TSC2 Deficiency Unmasks a Novel Necrosis Pathway That Is Suppressed by the RIP1/RIP3/MLKL Signaling Cascade

Tuberous sclerosis complex (TSC) is an autosomal-dominant multiorgan disorder caused by a germline-inactivating mutation in the tumor suppressor genes TSC1 or TSC2 (1). The incidence of TSC has been estimated as 1 in 6,000 persons worldwide, with the average age at diagnosis being 7.5 years (2, 3). Clinical manifestation of TSC is development of tumors and neoplastic lesions in kidney, lung, brain, heart and skin, among which renal cell carcinoma (RCC), renal angiomyolipoma (AML), pulmonary lymphangioleiomyomatosis (LAM), and brain tumors constitute the most common cause of TSC-associated deaths (3–6). In addition, sporadic forms of AML and LAM characterized by mutation in TSC1 or TSC2 genes may develop in patients who do not carry germline mutation in these genes (5). Despite well-defined etiology, effective therapeutic options for TSC-associated tumors are limited mainly to invasive surgery and in the case of LAM to lung transplant, with no pharmaceutical treatment capable of inducing long-term remission of tumor growth (5–7).

Loss-of-function mutation in the TSC1 or TSC2 gene has been defined as the main factor driving malignancy of TSC-associated tumors (6, 8). Expression of TSC1 and TSC2 leads to synthesis of their protein products hemartin and tuberin, respectively, which form heterodimeric complex inhibiting the small GTPase Ras homolog enriched in brain (Rheb), that is necessary for the activation of mTOR complex 1 (mTORC1; ref. 9). While in normal cells, mTOR-dependent signaling is tightly regulated by the availability of nutrition and growth factors, the lack of a functional TSC1/TSC2 complex in TSC-deficient cells results in constitutive hyperactivation of mTORC1 and its downstream targets to stimulate protein translation, cell growth, and metabolic reprogramming (9, 10). In addition, TSC1/TSC2 complex deficiency leads to Rheb-dependent inhibition of Raf/MEK/MAPK signaling and sensitizes cells to estrogen stimulation (11). In vivo studies also demonstrated a critical role for this pathway in the regulation of metastasis of TSC-associated tumor cells. Activation of Raf/MEK/MAPK signaling with estrogen elevated the number of cells circulating in blood, conferred resistance to matrix deprivation–induced apoptosis (anoikis), and promoted TSC-deficient cells' colonization in the lung, consistent with the higher incidence of LAM in the lungs of women (11, 12). Together, disruption of TSC1/TSC2 complex promotes the malignant phenotype of tumor cells via complex mechanisms involving multiple downstream targets and signaling pathways.

Dependence on mTOR hyperactivity for proliferation has been the target for therapies approved for AML and LAM treatment. Rapalogs (analogues of rapamycin) called sirolimus and everolimus that block mTORC1 activity show clinical efficacy in preventing LAM and AML progression and stabilizing respiratory and renal functions through cytostatic rather than cytotoxic effect (13–16). Consistent with this mechanism, when treatment is discontinued due to side effects associated with long-term drug administration (which include stomatitis, rash, fatigue, hyperglycemia, hyperlipidemia, and myelosuppression) tumor growth and further deterioration of renal and respiratory functions occurs (5, 17). Moreover, blocking mTOR activity triggers cytoprotective autophagy that confers resistance to cellular stress. Animal studies support this hypothesis by demonstrating that simultaneously blocking mTOR and autophagy is superior to targeting mTOR activity alone for preventing TSC-associated tumorigenesis (18, 19). The hypersensitivity of TSC2-null cells to oxidative stress was further substantiated by treatment with the alkaloid chelerythrine chloride that induces cell death in a reactive oxygen species manner (20). In addition, cell death caused by chelerythrine chloride was inhibited by necrostatin-1 (Nec-1), an inhibitor of RIP1 kinase activity, suggesting an important role for RIP1-dependent signaling in the regulation of TSC2-deficient cell survival. RIP1 kinase–dependent signaling leads to the activation of its downstream target RIP3 that later phosphorylates MLKL, resulting in formation of the necrosome and induction of programed necrosis called necroptosis (21).

