Epstein-Barr Virus-Encoded Latent Membrane Protein 1 Upregulates Glucose Transporter 1 Transcription via the mTORC1/NF-κB Signaling Pathways

ABSTRACT

Accumulating evidence indicates that oncogenic viral protein plays a crucial role in activating aerobic glycolysis during tumorigenesis, but the underlying mechanisms are largely undefined. Epstein-Barr virus (EBV)-encoded latent membrane protein 1 (LMP1) is a transmembrane protein with potent cell signaling properties and has tumorigenic transformation property. Activation of NF-κB is a major signaling pathway mediating many downstream transformation properties of LMP1. Here we report that activation of mTORC1 by LMP1 is a key modulator for activation of NF-κB signaling to mediate aerobic glycolysis. NF-κB activation is involved in the LMP1-induced upregulation of glucose transporter 1 (Glut-1) transcription and growth of nasopharyngeal carcinoma (NPC) cells. Blocking the activity of mTORC1 signaling effectively suppressed LMP1-induced NF-κB activation and Glut-1 transcription. Interfering NF-κB signaling had no effect on mTORC1 activity but effectively altered Glut-1 transcription. Luciferase promoter assay of Glut-1 also confirmed that the Glut-1 gene is a direct target gene of NF-κB signaling. Furthermore, we demonstrated that C-terminal activating region 2 (CTAR2) of LMP1 is the key domain involved in mTORC1 activation, mainly through IKKβ-mediated phosphorylation of TSC2 at Ser⁹³⁹. Depletion of Glut-1 effectively led to suppression of aerobic glycolysis, inhibition of cell proliferation, colony formation, and attenuation of tumorigenic growth property of LMP1-expressing nasopharyngeal epithelial (NPE) cells. These findings suggest that targeting the signaling axis of mTORC1/NF-κB/Glut-1 represents a novel therapeutic target against NPC.

IMPORTANCE Aerobic glycolysis is one of the hallmarks of cancer, including NPC. Recent studies suggest a role for LMP1 in mediating aerobic glycolysis. LMP1 expression is common in NPC. The delineation of essential signaling pathways induced by LMP1 in aerobic glycolysis contributes to the understanding of NPC pathogenesis. This study provides evidence that LMP1 upregulates Glut-1 transcription to control aerobic glycolysis and tumorigenic growth of NPC cells through mTORC1/NF-κB signaling. Our results reveal novel therapeutic targets against the mTORC1/NF-κB/Glut-1 signaling axis in the treatment of EBV-infected NPC.

INTRODUCTION

Various oncogenic viral proteins exert their carcinogenic properties by interfering with the host cellular signaling pathways associated with energy metabolism and cell growth. Elucidation of how these viral proteins modulate these cellular signaling pathways will contribute to our understanding of the mechanism of viral oncogenesis (1).

Nasopharyngeal carcinoma (NPC) is a special type of head and neck cancer commonly seen in southern China and regional countries, including Malaysia and Indonesia, as well as specific regions in North Africa (2). The prognosis of NPC patients is good if the disease is diagnosed at early stages and can be effectively treated by combined chemotherapy and radiotherapy. However, the long-term survival rate of NPC patients is still poor if the cancer is detected at advanced stages (2, 3). Epstein-Barr virus (EBV) infection in NPC is an early event and has been postulated to be an important step involved in NPC development (4). The EBV is a human gammaherpesvirus closely associated with both lymphoma (5) and epithelial malignancies, which include NPC (6) and gastric cancer (7). In NPC, EBV infection is predominantly latent in nature and the viral genes expressed include EBV nuclear antigen 1 (EBNA1), latent membrane proteins (LMP1, LMP2A, and LMP2B), EBV-encoded small RNAs (EBER1 and EBER2), and the BART microRNAs (6). LMP1 is well documented as a viral oncoprotein and has been postulated to play an essential role in NPC pathogenesis, particularly at its early stage of development (8). LMP1 is a potent activator of NF-κB signaling (9, 10). The structure of LMP1 contains a short N-terminal tail, a six-pass transmembrane domain, and two activation regions located at the cytoplasmic carboxyl tail, namely, C-terminal activating regions 1 and 2 (CTAR1 and CTAR2). Both CTAR1 and CTAR2 are involved in activation of NF-κB signaling (9, 11), but their roles in aerobic glycolysis are not well defined.

Aerobic glycolysis, also commonly referred as the Warburg effect, is frequently observed in tumor cells, even with a sufficient oxygen supply (12). The high demand of glucose metabolism for the growth of cancer cells requires accelerated glucose uptake into cells. The members in the family of glucose transporters (Glut-1 to -4), which transport extracellular glucose across the plasma membranes into cells, have attracted immense interest in their roles in cancer metabolism (13). Under normal physiological conditions, the expression level of Glut-1 is very low. High levels of Glut-1 were, however, observed in multiple types of tumors and have been widely used as an immunohistochemical marker to identify and evaluate the status of tumor progression in patients (13).

The EBV-encoded LMP1 is a potent activator of NF-κB. The role of LMP1 activation of NF-κB signaling in regulating activity of Glut-1 has been reported for B cell lymphomas (14). Its involvement in nasopharyngeal carcinoma is unclear. NF-κB signaling can be activated by both canonical and noncanonical pathways. The canonical activation of NF-κB signaling is dependent on I-κBα phosphorylation by IκB kinase β (IKKβ) and subsequent degradation by proteasome, resulting in emancipation and nuclear translocation of the p65/p50 dimer complex to activate transcription of NF-κB target genes (15). The noncanonical pathway of NF-κB involves the phosphorylation activation of IKKα, which mediates processing of p100 by NIK, and generation of the p52/RelB dimer, which is translocated to the nucleus to activate transcription of NF-κB target genes (15). Hence, both canonical signaling and noncanonical signaling of NF-κB require activation of IKK (15, 16). However, recent studies showed that besides being involved in NF-κB activation, the IKK components also contribute directly to activation of mTORC1 signaling and are involved in tumor angiogenesis, which is crucial for cancer progression (17, 18).

Mammalian target of rapamacin (mTOR) signaling is an evolutionarily conserved pathway of serine/threonine protein kinase. The mTOR signaling complex contains two different complexes, mTORC1 and mTORC2, based on the differences of their functional subunits (19). The mTORC1 signaling pathway is well investigated because of its important role in regulating energy metabolism and cell growth. The specific downstream effectors of mTORC1, p70s6K and 4E-BP1, control protein translation and ribosome biogenesis. The mTORC1 signaling pathway is commonly activated by phosphatidylinositol 3-kinase (PI3K)/AKT and other growth signaling events through different pathways, including direct phosphorylation of mTOR (20), phosphorylation of the mTOR substrate, Raptor (18), and phosphorylation of its negative regulator, the tuberous sclerosis (TSC) complex (21). The role of mTORC1 in aerobic glycolysis is well documented (22). The involvement of LMP1 and mTORC1 in regulating glucose metabolism in nasopharyngeal carcinoma is unclear. A recent study reported that LMP1 could activate mTORC1 signaling and that PI3K/AKT signaling was involved (23). Activation of mTORC1 correlated with LMP1 expression in NPC specimens. However, the details regarding the mechanisms involved in LMP1 activation of mTORC1, particularly in the regulation of glucose metabolism in NPC cells, remain to be defined. In this study, we have examined the key molecular events involved in LMP1-mediated mTORC1 activation and observed a novel property of LMP1 to upregulate Glut-1 transcription in immortalized nasopharyngeal epithelial cells, which is mediated by mTORC1/NF-κB signaling. While LMP1 could activate mTORC1 signaling through the PI3K/AKT pathway, we further delineated that the LMP1-CTAR2 plays an important role in mTORC1 activation through IKKβ to mediate phosphorylation of the Ser⁹³⁹ residue of TSC2. Our results showed that mTORC1/NF-κB signaling by LMP1 plays an essential role in modulating glucose metabolism in nasopharyngeal epithelial cells and may represent a novel therapeutic target for NPC treatment.

