The GSE90603, GSE63319, GSE165937 human glioblastoma datasets were acquired from the GEO database (https://www.ncbi.nlm.nih.gov/geo/). These datasets included miRNA sequencing data from both glioblastoma and normal tissues. The Venn diagram was processed using the Bioinformatics & Evolutionary Genomics (https://bioinformatics.psb.ugent.be/webtools/Venn/). The Volcano Plot was optimized using the OmicStudio tools at https://www.omicstudio.cn/tool. The GSE13030 with clinical information and matching gene expression was selected for survival analysis. The GSE154043 is composed of two subsets: GSE154041 (mRNA sequencing data) and GSE154042 (miRNA sequencing data), which was used to analyze the relationship between miR-137 and XIAP.

Predicted target genes of miR-137 were determined from the union of miRNA target predictions from TargetScan 8.0 (http://www.targetscan.org), PicTar (http://pictar.mdc-berlin.de/cgi-bin/PicTar vertebrate.cgi), PITA (https://genie.weizmann.ac.il/pubs/mir07/mir07_dyn_data.html) and microT (https://dianalab.e-ce.uth.gr/html/dianauniverse/index.php?r=microT_CDS). These genes were classified according to biological process by the PANTHER classification system (http://pantherdb.org/). The TCGA LGG (Brain Lower Grade Glioma) and GBM (Glioblastoma Multiforme) cohorts with the expression level of XIAP were acquired from the Gene Expression Profiling Interactive Analysis (http://gepia.cancer-pku.cn/). In addition, we analyzed the correlation between miR-137 and XIAP in TCGA datasets by using ENCORI (https://starbase.sysu.edu.cn).

Human embryonic kidney cell lines (HEK-293T) and human glioblastoma cell lines (U-87 MG, U251) were purchased from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). All the cell lines were maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco) supplemented with 10% fetal calf serum (FCS) (BIOIND, Kibbutz, Israel). All the cells were incubated in a humidified atmosphere of 5% CO₂ in air at 37°C.

Cells were seeded in 6-well plates a day before transfection. The transfection was performed when they were at 50–70% confluence. All transfections were performed using Lipofectamine 2000 reagent (Invitrogen) following the manufacturer’s instructions. Analyses on recipient cells or further research were performed 48 hours after transfection.

For Lentivirus packaging, HEK-293T cells were seeded in 60-mm dishes before transfection, then it was performed using a transient co-transfection system with 1 μg pMD2G, 3 μg psPAX2 and 4 μg pLenti6.3-Luciferase/miR-137. For lentiviral transduction, U-87 MG was plated on 24-well plates and allowed to attach overnight. The medium was replaced with 500 μL fresh lentiviral supernatant (LV-Luciferase or LV-miR-137) and 8 mg/mL polybrene (Sigma-Aldrich, St. Louis, MO, USA). The polybrene was used to assist the uptake of viral particles. After infection, cells were screened using 5 μg/mL blasticidin (Sigma-Aldrich, St. Louis, MO, USA) for 7 days and were used for the following experiments. The primer and oligo sequences are listed in Table 1.

Six-week-old male Balb/c nude mice were purchased from the Vital River Laboratories (Beijing, China). All animal procedures were approved by the Laboratory Animal Welfare and Ethics Committee of Fourth Military Medical University (FMMU). Approximately 1×10⁷ U-87 MG cells which stably expressing Luciferase or miR-137 and suspended in 200 μL PBS, were injected subcutaneously into the left flank of each mouse, twelve mice per group. Tumor volume was monitored using a vernier caliper once every three days. Once the tumors reached 200 mm³ as calculated by (π × length × width²)/6, each group were divided into two groups. Liposomes were packed with pLenti6.3-Luciferase or pLenti6.3-TRAIL according to co-incubation. Then they were injected into the tumor multi-directionally once every four days, meanwhile the tumor volume was measured. Mice were killed by spinal dislocation when they were treated five times. Then tumors were removed and snap-frozen in liquid nitrogen for further research or fixed in formalin for immunohistochemical analysis.