On the basis of the findings described above, we hypothesized that simultaneous targeting of two major mechanisms responsible for redox homeostasis, glutathione- and thioredoxin-dependent pathways, may synergistically induce cell death in TSC2-deficient cells due to their hypersensitivity to oxidative stress. To test this hypothesis, toxicity induced by the glutathione biosynthesis inhibitor l-buthionine-sulfoximine (BSO) and thioredoxin reductase inhibitor auranofin was assessed in vitro and in vivo using two different cellular models of TSC2 deficiency. The role of the RIP1/RIP3/MLKL-dependent pathway in modulating sensitivity of TSC2-deficient cells to oxidative and mitochondrial damage–inducing treatment was also investigated.

TSC2-wild-type and TSC2-deficient cell lines originated from a murine embryonic fibroblast (MEF) and human angiomyolipoma (621) used and described in previous reports (18–20) were a gift from Dr. Elizabeth Henske (Brigham and Women's Hospital, Boston, MA). Cell line authentication was performed for 621-derived cell lines using FTA Sample Collection Kit for Human Cell Authentication (ATCC) and for MEF-derived cell lines using multiplex PCR assay targeting short tandem repeat (STR) markers developed by Almeida and colleagues (22). In addition, cells tested negative for mycoplasma contamination using MycoProbe Detection Kit (R&D Systems) were confirmed for TSC2 expression status using Western blot analysis (primary antibody from Cell Signaling Technology). Cells were cultured in high-glucose DMEM supplemented with 10% FBS, 2 mmol/L glutamine, and 1% penicillin/streptomycin (Life Technologies) at 37°C in a humidified 5% CO₂ atmosphere and subcultured for up to 20 passages. BSO, auranofin, Nec-1, N-benzyloxycarbonyl-Val-Ala-Asp(O-Me) fluoromethyl ketone (z-VAD-FMK), α-ketoglutarate (α-KG), N-acetyl-l-cysteine (NAC), and chloroquine diphosphate salt (CQ) were obtained from Sigma-Aldrich. Necrosulfonamide (NSA) was obtained from Toronto Research Chemicals Inc., MHY1485 was obtained from EMD Millipore, and TNFα was obtained from R&D Systems.

The study was performed under a protocol approved by Institutional Animal Care and Use Committee at Lovelace Respiratory Research Institute (Albuquerque, NM). MEF-TSC2–deficient cells (3 × 10⁶) were inoculated bilaterally into the posterior flanks of 7-week-old athymic nude mice (Charles River Laboratories). The treatment began 10 days after the inoculation, when the tumor size was approximately 75 mm³. Combinations of BSO, auranofin, and Nec-1 dissolved in PBS were injected intraperitoneally for 5 consecutive days for 2 weeks. Control animals were injected with PBS according to the same schedule. The size of tumors was measured twice a week using a caliper, and the tumor volume was calculated according to the formula volume = (length × width²)/2. Necropsy of the animals followed by harvesting and weighing of tumors was performed 3 days after the last injection of the tested compounds.

For the crystal violet assay, cells were plated in a 48-well format (3–10 × 10³/well) and subjected to treatment with BSO and auranofin combinations 24 hours later. After 72 hours of treatment, cells were fixed with methanol at −20°C and stained with 0.1% crystal violet for 30 minutes. Cells were washed twice with water and solubilized by the addition of 10% acetic acid followed by shaking for 10 minutes. Absorbance was read using a plate reader at 595 nm. For propidium iodide (PI) exclusion test, cells were plated in a 24-well format (10–30 × 10³/well) and subjected to treatment with BSO, auranofin, NAC, Nec-1, and NSA combinations 24 hours later. After 24 hours of treatment, cells (both attached to the surface and floating in the culture medium) were collected by trypsinization, washed with PBS, and suspended in 10 μg/mL PI PBS solution. The relative number of living cells (low fluorescence–emitting population) and dead cells (high fluorescence–emitting population) was assessed by flow cytometric analysis at 535 nm.