RESULTS

LMP1 induces activation of mTORC1 signaling.

To investigate the relationship between LMP1 and mTORC1 signaling, we first examined mTORC1 activation in HONE1 cells stably expressing LMP1 by retroviral transduction (Fig. 1A). Activation of mTORC1 was clearly detected in HONE1 cells stably expressing LMP1 as indicated by the increased phosphorylation of mTOR at Ser²⁴⁴⁸ and the phosphorylation of downstream targets of activated mTORC1, p-p70s6K (Thr³⁸⁹) and p-4E-BP1 (Thr⁷⁰). We next examined the effect of transient expression of LMP1 to activate mTORC1 by transfecting LMP1-expressing plasmid into two different cell lines, HONE1 and 293T (Fig. 1B). Transient expression of LMP1 induced phosphorylation of mTOR (Ser²⁴⁴⁸) and mTORC1 downstream targets p70s6K (Thr³⁸⁹) and 4EBP1 (Thr⁷⁰) in a dose-dependent manner. Hence, both stable and transient expression of LMP1 effectively activated mTORC1 signaling in cells. In order to confirm the relevance of activation of mTORC1 by LMP1 in EBV-infected epithelial cells, we examined if knockdown of LMP1 may affect mTORC1 signaling in EBV-infected cells. For this part of the study, we used HONE1 cells stably infected with M81-EBV. M81-EBV was recently isolated from NPC origin and shown to have distinct biological properties (24). Stable infection with M81-EBV was established in our laboratory, and expression of LMP1 and mTORC1 activation was detected (Fig. 1C). Knockdown of LMP1 expression in HONE1-M81 cells by shLMP1 was accompanied by a decrease in mTORC1 signaling. This confirmed the involvement of LMP1 expression in induction of mTORC1 signaling in EBV-infected NPC cells.

FIG 1.

LMP1 induces activation of the mTORC1 signaling pathway. (A) HONE1-pLpcx and HONE1-LMP1 cells were lysed and analyzed by Western blotting using antibodies against LMP1 and various proteins in involved in mTORC1 signaling. (B) Various doses of 2117-LMP1 were transfected into HONE1 and 293T cells for 36 h and then examined by Western blotting to analyze for activation of the mTORC1 signaling pathway with specific antibodies. β-Actin expression was used as the loading control. (C) EBV-infected NPC cell line HONE1-M81 was infected with sh-LMP1 and control empty retroviral vector. The cellular RNA was extracted for qPCR analysis, and cell lysates were examined by Western blotting using antibodies against proteins involved in LMP1 and mTORC1 signaling pathways. Data are means ± SDs of triplicate measurements. **, P < 0.01.

LMP1 induces Glut-1 expression and aerobic glycolysis.

One of the important functions for mTORC1 is regulation of energy metabolism in cancer cells. Recent studies have shown that LMP1 promotes aerobic glycolysis in NPC cells (25, 26). Aerobic glycolysis is closely correlated with a higher rate of glucose uptake, which is regulated by glucose transporters (Gluts). We hypothesize that LMP1 may regulate expression of Gluts to increase glucose uptake for aerobic glycolysis. To confirm our hypothesis, we examined the transcription levels of Glut-1, -2, -3, and -4 in control and LMP1-transfected HONE1 cells and NP69 cells (an immortalized nasopharyngeal epithelial cell line) (Fig. 2A). Interestingly, the Glut-1 mRNA levels showed the most substantial increase in LMP1-transfected HONE1 and NP69 cells. Glut-2 expression was not detected in HONE1 or NP69 cells transfected with LMP1. Induction of Glut-3 and -4 was slightly elevated after LMP1 expression but at a much lower degree than that of Glut-1. As the induction of Glut-1 transcription by LMP1 was the most significant change among the 4 isoforms of Glut examined, we further examined the protein levels of Glut-1 by Western blotting in HONE1 and three immortalized nasopharyngeal epithelial (NPE) cell lines (Fig. 2B). The Glut-1 mRNA levels were also upregulated in the two immortalized NPE cells, NP460hTert and NP550hTert (Fig. 2C). These results clearly indicated that LMP1 expression upregulates Glut-1 expression at both protein and mRNA levels. We further examined if LMP1 expression was also associated with a higher rate of aerobic glycolysis in these cells by examining glucose consumption and lactate production. As shown in Fig. 2D and E, the LMP1-expressing cells had higher glucose consumption and produced more lactate than control cells. To explore the pathological relevance of LMP1 and Glut-1 expression in NPC, we assessed the expression of these proteins in NPC specimens. Figure 2F shows immunohistochemical staining for Glut-1 and LMP1 in NPC specimens. In the small number of NPC specimens examined (n = 16), an association of immunoreactivity scores of LMP1 and Glut-1 expression was observed (Fig. 2G). The expression patterns of LMP1 and Glut-1 are shown in two representative cases (Fig. 2F). High expression of Glut-1 and LMP1 could be observed at the membranes of NPC cells (case 2 in Fig. 2F). A more extensive study is warrant to further confirm the correlation of Glut-1 and LMP1 expression in NPC.

FIG 2.

LMP1 induces the expression of Glut-1 and increases glucose uptake. (A) NP69 and HONE1 cells were transfected with pcDNA and 2117-LMP1 expression plasmid. RNA was extracted 36 h later for RT-PCR analysis for Glut-1 to -4 gene transcription. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (B) Cells were lysed and the cell lysates were analyzed by Western blotting for LMP1 and Glut-1 expression using specific antibodies. β-Actin expression was used as the loading control. (C) RNA from NP460 and NP550 cells stably expressing LMP1 were extracted and analyzed by qPCR with Glut-1 primer. (D and E) The glucose consumption (D) and lactate production (E) of LMP1-transfected and control cells were determined. (F) Immunohistochemical staining revealed LMP1 and Glut-1 in formalin-fixed, paraffin-embedded NPC tissue sections. The images are of two representative NPCs. Images were acquired at a magnification of ×400. (G) Dot blot graph showing the immunoactivity scores of Glut-1 staining in NPC tumors with and without LMP1 expression. The median values of each group are shown by horizontal lines. Data are means ± SDs. *, P < 0.05; **, P < 0.01; ***, P < 0.005.

LMP1-induced NF-κB signaling is dependent on mTORC1 activation.

