Brain-expressed X-linked 2 Is Pivotal for Hyperactive Mechanistic Target of Rapamycin (mTOR)-mediated Tumorigenesis*

Background: mTOR signaling pathway is frequently activated in cancer. Results: Hyperactivation of mTOR stimulates STAT3/NF-κB-BEX2-VEGF signaling cascade. Conclusion: mTOR promotes tumorigenesis through up-regulation of STAT3/NF-κB-BEX2-VEGF signaling axis. Significance: The components in this mTOR-STAT3/NF-κB-BEX2-VEGF signaling cascade are candidate targets for the treatment of cancers associated with aberrant mTOR signaling. Frequent alteration of upstream proto-oncogenes and tumor suppressor genes activates mechanistic target of rapamycin (mTOR) and causes cancer. However, the downstream effectors of mTOR remain largely elusive. Here we report that brain-expressed X-linked 2 (BEX2) is a novel downstream effector of mTOR. Elevated BEX2 in Tsc2−/− mouse embryonic fibroblasts, Pten−/− mouse embryonic fibroblasts, Tsc2-deficient rat uterine leiomyoma cells, and brains of neuronal specific Tsc1 knock-out mice were abolished by mTOR inhibitor rapamycin. Furthermore, BEX2 was also increased in the liver of a hepatic specific Pten knock-out mouse and the kidneys of Tsc2 heterozygous deletion mice, and a patient with tuberous sclerosis complex (TSC). mTOR up-regulation of BEX2 was mediated in parallel by both STAT3 and NF-κB. BEX2 was involved in mTOR up-regulation of VEGF production and angiogenesis. Depletion of BEX2 blunted the tumorigenesis of cells with activated mTOR. Therefore, enhanced STAT3/NF-κB-BEX2-VEGF signaling pathway contributes to hyperactive mTOR-induced tumorigenesis. BEX2 may be targeted for the treatment of the cancers with aberrantly activated mTOR signaling pathway.

Aberrant vascularization is a feature of TSC lesions including facial angiofibroma, retinal hamartomas, cardiac rhabdomyomas, pulmonary lymphangioleiomyomatosis, renal angiomyolipoma, and liver hemangiomas (13). A variety of angiogenesis activating factors have been elucidated, such as basic FGF (bFGF), TGF-␣, TGF-␤, PDGF, and VEGF. The VEGF serum levels in some cancer patients are elevated (14), and VEGF is a therapeutic target for anti-angiogenesis (15). Studies have revealed that mTOR signaling pathway augments VEGF expression (16 -18). However, the underlying mechanism of mTOR up-regulation of VEGF is not completely understood.
Brain-expressed X-linked 2 (BEX2), one of the BEX family members, is abundantly expressed in the central nervous system (19). Evidence has shown that BEX2 is overexpressed in breast cancer (20). BEX2 promotes the growth of breast cancer cells and inhibits mitochondrial apoptosis (21). Reduction of BEX2 expression inhibits angiogenesis in vivo (22) and glioma cell migration and invasion (23). Conversely, BEX2 has been found as a tumor suppressor in human glioma because overexpression of BEX2 in glioma cells results in suppression of tumor growth in vitro and in vivo (24). Hence, the precise function, as well as the regulatory mechanisms, of BEX2 in the development of malignant diseases is still not clear.
In this study, we show that mTOR stimulated BEX2 expression. Hyperactive mTOR promoted carcinogenesis through up-regulation of STAT3/NF-B-BEX2-VEGF signaling cascade. This study not only unveils an effector in the mTOR signaling pathway, but also reveals a novel mechanism for augmented VEGF expression induced by hyperactive mTOR signaling. In addition, our work provides candidate targets for treatment of cancers associated with aberrant mTOR signaling.
Expression Profiling Analysis-RNA from WT and Tsc2 Ϫ/Ϫ MEFs was subjected to microarray on an Affymetrix GeneChip system with the Affymetrix Mouse Genome 430 2.0 Array, and data were analyzed using the Partek Express software (31).