Cells were plated in a 24-well format (10–30 × 10³/well) and subjected to treatment with BSO, auranofin, NAC, Nec-1, and NSA combinations 24 hours later. For ROS measurement, 16 hours after treatment, cells were incubated with 2 μmol/L 2′,7′-dichlorofluorescin diacetate (DCFDA, Sigma-Aldrich) diluted in complete culture medium for 15 minutes at 37°C. Next, the cells were collected from the culture dish, washed with PBS, and analyzed for green fluorescence intensity using a flow cytometer at 485 nm. To assess mitochondrial potential, 16 hours after treatment with BSO, auranofin, Nec-1, and NSA combinations, cells were incubated with 2 μmol/L JC-1 probe (Sigma-Aldrich) diluted in complete culture medium for 15 minutes at 37°C. Next, the cells were collected from the culture dish, washed with PBS, and analyzed for red fluorescence (535 nm) to green fluorescence (485 nm) ratio using flow cytometry. Alternatively, cells stained with JC-1 probe and counterstained for nuclei using 10 μg/mL of Hoechst 33342 were subjected to microscopic analysis at 535 nm and 485 nm using a fluorescent microscope.

Cells were plated in a 6-well format (1 × 10⁵/well), cultured overnight, and transfected with siRNA using Lipofectamine 3000 according to the manufacturer's protocol. Cells were reseeded onto a 24-well plate format 24 hours after transfection, subjected to treatment with BSO, auranofin, and α-ketoglutarate 48 hours after transfection, and analyzed for cell death 24 hours later using PI exclusion test. Two different control siRNAs (Life Technologies) and MLKL siRNAs (Life Technologies and GE Dharmacon) were used and gave identical results.

Cells were lysed using RIPA buffer (Cell Signaling Technology) supplemented with proteinase/phosphatase inhibitors cocktail (Cell Signaling Technology) and 1 mmol/L of phenylmethanesulfonylfluoride (PMSF; Sigma-Aldrich). After incubation on ice (20 minutes), lysates were centrifuged (15,000 × g, 10 minutes) at 4°C and supernatants were aliquoted and stored at −80°C. Total protein content was determined using Pierce BCA Protein Assay Kit (Thermo Scientific), and 70 μg of protein was electrophoretically fractionated on 4%–15% Mini-PROTEAN TGX Precast 10-well gel (Bio-Rad), blotted onto a nitrocellulose 0.45-μm membrane (Bio-Rad), blocked for 60 minutes in 5% nonfat milk in TTBS (TBS from Boston Bioproducts supplemented with 0.05% of Tween-20 from Sigma-Aldrich), and incubated overnight with primary antibody in 4°C. Polyclonal antibodies for activated caspase-3 (made in rabbit; Cell Signaling Technology), RIP1 (made in mouse; Becton Dickinson), RIP3 (made in rabbit; Abgent), p62 (made in mouse; BD Biosciences), LC3B (made in rabbit; Abgent), and β-actin (made in rabbit, Cell Signaling Technology) were used for primary detection. Primary antibodies were detected with goat anti-rabbit or anti-mouse horseradish peroxidase–conjugated secondary antibodies (Cell Signaling Technology) and visualized using SuperSignal West Pico Chemiluminescent Substrate kit (Thermo Scientific). Quantitative analysis of band intensity was performed using ImageJ.

Tissue sections were fixed with 10% neutral buffered formalin solution, sequentially incubated with 30% and 70% histologic grade ethanol, and processed for paraffin embedding. Sections (5 μm) were cut, immersed in several changes of xylene, rehydrated in graded alcohols, and washed in PBS. Antigen retrieval was then performed by incubating cells with 2 μg/mL proteinase K (Sigma-Aldrich) at 37°C for 15 minutes. Slides were washed with PBS, incubated with premixed TUNEL reagents (Roche Applied Science) in accordance with the manufacturer's protocol, washed with PBS, and immunostained for activated caspase-3 as described in the section “Immunostaining” (see below).