It is well documented that LMP1 activates NF-κB signaling to mediate multiple malignant phenotypes to facilitate tumorigenesis. Cross talk of NF-κB and mTORC1 signaling pathways has been reported (18, 27). Since LMP1 activates both NF-κB and mTORC1 signaling pathways, we hypothesize that there are functional interactions between these two signaling pathways. We first inhibited and activated canonical NF-κB signaling of LMP1, respectively, by knocking down the expression of NF-κB subunit p65 and I-κBα (inhibitor of canonical activation of NF-κB) by lentiviral shRNA expression vector, and examined for their impact on mTORC1 signaling (Fig. 3A). Neither inhibition nor activation of canonical activation of NF-κB had significant impact on the ability of LMP1 to phosphorylate mTOR and downstream substrates of activated mTORC1. In contrast, the specific inhibitor of mTORC1, rapamycin, as well as short hairpin RNA (shRNA) knockdown of Raptor (the functional unit of mTORC1) efficiently inhibited mTORC1 activation and abolished LMP1-induced phosphorylation of I-κBα (Fig. 3B and C). Furthermore, rapamycin also inhibited nuclear accumulation of the p65 subunit of NF-κB in LMP1-expressing HONE1 cells but had no significant effect on the levels of p65 subunit in the cytoplasmic compartments (Fig. 3D). These results suggest that mTORC1 activation by LMP1 is upstream of I-κBα phosphorylation in activation of canonical NF-κB signaling. Similarly, immunofluorescence staining also showed inhibition of nuclear accumulation of p65 subunit after rapamycin and shRaptor treatment in LMP1-expressing HONE1 cells (Fig. 3E). Finally, suppression of NF-κB signaling by rapamycin was also confirmed by promoter reporter assay for NF-κB activation using the 3κB luciferase reporter assay (Fig. 3F). The suppression and activation of NF-κB, respectively, in HONE1-LMP1 cells by transfection of shp65 and shI-κBα were also confirmed by the 3κB luciferase reporter assay for NF-κB activation (Fig. 3F). Taken together, these results support a regulatory role of mTORC1 in LMP1 activation of NF-κB signaling.

FIG 3.

LMP1-induced NF-κB signaling activation is dependent on mTORC1. (A) HONE1-LMP1 cells were infected with sh-pScramble, sh-p65, and sh-IκBα lentivirus. Forty-eight hours after infection, cells were harvested and the cell lysates were analyzed by Western blotting for expression of LMP1 and mTORC1 pathway proteins with their corresponding antibodies. (B and C) HONE-LMP1 cells were treated with rapamycin for 24 h (B) and shRaptor for 48 h (C) and analyzed by Western blotting for expression of LMP1 and mTORC1 proteins. β-Actin expression was used as the loading control. (D) Cells were treated with 100 nM rapamycin for 24 h and analyzed by Western blotting using specific antibodies. (E) Cells were fixed and stained with p65-specific antibody and observed under a fluorescence microscope. (F) Relative luciferase reporter activity for NF-κB activation in LMP1-expressing cells was evaluated after various treatments. Data are means ± SDs of triplicate measurements. *, P < 0.05; **, P < 0.01; ***, P < 0.005.

LMP1 upregulates Glut-1 transcription through activation of mTORC1/NF-κB signaling.

LMP1 was shown to upregulate the levels of both mRNA and protein of Glut-1 (Fig. 2), which is a novel property of LMP1 not previously reported. We examined if activation of Glut-1 by LMP1 is dependent on mTORC1 and NF-κB activation. Again, in LMP1-expressing HONE1 cells, expression of shRaptor and shp65 but not shI-κBα suppressed both protein levels and mRNA levels of Glut-1 (Fig. 4A, top and bottom, respectively). Furthermore, we also examined the effects of inhibition of PI3K/AKT, NF-κB, and mTORC1 on Glut-1 expression in HONE1-LMP1 cells. Treatments with LY294002, the NEMO-binding domain (NBD), and rapamycin all suppressed LMP1-induced upregulation of Glut-1 at both protein and mRNA levels (data not shown). We further examined the involvement of mTORC1/NF-κB signaling in glucose consumption and lactate production in LMP1-expressing cells. In LMP1-expressing cells, knocking down mTORC1 or suppressing NF-κB inhibited glucose uptake and lactate production, while activation of NF-κB accelerated these events (Fig. 4B). We also examined if the promoter activity of Glut-1 was affected by mTORC1 and NF-κB signaling in LMP1-expressing cells (Fig. 4C, top). Again, inhibition of either mTORC1 or NF-κB activity by specific chemical or genetic inhibitors effectively suppressed the LMP1-induced Glut-1 promoter activity. Furthermore, constitutive activation of NF-κB signaling by expression of wild-type NF-κB subunit p65 or its phosphomimetic mutant, p65S536E, enhanced the ability of LMP1 to activate Glut-1 promoter activity (Fig. 4C, bottom). All these results indicate that mTORC1 activation and NF-κB activation are closely involved in mediating LMP1-induced Glut-1 expression.

FIG 4.

LMP1 upregulates Glut-1 transcription through the activation of mTORC1-mediated NF-κB signaling. (A) HONE1 cells stably expressing LMP1 were infected with different lentiviruses expressing shRNA against raptor, p65, and I-κB and their control scramble shRNA vector for 48 h. The cellular protein was subjected to Western blot analysis and the RNA to qPCR analysis. (B) Glucose consumption and lactate production in LMP1-expressing HONE1 cells after various treatments were determined. (C) Relative luciferase reporter activity of NF-κB was evaluated after various treatments. (D) Luciferase reporter vectors used to measure Glut-1 promoter activity. 293T cells were cotransfected with LMP1 and different reporter plasmids and the luciferase activity was determined 48 h later. (E) Motif analysis of p65 binding sites in selected Glut-1 promoter sequences. Multiple-sequence alignment among different species revealed a consensus sequence for NF-κB binding. (F) 293T cells were cotransfected with LMP1 plus pGL3-Glut1 (#1) or pGL3-Glut3-MUT (#4) reporter plasmid. The luciferase activity was measured after 48 h. (G) HONE1 cells were transfected with pcDNA and LMP1 expression plasmids. The cell lysates were subjected to ChIP assay using an anti-p65 antibody. Normal rabbit IgG antibody served as the negative control. PCR was performed to amplify a region surrounding the putative NF-κB binding region and a nonspecific NF-κB binding region. Data are means ± SDs of triplicate measurements. *, P < 0.05; **, P < 0.01; ***, P < 0.005.

The p65 is a well-characterized subunit of NF-κB which is translocated to nucleus upon NF-κB activation to activate target gene transcription. We examined whether Glut-1 is a direct target gene of p65. Different Glut-1 luciferase reporter constructs were generated as shown in Fig. 4D, cotransfected with LMP1 expression plasmid into 293T cells, and examined for their respective luciferase reporter activities. The Glut-1 promoter reporter assay revealed higher luciferase activity in construct 1 and construct 2, containing 2,000-bp and 1,000-bp sequences upstream of the 5′ transcription start site (TSS), respectively, than in construct 3, which contains only the 500-bp sequence upstream of the 5′ TSS. These results indicated that the putative binding site of NF-κB may be localized between −1,000 and −500 bp upstream of the 5′ TSS of the Glut-1 promoter (Fig. 4D). We further scanned this region using prediction software, JASPAR, and found a conserved putative NF-κB binding site at −903 to −894 bp upstream of the 5′ TSS of the Glut-1 (Fig. 4E). To further confirm this binding activity of this region, we mutated residue −901 (G to A) and residue −900 (A to G) in this region and used the product as construct 4 (Fig. 4E) in the cotransfection assay with LMP1 expression plasmid. Indeed, the Glut-1 promoter activity was clearly attenuated in construct 4-transfected cells with the putative NF-κB-binding site mutated (Fig. 4F). To further confirm the direct binding of p65 to this consensus sequence of the Glut-1 promoter, we conducted a chromatin immunoprecipitation (ChIP) assay using antibody to put down p65-binding elements. The ChIP assay revealed a significantly increased binding of p65 to the putative NF-κB-binding region of the Glut-1 promoter in LMP1-expressing cells compared with the control cells. No significant binding of p65 to the nonspecific binding region (−1748 to −1615) of the Glut-1 promoter was detected (Fig. 4G). Taken together, these results support the notion that Glut-1 is a direct target gene of NF-κB activation.

