, supplementation with 0.1 mmol/L of NADPH (13) was sufficient to sustain the proliferation of Tsc2^−/− cells during treatment with the combination of chloroquine and 6-AN (Fig. 4D), suggesting that production of NADPH via PPP is essential to the proliferation of Tsc2^−/− cells.

The combination of chloroquine and 6-AN suppressed the proliferation of a novel Tsc2-null cystadenoma cell line, named 105K (Supplementary Fig. S6), similarly to the Tsc2^−/− MEFs. These cells were derived from a Tsc2^+/− C57Bl/6 mouse renal tumor, and confirmed to have loss of the second allele of Tsc2 gene (Supplementary Fig. S6), loss of tuberin expression, increased phospho-S6 levels, and decreased levels of autophagy (measured by p62 accumulation; Supplementary Fig. S6).

We found that chloroquine treatment led to an increase in ROS levels in Tsc2^−/− MEFs (Supplementary Fig. S4), which, together with cytosolic release of mitochondrial DNA due to impaired mitophagy (14), may activate NF-κB/Rel activity (15). Consistent with this reasoning, phosphorylation of p65/RelA was induced at 96 hours by the combination of chloroquine and 6-AN (Fig. 6A), associated with increased secretion of IL-6, a known transcriptional target of NF-κB (Supplementary Fig. S7). IL-6 was not detectable in Tsc2^+/+ control cells at baseline or after treatment with chloroquine/6-AN (Supplementary Fig. S7). Furthermore, we observed accumulation of cleaved caspase-1, a marker of inflammasome activation, under the combined chloroquine/6-AN treatment at 96 hours (Fig. 6B). Supplementation with 0.1 mmol/L of NADPH was sufficient to prevent accumulation of cleaved caspase-1 under chloroquine/6-AN treatment for 96 hours (Fig. 6C), consistent with the rescue of cell proliferation by NADPH (Fig. 4D).

To test the hypothesis that activation of NF-κB is necessary for the inhibition of cell proliferation by chloroquine and 6-AN, we treated the cells with parthenolide, which inhibits both p65/RelA and caspase-1 via alkylation (16). Parthenolide restored proliferation in cells treated with chloroquine and 6-AN (Supplementary Fig. S8). These data suggest that the combination of chloroquine and 6-AN inhibits the proliferation of Tsc2-null cells via ROS-dependent activation of NF-κB and caspase-1.

Chloroquine and its analog hydroxychloroquine, which block lysosome–autophagosome fusion and lysosomal function thus suppressing macroautophagy and chaperone-mediated autophagy, can inhibit cancer cell survival under stress conditions (1). We report here that prolonged (4 months) treatment with chloroquine suppressed the number and size of renal tumors in Tsc2^+/− mice, consistent with the decreased spontaneous renal tumors we previously observed in Tsc2/Beclin1 double heterozygous mice (7). We hypothesized that metabolic compensatory mechanisms are activated when autophagy is suppressed by chloroquine in Tsc2-deficient cells, thereby preventing a more complete therapeutic response. Metabolite profiling revealed striking changes in the metabolome of Tsc2-null cells treated with chloroquine, including increased levels of five PPP intermediate metabolites. Isotope labeling experiments revealed an increase in glucose oxidation under chloroquine treatment, which may reflect activation of the oxidative branch of the PPP. The pentose phosphate shunt, which is often upregulated in cancer, has both biosynthetic and oxidative functions, representing the main source of NADPH via its oxidative branch and supplying ribose-5 phosphate for nucleotide synthesis.

In parallel with the PPP upregulation, we found that chloroquine induced ROS production at a higher rate in Tsc2-null cells than in cells with intact TSC2. This could be due to impaired mitophagy and subsequent accumulation of dysfunctional mitochondria, which would increase the cell requirement for reducing power to buffer-elevated ROS production. TSC2-deficient cells may be particularly vulnerable to loss of reducing power because they are known to produce elevated amounts of ROS at baseline (17) and to be highly sensitive to increased ROS (18).

