Lnc-THOR silencing inhibits human glioma cell survival by activating MAGEA6-AMPK signaling

Long non-coding RNA THOR (Lnc-THOR) binds to IGF2BP1, essential for its function. We here show that Lnc-THOR is expressed in human glioma tissues and cells. Its expression is extremely low or even undetected in normal brain tissues, as well as in human neuronal cells and astrocytes. We show that Lnc-THOR directly binds to IGF2BP1 in established and primary human glioma cells. shRNA-mediated Lnc-THOR knockdown or CRISPR/Cas9-induced Lnc-THOR knockout potently inhibited cell survival and proliferation, while provoking glioma cell apoptosis. Contrarily, forced overexpression of Lnc-THOR promoted glioma cell growth and migration. Importantly, Lnc-THOR shRNA or knockout activated MAGEA6-AMPK signaling in glioma cells. AMPK inactivation, by AMPKα1 shRNA, knockout, or dominant-negative mutation (T172A), attenuated Lnc-THOR shRNA-induced A172 glioma cell apoptosis. Moreover, CRISPR/Cas9-induced IGF2BP1 knockout activated MAGEA6-AMPK signaling as well, causing A172 glioma cell apoptosis. Significantly, Lnc-THOR shRNA was ineffective in IGF2BP1 KO A172 cells. In vivo, Lnc-THOR silencing or knockout potently inhibited subcutaneous A172 xenograft tumor growth in mice. MAGEA6 downregulation and AMPK activation were detected in Lnc-THOR-silenced/-KO A172 tumor tissues. Taken together, Lnc-THOR depletion inhibits human glioma cell survival possibly by activating MAGEA6-AMPK signaling.


Introduction
Glioma is among the most aggressive human malignancies, causing significant human mortalities each year 1-3 . In the clinical practices, gliomas are commonly diagnosed at late/advanced stages with extremely poor prognosis 3 . Molecularly targeted therapies are essential for better glioma prognosis [4][5][6] . Our group has been exploring novel therapeutic targets for this devastating disease [7][8][9][10] . Noncoding RNAs (ncRNAs), including microRNAs, long noncoding RNAs (LncRNA), and circular RNAs, are originally known as transcriptional noise. Recent studies have implied that LncRNAs, and other ncRNAs, play pivotal roles in initiation and progression of human glioma 11 and many other cancers [12][13][14] .
Our previous studies have implied that forced activation of AMP-activated protein kinase (AMPK) can inhibit human glioma cells 21,22 . Thr-172 phosphorylation of AMPKα1 is essential for AMPK activation. AMPK activation inhibits mammalian target of rapamycin (mTOR) complex 1 (mTORC1), a key oncogenic cascade 23,24 . In human cancer cells, activated AMPK could also induce growth inhibition and cell-cycle arrest by stabilizing and activating p53 25 . Moreover, AMPK activation is shown to trigger autophagy and degradation of multiple growth factor receptors (i.e., epidermal growth factor receptor (EGFR) and platelet-derived growth factor receptor α (PDGFRα)), thereby causing cancer cell inhibition 26 .
AMPKα1 expression is often sequestered in human cancer cells. Pineda et al. showed that MAGEA6-TRIM28 complex is a cancer-specific ubiquitin ligase, responsible for degradation AMPKα1 only in cancer cells 27 . We have previously shown that MAGEA6 knockdown by targeted short hairpin RNA (shRNA) restored AMPKα1 expression, causing glioma cell death and apoptosis 21 . In the present study, we will show that Lnc-THOR-IGF2BP1 cascade is essential for MAGEA6 expression in glioma cells. Inhibition of Lnc-THOR-IGF2BP1 cascade will induce MAGEA6 downregulation, AMPKα1 expression, and AMPK signaling activation, inhibiting glioma cell survival in vitro and in vivo.