Total RNA was extracted from cultured cells using TRIzol reagent (Invitrogen) according to manufacturer’s instructions. 1 μg of RNA was employed to synthesize cDNA using the PrimeScript RT Reagent Kit Perfect Real Time (TaKaRa, Dalian, China) or the miScript II RT Kit (Qiagen, Hilden, Germany). For the detection of target gene mRNA levels, we employed the Fluorescent qRT-PCR using GAPDH as an internal control. The expression levels of miR-137 in glioblastoma cells and tissues were examined by qRT-PCR assays and normalized to the expression of U6 snRNA (sn-RNU6). All the primers were obtained from AuGCT (Beijing, China) and the reactions were run in triplicates. Relative expression levels of miRNA or mRNA were acquired and analyzed using the Bio-Rad C1000 Thermal Cycler (Bio-Rad, California, CA, USA). The primer sequences are listed in Table 1 and Supplementary Table S1.

HEK-293T cells were seeded in 48-well plates 24 h before transfection. Then cells were transfected with a mixture of 100 ng pGL3-XIAP-3’UTR, 20 μM miR-137 mimics or negative control (miR-NC), and 5 ng PRL-TK using Lipofectamine 2000 reagent and performed in three independent experiments. Firefly and Renilla luciferase activities were examined 48 h after transfection using a dual-luciferase reporter system (Promega, Madison, WI, USA).

Cells were trypsinized 48 h after transfection, and Western blot analysis was performed as previously described (11). The antibodies were used as follows: XIAP (CST, cat#14334S, Danvers, MA, USA), Caspase 3 (Proteintech, cat#19677-1-AP, Rosemont, IL, USA) and GAPDH (Proteintech, cat#10494-1-AP, Rosemont, IL, USA). The caspase 3 antibody can be used to analyze both the completed and cleaved forms of caspase-3. The intensity analysis of western blotting was quantified by Image J software. The protein expression data represents relative XIAP intensity from three separate experiments, and normalized to that of GAPDH. Standard IHC staining was performed as previously described (11). In brief, following deparaffinization and rehydration, the slides were incubated with anti-XIAP antibody (CST, cat#14334S, Danvers, MA, USA, 1:100). Images were captured with a microscope (BX51 Olympus, Tokyo, Japan) and processed with identical settings.

Cells were plated at 2.0×10³ cells per well in 96-well plates 24 h after transfection and allowed to adhere overnight prior to drug treatments. Cells were then treated with TRAIL (300 ng/mL) (Peprotech, Cranbury, NJ, USA), cell counting kit-8 assay (CCK-8) was performed (at 0, 24, 48, 72 h) to analyze cell density. For detection of caspase-3/7 activity, cells were treated with TRAIL for 24h and analyzed by using Caspase-Glo Assay kits (Promega, Madison, WI, USA) according to the manufacturer’s instructions. When using Z-VAD (10 μM) (Selleck, Houston, TX, USA), Necrosulfonamide/NSA (5 μM) (Selleck, Houston, TX, USA) and Necrostatin 1S/Nec-1S (20 μM) (MCE, Monmouth Junction, NJ, USA) to differentiate apoptotic and necroptotic cell death, cells were plated at 4.0×10³ cells per well in 96-well plates and cell density was measured 24 h after the treatment with those agents.

Cells were transfected with different mimics or genes 24 h before TRAIL treatment, then they were collected and suspended in PBS after 48 h TRAIL treatment. The obtained cells were stained with FITC Annexin-V Apoptosis Detection Kit I (BD Biosciences, Franklin Lakes, NJ, USA) and were analyzed by flow cytometry (FCM). Meanwhile, Cell apoptosis in xenograft tumor tissues was detected by TUNEL staining using an In Situ Cell Death Detection Kit, Fluorescein (Roche, Mannheim, Germany) according to the manufacturer’s protocol. The nuclei were then counterstained with DAPI (Sigma-Aldrich, St. Louis, MO, USA). Subsequently, samples were subjected to laser scanning confocal microscopy analysis (Nikon, Tokyo, Japan).

