mTORC1 enhancement of STIM1-mediated store-operated Ca²⁺ entry constrains tuberous sclerosis complex-related tumor development

Abstract

The protein complex of tuberous sclerosis complex (TSC)1 and TSC2 tumor suppressors is a key negative regulator of mammalian target of rapamycin (mTOR). Hyperactive mTOR signaling due to the loss-of-function of mutations in either TSC1 or TSC2 gene causes TSC, an autosomal dominant disorder featured with benign tumors in multiple organs. As the ubiquitous second messenger calcium (Ca²⁺) regulates various cellular processes involved in tumorigenesis, we explored the potential role of mTOR in modulation of cellular Ca²⁺ homeostasis, and in turn the effect of Ca²⁺ signaling in TSC-related tumor development. We found that loss of Tsc2 potentiated store-operated Ca²⁺ entry (SOCE) in an mTOR complex 1 (mTORC1)-dependent way. The endoplasmic reticulum Ca²⁺ sensor, stromal interaction molecule 1 (STIM1), was upregulated in Tsc2-deficient cells, and was suppressed by mTORC1 inhibitor rapamycin. In addition, SOCE repressed AKT1 phosphorylation. Blocking SOCE either by depleting STIM1 or ectopically expressing dominant-negative Orai1 accelerated TSC-related tumor development, likely because of restored AKT1 activity and enhanced tumor angiogenesis. Our data, therefore, suggest that mTORC1 enhancement of store-operated Ca²⁺ signaling hinders TSC-related tumor growth through suppression of AKT1 signaling. The augmented SOCE by hyperactive mTORC1-STIM1 cascade may contribute to the benign nature of TSC-related tumors. Application of SOCE agonists could thus be a contraindication for TSC patients. In contrast, SOCE agonists should attenuate mTOR inhibitors-mediated AKT reactivation and consequently potentiate their efficacy in the treatment of the patients with TSC.

INTRODUCTION

Tuberous sclerosis complex (TSC) is an autosomal dominant syndrome characterized by the growth of benign tumors in a wide range of organs such as brain, kidney, heart and lung. This disease is mainly caused by inactivating mutations of either TSC1 or TSC2 tumor suppressor genes.¹ TSC1 and TSC2 form a physical and functional protein complex that negatively regulates mammalian target of rapamycin (mTOR) through inactivation of a small GTPase, RAS homolog enriched in brain.² mTOR exists in rapamycin-sensitive mTOR complex 1 (mTORC1, composed of mTOR, DEPTOR, mLST8, PRAS40, and raptor) and rapamycin-insensitive mTOR complex 2 (mTORC2, composed of mTOR, DEPTOR, mLST8, mSin1 and rictor).³ mTORC1 can integrate various inputs, including growth factors, energy, nutrient, stress and oxygen to control cell survival, growth, proliferation and differentiation, while the mTORC2 pathway is insensitive to nutrients, but can be regulated by insulin-PI3K signaling and functions through several members of the AGC subfamily of kinases including AKT, SGK1 and PKCα.^4,5

mTORC1 has important roles in regulation of HIF1-, Myc- and SREBP-mediated glucose and lipid metabolism, as well as ULK1- and TFEB-mediated autophagy, in addition to S6K- and 4E-BP1-mediated protein synthesis.^6–10 Loss of the TSC1/TSC2 complex function activates mTORC1 signaling and drives tumor development in TSC. Thus, pharmacological inhibition of mTORC1 with rapamycin or RAD001 has shown promise in suppressing TSC-related tumor growth in clinical trials.^11,12 However, hyperactive mTORC1 also inhibits AKT1 through suppression of platelet-derived growth factor receptors and insulin receptor substrate 1, and enhancement of Grb10 phosphorylation.^13–18 Therefore, abolishment of this negative feedback regulation by repressing mTORC1 could compromise the efficacy of mTORC1 inhibitors in the treatment of TSC.

Calcium (Ca²⁺) is an important second messenger that controls various cellular processes by modulating Ca²⁺-regulated proteins and the corresponding signaling pathways. These processes range from short-term responses, such as muscle contraction and synaptic transmission, to long-term regulation of cell proliferation and apoptosis.¹⁹ Therefore, Ca²⁺ signaling is essential for the physiological function of cells and is finely controlled by a complicated transporter system. There are two main sources of intracellular free Ca²⁺: Ca²⁺ released from intracellular Ca²⁺ storage organelles, most notably the endoplasmic reticulum (ER), and Ca²⁺ influx from sources external to the cell. Stimulation of cell surface receptors leads to activation of PLC and consequent inositol 1,4,5-trisphosphate production, which results in rapid release of Ca²⁺ from ER through inositol 1,4,5-trisphosphate receptor channels. Depletion of ER Ca²⁺ induces Ca²⁺ influx across the plasma membrane. This Ca²⁺ influx is termed stored-operated Ca²⁺ entry (SOCE).^20,21 Recent studies have identified that stromal interaction molecule 1 (STIM1) and Orai1 (also named CRACM1) are responsible for SOCE.²¹ STIM1 is a single-pass transmembrane protein in the ER membrane and functions as the Ca²⁺ sensor in ER, while Orai1 is the pore-forming subunit of SOC channel in cell membranes.²² Upon ER Ca²⁺ depletion, STIM1 aggregates to form ‘puncta’ and rapidly translocates into ER-plasma membrane junctions, where it interacts with and activates Orai1 channels to induce SOCE.²³

