Exosomal transfer of long non-coding RNA SBF2-AS1 enhances chemoresistance to temozolomide in glioblastoma

See Additional file: Supplementary materials and methods for details on patient samples information. The study was approved by the Ethics Committee of Nanjing Medical University and written informed consent was obtained from all patients. 1

Five human GBM cell lines (U87, LN229, A172, T98, U251) and human embryonic kidney (HEK) 293 T cells were purchased from the Chinese Academy of Sciences Cell Bank (Shanghai, China), N3 primary culture cell were donated from Tian Tan Hospital and four drug-related cell lines (Pri GBM, N3S, Rec GBM, N3T3rd) were used from our previous study [13]. All GBM cells were maintained in high glucose DMEM medium supplemented with antibiotics (100 units/mL penicillin and 100 mg/mL streptomycin) and 10% fetal bovine serum (FBS). Normal human astrocytes (NHAs) were acquired from Sciencell Research Laboratories (Carlsbad, CA, USA) and cultured in astrocyte medium (Carlsbad, CA, USA). All cells were grown at 37 °C with 5% CO₂.

Exosomes were extracted from GBM cell culture through standard centrifugation steps as previously described [14]. See Additional file 1: Supplementary materials and methods for details on exosome isolation.

See Additional file: Supplementary materials and methods. 1

Western blot analysis was done as described previously [15]. Antibodies against caspase-3, ZEB1 (Cell Signaling Technology, USA); CD63, CD81, GM130, XRCC4 and γ-H2AX (Abcam, USA); and β-actin, GAPDH and tubulin (Santa Cruz, USA) were used for Western blot analysis.

See Additional file: Supplementary materials and methods for details on transfection and stable cell establishment. 1

Fish analysis was performed on human tissue and GBM cells using the method described previously [16]. IHC was performed on mice xenogeneic tumor tissue and human tissue as described previously [16].

The flow cytometric assay was performed as described previously [13]. The apoptotic cells were stained with Annexin V-FITC/PI and assessed by fluorescence activated cell sorting (FACS).

The colony formation assay was performed following the method described in our earlier study [17]. Cells were independently seeded onto 6-well plates followed by treatment with TMZ (200 μM, 24 h). Then the culture media were replaced by TMZ-excluded media. After 10–15 days of growth, cells were fixed with 4% paraformaldehyde and stained with crystal violet.

See Additional file: Supplementary materials and methods for details on immunofluorescent staining. 1

GBM cells were fixed in 4% paraformaldehyde for 15 min and stained with In Situ Cell Death Detection Kit, POD (Roche, Switzerland) according to the manufacturer’s instructions. Tunel-positive cells were observed with a Nikon ECLIPSE E800 fluorescence microscope.

See Additional file: Supplementary materials and methods for details on luciferase assay. 1

Chip assay was performed using Simple Chip Enzymatic Chromatin IP kit (Cell Signaling Technology, USA) according to the manufacturer’s instructions. Briefly, the chromatin of the GBM cells were treated with 3 μg anti-ZEB1 (Abcam). Immunoprecipitated DNA was subjected to RT-PCR using specific primers.

Comet assay was performed as described preciously [13]. Briefly, GBM cells that received TMZ treatment (200 μM) were analyzed by DNA Damage Detection Kit (KeyGEN BioTECH, China). Agarose-embedded cells were lysed and examined by agarose gel electrophoresis.

See Additional file: Supplementary materials and methods for details on Subcutaneous and orthotopic xenograft studies. All animal experiments were approved by the Institutional Committee for Animal Research of Nanjing Medical University and conducted in conformation with the Animal Welfare Act. 1

All statistical analyses were done using GraphPad Prism 5 Software (La Jolla, CA, USA). All values were shown as mean ± Standard Error of Mean (SEM) of three independent experiments. Student’s t-test or ANOVA was performed to evaluate pairwise comparison or multivariate analysis. All statistical analyses were performed using SPSS statistics 22 software. P value < 0.05 was judged to be statistically significant.