Tumor sections were fixed, paraffin embedded, rehydrated, and antigen retrieved according to the protocol described in “TUNEL” section (see above). For immunostaining of cells from in vitro culture, cells were collected from the culture dish via trypsinization, placed onto glass slides using a cytocentrifuge Cytospin 3 (Shandon), fixed with methanol at −20°C for 20 minutes, and washed with PBS. Tumor sections and cells were blocked in 1% BSA and 0.1% Triton X-100 in PBS. The slides were incubated with the primary antibody at 4°C overnight. Polyclonal antibodies made in rabbit for HMGB1 (Sigma-Aldrich) and activated caspase-3 (Cell Signaling Technology) were used for primary detection. Slides were next incubated with goat anti-rabbit FITC- or Alexa 647–conjugated secondary antibody for 1 hour in room temperature, counterstained for nuclei using 1 μg/mL DAPI (Sigma-Aldrich) solution in PBS, and coverslipped using Vectashield mounting medium (Vector Laboratories). Microscopic analysis was performed using a Zeiss fluorescent microscope equipped with Slidebook 5.0 software.

The results for each treatment group are summarized as the mean value ± SEM. Comparisons of results between groups were performed using Student t test, and among groups using one-way ANOVA test followed by Dunnett posttest (GraphPad Prism 5.04). Statistical significance was defined as a P value <0.05 (*), <0.01 (**), or <0.001 (***).

Two cellular models of TSC2 deficiency MEFs, and 621 cells isolated from a human angiomyolipoma, were used to investigate the role of TSC2 status in BSO- and auranofin-induced toxicity. Crystal violet assay demonstrated that cotreatment with BSO and auranofin reduced the number of MEF and 621 cells with TSC2 deficiency in a dose-dependent manner (53%, 80%, and 98% reduction for MEF-TSC2^−/− and 82%, 86%, and 91% reduction for 621-TSC2^−/− after 1 μmol/L/0.3 μmol/L, 1 μmol/L/1 μmol/L, and 1 μmol/L/3 μmol/L auranofin/BSO treatment, respectively). In contrast, wild-type cells were significantly less affected (19%, 38%, and 77% reduction for MEF-TSC2^+/+ and 20%, 39%, and 78% reduction for 621-TSC2^+/+ after 1 μmol/L/0.3 μmol/L, 1 μmol/L/1 μmol/L, and 1 μmol/L/3 μmol/L auranofin/BSO treatment, respectively; Fig. 1A). To determine whether reduction of TSC2-deficient cell number is caused by inhibition of proliferation or by cell killing, PI exclusion test was applied for direct detection of dead cells. While no increase in cell death was found after individual treatment with BSO or auranofin, the combination of these compounds increased the mortality from 5% to 21% and from 4% to 27% in MEF-TSC2^−/− and 621-TSC2^−/− cells, respectively, without affecting the viability of TSC2^+/+ cell lines (Fig. 1B). The mechanism of cell death induced in TSC2-deficient cells by BSO and auranofin was investigated. Western blot analysis revealed no activation of caspase-3, the major enzyme involved in apoptosis, while cotreatment with the pan-caspase inhibitor zVAD also failed to rescue cells from BSO/auranofin-induced cell death (Fig. 1C; Supplementary Fig. S1). In contrast, immunofluorescent staining demonstrated characteristics for necrosis as the main mechanism of cell killing as evident by nucleus-to-cytoplasm exclusion of DNA-binding protein HMGB1 after BSO/auranofin treatment (Fig. 1D).