STIM1-mediated SOCE is the predominant Ca²⁺ entry mechanism in non-excitable cells.^20,21 Disturbance of cellular Ca²⁺ homeostasis resulting from malfunction of SOCE may give rise to various diseases.²⁴ SOCE has also been found to affect cancer development.^25–31 However, the state of Ca²⁺ homeostasis in hyperactive mTOR cells and the roles of SOCE in TSC-related tumor progression have not been investigated. In this study, we examined the effect of TSC1/TSC2-mTOR signaling on cellular Ca²⁺ homeostasis and the role of STIM1-mediated SOCE in TSC-related tumor development.

RESULTS

Hyperactive mTORC1 signaling enhances SOCE

Dysregulation of Ca²⁺ signaling due to malfunction of specific Ca²⁺ transporters has been observed in tumor development.^25,26 However, the status of Ca²⁺ signaling was unknown in TSC-related cells. We examined cellular Ca²⁺ distribution by Ca²⁺ imaging in both wild-type (WT) (Tsc2^+/+) mouse embryonic fibroblasts (MEFs) and Tsc2-knockout MEF cells (Tsc2^−/− MEFs).³² Cells were first loaded with Fura 2-AM fluorescent dye, and then the ratio of F340/F380, an indicator of Ca²⁺ concentration, was recorded (Figure 1a). At resting state, the cytosolic free Ca²⁺ ([Ca²⁺]_c) was almost same in WT MEFs as in Tsc2 ^−/− MEFs (Figure 1a, Supplementary Figure 1a). When cyclopiazonic acid (CPA) was used to inhibit ER Ca²⁺-ATPase pump, Ca²⁺ released from ER ([Ca²⁺]_ER) was monitored. We found that there was no significant difference of [Ca²⁺]_ER between Tsc2^−/− MEFs and WT MEFs (Figure 1a, Supplementary Figure 1b). However, following ER Ca²⁺ depletion, Ca²⁺ influx through SOC into Tsc2 ^−/− MEF cells was increased compared with WT cells (Figures 1a and b). The mTORC1-specific inhibitor, rapamycin, blocked this elevation, but had no significant effect on [Ca²⁺]_c and [Ca²⁺]_ER (Figure 1a and Supplementary Figures 1a and b). After human TSC2 was introduced into Tsc2 ^−/− MEFs,³³ SOCE was decreased (Figures 1c and d) while [Ca²⁺]_c and [Ca²⁺]_ER remained unchanged (Figure 1c and Supplementary Figures 1c and d). These data suggest that mTORC1 is a positive regulator of SOCE.

Figure 1.

Hyperactive mTORC1 signaling enhances store-operated Ca²⁺ entry. MEF cells grown on glass coverslips were cultured in Dulbecco’s modified Eagle’s medium and then loaded with Fura2-AM to monitor Ca²⁺ fluorescence signaling. (a) Representative time course recording of intracellular Ca²⁺ fluorescence (F340/F380) in WT or Tsc2^−/− MEF cells pretreated with or without rapamycin (R) for 24 h. The bar above the tracings shows different perfusates with (black) or without (white) 1.3 mM CaCl₂. CPA was added to deplete the ER Ca²⁺ at the time indicated by the arrow. (b) Quantitation of Fura2-AM ratio changes induced by Ca²⁺ influx through SOC, from three to five separate preparations (n≥10 cells per preparation). Data represent mean±s.e.m. *P<0.05. (c) Representative time course recording of intracellular Ca²⁺ fluorescence (F340/F380) in Tsc2^−/− vector MEFs (Tsc2^−/− vec) and TSC2-restored Tsc2^−/− MEFs (Tsc2^−/− TSC2). (d) Quantitation of Fura2-AM ratio changes induced by Ca²⁺ influx through SOC, from three to four separate preparations (n≥10 cells per preparation). Data represent mean±s.e.m. *P<0.05.

mTORC1 positively regulates STIM1 expression

To explore how mTORC1 enhanced SOCE, we found that the protein level of STIM1 was increased in Tsc2^−/− MEF cells, and that elevation was reversed by rapamycin (Figure 2a and Supplementary Figure 2). These phenomena were also observed in Tsc1^−/− MEF cells in comparison with WT cells (Figure 2b). Moreover, restoration of TSC2 in Tsc2^−/− MEFs reduced STIM1 expression (Figure 2c). These data indicates that mTORC1 positively regulates STIM1 abundance.

Figure 2.

mTORC1 positively regulates STIM1 expression. (a–e) Total protein lysates were extracted from cells for immunoblotting. WT and Tsc2^−/− MEFs (a), WT and Tsc2 ^−/− MEFs (b), were treated with or without 10 nM rapamycin (R) for 24 h. (c) Tsc2^−/− MEFs were restored with human TSC2 by retroviruses. (d, e) Tsc2^−/− MEFs were transfected with small interfering RNA against rictor (d) or raptor (e) for 48 h. (f) Quantitative real-time PCR (QRT–PCR) analysis of STIM1 mRNA levels in WT and Tsc2 ^−/− MEFs treated with or without 10 nM rapamycin (R) for 24 h. Data represent mean±s.e.m. *P<0.05.