Immunoblotting Analysis-Cells were washed with cold PBS, harvested on ice in lysis buffer (2% SDS, 100 mM DTT, 10 mM Tris (pH 6.8), and 10% glycerol), and boiled for 10 min. Cell lysates were resolved by SDS-PAGE. Subsequently, proteins were transferred onto PVDF membrane (Millipore). The membranes were blocked in TBST (0.1 M Tris-Cl (pH 7.5), 0.9% NaCl, 0.5% Tween 20) containing 3% nonfat dry milk at room temperature and incubated with the indicated antibodies diluted with TBST containing 3% nonfat dry milk. After washing with TBST, the membranes were incubated with HRP-labeled secondary antibodies and then detected by chemiluminescence.
Chromatin Immunoprecipitation Assay-Chromatin immunoprecipitation was conducted to examine DNA-protein interactions with an anti-STAT3 antibody and a SimpleChIP enzymatic chromatin IP kit (Cell Signaling Technology) according to the manufacturer's protocol. The released DNA was purified and then used for analysis by PCR. The primer sequences used for PCR are as follows: the putative STAT3 binding site region of mouse BEX2, forward, ATGCTCTAACAACTGGACTT, and reverse, CTCGACTCAATAGATTTACTC; a nonspecific STAT3 binding region of mouse BEX2, forward, GGTGC-TGAATCTTTGAACA, and reverse, ATCCCTTTTGCT-AGCATC.
BEX2 Overexpression in WT MEFs-The coding sequence of BEX2 was inserted into PLXIN-hyg vector between XhoI and ClaI sites. PLXIN-hyg-BEX2 and PLXIN-hyg vector were transfected into retroviral packaging PT67 cells using Lipofectamine 2000. After cell selection with hygromycin B (100 g/ml), cell culture supernatants containing viruses were harvested and filtered with a 0.45-m filter for subsequent cell infection. Stably expressing cell lines were generated through selection with hygromycin B (100 g/ml).
VEGF ELISA-To determine the levels of VEGF secreted from cells, cells (30,000 -40,000 cells/well) were seeded in 12-well plates in triplicates. The next day, culture medium was replaced with 1 ml of fresh medium, and then cells were cultured in 5% CO 2 at 37°C for 48 h. The number of cells was counted by using a Vi-CELL (Beckman Coulter). Cell culture supernatants were collected, and the secreted VEGF levels in the supernatants were measured with a VEGF ELISA kit (R&D Systems, Minneapolis, MN). The levels of VEGF secretion were normalized to the number of cells (32).
Tumor Engraftment onto Chick Chorioallantoic Membrane-Fertilized chicken eggs were handled as described (33). Briefly, on embryonic day 10, 5,000,000 cells with overexpressed BEX2 or WT cells in 20 l of culture medium were deposited on chick chorioallantoic membrane. Digital pictures were taken under a stereomicroscope (Nikon SMZ1500) at days 3, 5, and 7 of tumor development.
Induction of Subcutaneous Tumors in Nude Mice-Immunodeficient nude mice (BALB/c, 4 -5 weeks old) were purchased from the Institute of Laboratory Animal Science, Chinese Academy of Medical Sciences and Peking Union Medical College. Eight male mice were in each cohort. The subcutaneous tumor model was established as described previously (25,31).
Human Kidney Tumor Assessment-The kidney angiomyolipoma tissue and adjacent normal tissue from a 16-year-old TSC patient with a frameshift mutation in the TSC2 gene (g.10059delC, p.S132SfsX50) were extracted with lysis buffer and then subjected to immunoblotting (32). All the procedures were performed under the permission of the Peking Union Medical College Hospital Ethics Board.