Chemicals and reagents
Puromycin and neomycin were obtained from Sigma-Aldrich (St. Louis, MO). Cell culture reagents were provided by Gibco-BRL (Grand Island, NY). The anti-MAGEA6 antibody (ab38495) was purchased from Abcam (Shanghai, China). All other antibodies were provided by Cell Signaling Tech (Shanghai, China). TRIzol reagents for RNA assays, Lipofectamine 2000, and other transfection reagents were obtained from Invitrogen (Shanghai, China).

Cell culture
Cultures of HCN-1a human neuronal cells, A172 and U251MG ("U251") human glioma cells, as well as the primary human astrocytes, were described earlier 22,28 . The human glioma cells, derived from two primary glioma patients, were provided by Dr. Cao 29,30 , which were named as "Pri-1/Pri-2," and cultured as previously described 30,31 . The protocols of studying human cells and tissues were approved by the Ethics Review Board of Shanghai Jiao-Tong University School of Medicine, according to Declaration of Helsinki.

Human tissues
As reported earlier 21 , a total of five glioma tissues, along with paired surrounding normal brain tissues, were acquired and stored in liquid nitrogen. Tissues were separated, thoroughly washed, minced, and homogenized by the tissue lysis buffer (BiYunTian, Wuxi, China). Written informed consent was obtained from each participant.
Quantitative real-time reverse transcriptase polymerase chain reaction (qPCR) As reported 21 , 500 ng RNA of each sample was applied in the reverse transcription (RT) reaction with specific RT primers and superscript III reverse transcriptase (Invitrogen). Afterwards, 100 ng obtained complementary DNA (cDNA) template was mixed with SYBR Master Mix (Applied Biosystem) and 200 nM primers. We utilized ABI Prism 7600H Fast Real-Time PCR system for qPCR assays. The primers are listed in Table 1. qPCR quantification was through 2 −ΔCt method using the following formula: 2 −(Ct of target gene − Ct of reference gene) . qPCR primers are listed in Table 1.

Lnc-THOR shRNA
A set of two shRNAs, against non-overlapping sequence of Lnc-THOR ("Seq1/2," designed and verified by Genechem, Shanghai, China), were individually inserted into GV248 construct. The construct, along with the lentivirus package plasmids (Genechem), were transfected to HEK-293 cells to generate Lnc-THOR shRNA lentivirus. The virus was enriched, filtered, and added to glioma cells (plated at a density of 1 × 10 5 cells/well into 6-well plates).
Cells were then subjected to selection by using puromycin (2.5 μg/mL, for 10-12 days). In stable cells, Lnc-THOR knockdown was verified by qPCR assay.

Cell proliferation assays
For the soft agar colony-formation assay, A172 cells (5000 cells of each treatment) were re-suspended in agar (0.5%)-containing complete medium (with fetal bovine serum (FBS)) and added on the top of 10-cm culture dishes. After incubation for 10 days, A172 cell colonies were stained and manually counted. The detailed protocol for the 5-ethynyl-2'-deoxyuridine (EdU) staining assay was reported earlier 32 .

Apoptosis assays
The detailed protocols of apoptosis assays, including Histone DNA enzyme-linked immunosorbent assay and Annexin V FACS, as well as terminal deoxynucleotidyl transferase-mediated dUTP-fluorescein nick end labeling (TUNEL) staining assay and caspase-3/caspase-9 activity assays, were described in previous studies 33,34 .
"Transwell" in vitro migration assay A172 glioma cells (3 × 10 5 cells in 300 μL medium) were seeded into the upper part of the "Transwell" chambers (12-μm pore size, BD Biosciences). The lower compartments were filled with complete medium with 10% FBS. After 48 h, on the upper surfaces the non-migrated A172 cells were removed. On the lower surfaces, the migrated cells were fixed, stained, and counted.

