A cytoplasmic long noncoding RNA LINC00470 as a new AKT activator to mediate glioblastoma cell autophagy

A primary tumor cell culture was performed as previously described [36]. Astrocytoma cell lines U251 and U87 were bought from cell banks of the Chinese Academy of Sciences (Shanghai, China). All astrocytoma cell lines were subjected to a short tandem repeat (STR) test. U251 and primary tumor cells were cultured in DMEM high-glucose medium with 10% FBS and a 1% antibiotic-antimycotic solution (Gibco, Grand Island, NY, USA), while U87 cells were cultured in MEM medium with 10% FBS and 1% antibiotic-antimycotic solution at 37 °C and 5% CO₂.

The following primary antibodies were used: AKT (rabbit, Proteintech, 10176-2-AP, WB1:1500, IP:1:250, RIP:1:100); FUS (rabbit, Abcam, ab23439, WB1:2000, IP1: 200, RIP1:100); phospho-Akt (Ser473) (rabbit, Cell Signaling, #4060, WB1:1500); phospho-Akt (Thr308) (rabbit, Cell Signaling, #13038, WB1:1500); hexokinase I (rabbit, Cell Signaling, #2024, WB1:1000); hexokinase II (rabbit, Cell Signaling, #2867, WB1:1000); Flag (mouse, Sigma-Aldrich, F1804, IP 1:200); GAPDH (mouse, Sangon, D190090, WB 1:5000); H3 (rabbit, Beyotime, AH433, WB 1:500); and p53 (mouse, Active Motif, 39739, WB 1:1000, RIP 1:150). MK-2206 2HCl (S1078) was purchased from Selleck.

Cell transfection was performed using Lipofectamine 3000 (Invitrogen-Life Technologies, Carlsbad, CA, USA) per the manufacturer’s instructions.

This procedure was carried out as previously described. The following primers were used: LINC00470: F: 5′-CGTAAGGTGACGAGGAGCTG-3′, R: 5′-GGGGAATGGCTTTTGGGTCA-3′; AKT: F: 5′-GAAGGACGGGAGCAGGC-3′, R: 5′-AAGGTGCGTTCGATGACAGT-3′; and GAPDH: F: 5′-AATGGGCAGCCGTTAGGAAA-3′, R: 5′-GCGCCCAATACGACCAAATC-3′.

Details of Western blotting were previously described [37]. Cell lysates were prepared with GLB buffer (10 mM Tris-HCl, pH = 7.5; 10 mM NaCl; 0.5% Triton X-100; 10 mM EDTA) supplemented with protease inhibitor cocktail (Bimake, Houston, TX, USA, B14001) and phosphatase inhibitor (Bimake, B15001). Cytoplasmic and nuclear proteins were prepared with a Nuclear and Cytoplasmic Protein Extraction Kit (Beyotime, p0028). Thirty-microgram proteins were subjected to electrophoresis in different percentages of gels according to the molecular weight of the detected proteins.

For the interaction of FUS and AKT, HEK293 cells were transfected with the indicated plasmids and extracted by the addition of lysis buffer. For the immunoprecipitation of endogenous FUS and AKT proteins extracted by the addition of lysis buffer, the soluble supernatants were incubated with the indicated antibodies for 1 h at 4 °C. The immunocomplexes were then precipitated with protein A-Sepharose CL-4B. The immunocomplexes were washed three times with lysis buffer, eluted by boiling in sample buffer for SDS-PAGE, and then subjected to immunoblot analysis.

GST fusion proteins containing various deletions of FUS cytoplasmic domain or deletions of AKT were expressed in U251 cells with the pGEX-4T-2 vector and were purified. The lysate was incubated for 1 h with GST-tagged proteins and glutathione-Sepharose 4B beads. The beads were subsequently washed three times in the lysis buffer containing 1 mM EDTA and 0.5 mM DTT. Precipitates were separated by SDS-PAGE and detected by Western blotting analysis.

This procedure was carried out as previously described [35].

Approximately 2 μg of the cell extract was mixed with agarose beads, which had already precipitated with the protein antibodies. Beads were washed briefly three times with GLB⁺ lysis, and the retrieved protein was detected by Western blotting. The co-precipitated RNAs were detected by RT-qPCR.

All animal experiments were approved by the Animal Care and Use Committee of Central South University. Mouse orthotopic xenograft model was performed as previously described [36]. Six-week SD mice were chosen. Injection of cyclophosphamide once every 2 days. One-centimeter incision was made on the midline, and a 1-mm burr hole was drilled at AP = + 1 mm and MR = − 3 mm from the bregma at the right hemisphere. Ten microliters of 10⁷ cell was infused into the brain at a depth of − 5 mm from the dura, at a speed of 1 μl/min.