The combination treatments in TSC2^+/+ cells reduced and increased protein levels of the autophagy biomarkers p62 and LC3B-II, respectively (Supplementary Fig. S2A). In contrast, no effect on these biomarkers was seen in TSC2^−/− cells (Supplementary Fig. S2A). The autophagy inhibitor chloroquine potentiated mortality of TSC2^−/− cells induced with BSO and auranofin, indicating that autophagy does not contribute to cell killing by the treatment. In contrast, the mTOR inhibitor rapamycin reduced cell death (Supplementary Fig. S2B). No significant effects by these agents on cell death were seen in TSC2^+/+ cells (Supplementary Fig. S2B).

Chemicals tested in this investigation have well defined role as free radical (BSO and auranofin) and mitochondrial damage (auranofin) inducing agents (23–26), therefore these cellular stress responses were evaluated. Combined treatment with BSO and auranofin increased ROS levels 2.7- and 2.0-fold in MEF-TSC2^−/− and 621-TSC2^−/− cells, respectively. No significant effects were observed in wild-type cells (Fig. 2A). The addition of the free-radical scavenger N-acetyl cysteine (NAC) prevented the elevation of ROS levels and cell death consistent with a ROS-dependent mechanism of cell killing (Fig. 2A and B). Measurement of mitochondrial electric potential using the JC-1 probe demonstrated that the relative number of functional mitochondria was reduced 23% and 26% in MEF-TSC2^−/− and 621-TSC2^−/− cells, respectively, with no reduction found in the wild-type cells (Fig. 2C). Fluorescent imaging of MEF-TSC2^−/− and 621-TSC2^−/− cells stained with JC-1 probe further revealed a decrease of red fluorescent signal associated with functional mitochondria and increase of green fluorescence associated with cytoplasm caused by BSO/auranofin treatment (Fig. 2D).

Medvetz and colleagues showed that cell death in MEF-TSC2^−/− cells caused by the ROS inducer chelerythrine chloride was mediated by RIP1-dependent signaling (20). Thus, the RIP1 kinase activity inhibitor Nec-1 and the RIP3/MLKL interaction inhibitor necrosulfonamide (NSA) were used to study the involvement of this pathway in BSO/auranofin-induced cell death. Unexpectedly, PI exclusion showed that Nec-1 and NSA significantly synergize with BSO and auranofin to induce killing in TSC2-deficient cells (increase of cell death up to 43%–73%, Fig. 3A). No significant increase in the number of dead cells was detected when Nec-1 or NSA were applied individually or combined with BSO or auranofin alone (not shown). A modest toxicity of BSO/auranofin/Nec-1 and BSO/auranofin/NSA combinations was also found in TSC2^+/+ cells; however, it did not exceed 16% of mortality in any of the experimental combinations (Fig. 3A). Similar to BSO/auranofin cotreatment, mortality induced by BSO/auranofin/Nec-1 and BSO/auranofin/NSA was not reduced by addition of the caspase inhibitor zVAD, substantiating necrosis as the mechanism of cell death (Supplementary Fig. S1). However, consistent with a role for Nec-1 and NSA in blocking necroptosis, these drugs abrogated the effect of TNFα combined with zVAD for induced cell death in TSC2-wild-type cells (Supplementary Fig. S3). Further studies revealed that Nec-1 and NSA potentiated BSO/auranofin-induced free-radical formation up to 4.3–5.1-fold in TSC2^−/− cell lines (Fig. 3B), and loss of mitochondrial potential was elevated up to 29%–55% (Fig. 3C). While the addition of Nec-1 or NSA further induced autophagy in TSC2-wild-type cells, no effect was detected in TSC2-deficient cells (Supplementary Fig. S2A), and chemical mTOR stimulation (rapamycin) and autophagy inhibition (chloroquine) further confirmed the BSO/auranofin/Nec-1/NSA-induced cell death is mTOR-dependent but autophagy-independent (Supplementary Fig. S2B).