Western blotting analysis
The detail protocol of western blotting assay was described in our previous studies 9, 10 . Briefly, for each treatment 40 μg of protein lysates (in each lane) were separated in denaturing 10-12% polyacrylamide gels and transferred to a polyvinylidene difluoride blots. After blocking (in 10% milk PBST solution) and three washes in TBST, blots were incubated with the indicated primary and secondary antibodies. Immuno-reactive proteins were detected by an enhanced chemiluminescence kit (Amersham, Shanghai, China) and analyzed through autoradiography. ImageJ software (NIH) was utilized for the quantification of the protein band, which was always normalized to the loading control.

AMPKα1 shRNA
As described 21 , the lentiviral AMPKα1 shRNA was added to A172 cells (plated at a density of 1 × 10 5 cells/ well into 6-well plates) for 48 h. Puromycin (2.5 μg/mL)containing complete medium was added to select stable cells for 5-6 days. Control cells were infected with the lentiviral scramble control shRNA ("sh-C"). AMPKα1 silencing in the stable cells was confirmed by western blotting.

AMPK activity assay
Following the treatments, 200 μg of total cellular lysates were first incubated with anti-AMPKα1 antibody. The AMPK activity was examined in the kinase assay buffer by adding AMP-[γ-32 P] ATP mixture and AMPK substrate SAMS (HMRSAMSGLHLVKRR) peptide 35 . Phosphocellulose paper was added afterwards, stopping the reactions. The AMPK radioactivity was examined by a scintillation counter, and its value was normalized to control level.

IGF2BP1 or AMPKα1 KO
A172 cells were seeded onto 6-well tissue culture plates at a density of 1 × 10 5 cells/well. The lenti-CRISPR/Cas9-IGF2BP1-KO-GFP construct (provided by Dr. Zhao 36 ) or the lenti-CRISPR/Cas9-AMPKα1-KO-GFP construct (from Dr. Li 37 ) was transfected to A172 cells through Lipofectamine 2000 protocol. FACS-mediated sorting of the GFP-positive cells were performed to select the monoclonal cells, which were then cultured in the puromycin-containing complete medium to achieve stable cells. IGF2BP1 or AMPKα1 KO in the stable cells was confirmed by western blotting and/or qPCR assays.

Ectopic IGF2BP1 overexpression
The recombinant adenovirus encoding IGF2BP1 expression construct (provided by Dr. Zhao 36 ) was added to cultured A172 cells (plated at a density of 1 × 10 5 cells/ well into 6-well plates) for 48 h. Cells were thereafter subjected to puromycin (2.5 μg/mL) selection for another 5-6 days. IGF2BP1 overexpression was confirmed by western blotting.

RNA immunoprecipitation (RIP)
RIP experiments were carried out through a described protocol 38 . Briefly, glioma cells were trypsinized, washed, and incubated with 0.3% formaldehyde and glycine 39 . Afterwards, glioma cells were washed, and resuspended, with the pellets dissolved in the RIP buffer 38 . The lysates were then incubated with the anti-IGF2BP1 antibody. Pellets were washed, re-suspended, and incubated with proteinase K-containing buffer. IGF2BP1-bound Lnc-THOR and MAGEA6 mRNA was tested by qPCR, with its level normalized to internal controls.

RNA Pull-Down assay
RNA Pull-Down was carried out using a previously described protocol 39 . In short, the biotin-labeled fulllength Lnc-THOR (provided by Dr. Wang 39 ) was folded in RNA structure buffer and incubated with cleared nuclei lysates of the glioma cells together with Dynabeads MyOne Streptavidin C1 magnetic beads ("Beads," again provided by Dr. Wang 39 ). Beads were washed, with the retrieved proteins examined by western blotting.

Xenograft assay
As reported 21 , the female severe combined immunodeficient (SCID) mice were housed under standard procedures. Lnc-THOR shRNA-bearing stable A172 cells, Lnc-THOR KO stable A172 cells, or the parental control A172 cells (5 × 10 6 cells in 200 µl of Matrigel gel, no serum, each mouse) were subcutaneously (s.c.) injected to the flanks. When the volume reached approximately 100 mm 3 for each tumor ("Day-0"), the recordings were started. Tumor volumes were calculated as described 21 .
All animal procedures were approved by IACUC of Shanghai Jiao-Tong University School of Medicine.