Mouse Brain Assessment-WT mice, neuronal specific Tsc1 knock-out (Tsc1 f/f Syn1Cre) mice, and neuronal specific Tsc1 knock-out mice treated with rapamycin were sacrificed on postnatal day 21, and brain tissues were extracted with lysis buffer and then subjected to immunoblotting as described previously (35). Total RNA was extracted from the whole brain tissues of WT, and neuronal specific Tsc1 knock-out mice were sacrificed on postnatal day 13 for qRT-PCR analysis.
Mouse Liver Assessment-Mice with hepatocyte-specific deficiency of Pten were generated by crossing Pten f/f mice with AlbCre transgenic mice (36). The liver tissues from a WT mouse (Pten ϩ/ϩ AlbCre, 9 months old) and a liver-specific Pten-deficient mouse (Pten f/f AlbCre, 9 months old) were extracted with lysis buffer and then subjected to immunoblotting (37). All animal protocols were approved by the Animal Center of the Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences and Peking Union Medical College, and were in accordance with the regulation of the Beijing Administration Office of Laboratory Animal on the care of experimental animals.
Statistical Analysis-Mouse tumor development and survival data were analyzed using the Kaplan-Meier log-rank test, and the two-tailed Student's t test was used to conduct the comparison between the groups. It was of statistical significance when p Ͻ 0.05.

Results
Loss of PTEN or TSC2 Potentiates BEX2 Expression-To identify novel targets of TSC2, we extracted mRNA from WT and Tsc2 Ϫ/Ϫ MEFs and conducted the gene expression profiling analysis (supplemental Table 1). We found that the abundance of BEX2 was higher in Tsc2 Ϫ/Ϫ MEFs than in WT MEFs (Table 1). Quantitative real time PCR revealed that the mRNA level of BEX2 in Tsc2 Ϫ/Ϫ MEFs was significantly higher than that of WT MEFs (Fig. 1A). Furthermore, the protein level of BEX2 in Tsc2 Ϫ/Ϫ MEFs was also increased (Fig. 1B). To examine whether the negative regulation of TSC2 on BEX2 exists in human cells, we knocked down TSC2 in human lung adenocarcinoma cell line A549. Reduction of TSC2 resulted in an increase of BEX2 expression in A549 cells (Fig. 1C). Moreover, the ectopically expressed human TSC2 in rat uterine leiomyoma cell line ELT3 (Tsc2-deficient cells) inhibited BEX2 expression (Fig. 1D). Similarly, BEX2 protein was much higher in Pten Ϫ/Ϫ MEFs than in WT MEFs (Fig. 1E). Furthermore, the protein level of BEX2 was increased in the mouse liver tissue with hepatocyte-specific Pten deletion (Fig. 1F). BEX2 was much higher in renal tumor tissues than in adjacent normal renal tissues from Tsc2 ϩ/Ϫ mice (Fig. 1G). Moreover, the protein level of BEX2 in renal tumor tissues from a 16-year-old female patient with TSC was higher when compared with adjacent normal renal tissues (Fig. 1H).