The glioblastoma samples (n = 10) and paired adjacent normal tissues (n = 10) were obtained from the department of Neurosurgery at Xijing Hospital in the Fourth Military Medical University (FMMU). Collection of clinical samples was approved by the Medical Ethics Committee of Xijing Hospital. All specimens were promptly snap-frozen using liquid nitrogen and stored at -80°C. The tissues were used for RNA isolation when needed and were used to analyze the expression level of miR-137 by using qRT-PCR assay.

Statistical analyses were performed using GraphPad Prism version 8.00 (GraphPad Inc., USA). All the data were exhibited as the mean ± SD of at least three independent experiments. The differences between groups were analyzed using two-sided Student’s t-test or two-way ANOVA, and a p value < 0.05 was considered statistically significant.

Given miRNA dysregulation frequently occurs in glioblastoma, we wondered to identify the functional miRNA(s) associated with progression of GBM. By combined utilization of bioinformatic analysis of raw data from a serial of GEO datasets, we identified twelve downregulated miRNAs in GBM tissues compared with normal brain tissues (Figure 1A). Among all of them, miR-129-5P, miR-139, miR-330, etc. have been reported to function as tumor suppressors in GBM previously (12–14). These evidences further confirmed the reliability of our screening strategy. In these selected miRNAs, miR-137 is one of top ten differential expressed miRNAs in all tested GEO datasets (Figure 1B). Accordingly, we focused on miR-137 for further analysis. We first analyzed the expression levels of miR-137 in the three GEO datasets, the results showed that the miR-137 expression in GBM tissues is significantly lower than that in normal tissues (Figure 1C). To further confirm this expression pattern, we collected fresh tissues and performed qRT-PCR to examine the expression level of miR-137 in GBM tissues and adjacent normal tissues from the GBM patients. The data showed a significant downregulation of miR-137 in cancer tissues compared with the paired normal tissues (Figure 1D). Subsequently, in order to test whether miR-137 can be a potential clinical prognostic indicator, we analyzed the relationship between the miR-137 expression and the survival rate in clinic. The result showed that the higher level of miR-137 is closely associated with the favorable prognosis in clinic (Figures 1E). These results indicated that miR-137 might be a potential tumor suppressor in glioblastoma.

It has been reported previously that GBM cells differentially respond to TRAIL-induced cell death (11, 15). Given that miR-137 plays a pivotal role in neuronal differentiation and downregulates in glioma genesis, we wondered to determine the biological role of miR-137 on the sensitivity of TRAIL-induced cell death in GBM cells. We transiently transfected TRAIL-low responsive GBM cell lines (U-87MG and U251 cells) with miR-137 mimics and negative control (miR-NC). MiR-137 was overexpressed in the two GBM cell lines respectively (Figure 1F). Then the transfected cells were incubated with the TRAIL recombinant proteins and cell density were measured. The synergistic inhibitory effects for miR-137 in combination with TRAIL were observed in the two cell lines, which suggested that TRAIL can further induce GBM cell death in the presence of miR-137 (Figure 1G).

To clarify the types of cell death that TRAIL induced in combination with miR-137, we assessed the effects of pan-caspase (Z-VAD), RIPK1 (Necrostatin 1S) and MLKL (Necrosulfonamide) inhibitors on cell density in the two GBM cell lines. All the two cell lines demonstrated rescue from inhibitory effects of miR-137 and TRAIL with pan-caspase inhibitors Z-VAD (Figure 2A). These suggested that GBM cell sensitivity to miR-137 with TRAIL is predominantly mediated by caspase-dependent apoptotic cell death. Then cells were assessed by the Annexin-V staining. Flow cytometry analysis revealed that overexpression of miR-137 raises the proportion of apoptotic cells evidently with regard to their respective controls (Figures 2B, C). The pro-apoptotic effects were also confirmed through caspase-3/7 activity and protein levels of caspase-3 (Figures 2D, E). Furthermore, the miR-137 inhibitor was introduced into U-87 MG and U251 cells, respectively (Figures 2F). As expected, knockdown of the endogenous miR-137 decreases the effects which were induced by TRAIL in GBM cells (Figures 2G–I). These findings suggested that overexpression of miR-137 sensitizes GBM cells to TRAIL-induced apoptosis.