Statistical analysis
All statistics were calculated by using the SPSS 18.0 statistical software (SPSS, Chicago, IL). Descriptive statistics including mean and standard deviation (SD) along with one-way analyses of variance were applied to determine significant differences. Two-tailed unpaired T test (Excel 2013) was applied to test significance between the two treatment groups. p < 0.05 was considered significant.

Lnc-THOR expression in human glioma tissues and cells
First, we tested the expression of Lnc-THOR in human glioma tissues. As described in our previous studies 21 , a total of five pairs of human glioma tissues ("T") and paired surrounding normal brain tissues ("N") were analyzed, and qPCR assay results in Fig. 1a show that Lnc-THOR levels are high in human glioma tissues, whereas its levels in normal brain tissues are, however, extremely low (Fig. 1a). Further studies show that Lnc-THOR is expressed in established (A172 and U251 lines) and primary human glioma cells (derived from two different patients, "Pri-1/-2") ( Fig. 1b). Its expression is almost undetected in the primary human astrocytes 22 and HCN-1a neuronal cells 22 (Fig. 1b). These results confirm unique Lnc-THOR expression in human glioma tissues and cells.

Lnc-THOR silencing or KO inhibits human glioma cell progression in vitro
In order to study the function of Lnc-THOR in human glioma cells, two lentivirus-encoded Lnc-THOR shRNAs, with non-overlapping sequences ("Seq1/2"), were individually transfected to A172 glioma cells. Following puromycin selection, the stable cells were established ("sh-Lnc-THOR" cells). Moreover, the lenti-CRISPR/Cas9 Lnc-THOR-KO construct (see "Methods" section) was transfected to A172 cells. Stable cells ("KO-THOR" cells) were established by FACS sorting of GFP cells and puromycin selection. Analyzing Lnc-THOR expression in the stable cells, by qPCR, confirmed that Lnc-THOR levels were dramatically downregulated in the stable cells with Lnc-THOR shRNA or Lnc-THOR KO construct (Fig. 2a). Lnc-THOR binds to IGF2BP1 to ensure mRNA stabilization of key pro-cancerous genes, including IGF2, Gli1, and Myc 15,16,18,20 . In A172 glioma cells, mRNA levels of IGF2, Gli1, and Myc were significantly downregulated in Lnc-THOR-silenced or Lnc-THOR-KO A172 cells (Fig. 2b). IGF2, Gli1, and Myc proteins were downregulated as well (Fig. 2c). Lnc-THOR shRNA or KO did not affect The genetically modified stable A172 cells, with Lnc-THOR shRNA ("sh-Lnc-THOR," with non-overlapping sequences, "Seq1/2"), scramble non-sense control shRNA ("sh-C"), or the lenti-CRISPR/Cas9 Lnc-THOR-KO construct ("KO-THOR"), were established, the expression of Lnc-THOR and listed genes in the stable cells and in parental control A172 cells ("Ctrl") were tested by qPCR and western blotting assays (a-c); Cells were further cultured for the indicated time, and cell viability was tested by MTT assay (d); Cell proliferation was analyzed by EdU staining (e) and soft agar colony-formation (f) assays; Cell migration was tested by "Transwell" assays (g). U251MG ("U251") and primary human glioma cells ("Pri-1/Pri-2") were transfected with lentiviral Lnc-THOR shRNA ("sh-Lnc-THOR," "Seq1") or "sh-C" and stable cells were established via puromycin selection. Lnc-THOR levels were tested by qPCR assay (h); Cell survival and proliferation were tested by MTT assay (i) and EdU staining assay (j), respectively. Western blotting assay of the IGF2BP1 protein retrieved by in vitro-transcribed Lnc-THOR in A172 cells and primary human glioma cells (k). qPCR analyses of Lnc-THOR expression enriched by the IGF2BP1 protein in A172 cells and primary human glioma cells (l). For all the in vitro function assays, the exact same amount of viable cells was initially seeded in each well/dish ("0 h") (same for all figures). Blot data were quantified and normalized to the corresponding loading control (c). Data are presented as mean ± SD (n = 5). *p < 0.05 vs.
Lnc-THOR-IGF2BP1 binding has been reported in other cancer cells 15,20,39 . To test the direct association between Lnc-THOR and the IGF2BP1 protein in glioma cells, we employed a Lnc-THOR pull-down assay 39 . Results demonstrated that the IGF2BP1 protein is co-precipitated with the in vitro-transcribed biotinylated Lnc-THOR (provided by Dr. Wang 39 ) in both A172 cells and primary human glioma cells ("Pri-1/-2") ( Fig. 2k). In addition, the RIP assay results show again the direct binding between Lnc-THOR and the IGF2BP1 protein in A172 cells and the primary human glioma cells (Fig. 2l).