All experiments were analyzed with GraphPad Prism 5 (La Jolla, CA, USA). Differences between the different groups were tested using Student’s t test or one-way ANOVA. The relationships between the LINC00470 expression and clinic-pathological parameters were examined using the χ² test. The expression of LINC00470 and patients’ survival time was analyzed by single factor and multiplicity factor analysis, and OS curves were calculated using the Kaplan-Meier method with the SPSS 15.0 program (SPSS Inc., Chicago, IL, USA). Data are expressed as the mean ± S.E.M. from at least three independent experiments. A probability value P < 0.05 was considered statistically significant.

A vector construct containing the full-length LINC00470 with EGFP tag was developed and assessed for LINC00470 expression. LINC00470 did not have any detectable protein-coding ability (Additional file 1: Figure S1A and B). To investigate biological processes associated with LINC00470 expression in GBM, Pearson correlation analysis between LINC00470 expression and whole genome profiling were performed in GBM samples by TCGA databases. A total of 1802 gene expressions that correlated with LINC00470 expression are shown in Fig. 1a. To investigate which canonical pathways were significantly dysregulated in GBM groups with LINC00470 expression, Fisher’s exact test was used to identify 20 canonical pathways in GBM that included PI3K-AKT signaling (Fig. 1a). These analyses indicated that LINC00470 may be associated with PI3K-AKT signaling. Then, we measured the expression levels of LINC00470, AKT, and p-AKT in GBM cell lines and primary cultured GBM cells by RT-qPCR (Additional file 2: Figure S2A) and Western blotting (Additional file 2: Figure S2B). We found a positive correlation between the expression of LINC00470 and p-AKT. In GBM, LINC00470 and AKT had no correlation (Additional file 2: Figure S2B). Overexpression of LINC00470 upregulated the expression of p-AKT^T308 and p-AKT^S473 (Fig. 1b). We designed three kinds of siRNA and selected the best interfering effects for follow-up study (Additional file 3: Figure S3), while knockdown LINC00470 reduced p-AKT^T308 and p-AKT^S473 levels (Fig. 1c). We also re-expressed LINC00470 in the LINC00470-KD cells. Expression of LINC00470 was elevated after LINC00470 was re-expressed in the LINC00470-KD cells (Fig. 1d upper) and enhanced the p-AKT^T308 and p-AKT^S473 level (Fig. 1d lower panel). However, neither overexpression nor knockdown of LINC00470 affected total AKT and PI3K expression (Fig. 1b, c and Additional file 4: Figure S4). Therefore, we proposed that LINC00470 modulates AKT activities possibly through a previously unidentified mechanism.

Bioinformatics (http://​starbase.​sysu.​edu.​cn/​browseRbpLncRNA.​php) predicted FUS, an RNA-binding protein associated with LINC00470, may also bind to AKT. An RIP assay verified the interaction between LINC00470 and FUS (Fig. 2a). A biotin-RNA pulldown assay further confirmed the binding between LINC00470 and FUS (Fig. 2b). Interestingly, RNA-binding protein immunoprecipitation of FUS, but not AKT, specifically retrieved LINC00470 (Fig. 2c). We also performed RIP assay in U87 cells; the result of RIP assay was consistent with that in U251cells (Additional file 5: Figure S5). These results indicated that FUS interacted with LINC00470, but there was no direct interaction between LINC00470 and AKT.

To examine the simultaneous existence of LINC00470, FUS, and AKT within the same complex, a two-step Co-IP assay was performed using HEK293 cell lysate from cells in which HA-AKT, Flag-FUS, and pcDNA3.1-LINC00470 were co-transfected. LINC00470 was found in the final immunoprecipitation, suggesting that LINC00470, FUS, and AKT form a ternary complex (Fig. 2d). At the same time, we also found LINC00470, FUS, and AKT can form a ternary complex in U251 cells (Additional file 6: Figure S6). In the absence of LINC00470, FUS and AKT were co-localized in the nucleus of HEK293 cells (Additional file 7: Figure S7 and Fig. 2e). However, overexpression of LINC00470 anchored FUS and AKT in the cytoplasm (Fig. 2e), while with knockdown of LINC00470 in U251 cells, FUS and AKT translocated from the cytoplasm to the nucleus (Fig. 2f). The interactions between FUS and AKT in the cytoplasm of U251 cells were also verified by Western blotting analysis (Fig. 2g left). An RNA pulldown assay showed that the interaction of LINC00470 and AKT disappeared after FUS knockdown in U251 cells (Fig. 2g right). We also found LINC00470 could not affect AKT activation after pcDNA3.1-LINC00470 was transfected into FUS-KD cells (Fig. 2h). The phosphorylated AKT was reduced in the cytoplasm in LINC00470-knockdown GBM cells (Fig. 2i). The data indicated that LINC00470 promoted the activation of AKT in the cytoplasm by interacting with FUS.