RNA interference technique was used to test whether knockdown of MLKL, a downstream mediator of RIP1/RIP3-dependent signaling, affects the sensitivity of TSC2-deficient cells to BSO/auranofin. Downregulation of MLKL protein level by RNA interference, confirmed with Western blot analysis, increased cell mortality up to 33% in MEF-TSC2^−/− cells with BSO/auranofin treatment, indicating that the signaling downstream from RIP1/RIP3 kinase activity is critical for TSC2-deficient cell sensitization (Fig. 4A). Knockdown experiments using 621-TSC^−/− were inconclusive because of high toxicity induced in this cell line by transfection (not shown). Finally, to test whether elevated sensitivity to test compounds is a result of mTOR hyperactivity, MEF- and 621-wild-type cells were treated with mTOR activator HMY1485. Chemical stimulation of mTOR sensitized MEF- and 621-wild-type cells, increasing mortality induced with BSO/auranofin, BSO/auranofin/Nec-1, BSO/auranofin/NSA up to 14.6%–16.9%, 16.2%–34.0% and 24.5%–52.7%, respectively (Fig. 4B).

Among other functions, RIP1/RIP3/MLKL activity is known to stimulate glutaminolysis, which is indispensable for TSC2-deficient cells to survive stress conditions as described previously (27, 28). Thus, the hypothesis that inhibition of RIP1/RIP3/MLKL signaling sensitizes TSC2-deficient cells to BSO/auranofin via inhibition of glutaminolysis was tested. Treatment of MEF- and 621-TSC2-deficient cells with Nec-1 or NSA caused 18%–48% reduction in levels of α-ketoglutarate, a critical product of glutamine conversion (Fig. 5A). Furthermore, supplementation of culture medium with exogenous α-ketoglutarate reduced cell death induced with BSO/auranofin/Nec-1 and BSO/auranofin/NSA by 19%–40% and 13%–27%, respectively (Fig. 5B), and abrogated the sensitizing effect of MLKL knockdown by BSO/auranofin treatment (Fig. 5C). Recent reports demonstrated that α-ketoglutarate may regulate autophagy via various mechanisms including Akt/mTOR signaling (29, 30); however, no changes in p62 and LC3B-II protein levels were found in our experimental setting (Fig. 5D). Finally, expression analysis performed using Western blot analysis revealed that RIP1, RIP3, and MLKL protein levels are elevated 1.9–7.6-fold in the 621-TSC2–deficient cells compared with the wild-type cells, with the similar trend in RIP1 and RIP3 expression changes observed in the MEF model (Fig. 5E).

Mouse xenograft model of MEF-TSC2^−/− cells, which form fast growing tumors in contrast to low tumorigenic 621 cells, was used to test the antitumor efficacy of BSO, auranofin, and Nec-1. While there was no therapeutic effect of any of these compounds applied alone, treatment with BSO/auranofin and BSO/auranofin/Nec-1 significantly reduced the growth of TSC2-deficient tumors as expressed by a reduction of tumor volume and mass by 37%–78% compared with sham-treated animals (Fig. 6A and B). TUNEL/caspase-3 double staining was performed in tumor tissue sections to evaluate the extent and the mechanism of cell death. While no significant increase in the number of TUNEL-positive cells was observed in tumors subjected to BSO/auranofin treatment (not shown), BSO/auranofin/Nec-1 increased positivity by 5-fold compared with sham (Fig. 6C and D). While 72% of dead cells in sham tumors stained positive for activated caspase-3, the majority (79%) of TUNEL-positive cells detected in BSO/auranofin/Nec-1–treated tumors were negative for activated caspase-3, further substantiating the dominant role for necrosis versus apoptosis induced by this combined treatment (Fig. 6C and E).