Lnc-THOR silencing or depletion inhibits A172 xenograft tumor growth in vivo
As described in our previous study 21 , an A172 tumor xenograft SCID mice model was established to study the potential activity of Lnc-THOR in vivo. The genetically modified stable A172 cells with Lnc-THOR shRNA ("sh-Lnc-THOR," "Seq-1") or lenti-CRISPR/Cas9 Lnc-THOR-KO construct ("KO-THOR"), as well as the parental control A172 cells ("Ctrl"), were inoculated via s.c. injection to the flanks of the SCID mice. Tumor volumes were recorded, and the tumor growth curve results in Fig. 7a demonstrated that A172 tumor growth was significantly inhibited with Lnc-THOR silencing or depletion. Estimated daily tumor growth, calculated by (estimated tumor volume at Day-35 − estimated tumor volume at Day-0)/35), was also significantly decreased in "sh-Lnc-THOR" tumors and "KO-THOR" tumors (Fig. 7b). Moreover, "sh-Lnc-THOR"   Fig. 7 Lnc-THOR silencing or depletion inhibits A172 xenograft tumor growth in vivo. Parental control A172 cells ("Ctrl"), the genetically modified stable A172 cells with Lnc-THOR shRNA ("sh-Lnc-THOR," "Seq1"), or the lenti-CRISPR/Cas9 Lnc-THOR-KO construct ("KO-THOR") were injected s.c. to the flanks of female SCID mice. When the tumors reached 100 mm 3 (10 mice per group), recordings were initiated ("Day-0"). Tumor volumes (a, in mm 3 ) were monitored every 7 days; Daily tumor growth was calculated as described (b). At Day-35, tumors of all three groups were isolated and weighed individually (c). At Day-7 and Day-14, one tumor per group was isolated, and the expression of Lnc-THOR (d) and listed proteins (e) in the fresh tumor lysates was tested. Blot data were quantified and normalized to the corresponding loading control (e). Data are presented as mean ± SD. *p < 0.05 vs. "Ctrl" group (a-c, n = 10) tumors and "KO-THOR" tumors weighed significantly lower than "Ctrl" tumors (Fig. 7c). The body weights of the SCID mice were not significantly different between the three groups (data not shown). These results confirmed that Lnc-THOR silencing or depletion inhibited A172 xenograft tumor growth in vivo.
In order to test signaling changes in vivo, at Day-7 and Day-14, one tumor per group was isolated (total six tumors). Fresh tumor lysates were achieved and tested. When compared to "Ctrl" tumors, Lnc-THOR levels were significantly decreased in the "sh-Lnc-THOR" tumors and "KO-THOR" tumors ( Fig. 7d), where MAGEA6 downregulation, AMPKα1 upregulation, and AMPK activation, as well as p-S6K1 inhibition, were detected (Fig. 7e). Total S6K1 and IGF2BP1 expression was unaffected by Lnc-THOR silencing or KO in tumor lysates (Fig. 7e). These results in vivo are therefore in line with the in vitro findings.