Our previous experimental results, have demonstrated that miR-151a-3p promotes TMZ sensitivity in GBM cells [18]. Recent studies have shown that lncRNAs can act as sponges or function as competing endogenous RNA (ceRNA) to regulate miRNA expression [19, 20]. Thus, we considered whether miR-151a-3p is regulated by lncRNA. We used the bioinformatic website (DIANA TOOLS, http://diana.imis.athena-innovation.gr/↗) to find potential lncRNAs that could regulate miR-151a-3p. We chose the top 10 lncRNAs predicted in the website and verified their expression levels in N3 GBM cells by qRT-PCR (Additional file 2 Figure S1A). The result showed that five lncRNAs (RP11-111F5.4, PSMD-AS1, RP11-967 K21.1, SBF2-AS1, ARHGEF26-AS1) had higher expression in N3 GBM cells. We then used the dual luciferase reporter assays to verify the prediction analysis. HEK293T cells were transfected with luciferase plasmid containing the sequence of miR-151a-3p together with plasmids carrying the lncRNAs or control sequence. We found that lncSBF2-AS1 had the strongest ability to suppress miR-151a-3p-driven luciferase activity (Additional file 2 Figure S1B). Meanwhile, Knockdown of SBF2-AS1 in N3 GBM cells had the highest caspase-3 activity (Additional file 2 Figure S1C). Then we examined the levels of SBF2-AS1 in normal human astrocytes (NHA) and five GBM cell lines (U87, LN229, A172, T98, U251) using qRT-PCR. All GBM cells were found to highly express SBF2-AS1 in comparison with NHAs (Additional file 2 Figure S1D). Next, we analyzed lncSBF2-AS1 in TCGA database of GBM and found that lncSBF2-AS1 level was higher in GBM than low-grade gliomas (LGG), and high SBF2-AS1 expression correlated with poor survival in GBM patients (Additional file 2 Figure S1E and F). Thus, we considered lncSBF2-AS1 as the upstream of miR-151a-3p and lncSBF2-AS1 was chosen as a candidate for further investigation.

To confirmed the effect of SBF2-AS1 on chemoresistance, we first analyzed TMZ-resistance and TMZ-sensitive GBM cell viability in response to TMZ (Fig. 1a and b). The levels of SBF-AS1 in TMZ-sensitive cell lines and TMZ-resistant cell lines, analyzed by qRT-PCR, showed that TMZ resistant cells had higher SBF2-AS1 levels (Fig. 1c). Then we measured SBF2-AS1 levels in 20 primary GBMs and 20 recurrent GBMs by qRT-PCR (Fig. 1d). Fluorescence in situ hybridization (FISH) assays employing SBF2-AS1 probe showed that recurrent GBM tissues displayed higher level of SBF2-AS1, whereas primary GBM tissues exhibited lower SBF2-AS1 signal (Fig. 1e). Together, these results suggest that lncSBF2-AS1 is upregulated in TMZ-resistant GBM cell lines and tissues, pointing towards a possible correlation between SBF2-AS1 and TMZ resistance.

To determine whether lncSBF2-AS1 is associated with TMZ resistance, we overexpressed SBF2-AS1 in N3S and Pri GBM cells (TMZ sensitive cell line, Fig. 1a and b and Fig. 2a). We found that Pri GBM and N3T3rd cells transfected with vector control (VC) showed a significant decrease in colony-forming ability after TMZ treatment, whereas overexpression of SBF2-AS1 in Pri GBM and N3S cells rescued the colony-forming capacity of TMZ (Fig. 2b). We also tested apoptotic rate of TMZ-sensitive cells overexpressing SBF2-AS1 or VC cells in response to TMZ. As shown in Fig. 2c, d and e, a marked decrease in number of apoptotic cells with a co-dependent reduction in apoptosis related protein, cleaved caspase-3 [21], was observed in N3S and Pri GBM cells overexpressing SBF2-AS1 compared with VC cells. These findings indicate that SBF2-AS1 overexpression reverses TMZ-induced apoptosis and growth inhibition in TMZ sensitive GBM cells.