As mTOR exists in rapamycin-sensitive mTORC1 and rapamycin-insensitive mTORC2, we knocked down raptor and rictor, respectively, in Tsc2^−/− MEFs, and found that reduction of raptor decreased STIM1 abundance, whereas depletion of rictor did not change STIM1 level (Figures 2d and e), further confirming mTORC1 regulation of STIM1 abundance. To elucidate whether the enhanced STIM1 protein was due to its higher mRNA level, we used quantitative real-time PCR (QRT–PCR) to analyze STIM1 mRNA levels. Compared with WT MEFs, a substantial mTORC1-dependent elevation in STIM1 mRNA was observed in Tsc2 ^−/− MEFs (Figure 2f), suggesting that mTORC1 may increase STIM1 through stimulation of STIM1 transcription or inhibition of its mRNA degradation.

In addition to MEF cells, we evaluated the regulation of mTORC1 on STIM1 expression in cancer cell lines. Even though the levels of STIM1 differed significantly in the panel of cancer cell lines we examined, rapamycin reduced the abundance of STIM1 to various extents in all lines (Figure 3a). To investigate whether this newly discovered mTORC1 regulation of STIM1 exists in vivo, we checked the mouse kidney tumors driven by hyperactive mTORC1 due to Tsc2 exon 3 deletion (Tsc2^del3/+).³⁴ STIM1 was much higher in these kidney tumor samples than in adjacent kidney tissues (Figure 3b). Likewise, STIM1 was upregulated in the kidney angiomyolipoma tissue compared with matched adjacent kidney tissue from a TSC patient with TSC2 mutation (Figure 3c). Together, STIM1 is an effector of mTORC1 signaling, and upregulation of STIM1 may contribute to mTORC1 enhancement of SOCE.

Figure 3.

mTORC1 positively regulates STIM1 expression in cancer cell lines and tumor tissues. Protein lysates from cells or tissues were subjected to immunoblotting. (a) Cancer cell lines treated with or without 10 nM rapamycin (R) for 24 h. (b) Age-matched kidneys from two WT mice and kidney tumors from two Tsc2^del3/+ mice. (c) Kidney tumor tissue (T) and adjacent normal tissue (N) from a TSC patient.

Inhibition of SOCE by silencing STIM1 accelerates TSC-related tumor development

As Ca²⁺ is critical for many of the cellular processes and altered Ca²⁺ signaling is implicated in the progression of tumor development,^25,26 we reasoned that mTORC1 enhancement of SOCE might have important roles in TSC-related tumor formation. To test this hypothesis, we first knocked down STIM1 by lentivirus-expressing small hairpin RNA (shRNA) in Tsc2-null MEFs (Figure 4a). The effect of depleted STIM1 on SOCE was then examined by Ca²⁺ imaging. SOCE was impaired in Tsc2-null cells when STIM1 was depleted (Figures 4b and c). To test the in vivo consequence of the impaired SOCE in cells lacking Tsc2, we examined the tumorigenic potential of Tsc2^−/− MEFs expressing shRNA for STIM1, in comparison with scrambled shRNA. Interestingly, inhibiting SOCE increased the tumorigenic capacity of Tsc2-null MEFs in a nude mouse model, as assessed by tumor initiation and survival of tumor bearing mice after subcutaneously inoculating 10⁶ cells (Figures 4d and e). We further confirmed these results in NTC/T2 null cells, which were cultured from a subcutaneous tumor formed by Tsc2^−/− MEF cells in a nude mouse and had acquired greater potential of tumorigenesis.³⁵ Silencing SITM1 accelerated tumor development from NTC/T2 null cells (Supplementary Figures 3a–c). These data indicate that STIM1-mediated SOCE constrains TSC-associated tumor development.

Figure 4.

Attenuation of SOCE by knockdown of STIM1 accelerates TSC tumor development. Tsc2^−/− MEFs were transduced with lentiviruses expressing shRNA for STIM1 or scramble shRNA as control. (a) Cells were harvested for immunoblotting. (b, c) SOCE was measured by Ca²⁺ imaging. Data represent mean±s.e.m. *P<0.05. Cells were inoculated subcutaneously into nude mice, and then tumor formation (d) and survival of the mice (e) were analyzed.