Discussion
The results of the current study indicate that Lnc-THOR could possibly be a novel and important therapeutic target of human glioma. Lnc-THOR is uniquely expressed in human glioma tissues and cells. Its expression is extremely low or even undetected in normal brain tissues, as well as in normal neuronal cells/astrocytes. In established (A172 cell line) and primary human glioma cells, Lnc-THOR shRNA or KO potently inhibited cell survival and proliferation, while provoking cell apoptosis. Contrarily, forced overexpression of Lnc-THOR can further promote glioma cell growth and migration. In vivo, A172 xenograft tumors with Lnc-THOR silencing or KO grew significantly slower than control tumors in SICD mice. These results are in line with recent findings proposing Lnc-THOR as a novel therapeutic oncotarget for many human cancers [16][17][18][19][20]45 .
MAGE-TRIM28 complex is a cancer-specific AMPKα1 ubiquitin ligase 21,27,40,41 . We have previously shown that MAGEA6, one of the key AMPKα1's ubiquitin ligase 21,27,40,41 , is uniquely expressed in human glioma tissues and cells, responsible for AMPKα1 degradation and AMPK inhibition. Contrarily, MAGEA6 silencing/depletion restored AMPKα1 expression and induced AMPK activation, causing downstream mTORC1 inactivation and glioma cell death 21 . The regulatory mechanism of MAGEA6 expression in glioma is elusive. The results of this study suggest that Lnc-THOR-IGF2BP1 association is important for MAGEA6 expression in glioma cells. The RIP results show that MAGEA6 mRNA directly binds to the IGF2BP1 protein in A172 cells and the primary glioma cells. Significantly, Lnc-THOR silencing/KO or IGF2BP1 KO induced MAGEA6 degradation (both mRNA and protein), AMPKα1 protein accumulation, and AMPK activation in A172 glioma cells. These results suggest that Lnc-THOR-IGF2BP1 complex is important for MAGEA6 expression, causing AMPKα1 degradation and AMPK inactivation in human glioma cells.
Our study 21 and others have implied that forced activation of AMPK signaling can induce human cancer cell apoptosis via regulating its downstream effectors, including mTORC1 inhibition 23,24,42 , autophagy induction [42][43][44] , and RTK (EGFR, PDGFR, etc) degradation 10,26 . We provided evidences here to support that AMPK activation mediates, at least in part, Lnc-THOR-depletion-induced glioma cell death. In A172 cells, Lnc-THOR silencing/KO induced MAGEA6 degradation, AMPKα1 elevation, and AMPK signaling activation, causing mTORC1 inhibition and EGFR-PDGFR degradation and eventually cell apoptosis. Similarly, IGF2BP1 KO also activated MAGEA6-AMPK signaling in A172 cells. Importantly, AMPK inactivation, by AMPKα1 shRNA, KO, or dominant-negative mutation, attenuated Lnc-THOR shRNA-induced A172 cell apoptosis. Significantly, AMPK blockage failed to completely reverse Lnc-THOR shRNA-induced cytotoxicity in A172 glioma cells, suggesting that both AMPK-dependent and AMPKindependent mechanisms are responsible for Lnc-THORsilencing-induced glioma cell death (see Fig. 8, the proposed signaling pathway of the study). Therefore, although further studies are needed to explore the detailed underlying mechanisms, here we propose that Lnc-THOR-IGF2BP1 association is vital for MAGEA6 expression and AMPK inactivation in human glioma cells (Fig. 8).

Conclusion
Lnc-THOR depletion activates MAGEA6-AMPK signaling and inhibits human glioma cell survival.