To examine whether lncSBF2-AS1 confers TMZ resistance in vivo, Pri GBM cells expressing SBF2-AS1 or VC was subcutaneously injected into immunocompromised mice. After GBM engraftment was confirmed, we treated mice with TMZ (66 mg/kg/day, 5 days per week for three cycles) or DMSO. We observed that mice injected with Pri GBM-VC had a marked decrease in tumor volume and weight in response to TMZ treatment, whereas xenografts overexpressing SBF2-AS1 had unconspicuous effects on tumor growth when treated with TMZ (Fig. 2f, g and h). IHC staining was performed to verify the expression of cleaved caspase-3. TMZ treated xenografts overexpressing SBF2-AS1 had a significantly decreased level of cleaved caspase-3 compared with control xenografts (Fig. 2i). Hence, our data indicated that overexpression of SBF2-AS1 can lead to the development of acquired TMZ resistance.

According to the above findings, we determined whether depletion of SBF2-AS1 could cause sensitization to TMZ in GBM cells. We transfected TMZ-refractory recurrent GBM cells and N3T3rd cells (TMZ resistant cell line, Fig. 1a and b) with three separate luciferase-encoding SBF2-AS1 shRNA or control shRNA. The expression of SBF2-AS1 in recurrent GBM cells and N3T3rd cells was analyzed by qRT-PCR (Fig. 3a). According to the knockdown efficiency, we chose shSBF2-AS1–2 for subsequent experiments. SBF2-AS1-knockdown recurrent GBM cells and N3T3rd cells showed a significant decrease in colony-forming ability (Fig. 3b) and the delay in apoptosis was remarkably increased (Fig. 3c, d and e).

Then we investigated the role of SBF2-AS1 deletion on TMZ resistance in vivo. SBF2-AS1-depleted Rec GBM cells or control cells were inoculated subcutaneously into immunocompromised mice. The tumor-bearing mice were given DMSO or TMZ treatment (66 mg/kg/day). As we expected, the growth ability of tumor generated by shSBF2-AS1-expressing GBM cells were significantly lower than those produced by control GBM cells after exposure to TMZ (Fig. 3f, g and h). We also used IHC to measure cleaved caspase-3. SBF2-AS1-knockdown xenografts had an obviously high expression level of cleaved caspase-3 in comparison with control xenografts in response to TMZ (Fig. 3i). These results confirm that inhibition of lncSBF2-AS1 endows TMZ-refractory recurrent GBM cells with chemotherapy sensitivity in vitro and in vivo.

Several studies have shown that transcription factors can bind to specific regions of promoters to regulate gene expression [22]. Therefore, we explored if transcription factors can regulate SBF2-AS1. Potential transcription factors were predicted using the JASPAR database (http://jaspar.binf.ku.dk/↗), and luciferase reporter assays were performed to determine the most interactive transcription factor from the top five predictions. Briefly, HEK293T cells were transfected with luciferase plasmid containing the sequence of the SBF2-AS1 promoter region together with plasmids containing the transcription factors or control sequence. The results showed that ZEB1 revealed maximum luciferase activity (Additional file 3 Figure S2A). Then we analyzed ZEB1 expression in TMZ-resistance cells and TMZ-sensitive cells (Additional file 3 Figure S2B). Based on luciferase activities and ZEB1 protein levels in GBM cells, we chose ZEB1 as the object for further study. ZEB1(Zinc finger E-box binding homeobox 1), the key factor of EMT (Epithelial-mesenchymal transition), have been reported as a important molecule for GBM chemoresistance and recurrence [23]. We selected 20 recurrent GBM tissues and correlation analysis between ZEB1 and SBF2-AS1 indicated that the expression level of SBF2-AS1 is correlated with that of ZEB1 (Fig. 4a). To further confirm our hypothesis, we knocked down or overexpressed ZEB1 in Rec GBM or Pri GBM cells, respectively (Fig. 4b), and detected SBF2-AS1 level by qRT-PCR (Fig. 4c). The results showed that SBF2-AS1 level was positively correlated with ZEB1 level. By scanning the promoter region of SBF2-AS1 with the JASPER database, we found four putative ZEB1 binding sites (Fig. 4d). Chromatin immunoprecipitation assay (CHIP) was performed to elucidate the binding of ZEB1 and SBF2-AS1 promoter region. We observed PCR products in the ZEB1 immunoprecipitation in the Site 2 group, which confirmed that ZEB1 bound to SBF2-AS1 promoter at the − 684 to − 676 bp region (Fig. 4e). These data indicated that transcription factor ZEB1 could regulate SBF2-AS1 at transcription level.