Inhibition of SOCE by dominant-negative Orai1 promotes TSC-related tumor development

It has been well established that Orai1, the downstream effector of STIM1, is the pore-forming component of SOC channel in the plasma membrane. To assess more directly the role of SOCE in tumor development, human dominant-negative Orai1-E106Q (hDN-Orai1) was used to block SOC.³⁶ We first expressed hDN-Orai1 and then examined the SOC activity in Tsc2-null MEF cells (Figure 5a). Ca²⁺ imaging results showed that hDN-Orai1 abolished Ca²⁺ entry in these cells (Figures 5b and c). We verified the effect of hDN-Orai1 on SOC by nuclear factor of activated T-cells translocation assays, as nuclear factor of activated T-cells translocates from cytoplasm to the nucleus in response to sustained elevation of cytosolic Ca²⁺ via SOCE.^29,37 At resting state, green fluorescent protein-tagged nuclear factor of activated T-cells localized exclusively in the cytoplasm of both control and hDN-Orai1-expressing cells (Supplementary Figure 4a). Following treatment with CPA, nuclear translocation of nuclear factor of activated T-cells-green fluorescent protein took place in nearly 99.4% of control cells, but in only 19.7% of hDN-Orai1-expressing cells (Supplementary Figure 4b), indicating that hDN-Orai1 successfully blocked SOCE in Tsc2-null MEF cells. Next, we examined the effect of hDN-Orai1 on the tumorigenic potential of Tsc2^−/− MEFs, and found that hDN-Orai1-expressing cells exhibited accelerated tumor growth in nude mice in comparison with control cells, as assessed by tumor initiation and mouse survival (Figures 5d and e). These data suggest that SOCE indeed restricts TSC-related tumor growth.

Figure 5.

Blocking SOCE by ectopically expressing hDN-Orai1 promotes TSC tumor development. Tsc2^−/− MEFs were transduced with retroviruses for hDN-Orai1 expression or control vector. (a) RT–PCR analysis to detect the expression of hDN-Orai1. (b, c) SOCE was measured by Ca²⁺ imaging. Data represent mean±s.e.m. *P<0.05. Cells were inoculated subcutaneously into nude mice, and then tumor formation (d) and survival of the mice (e) were analyzed.

Inhibition of SOCE increases AKT1 activity in Tsc2-null cells

TSC is a benign tumor syndrome with hyperactive mTOR signaling and consequential decreased AKT1 activity due to the negative feedback regulation of mTOR on AKT1.^14,16–18 We speculated that inhibition of SOCE might disrupt this feedback loop and then restore AKT1 activity, which in turn could accelerate TSC-related tumor growth. As expected, SKF-96365, an inhibitor of SOC, increased the phosphorylation of AKT1 in Tsc2^−/− MEFs but had no obvious effect in WT MEFs (Figure 6a). In addition, CPA, an inducer of SOCE, decreased the phosphorylation of AKT1, and SKF-96365 reversed the suppression of CPA on AKT1 activation (Figure 6b). Consistently, inhibition of SOCE by silencing STIM1 in Tsc2^−/− MEF cells increased the phosphorylation of AKT1. Serum stimulation for 10 min after 24 h-starvation also potentiated AKT1 activation in STIM1-knockdown Tsc2^−/− MEF cells (Figure 6c). Moreover, hDN-Orai1-expressing Tsc2^−/− MEFs had higher phosphorylation of AKT1 and its substrates, GSK3β and BAD, while ERK signaling remained unchanged (Figure 6d). These data demonstrate that SOCE negatively regulates AKT1, and the restoration of AKT1 activity may contribute to the acceleration of Tsc2^−/− cell tumor development if SOCE is impaired.

Figure 6.

Inhibition of SOCE increases AKT1 activity in Tsc2-null cells. Protein lysates from cells were subjected to immunoblotting. (a) WT and Tsc2^−/− MEFs were treated with 10 μM SKF 9636 to block SOC at indicated time points. (b) Tsc2^−/− MEFs were pretreated with or without 10 μM SKF 9636 to block SOC for 5 min, followed by addition of 10 μM CPA to induce SOCE at indicated time points. (c) WT MEFs transduced with lentiviruses for scramble shRNA and Tsc2^−/− MEFs transduced with lentiviruses expressing the shRNA for STIM1 or scramble shRNA were cultured in complete media treated with or without 10 nM rapamycin (R) for 24 h, or starved for 24 h and then stimulated with 10% serum for 10 min. (d) Tsc2^−/− MEFs were transduced with retroviruses for hDN-Orai1 expression or control vector.

Inhibition of SOCE promotes TSC-related tumor angiogenesis

Angiogenesis is essential to support tumor development by providing nutrients and oxygen, and replenishing the microenvironment. We noticed that after Tsc2^−/− MEF cells expressing shRNA against STIM1 were injected subcutaneously into nude mice, the established solid tumor masses were surrounded by pool of blood, and for Tsc2^−/− MEF cells expressing hDN-Orai1, the tumors ulcerated much earlier. These findings prompted us to hypothesize that suppression of SOCE might affect the permeability of tumor vessels. As vascular endothelial growth factor ((VEGF), also named vascular permeability factor) can increase vascular permeability and promote tumor angiogenesis,³⁸ we checked VEGF expression by western blot and found that inhibiting SOCE either by knocking down STIM1 or expressing hDN-Orai1 increased VEGF abundance in Tsc2^−/− MEF cells (Figures 7a and b). The secreted VEGF in cell culture supernatants was also increased when SOCE was blocked in Tsc2^−/− MEF cells (Figures 7c and d). Moreover, knockdown of STIM1 in WT MEFs and NIH3T3 cells increased VEGF secretion (Supplementary Figure 5a and b), indicating that SOCE suppresses VEGF expression. We also found that the serum VEGF concentration from nude mice with tumors derived from hDN-Orai1-Tsc2^−/− MEF cells was higher than that of control mice (Supplementary Figure 5c). Histologically, vascular endothelial cell staining by anti-Von Willebrand factor (vWF) antibodies revealed that tumors established from SOCE-impaired Tsc2 ^−/− MEF cells formed more microvasculature than control cell-derived tumors did (Figures 7e and f). These data suggest that inhibition of SOCE may accelerate TSC-related tumor development through VEGF-driven tumor angiogenesis and increased vascular permeability, both of which can facilitate nutrients and oxygen supplies for tumor growth.