Next, a series of LINC00470 deletion mutants were constructed to determine the nucleotides in LINC00470 that bind to FUS. An RNA pulldown assay showed that there was an interaction between FUS and LINC00470 mutants (1–300 nt, 1–710 nt, 1–1500 nt, 1–2231 nt, and 100–2231 nt), but no interaction between FUS and other LINC00470 deletion mutants (1–100 nt, 300–2231 nt, 710–2231 nt, 1500–2231 nt, 300–710 nt, 300–1500 nt, and 710–1500 nt) (Fig. 3a), suggesting that the 100–300 nt region of LINC00470 was responsible for its binding to FUS. We also found FUS bound to LINC00470 through its RNA recognition domain (RRM) (Fig. 3b). Confocal fluorescence microscopy indicated that LINC00470 and FUS were mainly co-localized in the cytoplasm in HEK 293 cells after overexpression of LINC00470 (Fig. 3c).

FUS was reported to be continuously shuttling between the nucleus and the cytoplasm [38]. FUS contains multiple post-translational modification sites in the RRM and GGR domain, and post-translational modifications of FUS have profound effects on its binding capacity of DNA, RNA and proteins, changes in protein stability, or subcellular localization [39, 40]. We speculated that LINC00470 may impact FUS subcellular localization. The nuclear localization of FUS was explored by transient expression of the GFP-FUS fusion plasmid in HEK293 cells (Fig. 3d). When HEK293 cells were co-transfected by pcDNA3.1-LINC00470 and the GFP-FUS fusion plasmid, LINC00470 led to the translocation of FUS from the nucleus to the cytoplasm (Fig. 3d). In GBM cells, the expression of FUS was increased by LINC00470 overexpression and the FUS level was significantly increased in cytoplasm; however, its expression was decreased in the nucleus (Fig. 3e). These data suggested that LINC00470 anchored FUS in the cytoplasm and promoted its expression in the cytoplasm. FUS immunoprecipitation from U251 cells after overexpression of LINC00470 was immunoblotted with anti-phospho-T (threonine phosphorylation) antibodies to assess the level of FUS phosphorylation; LINC00470 promoted phosphorylation of FUS at threonine residues (Fig. 3f). In addition, we also found that LINC00470 was mainly located in the cytoplasm in GBM cells by RNA fluorescence in situ hybridization (Fig. 3g).

The “Scansite 2.0” software was utilized to identify a docking domain (GGR domain) in FUS, which is an AKT kinase-binding site. GFP-FUS and RFP-AKT expression plasmids were co-transfected into HEK293 cells, CO-IP and immunofluorescence suggested there were interactions between FUS and AKT, and both were co-localized in the nucleus of HEK293 cells (Fig. 4a, b). In addition, we confirmed that endogenous AKT interacted with FUS in the cytoplasm of U251 cells (Fig. 4c, d). Next, a fusion protein of the GGR domain mutation in FUS (GST-FUS-GGR domain) was constructed. A GST pulldown assay indicated that AKT was precipitated with the GST-FUS-GGR peptide (Fig. 4e left) and FUS mainly bound with the N domain of AKT (Fig. 4e right). Then, we analyzed the changes in AKT protein levels after silencing FUS. Knockdown of FUS did not affect AKT expression. Similarly, FUS expression was not affected by silencing AKT (Fig. 4f). However, we observed that FUS influenced the subcellular localization of AKT by promoting AKT nuclear translocation and increased AKT activation in the nucleus (Fig. 4g).

AKT, which is frequently dysregulated in cancer, is a well-established regulator of glucose metabolism [41]. Its regulation on metabolic processes is required for tumor proliferation, apoptosis, and autophagy [42‐44]. Enforcing or silencing LINC00470 expression in GBM cells increased or reduced glycolysis uptake and lactate production, respectively (Fig. 5a, b). Hexokinases catalyze the first and irreversible step of glucose metabolism, i.e., the ATP-dependent phosphorylation of glucose to yield glucose-6-phosphate [45]. Overexpression of LINC00470 increased the total hexokinase activity in U251 cells compared to controls, and HK activity was inhibited after knockdown of LINC00470 in U251 cells (Fig. 5c). HK1 is a major isoform of HK and is the first key enzyme in the glycolysis pathway [45]. Importantly, the protein expression level of HK1 was markedly increased in response to LINC00470 overexpression (Fig. 5d). In contrast, we found that HK2, another major isoform of HK, was not changed statistically significantly in LINC00470-overexpressed cells (Fig. 5d).