Therefore, these data suggest that the PPP becomes a crucial metabolic survival mechanism in TSC2-null cells when they encounter the increase in oxidative stress due to further autophagy suppression. To test this hypothesis, we used the antimetabolite 6-AN, which inhibits the PPP-dependent NADPH supply (19), in combination with chloroquine. The chloroquine/6-AN combination dramatically suppressed the proliferation of Tsc2-null cells, which was rescued by supplementation of NADPH. Similar results were seen with spautin-1 (12), a recently reported inhibitor of Beclin-1, and with downregulation of Atg5 and Lamp2a genes, supporting the hypothesis that the effect of chloroquine in these experiments is autophagy-mediated. Spautin-1 has been found to induce cell death under glucose deprivation conditions in breast cancer cells (12), further underscoring the links between the metabolic status of tumor cells and their response to autophagy inhibition.

Notably, the combination of chloroquine and 6-AN triggered activation of the NF-κB and inflammasome pathways in a TSC2-dependent manner, and parthenolide (16), which inhibits both NF-κB and caspase-1, rescued proliferation under these conditions. Inflammasomes are multiprotein complexes that mediate the cleavage and activation of caspase-1, thereby stimulating an inflammatory response and/or programmed cell death (20). Multiple factors can activate the inflammasome, including autophagy inhibition, mitochondrial DNA release, increased ROS, and activation of NF-κB (15, 21), thus affecting cell survival/proliferation. Recent work has linked the NF-κB pathway to the immune response and cellular senescence and has suggested that NF-κB may serve a tumor-suppressor function in some circumstances. Together, our results lead us to the conclusion that the inhibition of autophagy and the PPP leads to depletion of NADPH, increased ROS production, and activation of NF-κB/Rel. We propose that this ultimately suppresses cell proliferation via activation of the inflammasome (Fig. 6D).

Our results point toward novel therapeutic strategies for neoplasms with mTORC1 activation by targeting their intrinsic dependence on autophagy and the metabolic compensatory mechanisms triggered by autophagy inhibition. Thus, in TSC and related diseases, including lymphangioleiomyomatosis (LAM), our data provide a first proof-of-concept that therapeutic targeting of cells with hyperactive mTORC1 can be accomplished by targeting their metabolic vulnerabilities without the utilization of mTORC1 inhibitors. We note that the autophagy inhibitors chloroquine and hydroxychloroquine (Plaquenil) are currently being tested in clinical trials for multiple types of cancer and for LAM, and a novel derivative of chloroquine with higher efficacy in vitro and in vivo (22) has been recently synthesized. Furthermore, more than 100 cancer trials using “Rapalogs” and other mTORC1 inhibitors are currently enrolling patients, yet durable clinical responses are uncommon. Therefore, novel approaches targeting the metabolism of tumors with mTORC1 activation may have broad clinical impact.

D.J. Kwiatkowski is a consultant/advisory board member of Novartis. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A.A. Parkhitko, C. Priolo, J.J. Yu, D.J. Kwiatkowski, J.W. Locasale, A.M.K. Choi, E.P. Henske

Development of methodology: A.A. Parkhitko, C. Priolo, K. Mizumura, W. Xu, J.W. Locasale, J.M. Asara, E.P. Henske

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.A. Parkhitko, J.L. Coloff, J.J. Wu, W. Xu, I.A. Malinowska, T. Finkel, E.P. Henske

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.A. Parkhitko, C. Priolo, J. Yun, J.J. Wu, I.A. Malinowska, D.J. Kwiatkowski, J.W. Locasale, J.M. Asara, A.M.K. Choi, E.P. Henske

Writing, review, and/or revision of the manuscript: A.A. Parkhitko, C. Priolo, I.A. Malinowska, J.J. Yu, D.J. Kwiatkowski, J.W. Locasale, J.M. Asara, E.P. Henske

Study supervision: C. Priolo, J.W. Locasale, A.M.K. Choi, E.P. Henske

The authors thank Min Yuan and Susanne Breitkopf for help with mass spectrometry experiments. This research was conducted in partial fulfillment of A.A. Parkhitko's doctoral degree requirements at the Department of Molecular Biology, Russian State Medical University, under the supervision of Dr. Olga Favorova.

This work was supported by the LAM Foundation, the Adler Foundation, the Tuberous Sclerosis Alliance, the National Institute of Diabetes and Digestive and Kidney Diseases (to E.P. Henske); grants NIH 5P01CA120964-04 and NIH DF/HCC, Cancer Center Support Grant P30CA006516-46 (to J.M. Asara); the National Heart, Lung, and Blood Institute (to J.J. Yu); and the National Cancer Institute (to D.J. Kwiatkowski).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.