Figure 7.

Inhibition of SOCE promotes TSC tumor angiogenesis. Tsc2 ^−/− MEFs were transduced with lentiviruses expressing shRNA for STIM1 or scramble shRNA as control (a, c and e); transduced with retroviruses for hDN-Orai1 expression or control vector (b, d and f). (a, b) Total protein lysates were extracted for immunoblotting. (c, d) Cell culture supernatants were analyzed for VEGF by enzyme-linked immunosorbent assay assay. Data represent mean±s.e.m. *P<0.05. RU, relative unit. (e, f) H&E staining and vWF immunohistochemistry of xenografted subcutaneous tumors from mice.

mTORC1-STIM1-SOCE-AKT1 signaling regulation in TSC-related rat uterine leiomyoma cells

To exclude the possibility that the negative effect of mTOR regulation of SOCE on tumor development was unique to the MEF cells in our study, we examined this novel mTOR-mediated Ca²⁺ signaling cascade in Tsc2-null ELT-3 cells, which were cultured from a uterine leiomyoma in an Eker rat.^39,40 mTORC1 upregulation of STIM1 expression was observed in these cells (Figure 8a), as STIM1 level was reduced by rapamycin. When SOCE was inhibited in ELT-3 cells by knocking down STIM1 (Figures 8b and c), AKT1 activity was upregulated and more VEGF was secreted from these cells (Figures 8d and e). Furthermore, blocking SOCE potentiated the tumorigenic capacity of ELT-3 cells, as reduced survival was observed in the nude mice bearing subcutaneous tumors originated from STIM1-depleted ELT-3 cells (Figure 8f). In agreement with the findings from MEFs, mTORC1-regulated STIM1-SOCE signaling also acts as a tumor suppressor in uterine leiomyoma development caused by loss of TSC2.

Figure 8.

mTORC1-STIM1-SOCE-AKT1 signaling cascade in TSC-related rat uterine leiomyoma cells. (a) Total protein lysates were extracted from ELT-3 cells treated with or without 10 nM rapamycin (R) for 24 h for immunoblotting. ELT-3 cells were transduced with lentiviruses expressing the shRNA for STIM1 or scramble shRNA as control (b–f). (b, c) SOCE was measured by Ca²⁺ imaging. (d) Total protein lysates were extracted for immunoblotting. (e) Cell culture supernatants were analyzed for VEGF by enzyme-linked immunosorbent assay. Data represent mean±s.e.m. *P<0.05. RU, relative unit. (f) Cells were inoculated subcutaneously into nude mice, and then survival of these mice was analyzed.

DISCUSSION

In the present study, we have demonstrated that mTORC1 signaling positively regulates SOCE, the predominant Ca²⁺ entry mechanism in non-excitable cells. The ER Ca²⁺ sensor STIM1 is a key effector of mTORC1 in regulating SOC-induced Ca²⁺ signaling. Inhibition of SOCE increases AKT1 activity and angiogenesis, and in turn may exacerbate TSC-related tumor development.

The receptor tyrosine kinase-PI3K/PTEN-AKT-TSC1/2-mTORC1 pathway has important roles in regulating cell survival, growth, proliferation and differentiation.^4,41 Because of recurrent gain-of-function mutations of proto-oncogenes or loss-of-function mutations of tumor suppressors, mTORC1 signaling is frequently augmented in a variety of human cancers.⁴² We determined that Tsc2-deficient MEF cells and rat uterine leiomyoma cells exhibited robust SOCE in response to ER Ca²⁺ depletion, and the elevation of Ca²⁺ influx through SOC was mTORC1-dependent. Our data thus suggest that mTORC1 is a positive regulator of SOCE. In addition, STIM1 was upregulated and contributed to the enhancement of SOCE in hyperactive mTORC1 cells. These results are consistent with a recent publication describing the upregulation of STIM1 and Orai1 via AKT/mTOR signaling in human pulmonary arterial smooth muscle cells.⁴³ Taking together the findings from MEFs, mouse, rat and human tumor cell lines, as well as TSC-related kidney tumors from mouse and human, mTORC1 regulation of store-operated Ca²⁺ signaling may function widely in different tissues for various physiology processes, and aberrations of this signaling cascade could contribute to broad pathological conditions.