Upregulation of Glut-1 mediates malignant properties in LMP1-expressing cells.

Increased glucose uptake and metabolism are considered to have proliferative advantages for cancer cells with altered energetic metabolism which require higher consumption of glucose for energy and intermediate metabolites. Our results showed that LMP1 increased aerobic glycolysis in cells, which was accompanied by activated mTORC1/NF-κB signaling and Glut-1 expression. We further examined if upregulation of Glut-1 by mTORC1/NF-κB activation is involved in the increased aerobic glycolysis and enhanced malignancy in LMP1-expressing NP69 cells. We used a specific chemical inhibitor, STF-31, to block the enzymatic activity of nicotinamide phosphoribosyltransferase (NAMPT), one of the key enzymes for glucose metabolism (28, 29), as well as shGlut-1 expression to abolish Glut-1 transcription and examined their effects on aerobic glycolysis. STF-31 showed no effect on Glut-1 protein expression, while shGlut-1 abolished Glut-1 expression (Fig. 5A). However, treatment with the metabolic inhibitor STF-31 and knockdown of Glut-1 mRNA expression by shRNA in LMP1-expressing cells led to a marked increase of apoptosis as indicated by the activity of caspase 3 (Fig. 5A), reduction of proliferation as indicated by thymidine incorporation (Fig. 5B), decrease of glucose consumption (Fig. 5C), and lactate production (Fig. 5D). Taken together, these data indicate that Glut-1 plays essential roles in cell growth and aerobic glycolysis of LMP1-expressing cells.

FIG 5.

Upregulation of Glut-1 mediates LMP1-induced cell proliferation and tumorigenesis. (A) NP69 cells were transfected with LMP1 expression plasmid and treated with STF-31 or infected with lentivirus expressing shGlut-1 for 48 h. The cells were then lysed and subjected to Western blotting using specific antibodies against Glut-1 and cleaved caspase 3. β-Actin expression was used as the loading control. (B) Cells were treated with STF-31 or infected with lentivirus expressing shGlut-1 for different time points and then assayed for [³H]thymidine uptake. (C and D) Glucose consumption (C) and lactate production (D) of NP69 cells after the indicated treatments were determined. (E and F) Representative results of colony formation (E) and anchorage-independent growth in soft agar (F) after various treatments. Histograms indicate the number of colonies formed and degree of anchorage-independent cell growth. Data are means ± SDs of triplicate measurements. *, P < 0.05; **, P < 0.01; ***, P < 0.005.

The activation of Glut-1 expression is also involved in modulating colony formation assay (Fig. 5E) and anchorage-independent growth of LMP1-expressing cells (Fig. 5F) in LMP1-expressing cells after treatment with STF-31 and knocking down Glut-1 mRNA expression. The immortalized NP69 cells were again used in this part of the study. Its susceptibility to transformation actions of LMP1 has been well documented in our earlier studies (8, 30). Treatment of NP69 either with STF-31 or shGlut-1 suppressed the ability of LMP1 to induce colony formation and anchorage-independent growth in soft agar. Similar results were also obtained if the mTORC1/NF-κB signaling axis was compromised (data not shown). Taken together, these observations demonstrate that Glut-1 expression, mediated by mTORC1/NF-κB, is involved in LMP1-induced growth, colony formation, and anchorage independence of immortalized NPE cells, which facilitate their tumorigenic properties.

LMP1-induced mTORC1 activation via the AKT/ERK/IKK signaling axis.

As mentioned above, the underlying signaling pathways of LMP1-induced mTORC1 are largely undefined. The PI3K/AKT is a well-documented upstream site of mTORC1 signaling which is also involved in LMP1 activation of mTORC1 (23). Previous studies have shown that LMP1 is capable of activating multiple intracellular signaling pathways, including the PI3K/AKT, mitogen-activated protein kinase kinase (MEK)/extracellular signal-regulated kinase (ERK), and IKK/NF-κB signaling pathways (11). However, the role of IKK in LMP1-induced mTORC1 signaling is undefined. We first examined the involvement of these LMP1-signaling pathways in mediating LMP1-activated mTORC1 signaling. As shown in Fig. 6A, transient expression of LMP1 readily activated AKT, ERK, and IKK in 293T and HONE1 cells. We also observed that U0126 and LY294002 could effectively inhibit LMP1-induced activation of IKK complex (data not shown). U0126 effectively inhibited ERK activation but not AKT activation in LMP1-expressing cells, whereas LY294002 inhibited both ERK and AKT activation (data not shown). This would imply that PI3K/AKT signaling acts upstream of MEK/ERK signaling to activate IKK in LMP1-expressing cells.

FIG 6.

LMP1-induced mTORC1 signaling activation via the AKT/ERK/IKK signaling axis. (A) HONE1 and 293T cells were transiently transfected with LMP1 and examined after 36 h. The cells were lysed and subjected to Western blot analyses for activation of multiple signaling pathways as indicated. (B) HONE1 cells stably expressing LMP1 (HONE1-LMP1) were treated with U0126, LY294002, and rapamycin for 24 h. After that, an [³H]thymidine uptake assay was performed to estimate cell proliferation. (C) HONE1 cells were transfected with LMP1 expression vector and treated with U0126, NBD, and rapamycin for 48 h. After treatment, the cells were collected, lysed, and subjected to Western blot analyses for expression of LMP1 and mTORC1 pathway-related proteins. β-Actin expression was used as the loading control. (D) HONE1 cells transiently transfected with LMP1 cells either treated with rapamycin or infected with lentivirus expressing shRNA against Raptor were examined after 48 h. The cells were lysed and examined by Western blotting for IKK proteins using specific antibodies. β-Actin expression was used as the loading control.

HONE1 cells stably expressing LMP1 exhibited a higher proliferative rate, as indicated by the increased thymidine uptake compared to that of control HONE1 cells (Fig. 6B). Rapamycin, a specific inhibitor of mTORC1, suppressed thymidine uptake in HONE1-LMP1 cells induced by LMP1, suggesting the involvement of mTORC1 activation to support growth of these LMP1-expressing HONE1 cells (Fig. 6B). Similarly, U0216 and LY294002 were also shown to suppress the increased thymidine uptake in HONE1-LMP1 cells, suggesting that MEK/ERK and PI3K/AKT signaling pathways are also involved in the growth of HONE1-LMP1 cells either directly or via mTORC1 activation. Transient expression of LMP1 in HONE1 cells and treatment with U0126 effectively inhibited LMP1-induced mTORC1 activation (Fig. 6C). Similarly, NBD and rapamycin, which are specific inhibitors of IKK complex and mTORC1, respectively, also inhibited mTORC1 activation in LMP1-expressing cells (Fig. 6C). Interestingly, inhibition of mTORC1 signaling by rapamycin treatment or knockdown of Raptor, the key component of mTORC1, suppressed LMP1-induced IKKβ activation, suggesting a role for mTORC1 in mediating LMP1 activation of the IKK complex and NF-κB signaling through undefined pathways (Fig. 6D). All these findings suggest that the AKT/ERK/IKK signaling axis is involved in LMP1-induced mTORC1 and IKKβ activation to mediate LMP1-induced NF-κB signaling.

CTARs of LMP1 are involved in LMP1-induced mTORC1 activation.