In addition, we checked the relationship between ZEB1 and SBF2-AS1 in vitro. The apoptosis of cells was significantly decreased in ZEB1-overexpressed Pri GBM cells as opposed to ZEB1-knockdown Rec GBM cells (Fig. 4f, g and h). These findings indicated that ZEB1 could regulate the level of SBF2-AS1 and was involved in SBF2-AS1-mediated apoptosis of GBM cells.

As mentioned above, there might be a connection between SBF2-AS1 and miR-151a-3p. Increasing evidence shows that lncRNAs can function as ceRNAs for miRNAs or interact with RNA-binding proteins to regulate target gene expression [24, 25]. To further evaluate the molecular mechanism by which SBF2-AS1 promotes TMZ resistance in GBM cells, we firstly used FISH and subcellular fractionation to analyze its distribution. The results showed that SBF2-AS1 was primarily observed in the cytoplasm (Fig. 5a and Additional file 4 Figure S3A), indicating that SBF2-AS1 might function as a ceRNA to regulate target expression. To confirm that miR-151a-3p binds directly to SBF2-AS1, we set up a dual luciferase reporter assay. The observed miR-151a-3p-mediated suppression of luciferase activity was inhibited by mutation (MUT) of SBF2-AS1 (Fig. 5b). Moreover, RNA-binding protein immunoprecipitation (RIP) assay showed that in control GBM cells, the amount of lncSBF2-AS1 and miR-151a-3p immunoprecipitated with Ago2 was higher than respective IgG group, and knockdown of miR-151a-3p restrained the enrichment of SBF2-AS1 and miR-151a-3p in Ago2 precipitates (Fig. 5c and d). Then, we analyzed the correlation between lncSBF2-AS1 and miR-151a-3p levels in 20 pairs of recurrent GBM cancer tissues by using qRT-PCR. The result showed an inverse correlation between lncSBF2-AS1 and miR-151a-3p levels (Additional file 4 Figure S3B). Collectively, these data demonstrate that SBF2-AS1 sponges miR-151a-3p by acting as a ceRNA.

To investigate whether miR-151a-3p participated in SBF2-AS1-mediated apoptosis of GBM cells, miR-151a-3p was knocked down or overexpressed in SBF2-AS1-depleted GBM cells. Then the apoptosis of GBM cells in response to TMZ was determined. The results showed that overexpression of miR-151a-3p promoted the suppressive effect on TMZ resistance, which was induced by SBF2-AS1 knockdown in GBM cells, while inhibition of miR-151a-3p reversed the SBF2-AS1 knockdown-mediated apoptosis in GBM cells in response to TMZ (Fig. 5e and f). Together, these results showed an involvement of miR-151a-3p in lncSBF2-AS1-mediated TMZ resistance in GBM cells.