Ca²⁺ is a ubiquitous and essential second messenger in cells. Changes in Ca²⁺ transporters/Ca²⁺ signaling lead to disruption of biological processes that are intimately involved in tumor progression, such as cell proliferation, survival, differentiation, mobility, invasion and angiogenesis.^25,26 Recent studies have shown that Ca²⁺ influx through TRPC and/or Orai1 activated by SPCA2 or STIM1 is critical for the development of liver, breast and cervical cancers.^28–31 In our study, SOCE is elevated by mTORC1 signaling, and unexpectedly, inhibition of SOCE promotes instead of inhibits TSC-related tumor growth in nude mouse model. It is noteworthy that before the identification of its role as an ER Ca²⁺ sensor, STIM1 was originally proposed as a tumor suppressor in rhabdomyosarcoma tumors⁴⁴ and was defined as an anti-metastasis gene in melanoma cells.⁴⁵ Moreover, SOC-induced Ca²⁺ influx was downregulated to protect prostate cancer cells from apoptosis in the late androgen-independent stage.⁴⁶ Taken together, SOCE may acts as negative regulator for tumor development.

Loss of TSC1 or TSC2 leads to hyperactive mTOR signaling, which is the main cause of tumor growth in TSC patients, and the decreased AKT1 activity due to the negative feedback regulation of mTOR may account for the benign nature of TSC tumors.¹⁸ Compromised expression of insulin receptor substrate 1 and platelet-derived growth factor receptors reduce the signaling input from PI3K to AKT1.^14,15 Recent findings reveal that Grb10 is another mediator of this feedback inhibition.^16,17 Our data demonstrate that impairment of SOCE restores AKT1 activity in Tsc2-null cells, suggesting that enhanced SOCE by hyperactive mTORC1 also has a critical role in the negative feedback regulation. Increased AKT1 activity by disruption of SOCE may contribute to the malignancy of TSC tumors in the animal models used in our study. How SOCE suppresses AKT1 or inhibition of SOCE disturbs the feedback regulation of AKT1 by mTORC1 remains to be elucidated.

SOCE is essential for platelet procoagulant activity. STIM1^−/− or Orai1^−/− platelets exhibit deficiencies in thrombus formation.⁴⁷ Constitutively active STIM1 elevates Ca²⁺ concentration and promotes maturation of platelets. The increased platelet consumption also leads to macrothrombocytopenia and bleeding in mice.⁴⁸ Although enhanced production of VEGF has been observed in Tsc1- or Tsc2-knockout cells and mice,⁴⁹ here we demonstrated that depletion of STIM1 or ectopic expression of hDN-Orai1 in Tsc2^−/− MEF cells further increased the production of VEGF and then caused tumor bleeding in nude mice, suggesting that SOCE may affect tumor vascular stability and permeability. Therefore, dysregulation of SOCE in both tumor cells and platelet cells may account for human cancer hemorrhage. Our observations show that inhibition of SOCE enhances tumor angiogenesis and vascular permeability, both of which facilitate nutrient and oxygen supplies for tumor growth.

In summary, we have found that loss of Tsc2 enhances SOCE through mTORC1-dependent upregulation of STIM1 and SOCE is a potent suppressor of AKT1. Inhibition of SOCE restores AKT1 activity and enhances angiogenesis, thus unleashing the constraint of SOCE on TSC tumor development. We speculate that the frequently altered mTORC1 signaling may potentiate SOCE in various human cancers, but the exact role of SOCE in cancer development needs to be further determined, especially in cancers that are refractory to mTORC1-mediated negative regulation of AKT1 signaling, such as cancers with PTEN loss or constitutively active mutant PI3K/AKT1. SOCE inhibition of AKT1 may contribute to benign tumor growth in TSC patients, and the removal of mTORC1-STIM1-SOCE suppression on AKT1 would compromise the efficacy of mTOR inhibitors in the treatment of TSC patients. Patients with TSC should, therefore, avoid taking antagonists of SOCE. In contrast, SOCE agonists could neutralize AKT1 reactivation exerted by mTOR suppression and consequently potentiate the efficacy of mTOR inhibitors in the treatment of the patients with TSC.

MATERIALS AND METHODS

Reagents

Ethylene glycol tetraacetic acid, Rapamycin, Fura 2-AM, hygromycin B, SKF-96365 and CPA were obtained from Sigma-Aldrich (St Louis, MO, USA). Dulbecco’s modified Eagle’s medium, fetal bovine serum and Lipofectamine 2000 were from Invitrogen (Carlsbad, CA, USA). All restriction enzymes were from Takara (Kyoto, Japan).

Antibodies

Antibodies against TSC2, ERK, phospho-ERK and β-actin were from Santa Cruz Biotechnology (Santa Cruz, CA, USA), and antibodies against AKT1, phospho-AKT1 (Ser473), phospho-GSK3β (Ser9), phospho-BAD (Ser136) were from Cell Signaling Technology (Danvers, MA, USA). The mouse anti-STIM1 antibody was obtained from BD Biosciences (San Jose, CA, USA). vWF antibody was from Abcam (Cambridge, MA, USA). VEGF antibody was from Millipore (Billerica, MA, USA). Anti-phospho-S6 (Ser235/236) and anti-S6 antibodies have been described previously.³²

Cell lines and culture

MEF cells and NTC/T2 null cells have been described previously.^14,35 PT67 cells were from Clontech (Mountain View, CA, USA). MCF7, HCT-116, A549, HepG2, MDA-MB-468, MDA-MB-231, PC3, CaSki, SiHa, B16 and HEK 293 T cell lines were from American Type Culture Collection (Manassas, VA, USA). Tsc2-deficient ELT-3 cells were established from a uterine leiomyoma in an Eker rat with a germline insertion in Tsc2 gene.^39,40 ELT-3 cells were maintained and propagated in Dulbecco’s modified Eagle’s medium/F12 (1:1) with 10% fetal bovine serum. The other cell lines used in this study were cultured in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum.