These studies have identified a targeted multidrug approach that alters redox homeostasis and induces mitochondrial damage to synergistically and potently inhibit the growth of TSC2-deficient tumors in an mTOR-dependent manner. Furthermore, a novel TSC context–dependent function for RIP1/RIP3/MLKL signaling through activation of glutaminolysis that sustains the viability of TSC2-deficient cells subjected to oxidative and mitochondrial stress was identified. Inhibition of RIP1/RIP3/MLKL-regulated glutaminolysis disrupts mitochondrial homeostasis in TSC2-deficient cells constituting a new form of regulated necrosis (Fig. 7). Autophagy does not contribute to killing TSC2-deficient cell by this new regulated cell death mechanism; rather the basal autophagy activity protects TSC2-deficient cells against the toxic effects of the tested drugs. In contrast, mTOR hyperactivation associated with TSC2 deficiency is required to induce necrosis by BSO, auranofin, and RIP1/MLKL inhibitors in TSC2-deficient cells.

The contribution of TSC2 deficiency to sensitivity to oxidative stress has been demonstrated in variety of malignant and nonmalignant cellular models. Elevated oxidative stress, accompanied by increased expression of stress markers such as CHOP and HO-1, was found in hippocampal neurons isolated from TSC2-deficient rat embryos at increased vulnerability of cells to death via the mitochondrial death pathway (31). This effect was reversed by rapamycin, indicating a critical role for mTOR hyperactivation in this process. Furthermore, whole organism overexpression of Rheb, which mimics TSC-null phenotype, sensitized Drosophila melanogaster to oxidative stress induced with H₂O₂ via mTOR/S6K hyperactivation–dependent mechanism and promoted early senescence of their locomotor behavior (32). Clinical validation of this phenomenon was demonstrated through detection of elevated levels of oxidized DNA in angiomyolipoma tissue samples isolated from TSC patients (33). Moreover, human TSC–associated brain neoplastic lesions showed that elevated expression of the gene coding for glutamate-cysteine ligase catalytic subunit (GCLC) is increased in TSC2-deficient cells in response to elevated free-radical levels, and its knockdown results in growth arrest and cell death of brain tumor–derived cells (34). The mechanism of increased ROS production in TSC2-deficient cells is not fully understood, but has been associated with the accumulation of impaired mitochondria, the main source of free-radical species. In TSC-deficient cells removal of dysfunctional mitochondria via mitophagy is disabled because of mTOR hyperactivity and its inhibitory effect on autophagosome assembly (35, 36). In addition, mTOR activates transcription of genes for mitochondrial biogenesis including peroxisome proliferator-activated receptor γ coactivator-1α (PGC-1α; ref. 37). Thus, the fact that elevated levels of free radicals block the ability of TSC2-deficient cells to accommodate additional oxidative stress induced by exogenous factors led to the identification of chelerythrine chloride (20). Chelerythrine chloride–induced cell killing was ROS-dependent, and was rescued by the free-radical scavenger N-acetyl-cysteine (NAC) in TSC2-deficient cells. Our studies greatly expand this approach by demonstrating that simultaneous targeting of redox homeostasis via glutathione depletion (BSO) and mitochondrial integrity via thioredoxin reductase inhibition (auranofin) is highly synergistic in TSC2-deficient cell killing at relatively low doses of both drugs. The combination of BSO and auranofin reduces viability and clonogenic survival of head and neck cancer cells, and sensitizes lung and head and neck cancer cells to cisplatin and erlotinib, respectively (38–39). However, as high as 100 µmol/L BSO in combination with 0.5-5 μmol/L auranofin was used to reduce viability while 1 μmol/L BSO combined with 1 μmol/L auranofin was sufficient to induce death of TSC2-deficient cells.