Next, we explored the molecular mechanisms underlying the LINC00470 that affects the activity of HK1. Inhibiting the activity of AKT with MK-2206 resulted in downregulated of HK1 (Fig. 5e). Additionally, we transfected both pcDNA3.1 and pcDNA3.1-LINC00470 vectors into U251 cells, then analyzed the protein levels of HK1 in the cytoplasm and nucleus. HK1 expression increased in the cytoplasm under different experimental conditions; concomitantly, HK1 expression in the nucleus did not change significantly (Fig. 5e). To determine how HK1 protein changed, we treated U251 cells with cycloheximide (CHX) and analyzed the stability of HK1 in response to LINC00470 overexpression. The half-life of HK1 was much longer in LINC00470-overexpressed cells than that in controls (Fig. 5f). We further explored the mechanism of AKT-mediated HK1 regulation and found lower HK1 ubiquitination levels in LINC00470-transfected cells treated with MG132, and a restoration experiment was performed by knocking down AKT. We found that the ubiquitination level of HK1 was rescued (Fig. 5g). Together, these observations suggested that LINC00470 affected the ubiquitination and expression of HK1 through activating AKT.

The above data suggested that LINC00470 plays an important role in GBMs. Accordingly, the primary cultured cells were used to evaluate the functions of LINC00470. As shown in Additional file 7: Figure S7, there was relatively low expression of LINC00470 in PG-1 and PG-2 cells and relatively high expression of LINC00470 in PG-3 and PG-4 cells. Therefore, we expected overexpression of LINC00470 in PG-1 and PG-2 cells and knockdown of LINC00470 in PG-3 and PG-4 cells (Fig. 6a). We found overexpression of LINC00470 contributed to the proliferation of PG1 and PG2 cells by CCK-8 assay (Fig. 6b and Additional file 8: Figure S8). Knockdown of LINC00470 in PG-3 and PG-4 cells decreased cell proliferation (Fig. 6b and Additional file 8: Figure S8). Autophagy primarily promoted the progression of cancers [46, 47]. We also found that, in PG-1 cells, overexpression of LINC00470 inhibited the levels of autophagy (Fig. 6c, d), and in PG-3 cells, knockdown of LINC00470 promoted the levels of autophagy (Fig. 6c).

To evaluate whether glycolysis activation serves as an upstream mechanism for LINC00470-mediated autophagy, we monitored the markers of autophagy by knockdown of HK1, FUS, and AKT, respectively. The results showed that when HK1, FUS, or LINC00470 was knocked down, the autophagy level of GBM did not decrease. These results suggested that LINC00470 affected autophagy that was required for HK1, FUS, and AKT (Fig. 6e). At the same time, we applied an intracranial orthotopic transplanted model to evaluate whether LINC00470 mediated GBM tumorigenesis. Compared to mice transplanted with U251-sh-control cells, mice transplanted with U251-sh-LINC00470 cells exhibited longer survival (Fig. 6f), gained more weight (Fig. 6g), and had smaller tumors (Fig. 6h). Knockdown of LINC00470 significantly increased the autophagy levels and decreased the expression of Ki-67 and LINC00470 in an intracranial orthotopic transplanted model (Fig. 6i, j).

To further evaluate the clinical significance of LINC00470 in astrocytomas, including GBMs, we found that the levels of LINC00470 were significantly increased in astrocytoma tissues (n = 60) compared with normal brain tissues (n = 12) by RT-qPCR (Fig. 7a), especially in high-grade astrocytomas (Fig. 7b). We next measured LINC00470 levels in a panel of 75 astrocytoma tissues and 15 normal brain tissues by in situ hybridization (Fig. 7c). The results were consistent with those of RT-qPCR (Fig. 7a).

Subsequently, we conducted a univariate cox regression analysis using clinical variables for astrocytoma patients and found that expression of LINC00470, astrocytoma grade, patients’ age, and the astrocytoma location were statistically associated with overall survival (Table 1). The multivariate cox proportional hazards model indicated that LINC00470 expression and astrocytoma grades were independently associated with overall survival (hazard ratio [HR] = 2.876, P = 0.02; HR = 1.892, P = 0.044; respectively) (Table 2). The results showed that LINC00470 was an independent prognostic factor in astrocytoma patients.

The patients were divided into high or low LINC00470 expression groups according to the ISH scores. Kaplan-Meier analysis of the 75 patients with astrocytoma revealed that high LINC00470 expression levels significantly correlated with shorter survival times (Fig. 7d). High LINC00470 expression was significantly associated with a poor prognosis of astrocytoma patients.