Measurement of intracellular Ca²⁺

Cells seeded on coverslips were loaded with 3 μM Fura 2-AM by incubation in Hank’s balanced salt solution (136.9 mM NaCl, 4.2mM NaHCO₃, 0.3 mM Na₂HPO₄, 5.4 mM KCl, 0.4 mM KH₂PO₄, 0.5 mM MgCl₂, 0.4 mM MgSO₄, 5.5 mM D-Glucose, 1.3 mM CaCl₂, 8.0 mM Hepes, pH7.2) for 30 min in the dark at room temperature, followed by washing and an additional 30 min incubation to ensure full de-esterification. Coverslips were then transferred to a perfusion chamber. For every coverslip, at least 10 cells were measured simultaneously (excited at 340 and 380 nm with emission at 510 nm) using a ×20 objective xenon lamp (Lambda DG4, Sutter Instrument Company, Novato, CA, USA) equipped with a Nikon epifluorescence microscope (TE2000-U; Nikon, Tokyo, Japan) and band-pass filters for wavelengths of 340 and 380 nm. Based on the equation: ${[{\text{Ca}}^{2+}]}_{\mathrm{c}}=K_{\mathrm{d}}\times (S_{\mathrm{f}2}/S_{\mathrm{b}2})\times (R-R_{min})/(R_{max}-R)$, the [Ca²⁺]_c can be represented by the ratio (R) of 340/380 nm fluorescence images. Cells were first perfused with Hank’s balanced salt solution containing 1.3 mM CaCl₂ to determine the resting cytosolic free Ca²⁺ concentration ([Ca²⁺]_c). Then, the buffer was replaced by Ca²⁺-free and 2 mM ethylene glycol tetraacetic acid-containing Hank’s balanced salt solution to remove extracellular Ca²⁺. Once the 340 nm/380 nm ratio reached a steady state, CPA was added to achieve final 10 μM concentration and Ca²⁺ released from ER ([Ca²⁺]_ER) was monitored. When ER Ca²⁺ was depleted, 1.3 mM CaCl₂ was added back into the perfusate to record Ca²⁺ influx (SOCE).

Immunoblotting

Whole cells were lysed in lysis/loading buffer containing 2% SDS, 10% glycerol, 10 mM Tris (pH 6.8) and 100 mM dithiothreito, boiled for 10 min, and then subjected to immunoblotting as described previously.⁴¹

RT–PCR

Total RNA was extracted from cells using Trizol (Invitrogen) following the manufacturer’s instructions. One microgram of RNA was reversely transcribed using the PrimeScriptTM RT Reagent Kit (Takara) in a 20 μl reaction. After 20-fold dilution, 5 μl of cDNA was used as template in a 20 μl PCR reaction. Oligonucleotide primers were synthesized to detect murine β-actin and human Orai1 E106Q mutant stably expressed in Tsc2^−/− MEF cells. Primers were designed specifically for human Orai1 to prevent coamplification of murine Orai1. The sequences were as follows: Mouse β-Actin forward: 5′-GGCCCAGAGCAAGAGAGGTATC-3′ reverse: 5′-GCCATC TCCTGCTCGAAGTCTAG-3′ Human Orai1 forward: 5′-CGCTGACCACGAC TACCCAC-3′ reverse: 5′-GGACGGCGAAGACGATAAAG-3′.

Quantitative real-time PCR (QRT–PCR)

cDNA was prepared as above for RT–PCR. After 20-fold dilution, 5 μl of cDNA was used as template in a 20 μl QRT–PCR reaction. Amplification was performed for 40 cycles using TransStart Green quantitative PCR SuperMix (TransGen Biotech, Beijing, China). Oligonucleotide primers were synthesized to detect Orai1 and STIM1 with β-actin as internal control. Primers were designed on exon junctions to prevent coamplification of genomic DNA, and the sequences as follows: Mouse β-Actin forward: 5′-GGATGCACCCGCCTAAGG-3′ reverse: 5′-AGAGGGAAATCGTGCGTGAC-3′ Mouse STIM1 forward: 5′-CGTCCGAAACATCCATAAGC-3′ reverse: 5′-CACCT CATCCACAGTCCAGTTG-3′.

RNA interference

Small interfering RNA oligonucleotides for negative control (5′-TTCTCCGAACGTGTCACGT-3′), Raptor (5′-AAGGACAACGGTCACAAGTAC-3′) and Rictor (5′-AAGCCCTACAGCCTTCATTTA-3′) were synthesized by Gene-Pharma Company Ltd (Shanghai, China). Cells were seeded in 12-well plates and transfected with small interfering RNAs using Lipofectamine 2000 following the manufacturer’s instructions. Forty-eight hours later, cell lysates were harvested for immunoblotting.