mTOR Positively Regulates BEX2 Expression-Because deletion of either TSC2 or PTEN leads to mTOR activation (7) and BEX2 overexpression, we explored whether there is a regulatory relationship between mTOR and BEX2. We treated WT and Tsc2 Ϫ/Ϫ MEFs with the mTOR inhibitor, rapamycin. Rapamycin inhibited BEX2 expression in WT and Tsc2 Ϫ/Ϫ MEFs (Fig. 2, A and B). A similar change was presented in WT and Pten Ϫ/Ϫ MEFs with rapamycin treatment (Fig. 2C). To test the in vivo relevance of this finding, we examined the expression of BEX2 in the brain tissues dissected from WT mice and the neuronal specific Tsc1 knock-out mice (Tsc1 f/f Syn1Cre). The level of BEX2 mRNA was elevated in the brain tissues derived from the neuronal specific Tsc1 knock-out mice over WT mice (Fig. 2D). Moreover, the BEX2 protein level was also increased in the neuronal specific Tsc1 knock-out mice, and the augmented BEX2 was suppressed by rapamycin administration (Fig. 2E). BEX2 expression declined after treatment with rapamycin in human liver cancer cell lines (HepG2, Bel-7402, SK-HEP-1), lung adenocarcinoma cell line A549, and rat uterine leiomyoma cell line ELT3 (Tsc2-deficient cells) (Fig. 2F). To validate mTORC1 in the regulation of BEX2, we knocked down mTOR, Raptor, and Rictor expression in Tsc2 Ϫ/Ϫ MEFs, respectively. Cells transfected with mTOR and Raptor siRNAs had a remarkable decrease in the expression of phospho-S6 and BEX2 (Fig. 2G). However, there was no dramatic change in both phospho-S6 and BEX2 expression after knockdown of Rictor, suggesting that BEX2 expression is controlled by mTORC1 instead of mTORC2. Taken together, these data show that mTORC1 is a positive regulator of BEX2 expression. Reduction of BEX2 Inhibits the Tumorigenic Capacity of Cells with Activated mTOR-To investigate the role of BEX2 in the tumorigenesis of cells with active mTOR, we evaluated the tumorigenicity of Tsc2 Ϫ/Ϫ MEFs expressing shBEX2 or scramble shRNA (shV) in a nude mouse model. Depletion of BEX2 significantly attenuated tumor initiation and progression of Tsc2 Ϫ/Ϫ MEFs in nude mice (Fig. 3A). Similarly, knockdown of BEX2 significantly compromised the tumorigenesis of Pten Ϫ/Ϫ MEFs in nude mice (Fig. 3B).
To explore whether STAT3 directly transactivates BEX2 gene transcription, we identified a putative STAT3 binding sequence (Ϫ569/Ϫ561; TTCCAGGAA) within the promoter of mouse BEX2 gene through the analysis of the 5Ј-flanking sequence of the BEX2 gene upstream of the transcription start site (Fig. 4F, left). ChIP analysis revealed that the binding of STAT3 to the putative binding site in the promoter of BEX2 gene was increased in Tsc2 Ϫ/Ϫ MEFs when compared with in WT cells. Moreover, the interaction between STAT3 and the BEX2 promoter was decreased after rapamycin treatment (Fig.  4F, right). Taken together, STAT3 directly promotes the transcription of BEX2 downstream of mTOR.
NF-B is part of a family of transcription factors, including RelA/p65, c-Rel, RelB, NF-B1/p50, and NF-B2/p52 (41), and is constitutively activated in multiple types of tumors (42). NF-B has been reported locating in the downstream of mTOR (43). To investigate whether BEX2 expression is controlled by NF-B, we transfected siRNA targeting p65 into Tsc2 Ϫ/Ϫ MEFs, Pten Ϫ/Ϫ MEFs, and SK-HEP-1 cells to reduce NF-B activity. Knockdown of p65 suppressed BEX2 expression in these cells (Fig. 4G). Moreover, inhibition of NF-B with the NF-B inhibitor, CAPE, resulted in a remarkable reduction of BEX2 protein level in Tsc2 Ϫ/Ϫ MEFs, Pten Ϫ/Ϫ MEFs, and A549 cells (Fig. 4H). However, the mRNA levels of BEX2 in Tsc2 Ϫ/Ϫ MEFs or Pten Ϫ/Ϫ MEFs were not significantly different in the absence or presence of CAPE (Fig. 4I). Hence, these results demonstrate that NF-B positively regulates BEX2 protein level. Our previous study suggests that STAT3 has no regulatory cross-talk with NF-B (25). In this study, inhibition of NF-B or STAT3 in Tsc2 Ϫ/Ϫ MEFs was unable to affect the activity of either NF-B or STAT3 (Fig. 4J). Therefore, STAT3 and NF-B stimulate BEX2 expression in parallel downstream of mTOR.