As lncSBF2-AS1 can sponge miR-151a-3p and XRCC4 is a miR-151a-3p target gene [13], we determined whether lncSBF2-AS1 can regulate the expression of XRCC4 by sponging miR-151a-3p. We found knockdown of lncSBF2-AS1 also reduced XRCC4 protein levels in N3T3rd and recurrent GBM cells (Fig. 6a). To determine whether miR-151a-3p plays an important role between lncSBF2-AS1 and XRCC4, we co-transfected miR-151a-3p inhibitor and shSBF2-AS1 into recurrent GBM cells. Indeed, the suppression of XRCC4 protein levels induced by shSBF2-AS1 was effectively reversed by the miR-151a-3p inhibitor (Fig. 6b). Then, we analyzed the correlation between lncSBF2-AS1 and XRCC4 expression in 20 recurrent GBM cancer tissues. The result showed a positive correlation between lncSBF2-AS1 and XRCC4 expression, consistent with the existence of an lncSBF2-AS1-miR-151a-3p-XRCC4 regulatory axis (Fig. 6c). Together, these data suggested that lncSBF2-AS1 regulates the expression of XRCC4 by post-transcriptional modulation of miR-151a-3p.

Recent studies have reported that TMZ resistance can be promoted by DNA double-strand break (DSB) repair [26–28] and lncRNAs play an essential role in DSB [29, 30]. Thus, we wondered whether improved DSB repair is directly dependent on lncSBF2-AS1. A sensitive marker of DSBs, γ-H2AX [31], was assessed by confocal microscopy. We found that although the average number of γ-H2AX focis were slightly different in control shRNA (shctrl) and shSBF2-AS1 groups, the number of γ-H2AX focis in shSBF2-AS1 cells showed a significant increase within 8 h after TMZ removal, whereas shctrl cells had a constantly low number of γ-H2AX (Fig. 6d), even at 12 h. We next employed comet assay to survey DNA repair progression. The tail was still detectable in SBF2-AS1-depleted GBM cells 12 h after TMZ withdrawal. Meanwhile, the process of DSB clearance was completed within 12-h after TMZ removal in control cells (Fig. 6e). Furthermore, using Western blot assay, we found persistence of high expression level of γ-H2AX in GBM cells lacking SBF2-AS1, but not control cells 12 h after the end of TMZ treatment (Fig. 6f). These data showed that depletion of lncSBF2-AS1 delayed the repair process of TMZ-induced DNA damage.

Cell-secreted exosomes and their contents can be internalized by surrounding cells [32, 33]. Nucleotide is one of the inclusions of exosomes [34] and lncRNAs transferred by exosomes promote drug resistance [35, 36]. To investigate the manner of SBF2-AS1 intercellular delivery, the level of SBF2-AS1 was analyzed by qRT-PCR. We found that the level of SBF2-AS1 in the culture medium showed only a slight change upon RNase treatment but markedly decreased after treated with RNase and Triton X-100 synchronously (Fig. 7a), indicating that extracellular SBF2-AS1 was mainly packed in membranes. Then we isolated exosomes derived from TMZ-resistant glioma cells and parental cells. The characteristics were analyzed, and quantification was done using transmission electron microscopy (TEM), nanoparticle tracking analysis (NTA) and Western blot analysis. A typical lipid bilayer membrane was observed, and the size of exosomes were confirmed by TEM and NTA (Fig. 7b and c). A Western blot analysis showed the proteins CD63 and CD81 were positively expressed in exosomes and GM130 was mainly present in the cell lysate [37] (Fig. 7d).

Next, we determined whether the exosomes of TMZ-resistant cells (Rec GBM and N3T3rd cells) transferred SBF2-AS1 to parental cells (Pri GBM and N3S cells). Rec GBM and N3T3rd cells were co-transfected with fluorescein amidite (FAM)-tagged SBF2-AS1 and exosomal marker CD63. The exosomes were isolated from culture medium collected from Rec GBM and N3T3rd cells. Parental cells were treated with purified exosomes. SBF2-AS1 wrapped by exosomes and internalization of exosomes were observed by confocal microscopy (Fig. 7e). To further confirm the effect of exosomal lncSBF2-AS1, we knocked down SBF2-AS1 in TMZ-resistant cells and the level of lncSBF2-AS1 in exosomes echoed (Additional file 5 Figure S4A). Then the exosomes were used to treat parental cells and we found that the apoptosis of parental cells, treated with Rec GBM and N3T3rd exosomes showed a significant decrease after TMZ treatment, however, knockdown of exosomal SBF2-AS1 reversed the apoptosis of GBM cells (Fig. 7f, g and h). Meanwhile, the lncSBF2-AS1 transfer of TMZ-resistant exosomes downregulated XRCC4 expression in Pri GBM and N3S cells (Additional file 5 Figure S4B), and exosomal lncSBF2-AS1 accelerated the DNA damage repair in GBM recipient cells after TMZ treatment (Additional file 5 Figure S4C and D). Thus, our data indicated that exosomal lncSBF2-AS1 could enhance chemoresistance to TMZ in vitro.