We next determined the key functional domain(s) of LMP1 involved in mTORC1 activation using different LMP1 mutants defective in key signaling domains, C-terminal activating region 1 (CTAR1) and CTAR2, of LMP1 (Fig. 7A). Wild-type LMP1 and LMP1 mutants with defective CTAR1 (LMP1-3A), CTAR2 (LMP1-8C), and combined defective CTAR1 and CTAR2 (LMP1-3A/8C) domains were used. The properties of these mutants in defining signaling activities of LMP1 have been previously described (31). We first examined the ability of these wild-type and mutant LMP1 plasmids in stimulation of [³H]thymidine uptake into HONE1 cells, which reflects their impact on cell proliferation. As shown in Fig. 7B, wild-type LMP1 could stimulate thymidine uptake upon transfection into HONE1 cells. The ability to stimulate thymidine uptake was diminished in all the LMP1 mutants examined, particularly in mutant LMP1-8C, which harbors a defective CTAR2 domain, and LMP1-3A/8C, which is defective in both CTAR1 and CTAR2 domains. Hence, the CTAR2 region in the LMP1 is more dominant in stimulating cellular incorporation of [³H]thymidine. Interestingly, in HONE1 and 293T cells, mutation in CTAR2 also diminished the ability of LMP1 to activate mTORC1 more than mutation in CTAR1 (Fig. 7C). LMP1 with defective CTAR1 and CTAR2 lost most of the ability to activate mTORC1. Taken together, these results established that the CTAR2 domain of LMP1 is the key domain of LMP1 involved in mTORC1 activation.

FIG 7.

C-terminal activating regions (CTARs) of LMP1 are involved in LMP1-induced mTORC1 activation. (A) Schematic illustrations of wild-type LMP1 and its mutants used for thymidine incorporation, mTORC1 activation, and glucose metabolism. WT, wild-type LMP1; 3A, LMP1 harboring CTAR1 mutation; 8C, LMP1 harboring CTAR2 deletion; 3A/8C, LMP1 harboring CTAR1 mutation and deletion of CTAR2. (B) HONE1 cells were transfected with different LMP1 constructs and examined for [³H]thymidine uptake, which reflects cell proliferation. (C) HONE1 and 293T cells were transfected with different types of LMP1 constructs. The cells were lysed and analyzed by Western blotting for LMP1 and mTORC1 proteins. (D) HONE1 cells were transfected with different types of LMP1 constructs and then analyzed for activation of the AKT, MEK/ERK, and IKK pathways using specific antibodies. β-Actin was used as the loading control. (E to H) HONE1 cells were transfected with different LMP1 constructs and analyzed for protein expression of Glut-1(E), RNA levels of Glut-1 (F), glucose consumption (G), and lactate production (H) after transfection with wild-type and mutant LMP1s. Data are means ± SDs of triplicate measurements. *, P < 0.05; **, P < 0.01; ***, P < 0.005.

As the AKT/ERK/IKK signaling pathways are upstream events of LMP1 activation of mTORC1, the abilities of CTAR1 and/or CTAR2 mutants to act on these upstream pathways of mTORC1 were also examined (Fig. 7D). Our transfection experiments showed that while both CTAR1 and CTAR2 are involved in the activation of AKT, MEK, ERK, and IKK, there are differences in their abilities to activate these signaling pathways. CTAR1 (as revealed by the LMP1-3A mutant) plays a more dominant role over CTAR2 (as revealed by the LMP1-8C mutant) in activation of AKT, MEK, and ERK. Differential abilities of CTAR1 and CTAR2 to activate IKK were observed. While both CTAR1 and CTAR2 are involved in the activation of IKK, phosphorylation of IKKα was more affected by CTAR1 mutation, while phosphorylation of IKKβ was more affected by CTAR2 mutation (Fig. 7D). The ability to activate canonical activation of NF-κB, as indicated by the phosphorylation of I-κB in Fig. 7D, was also more affected by mutation in CTAR2, which confirms previous reports of the dominant role of CTAR2 to activate NF-κB (31). With respect to Glut-1 expression, CTAR2 of LMP1 was also shown to be more dominant in activation of protein expression and mRNA transcription of Glut-1 (Fig. 7E and F); a similar pattern was observed with mTORC1 activation. Furthermore, glucose consumption and lactate production were also more significantly affected by mutation in CTAR2 than in CTAR1 (Fig. 7G and H). Taken together, these results showed that both CTAR1 and -2 of LMP1 are involved in activation of mTORC1 and transcription of Glut-1. The CTAR2 signaling domain plays a more dominant role over CTAR1 in mTORC1 and Glut-1 activation.

IKKβ is the main contributor to LMP1-induced mTORC1 activation.

To investigate if LMP1 activation of mTORC1 is indeed dependent on the IKK components, expression plasmids of Flag-tagged IKKα and IKKβ carrying a kinase mutation (KM) were used as dominant negative mutants to block the kinase activity of individual IKK components. The IKK mutants were transfected either singly or together with the LMP1 expression plasmid into HONE1 cells and examined for mTORC1 activation as visualized by levels of p-mTOR, p-p70s6K, and p-4EBP1. As shown in Fig. 8A, both IKKα-KM and IKKβ-KM plasmids suppressed LMP1-induced NF-κB activation as indicated by the decreased levels of p-I-κBα. IKKβ-KM was more effective in suppressing phosphorylation of I-κBα. Expression of both IKKα-KM and IKKβ-KM mutants completely abolished the mTORC1 signaling induced by LMP1, which further confirms that LMP1 activation of mTORC1 is dependent on IKK activation. Furthermore, the IKKβ-KM mutant was more effective in inhibition of mTORC1 signaling than the IKKα-KM mutant, suggesting a more dominant role of IKKβ over IKKα in mediating LMP1-induced mTORC1 activation. To further confirm the role of IKKβ in inhibition of mTORC1 activation, we used an IKKβ-specific inhibitor, PS1145, to treat HONE1-LMP1 cells. As shown in Fig. 8B, PS1145 could effectively inhibit phosphorylation of IKKβ but not IKKα. We also observed that both mTORC1 signaling and NF-κB pathway activity were reduced after IKKβ inhibitor treatment. Overexpression of wild-type IKKs, particularly IKKβ, in HONE1-LMP1 cells increased expression of Glut-1 at both protein and mRNA levels, while expression of KM mutants of either IKKα or IKKβ abolished the LMP1-induced upregulation of Glut-1 expression (Fig. 8C and D). Similar patterns for glucose metabolism were observed in HONE1-LMP1 cells when the wild-type and mutant IKKs were expressed with IKKβ, and IKKβ inhibitor showed a more dominant role over IKKα in mediating glucose consumption and lactate production (Fig. 8E to H).

FIG 8.

IKKβ is the main contributor to LMP1-induced mTORC1 activation. (A) HONE1 cells were cotransfected with LMP1 and Flag-IKKα and -β KM mutants for 36 h and then analyzed by Western blotting for activation of the NF-κB and mTORC1 pathways. β-Actin expression was used as the loading control. (B) HONE1-LMP1 cells were treated with PS1145 for 24 h and then analyzed by Western blotting for detection of the mTORC1 and NF-κB pathways. (C and D) HONE1 cells were transfected with different IKK mutant constructs and then examined for expression of Glut-1 protein by Western blotting (C) and Glut-1 mRNA transcription by qPCR (D). (E to H) Glucose consumption (E and G) and lactate production (F and H) were measured after the indicated treatments. Data are means ± SDs of triplicate measurements. *, P < 0.05; **, P < 0.01.

IKKβ contributes to mTORC1 activation via phosphorylation of TSC2 at Ser⁹³⁹.