One pathway recently described to regulate redox and mitochondrial homeostasis is the RIP1/RIP3/MLKL pathway. While initially defined as responsible for the induction of programmed necrosis (necroptosis) after stimulation with TNFα and under hypoxia–reperfusion stress (40–42), it was further found to play an oncogenic function. We recently reported that RIP1 overexpression is found in human lung tumors and mouse lung exposed to cigarette smoke, and that RIP1 potentiates neoplastic transformation of lung epithelial cells by suppressing carcinogen-induced oxidative stress (43). Depletion of RIP1 in lung cancer cells decreased mitochondrial oxidative phosphorylation by suppressing the expression of PGC-1α and promoted excessive aerobic glycolysis, which in turn accelerated DNA damage and inhibited cell proliferation (44). RIP1 has also been shown to suppress basal autophagy (45); however, this does not occur in the context of TSC deficiency. Reports regarding TSC2-deficient cells are limited to the observation that inhibition of RIP1 activity using Nec-1 protects from cell death induced with chelerythrine chloride (20). However, an opposite effect was shown in our current study where inhibition of RIP1 kinase activity, disruption of RIP3/MLKL binding, or MLKL knockdown sensitizes cells to BSO and auranofin in a TSC2-dependent manner. The difference in outcome between studies may relate to other effects by chelerythrine chloride that include inhibition of protein kinase C (46), making the nature of cellular stress induced with this agent highly pleiotropic. In contrast, BSO and auranofin induce oxidative and mitochondrial damage in a highly specific manner by glutathione depletion and thioredoxin reductase inhibition. In addition, the concentration of Nec-1 used by Medvetz and colleagues was 3-fold higher than used in our study, which may contribute to off-target effects such as inhibition of indoleamine-pyrrole 2,3-dioxygenase (IDO) by high dose of Nec-1 (47). RIP1 and RIP3 proteins are involved in regulation of multiple processes (including NF-κB signaling, activation of inflammasome via NLRP3, activation of caspase-8, and secretion of IL8) based on the mechanisms independent from their kinase activity (21, 48). Therefore, targeting of kinase-dependent signaling rather than their “scaffold” functions provides a therapeutic benefit against TSC2-deficient cells. Although the mechanism of the sensitizing effect of RIP1/RIP3/MLKL inhibition remains to be defined, experiments using the α-ketoglutarate supplementation and the measurement of its endogenous levels described in this study support a critical role for glutamine metabolism regulation for cell survival. Previously, Choo and colleagues found that TSC2-deficient cells are highly dependent on glutaminolysis for survival in stress conditions induced by glucose deprivation (28). Simultaneously, RIP3/MLKL complex was demonstrated to positively regulate glutamine conversion by activating glutamate-ammonia ligase (GLUL) and glutamate dehydrogenase-1 (GLUD1; ref. 27). Knowing that glutaminolysis is the process that exclusively occurs in and contributes to mitochondria function, it is likely that the synergism in cell killing induced by RIP1/RIP3/MLKL inhibition and BSO/auranofin treatment is the result of targeting mitochondrial homeostasis via three distinct mechanisms: redox-, structural-, and metabolic-based. This finding is of high clinical importance for developing new treatment strategies for TSC patients, as numerous studies have identified small-molecule drugs targeting glutamine metabolism. One inhibitor of glutaminase (GLS) called CB-839 is currently under evaluation in a phase I clinical trial in patients with advanced lung, breast, and renal tumors (49). In addition, combining the CB-839 analogue BPTES with a HSP90 inhibitor induced cell death selectively in mTOR-driven tumors, supporting clinical benefit of targeting glutaminolysis specifically in this type of malignancy. Thus, our study provides a novel target, distinct from mTOR but associated with its hyperactivation, that may be used to develop more effective alternative therapies against TSC-associated tumors, and that extend beyond tuberous sclerosis to include cancers with high incidence of spontaneous TSC1/2 gene mutations (50), like bladder, endometrial, and skin malignancies.

No potential conflicts of interest were disclosed.

Conception and design: P.T. Filipczak, A. Salzman, Y. Lin, S.A. Belinsky

Development of methodology: A. Salzman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.T. Filipczak, W. Chen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P.T. Filipczak, W. Chen, A. Salzman, Y. Lin

Writing, review, and/or revision of the manuscript: P.T. Filipczak, W. Chen, A. Salzman, Y. Lin, S.A. Belinsky

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.L. Thomas, J.D. McDonald

Study supervision: S.A. Belinsky

We thank Ms. Elise Calvillo of LRRI for the graphics artwork.

This work was supported by Lovelace Respiratory Research Institute internal funding.

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.