To investigate whether exosomal lncSBF2-AS1 is necessary for TMZ resistance in vivo, mice were orthotopically injected with 2.5 × 10⁵ luciferase-labeled Pri GBM cells following pre-incubation with PBS, Rec GBM-exo or Rec GBM-exo (shSBF2-AS1) for 2 days. The tumor-bearing mice were exposed to DMSO or TMZ (66 mg/kg/day) for 5 days each week for three cycles and analyzed by bioluminescence imaging. Control xenografts showed tumor progression. We found that xenografts carrying Rec exosomes treated Pri GBM cells displayed a significant advance of tumor growth in response to TMZ. Meanwhile, SBF2-AS1-depleted Rec exosomes treated Pri GBM cells demonstrated a regression in tumor progression after TMZ treatment (Fig. 8a). The findings were also confirmed by the survival curves (Fig. 8b), bioluminescence assay (Fig. 8c) and TUNEL staining (Fig. 8d). Moreover, we performed FISH and IHC on xenograft tissues receiving TMZ therapy. Tumors with high SBF2-AS1 levels (Rec-exo treated) tended to express lower miR-151a-3p and lower protein levels of cleaved caspase-3 and γ-H2AX and higher protein level of XRCC4 (Fig. 8e). Higher protein levels of cleaved caspase-3 and γ-H2AX and higher miR-151a-3p level together with lower XRCC4 level was observed in low SBF2-AS1 groups (PBS treated and Rec GBM-exo (shSBF2-AS1) treated) (Fig. 8e). These data indicate that exosomal lncSBF2-AS1 promoted development of acquired resistance to TMZ in vivo.

Exosomal RNAs can be transferred to neighboring cells or cells in other organs through the circulation system [38]. Furthermore, exosomal lncRNA has been considered as a potential biomarker of tumor chemoresistance [39]. Thus, exosomal lncSBF2-AS1 secreted by recurrent GBM cells may be detected in circulation. To confirm our assumption, we extracted serum exosomes collected from recurrent GBM patients and compared them with primary GBM patients (patient information are listed in Additional file 6: Table S3). Purified exosomes were authenticated by TEM analysis and Western blotting analysis (Fig. 9a and b). qRT-PCR assay was performed, and we found that exosomal lncSBF2-AS1 was enriched in the serum of recurrent GBM patients instead of primary GBM patients (Fig. 9c). Recurrent GBM patients with higher serum exosomal SBF2-AS1 levels had worse prognosis, forecasting a poor response to TMZ treatment (Fig. 9d). IHC staining was performed to evaluate the correlation between exosomal lncSBF2-AS1 and apoptosis marker cleaved caspase-3 in recurrent GBM patients. The results showed that exosomal lncSBF2-AS1 levels in the serum of recurrent GBM patients were positively correlated with cleaved caspase-3 expression (Fig. 9e). Altogether, our data indicates that exosomal lncSBF2-AS1 may serve as a promising diagnostic biomarker for recurrent GBM patients.

This work was supported by National Nature Science Foundation of China (No. 81402056 and 81472362) and National Key Reserch and Development Plan (No. 2016YFC0902500).

All data generated or analyzed during this study are included either in this article or in the additional files.

The study was approved by the Ethics Committee of Nanjing Medical University and written informed consent was obtained from all patients.

We have obtained consents to publish this paper from all the participants.

The authors declare that they have no competing interest.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

All data generated or analyzed during this study are included either in this article or in the additional files.