It is well documented that the TSC1/2 complex is a major mediator for mTORC1 activation. TSC1/2 responses to signals from a large number of distinct modulators regulate mTORC1 activity. The TSC1/2 complex could be phosphorylated at different sites under different conditions by different upstream kinases (19, 32). Regulation of TSC2 by IKK components in LMP1-mediated activation of mTORC1 has not been previously examined. Our results indicate that IKKβ also plays a more dominant role than IKKα in mediating LMP1-induced mTORC1 activation. Previous studies have shown that IKKβ suppresses TSC1 to activate mTORC1 (17). In this study, we examined if regulation of TSC2 by IKK is involved. We first cotransfected pcDNA-LMP1 with either Flag-tagged wild-type IKKβ or IKKβ mutant with defective kinase (KM) plasmids into HONE1 cells and detected the phosphorylation of various TSC2 activation residues (Ser⁹³⁹, Thr¹⁴⁶², and Ser¹³⁸⁷) by specific antibodies. As shown in Fig. 9A, LMP1 induced TSC2 phosphorylation at Ser⁹³⁹ and Thr¹⁴⁶² but inhibited phosphorylation at Ser¹³⁸⁷. Furthermore, expression of IKKβ-KM mutant, but not wild-type IKKβ, blocked LMP1-induced phosphorylation of TSC2 at Ser⁹³⁹. The phosphorylation of the other two TSC2 activation sites, Thr¹⁴⁶² and Ser¹³⁸⁷, was not affected. These observations suggest that TSC2 phosphorylation at Ser⁹³⁹ may play a major role for IKKβ to mediate LMP1-induced mTORC1 activation. The phosphorylation of TSC2 Ser⁹³⁹ by IKKβ during LMP1 activation of mTORC1 was also confirmed in 293T cells (Fig. 9B). By using IKKβ inhibitor, we also confirmed that IKKβ mediates TSC2 phosphorylation at Ser⁹³⁹ (Fig. 9C).

FIG 9.

IKKβ phosphorylates TSC2 at Ser⁹³⁹ to activate mTORC1. (A and B) HONE1 (A) and 293T (B) cells were cotransfected with LMP1 and wild-type IKKβ (Flag-IKKβ WT) and mutant IKKβ (Flag-IKKβ-KM) constructs for 36 h. Different phosphorylation sites of TSC2 and other relevant proteins were then analyzed with corresponding antibodies in transfected cells. β-Actin expression was used as the loading control. (C) HONE1-LMP1 cells were treated with PS1145 for 24 h and then analyzed by Western blotting for detection of the different phosphorylation sites of TSC2 by the specific antibodies. (D) HONE1 cells were cotransfected with different phosphorylation site-mutated TSC2 plasmids and 2117-LMP1 for 36 h. The cells were collected and cell lysates were subjected to Western blot analyses. The phosphorylation of mTORC1 components was used to monitor mTORC1 signaling activity. β-Actin expression was used as the loading control.

To further confirm that Ser⁹³⁹, but not Thr¹⁴⁶², phosphorylation of TSC2 is important for LMP1-induced mTOR activation, we transfected TSC2 mutants with their phosphorylation sites mutated (pcDNA3-Flag-TSC2^S939A, pcDNA3-Flag-TSC2^T1462A, and pcDNA3-Flag-TSC2^S939A/T1462A into HONE1 cells and monitored the phosphorylation of mTORC1 (Ser²⁴⁴⁸) and its downstream events, phosphorylation of p70s6k and 4EBP1. As shown in Fig. 9D, the expression of the pcDNA3-Flag-TSC2^S939A mutant and pcDNA3-Flag-TSC2^S939A/T1462A double mutant was more effective than expression of the pcDNA3-Flag-TSC2^T1462A mutant in suppression of mTORC1 activation. These results support the idea that phosphorylation of TSC2 Ser⁹³⁹ plays a more important role in mediating LMP1-induced mTORC1 activation via IKKβ activation.

DISCUSSION

Aerobic glycolysis is an important hallmark of cancer cells. The involvement of LMP1 in aerobic glycolysis was reported in recent studies involving activation of the FGFR1 signaling pathway and c-Myc-mediated HK2 expression (25, 26). LMP1 also upregulates IKKβ/NF-κB signaling to modulate AKT-induced plasma membrane trafficking of Glut-1 in B cell lymphoma (14). Translocation of Glut-1 to the cell membrane by LMP1 was also observed in a recent study with immortalized nasopharyngeal epithelial cells (NP69) stably expressing LMP1 (25). In this study, LMP1 could also directly upregulate Glut-1 transcription to drive aerobic glycolysis in multiple immortalized nasopharyngeal epithelial cell lines and the cancer cell line HONE1 via mTORC1 activation and NF-κB signaling. Interestingly, upregulation of Glut-1 expression by LMP1 was not observed in B cell lymphoma in an earlier study (14). The reasons for this apparent discrepancy are unclear at this stage. It remains to be determined if the ability of LMP1 to directly upregulate transcription and expression of Glut-1 is cell type specific and could be detected in epithelial carcinoma cells but not in lymphoma cells. However, the different strains of LMP1 and experimental approaches used in the two studies may also be accountable for the discrepancy observed in the action of LMP1 on Glut-1 transcription. In our study, the 2117-LMP1 used is the common LMP1 variant detected in nasopharyngeal carcinoma (NPC) from Hong Kong patients (33). The 2117-LMP1 was reported to have more potent activity in NF-κB activation but lower cellular toxicity than the B95.8-derived LMP1 sequences commonly used in B cell lymphoma studies (34). Moreover, dual mechanisms of activation of Glut-1 involving direct activation and membrane relocation of Glut-1 are involved in the glucose uptake in brown adipose tissue upon stimulation of β₃ adrenoceptors involving cAMP1 and mTOR activation, respectively (35). More studies are warranted to further dissect the mechanisms involved in the upregulation of Glut-1 transcription and membrane translocation in LMP1 activation of glucose uptake in nasopharyngeal epithelial cells.

Here we report that the upregulation of Glut-1 transcription is dependent on LMP1 activation of mTORC1 and NF-κB signaling. Furthermore, expression of Glut-1 and LMP1 could be detected in NPC specimens by immunocytochemistry, supporting the relevance of this signaling pathway in NPC pathogenesis. NF-κB signaling is well documented for its involvement in mediating multiple pathological properties of LMP1, including cell proliferation (11), tumorigenic transformation (36), and cell invasion and motility (37). We have recently reported that activation of the epidermal growth factor receptor (EGFR)/mTORC1/NF-κB signaling axis is common in telomerase-immortalized nasopharynageal epithelial cells (38). Our findings that Glut-1 expression could be upregulated by LMP1 through the mTORC1/NF-κB activation suggest that Glut-1 activation may facilitate growth of immortalized nasopharyngeal epithelial cells, which facilitates their subsequent tumorigenic transformation. Upregulation of Glut-1 transcription also mediates multiple malignant properties of LMP1 in nasopharyngeal epithelial cells. Our study supports a role for EBV in malignant transformation of precancerous nasopharyngeal epithelium to cancer cells (6). The recent report on genomic profiling of NPC using whole-exome sequencing showed that loss-of-function mutations of multiple negative regulators, including CYLC, TRAF3, and NFKBIA, upstream of NF-κB signaling activation are common in NPC, supporting an important role for NF-κB activation in the development of NPC (39). Interestingly, a mutually exclusive relationship between somatic mutation of these negative regulator of NF-κB and LMP1 expression in NPC was observed. This observation supports the postulation that NF-κB activation in NPC, mediated by either somatic mutation of negative regulators of NF-κB or expression of EBV-encoded LMP1, facilitates NPC pathogenesis. We propose that activation of aerobic glycolysis by NF-κB, by either LMP1 or non-LMP1 mechanisms, supports growth and survival of NPC cells.