BEX2 Mediates mTOR Augmentation of VEGF Expression-Previous studies have shown that activation of PI3K-AKT-mTOR signaling pathway increases VEGF expression (16,17,44). Deficiency in either Tsc2 or Pten indeed led to up-regulation of VEGF, and inhibition of mTOR by rapamycin suppressed the expression of VEGF (Fig. 5, A and B). To determine whether there was a potential relationship between BEX2 and VEGF, we examined the abundance of VEGF in cell lysates and cell culture supernatants in Tsc2 Ϫ/Ϫ MEFs transfected with shRNA targeting BEX2. Suppression of BEX2 expression reduced VEGF expression (Fig. 5C) and secretion (Fig. 5D) in Tsc2-null MEFs. In addition, BEX2 overexpression increased the expression of VEGF in WT MEFs (Fig. 5E). To seek the relevance of this finding in human cancer cells, we knocked down BEX2 expression in human hepatocellular carcinoma SK-HEP-1 cells. Reduction of BEX2 suppressed VEGF expression in SK-HEP-1 cells (Fig. 5F). Moreover, we examined the effect of BEX2 on angiogenesis through chick chorioallantoic membrane assay. BEX2-overexpressing cells formed more blood vessels, with severe bleeding, than the control cells on day 7 (Fig. 5G). Thus, BEX2 mediates mTOR up-regulation of VEGF production and angiogenesis.
mTOR Regulates STAT3/NF-B-BEX2-VEGF Signaling in Vitro and in Vivo-It was reported that STAT3 and NF-B positively modulate VEGF expression (45,46). Likewise, inhibition of STAT3 or NF-B with JSI-124 or CAPE decreased expression of VEGF in SK-HEP-1, HepG2, and A549 cells, respectively (Fig. 6, A and B). We next investigated whether this newly identified mTOR regulation of STAT3/NF-B-BEX2-VEGF signaling network is present both in vitro and in vivo. Inhibition of mTOR by rapamycin down-regulated p-S6, p-STAT3, p-p65, BEX2, and VEGF in HepG2, SK-HEP-1, ELT3, and A549 cells (Fig. 6C). Moreover, there are concurrent increases in the protein levels of p-S6, p-STAT3, p-p65, BEX2, and VEGF in kidney cystadenomas of four Tsc2 ϩ/Ϫ mice when compared with that in paratumor tissues (Fig. 6D). Taken together, these data indicate that the mTOR-STAT3/NF-B-BEX2-VEGF signaling network exists in vitro and in vivo.

Discussion
RTK-PI3K/PTEN-AKT-TSC1/2-mTOR signaling pathway is frequently deregulated in cancer, but the underlying mechanisms remain less clear. In this study, we have elucidated that BEX2 is a novel downstream target of mTOR, and activation of mTOR promotes tumorigenesis through up-regulation of STAT3/NF-B-BEX2-VEGF signaling cassette.
As the significance of mTOR signaling in physiology and diseases has been increasingly appreciated, the molecular events downstream of mTOR are under intensive investigation. In addition, the underlying mechanism of mTOR up-regulation of angiogenesis is not completely understood. Evidence has shown that BEX2 is overexpressed in breast cancer (20), indicating that BEX2 may play a role in cancer development. Reduction of BEX2 expression inhibits angiogenesis in vivo (22). Our microarray analysis revealed that the abundance of BEX2 was ϳ8.5-fold higher in Tsc2 Ϫ/Ϫ MEFs than in WT MEFs (Table 1). Therefore, BEX2 was chosen for mechanistic study of tumorigenesis caused by aberrant activation of mTOR.
Because the abundance of BEX2 is higher in TSC-associated kidney angiomyolipoma, the enhanced expression of BEX2 may be involved in TSC development. We speculate that BEX2 is a candidate target for the treatment of TSC.