Plasmids and viruses

pLL3.7-shRNA expression vector was used to knockdown STIM1; the target sequence was 5′-GAGTCTACCGAAGCAGAGT-3′. Lentiviruses were generated by co-transfecting pLL3.7 and the packaging vectors (VSVG, REV and pMDL) into 293T cells. Forty-eight hours later, viruses were collected and filtered through a 0.45 μm filter for cell infection. The infection rate was evaluated by the expression of green fluorescent protein after incubation with viruses for 2 days.

Using Addgene (Cambridge, MA, USA) plasmid 19757 as template for human WT Orai1, dominant-negative Orai1 (E106Q) was mutated with primers 5′-GGTGGCAATGGTGGCGGTGCAGCTGGACG-3′ and 5′-CGTCCA GCTGCACCGCCACCATTGCCACC-3′, and amplified with primers of 5′-G TAGGATCCACCATGCATCCGGAGCCCGCC-3′ and 5′-TACATCGATCTAGGCAT AGTGGCTGCCGGG-3′, and then cloned into pLXIN-hyg vector at Xho I/Hpa I to obtain pLXIN-hyg-hDN-Orai1. Retroviruses were generated by transfected pLXIN-hyg plasmids into the packaging cell line PT67 using Lipofectamine 2000. Forty-eight hours later, conditioned culture medium containing viruses was filtered through a 0.45 μm filter, and then used to transduce cells. The transduced cells were selected with 100 μg/ml hygromycin B.

Quantification of VEGF secretion by enzyme-linked immunosorbent assay

To detect secreted VEGF levels from cells, a total of 6–8 × 10⁴ cells per well were seeded in 12-well plates overnight, then changed with 1 ml fresh medium and cultured for 2 days. Cell numbers were counted before measurements using a hemocytometer, and secreted VEGF levels were determined in cell-free culture supernatants using an enzyme-linked immunosorbent assay kit (R&D Systems, Minneapolis, MN, USA). Then, the secretion of VEGF was normalized to cell numbers.

To detect the serum VEGF levels from nude mice bearing tumors, 0.5–1 ml blood was collected and allowed to clot for 20–30 min at room temperature. The clotted material was removed by centrifugation at 3000 r.p.m. for 15 min at 4 °C. Then, the supernatants were obtained as sera to detect VEGF levels using an enzyme-linked immunosorbent assay kit (R&D Systems).

Mouse kidney tumor assessment

The normal kidney tissues from two WT mice (22- and 23-months-old) and kidney tumors from two heterozygous Tsc2 exon 3 deletion (Tsc2^del3/+) mice (19- and 22-months-old) were sonicated and extracted for immunoblotting using lysis buffer.³⁴ Animal protocols were approved by the Center for Animal Resources and Comparative Medicine (Boston) and were compliant with federal, local and institutional guidelines on the care of experimental animals.

Human kidney tumor assessment

Partial nephrectomy of the right kidney was performed for a 16-year-old female TSC patient with a frameshift mutation of TSC2 (g.10059delC, p.S132SfsX50) in the Department of Urology, Peking Union Medical College Hospital. Kidney tumor and adjacent normal kidney samples were freshly obtained from a resected angioleiomyolipoma for immunoblotting. All the procedures were performed under the permission of the Peking Union Medical College Hospital Ethics Board.

Induction of subcutaneous tumors in nude mice

Subcutaneous tumors were established as described previously.^14,41,50 Immunodeficient nude mice (BALB/c, 6–8-weeks-old) were obtained from the Institute of Laboratory Animal Sciences, Chinese Academy of Medical Sciences and Peking Union Medical College: 6–8 mice were used in each cohort.

Tumor growth and mouse survival were assessed after subcutaneous inoculation of 1 × 10⁶ MEFs or 5 × 10⁶ ELT-3 cells into the right posterior back region. Mice were euthanized when tumor size was >1000 mm³, there was ulceration over the tumor, or weight loss of >10% occurred. The animal protocol was approved by the Animal Center of the Institute of Basic Medical Sciences of the Chinese Academy of Medical Sciences and Peking Union Medical College and was in compliance with the regulations of the Beijing Administration Office of Laboratory Animals on the care of experimental animals.

Histology

Nude mouse tumor samples were fixed in 4% buffered paraformaldehyde at room temperature and embedded in paraffin. Five micrometer-thick sections were stained with H&E according to standard protocols. Immunohistochemistry for detection of the vWF signal was performed using a standard protocol. Anti-vWF antibody was used at 1:500 and incubated in a humidified chamber overnight at 4 °C, and horseradish peroxidase-goat anti-rabbit IgG (Zhongshan Jinqiao, Beijing, China) was used as the secondary antibody. Diaminobenzidine was used to develop the signal. Sections were lightly counterstained with hematoxylin.

Statistics

The Kaplan–Meier log-rank test was used for analysis of mouse tumor development with GraphPad Prism software (GraphPad Software, La Jolla, CA, USA). Comparison between the groups of data was conducted using the two-tailed Student’s t-test. Differences were considered significant when P<0.05.