In this study, we observed that NF-κB activation by LMP1 is dependent on mTORC1 activation. A recent study has reported that LMP1 could activate mTORC1 signaling in NPC and that PI3K/AKT activation is involved (23). Details of the activation mechanisms of mTORC1 by LMP1 and its involvement in NF-κB activation and aerobic glycolysis are largely undefined. In this study, we showed that LMP1 activation of mTORC1 is essential for NF-κB signaling. The CTAR2 signaling domain of LMP1 plays a more dominant role in the activation of IKKβ and phosphorylation of TSC2 Ser⁹³⁹ to mediate mTORC1 signaling and NF-κB activation, which modulates Glut-1 transcription. mTOR is an evolutionally conserved serine/threonine family protein kinase for regulating cell growth and proliferation via distinct pathways, including regulation of protein biosynthesis and ribosome biogenesis (19, 40). LMP1 is an important EBV-encoded oncoprotein commonly expressed in NPC, particularly in early preinvasive lesions (4, 8). Activation of mTORC1 in NPC correlates with poor survival of NPC patients (23). A major function of mTORC1 is regulating energy metabolism to support growth of tumor cells. Our results reveal a close interactive relationship among LMP1, mTORC1, NF-κB, and Glut-1 in promoting aerobic glycolysis in NPC. Members of the Glut family are involved in the transportation of extracellular glucose across the plasma membrane into cells to mediate glucose metabolism and support growth of human cancer cells (13, 41). In a number of cell types, the transport of glucose into cells via Glut-1 is the rate-limiting step for glucose metabolism (42). Glut-1 inhibitors with potential applications in cancer treatment have recently been developed. STF-31, used in this study, is one of the promising Glut-1 inhibitors which have been shown to selectively kill renal carcinoma cells (25, 43). Interestingly, there are reports showing that high glucose levels also enhance virus infection of host cells, supporting viral replication and maintenance of latency infection (44–46). Activation of Glut-1 transcription is a novel property of LMP1 not previously reported. The ChIP assay revealed direct binding of the p65 subunit of NF-κB to the promoter of Glut-1, suggesting that LMP1-activated NF-κB signaling is directly involved in driving up the transcription of Glut-1. Glut-1 transcription could also be upregulated by transcription factors like Sp1 (47) and HIF-1α (48), but their involvement in LMP1 upregulation of Glut-1 expression remains to be investigated. Defining the molecular pathways of Glut-1 upregulation by LMP1 in NPC cells may reveal novel therapeutic targets for NPC treatment. Indeed, a recent study has reported that targeting LMP1-mediating glycolysis sensitizes NPC to radiation therapy (26).

We have elaborated the essential role of IKK in mediating LMP1 activation of mTORC1. IKK activation is the upstream event of NF-κB activation. Constitutive NF-κB activation is common in EBV-associated NPC and is involved in NPC pathogenesis (43). Activation of NF-κB is a major signaling pathway of the EBV-encoded LMP1. LMP1 activates NF-κB signaling through two distinct pathways: the canonical pathway (major pathway) through CTAR2/TRAF6/TAK1/IKKβ and the noncanonical pathway (minor pathway) through CTAR1/TRAF3/NIK/IKKα (31). A causal relationship between IKK activation and mTORC1 signaling has been reported in previous studies (18, 49, 50). In this study, we showed that the IKK complex plays an essential role in mediating LMP1 activation of mTORC1. The involvement of IKK in LMP1 activation of mTORC1 has not been defined in previous studies. Phosphorylation of TSC1 and phosphorylation of TSC2 are key upstream regulatory events of mTORC1. Several studies have reported that IKKα could directly activate mTORC1 signaling by phosphorylation of TSC1 (18, 27, 51). A recent study showed that IKKβ could also activate mTORC1 by phosphorylation of TSC1 at Ser⁵¹¹ (17). TSC2 Ser⁹³⁹ and Thr¹⁴⁶² could be phosphorylated by AKT, and Ser¹³⁸⁷ could be phosphorylated by AMP-activated protein kinase (AMPK) (52, 53). In our study, we have examined phosphorylation of multiple sites of TSC2 after expression of LMP1 and revealed a novel role of LMP1-CTAR2 in activation of IKKβ to phosphorylate TSC2 Ser⁹³⁹ in LMP1-induced mTORC1 activation.

Interestingly, inhibition of mTORC1 also effectively decreased the phosphorylation level and activation of IKKβ by LMP1, which suggests that the mTORC1 activation may be involved in a positive-feedback loop for LMP1-mediated NF-κB signaling. The underlying mechanisms involved are poorly defined. Nonetheless, a recent report showed that IKK/NF-κB activity could be mediated by TSC2 through mTORC1 activation in PTEN-null prostate cancer cells, suggesting that mTORC1 could serve as an upstream modulator for IKK/NF-κB activation (54). Another study also reported that mTORC1 activation regulates NF-κB activation, which involves upregulation of EGFR and IKK activation, possibly through a positive-feedback loop regulation (55). Feedback mechanisms of downstream events of NF-κB activated by LMP1, e.g., EGFR overexpression, may be involved to modulate upstream events, e.g., PI3K/AKT, to induce mTORC1 activation, hence forming a positive-feedback loop in mTORC1/NF-κB activation in LMP1-expressing NPC cells to promote tumorigenesis. A schematic diagram summarizing our findings and proposed signaling events is shown in Fig. 10.

FIG 10.

Schematic diagram of LMP1-induced mTORC1/NF-κB/Glut-1 activation in NPC cells. The viral oncoprotein LMP1 activates mTORC1 through AKT/ERK/IKK signaling axis. Both CTAR1 and CTAR2 of LMP1 are involved in activation of mTORC1 signaling. LMP1-CTAR1 regulates mTORC1 activation through AKT-ERK-IKKα, while LMP1-CTAR2 regulates mTROC1 activation via IKKβ. The CTAR2/IKKβ plays a more dominant role in mTORC1 activation through phosphorylation of Ser⁹³⁹ of TSC2. Furthermore, mTORC1 activation induced by LMP1 also modulates NF-κB signaling by undefined pathways (dotted arrows), which regulates aerobic glycolysis and promotes tumorigenesis through upregulation of Glut-1 transcription and expression.

In conclusion, we demonstrated that LMP1 activation of mTORC1 signaling contributes to accelerate aerobic glycolysis and enhance malignant properties in NPC cells. Upregulation of the NF-κB/Glut-1 signaling pathway is involved. Our results may provide novel therapeutic targets against the mTORC1/NF-κB/Glut-1 signaling cascade in treatment of NPC.

MATERIALS AND METHODS

Chemicals, antibodies, and plasmids.

The mTORC1-specific inhibitor (rapamycin) and IKKβ inhibitor (PS1145) were purchased from Sigma (St. Louis, MO). The Glut-1 inhibitor STF-31 was purchased from Selleckchem (Houston, TX). U0126, LY294002, and NBD were purchased from Calbiochem (San Diego, CA). Antibodies against p-mTOR, mTOR, p-p70s6K, p70s6K, p-4E-BP1, 4EB-P1, p-AKT, AKT, p-I-κBα, p-IKKα/β, IKKβ, p-ERK, ERK, p-MEK, MEK, p-TSC2, and TSC2 were purchased from Cell Signaling Technology (Danvers, MA). Antibodies against β-actin and horseradish peroxidase (HRP)-linked secondary antibodies were purchased from Santa Cruz Biotechnology (TX), and anti-Glut-1 antibody was purchased from Abcam (MA). Flag-tagged IKK (K44M) plasmids were kindly provided by D. Y. Jin (University of Hong Kong). Wild-type and mutated LMP1 constructs were kindly provided by Z. G. Wu (Hong Kong University of Science and Technology). Their signaling properties have been previously described (31). The TSC2-related mutation plasmids were obtained from Addgene (52).