There is limited characterization on the regulatory mechanisms upstream of BEX2. Our data show that STAT3, a downstream effector of mTOR, is a positive regulator of BEX2, and BEX2 is thus a novel effector of STAT3. Moreover, STAT3 up-regulates BEX2 expression by directly binding to the promoter of BEX2 gene (Fig. 4, A-F). In addition, NF-B positively regulates BEX2 expression. However, NF-B does not influence the transcription of BEX2 gene (Fig. 4, G-I), and therefore NF-B regulation of BEX2 is likely a post-transcriptional event. Based on our previous study (25) and the finding in this study (Fig. 4J), there is no functional cross-talk between STAT3 and NF-B. We propose that STAT3 and NF-B positively modulate BEX2 expression downstream of mTOR in parallel.
The role of BEX2 in tumorigenesis is controversial. BEX2 is considered as a proto-oncogene in breast cancer (21) and a tumor suppressor in glioma (24). Our data demonstrate that reduction of BEX2 inhibits tumorigenesis of Tsc2 Ϫ/Ϫ or Pten Ϫ/Ϫ MEFs (Fig. 3), supporting BEX2 as an mTOR-regulated proto-oncogene. Thus, BEX2 is a promising target that may be harnessed for the treatment of cancers associated with aberrant mTOR signaling.
It is well known that hyperactive mTOR signaling leads to tumorigenesis with augmented angiogenesis, and VEGF has mTOR Up-regulation of BEX2 OCTOBER 16, 2015 • VOLUME 290 • NUMBER 42 been established as a major stimulator of angiogenesis (18,47).
Here we have demonstrated that BEX2, as a novel downstream target of mTOR, positively controls VEGF expression, and BEX2 overexpression in WT MEFs promotes angiogenesis in the chick chorioallantoic membrane (Fig. 5). STAT3 and NF-B augment BEX2 expression induced by mTOR (Fig. 4). Moreover, STAT3 and NF-B are positive upstream regulators of VEGF (Fig. 6, A and B). In addition, the mTOR-STAT3/NF-B-BEX2-VEGF signaling cascade was shown to exist in human cancer cell lines in vitro and renal tumors of Tsc2 ϩ/Ϫ mice in vivo (Fig. 6, C and D). Taken together, mTOR may promote tumorigenesis through enhanced STAT3/NF-B-BEX2-VEGF signaling cassette (Fig. 6E). As a target gene of c-Jun, BEX2 reciprocally regulates c-Jun in breast cancer (48). mTOR signaling pathway augments VEGF expression through up-regulation of HIF-1␣ (18,49). c-Jun positively regulates VEGF expression by stabilizing HIF-1␣ (50). Hence, we postulate BEX2 enhances VEGF through up-regulation of c-Jun-HIF-1␣ signaling cascade.
In summary, we have illustrated that BEX2 plays an important role in tumorigenesis caused by aberrant activation of mTOR. Hyperactivation of mTOR stimulates STAT3/NF-B-BEX2-VEGF signaling cascade. The components in the newly established mTOR-STAT3/NF-B-BEX2-VEGF cascade are potential targets for the treatment of cancer with aberrant RTK-PI3K/PTEN-AKT-TSC1/2-mTOR signaling.
Author Contributions-Z. H. designed and carried out most of the experiments, analyzed data, and wrote most of the paper. Y. W. performed experiments, analyzed data, and revised the paper. F. H. performed experiments and analyzed data. R. C. provided human kidney tumor tissue samples. C. L. provided ELT-3 cells stably expressing TSC2. F. W. performed the gene expression profiling analysis. J. G. and D. J. K. provided brain tissues of neuronal specific Tsc1 knockout and WT control mice. J. L. and J. W. B. performed chick chorioallantoic membrane assay for angiogenesis. P. T. supervised some of the study. H. Z. and X. Z. designed experiments, analyzed data, and wrote the paper.