Cell lines and cell culture.

HONE1 cells (an EBV-negative NPC cell line) were cultured in RPMI 1640 medium (Sigma, St. Louis, MO), and 293T cells were cultured in Dulbecco modified Eagle medium (DMEM; Sigma). Both media were supplemented with 10% (vol/vol) fetal bovine serum (GIBCO), 100 U/ml of penicillin, and 100 U/ml of streptomycin, and the cells were cultured at 37°C in a humidified atmosphere with 5% CO₂. The immortalized NP69 nasopharyngeal epithelial cell line was established by our laboratory and was cultured in keratinocyte serum-free medium (Invitrogen, Carlsbad, CA) (56).

Lentivirus preparation.

Lentivirus preparation was performed as described previously (38). Briefly, the target gene constructs in the lentivirus packaging vectors psPA × 2 and pMD2G were transfected into the 293T packaging cell line using the X-tremeGENE HP DNA transfection reagent (Roche) according to the manufacturer's instructions. Three days after transfection, the supernatant was collected, filtered through a 0.45-μm filter, and used for infecting cells.

[³H]thymidine incorporation assay.

Cells were plated in a 24-well plate, incubated in a 37°C incubator with CO₂, and used for experiments at 60% cell confluence. After various treatments, 1 μCi of [³H]thymidine was added to each well and the cells were further incubated for 4 h for incorporation of [³H]thymidine. The thymidine incorporation assay was performed as previously described (38).

Promoter reporter constructs and luciferase assay.

Human Glut-1 promoter fragments were generated by PCR amplification using human genomic DNA as the template and cloned into the NheI/XhoI (New England BioLabs [NEB], USA) sites of the luciferase reporter plasmid pGL3-basic (Promega, Madison, WI). The primer sequences used to generate the Glut-1 promoter sequences are as follows: forward primer 1, 5′-TAAGCAGCTAGCCTGCTCACTCATTCGTGCAT-3′; forward primer 2, 5′-TAAGCAGCTAGCACAATACCAACCAGT-3′; forward primer 3, 5′-TAAGCAGCTAGCGTTCAAACCCGAGGTCTA-3′; and reverse primer, 5′-TGCTTACTCGAGGATCGGCTCGTTCTCTCT-3′. The potential NF-κB binding site on the promoter region of the human Glut-1 gene was mutated by PCR amplification with a site-mutated primer as follows: forward primer 4, 5′-TATGAGTAAATGAGATTCCCACACCAATCT-3′, and reverse primer, 5′-AGATTGGTGTGGGAATCTCATTTACTCATA-3′. Cells were cultured in triplicates at 80% confluence in 12-well plates and transfected with the various promoter constructs (500 ng/well) in combination with the Renilla plasmid (100 ng/well) as an internal control. Luciferase activity was detected with a dual-luciferase reporter assay system (Promega).

Western blotting.

Cells were treated as indicated in Results. After treatment, cells were harvested, washed with phosphate-buffered saline (PBS), and lysed in ice-cold radioimmunoprecipitation assay (RIPA) lysis buffer for 30 min. After protein quantification, 20-μg quantities of denatured protein samples were subjected to SDS-PAGE. Western blot analysis was performed as described previously (38).

Immunofluorescence staining and microscopy.

Cells grown on coverslips were fixed with 4% formaldehyde at room temperature for 10 min. After being rinsed with ice-cold PBS three times, cells were permeabilized with 0.1% Triton X-100 for 10 min and then blocked with 3% bovine serum albumin (BSA) for 1 h at room temperature. Coverslips were incubated overnight at 4°C with primary antibodies against p65, followed by fluorescence dye-conjugated secondary antibody for 1 h in the dark. The cells were then washed repeatedly with PBS and counterstained with 4′,6-diamidino-2-phenylindole (DAPI), and fluorescence images were acquired by OLYMPUS BX51.

Immunohistochemical staining.

Immunohistochemical staining was performed as described previously (25, 43).

Measurements of glucose consumption and lactate production.

Cells (5 × 10⁵) were seeded in a 6-well plate and used for experiments after 6 h of incubation. After treatment, the medium was discarded and replaced with fresh medium, and cells were incubated for another 8 h. The culture medium was then collected for determination of glucose consumption and lactate levels using commercial assay kits for glucose and lactate (Eton Bioscience, San Diego, CA). Data were normalized to final cell counts per well. Glucose consumption in the supernatant was calculated as the difference in glucose concentrations in incubated medium and fresh medium.

RT-PCR screening.

Cells at 80% confluence were harvested for total RNA extraction by TRIzol (Roche). The extracted RNA was quantified and converted into cDNA. Quantitative real-time PCR (qRT-PCR) of all samples was performed using a LightCycler 480 Probe Master with the human universal probe library set (Roche Applied Science, Switzerland). Specific forward and reverse primers were designed with the assistance of ProbeFinder software (Roche Applied Science). The primer sequences used in qRT-PCR screening are as follows: 5′-GGTTGTGCCATACTCATGACC-3′ (forward) and 5′-CAGATAGGACATCCAGGGTAGC-3′ (reverse) for Glut-1, 5′-CCCTGTCTGTATCCAGCTTTG-3′ (forward) and 5′-TGTTTGCTACTAACATGGCTTTG-3′ for Glut-2, 5′-TTCTGATATTGCCGCACTAGG-3′ (forward) and 5′-AGTCTGAGGTTGGGGGAACT-3′ (reverse) for Glut-3, and 5′-TTTCTCAGTGGGACAAACCA-3′ (forward) and 5′-AAGGTGGTGGGAAACTGGAT-3′ (reverse) for Glut-4.

ChIP assay.

Chromatin immunoprecipitation (ChIP) was performed using a ChIP assay kit (Covaris, MA) according to the manufacturer's recommendations. The antibodies used were normal rabbit IgG against p65. The immunoprecipitated DNA was purified and analyzed by qRT-PCR. The primers for the Glut-1 promoter domain containing the putative NF-κB binding region and nonspecific binding region used in the ChIP assays are as follows: 5′-ATTACAGGCGTGAGCCA-3′ (forward) and 5′-CACTTCATGAGCTTATCTGG-3′ (reverse) for the putative binding region and 5′-CATCTGTCTGAAATTGTCATGG-3′ (forward) and 5′-ACAAGCTGAGCAAATGTG-3′ (reverse) for the nonspecific binding region. The data were plotted as the ratio of immunoprecipitated DNA (with subtraction of nonspecific binding to IgG) to total input DNA.

Colony formation and anchorage-independent growth assay.

For the colony formation assay, cells were seeded at 500/well in 12-well plates after treatment and incubated for 14 days. Then they were washed with PBS and stained with 0.1% crystal violet for 15 min and counted for number of colonies. Colonies containing more than 50 cells were counted and included in the calculation. For the anchorage-independent growth assay, a 0.6% (wt/vol) bottom layer of low-melting-point agarose was prepared in 6-well plates. A layer of 0.3% agarose containing 5 × 10⁴ cells was placed on top of the 0.6% low-melting-point agarose. The anchorage-independent colonies in the 0.3% agarose were counted after 3 weeks.

Statistical analysis.

Data are expressed as means ± standard deviations (SDs) from three independent triplicates and were analyzed with GraphPad Prism 5 statistical software. The differences between experimental groups were analyzed with the Dennett t test; P values of <0.05 were considered statistically significant.