To identify the downstream target(s) of miR-137 which can regulate sensitivity of TRAIL-induced cell death in the GBM cells, we firstly used four prediction websites (PITA, TargetScan, PicTar and microT) to analyze the putative target genes of miR-137 (Figure 3A). The predicted target genes from the algorithms were further analyzed according to the PANTHER classification system. Twelve genes were clustered in the apoptosis signaling pathway (Figure 3B). The reverse relationship between miR-137 and these genes were analyzed in TCGA database. Among them, we focused on XIAP (X-chromosome Inhibitory apoptotic protein), which functionally antagonizes apoptotic stimulation in transformed cells. More importantly, we found that miR-137 expression level is inversely correlated with XIAP expression in five types of cancers (Figure 3C). Using qRT-PCR, we confirmed that the mRNA level of XIAP is downregulated in miR-137-overexpressed U251 cells (Figure 3D). Additionally, we found that the expression level of XIAP significantly decreases in LGG and GBM samples compared with the control in the TCGA database (Figure 3E). These data suggested that XIAP is suppressed by mir-137 in GBM cells, and it may be a target of miR-137.

According to the silicon analysis, two predicted binding sites of miR-137 present on the 3’UTR region of XIAP (Figure 4A). To determine whether miR-137 recognizes these two putative sites (2905–2911 bp and 3720–3726 bp), we constructed luciferase reporter plasmids containing wild-type or target site 1- or site 2-mutant 3’UTR of XIAP. The indicated plasmids were co-transfected into HEK-293T cells with miR-NC or miR-137 mimics. The transfection efficiency was normalized by co-transfection with renilla reporter vector. Overexpression of miR-137 significantly decreased the normalized luciferase activity of wild-type XIAP 3’UTR, but had slight effect on any mutation of XIAP 3’UTR (Figure 4B). Subsequently, we attempted to validate whether miR-137 could downregulate the expression level of XIAP in GBM cells. Using qRT-PCR analysis, we found that the mRNA level of XIAP significantly decreases when the miR-137 was overexpressed in the GBM cells (Figure 4C). Consistent with the reduction of the XIAP mRNA level, Western blot analysis also showed that enforced expression of miR-137 dramatically reduced the XIAP protein level (Figure 4D). Meanwhile, the miR-137 inhibitor was introduced into U-87 MG and U251 cells, respectively. As expected, knockdown of the endogenous miR-137 increased XIAP mRNA and protein levels in GBM cells (Figures 4E, F).

To further determine the correlation between the expression levels of miR-137 and XIAP, we acquired the miRNA and mRNA sequencing data of GBM samples from the GEO database. The expression level of miR-137 is reversely correlated with the expression of XIAP (Figure 5A). More strikingly, by mining the data from TCGA, we also found the negative correlation between miR-137 and XIAP in multiple types of cancers, including breast cancer (BRCA), bladder urothelial cancer (BLCA), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), stomach adenocarcinoma (STAD) and head and neck squamous cell carcinoma (HNSC) (Figure 5B). All the data showed that there is a general negative correlation between miR-137 and XIAP in cancer samples. Thus, these results indicated that XIAP is a bona fide target of miR-137 in GBM cells, or even in other types of cancers.

Given that XIAP is a negative regulator of apoptosis in cancer cells and is a direct target of miR-137 in GBM cells, we hypothesized that a miR-137-XIAP axis could play a functional role in regulating sensitivity of GBM cells to TRAIL-induced cell death. To test the hypothesis, we simultaneously transfected both XIAP-expressing plasmid and miR-137 mimics into GBM cells. Western blot and qRT-PCR analyses validated the rescued expression of exogenous XIAP in GBM cells transfected with both miR-137 and XIAP-expressing plasmid (Figures 6A, B). Furthermore, cell density and Annexin-V staining assay showed that the exogenously expressed XIAP abrogated the miR-137-triggerred TRAIL sensitivity in GBM cells (Figures 6C–E). Meanwhile, caspase-3/7 assay also confirmed the effects (Figure 6F). The results demonstrated that XIAP is potential negative regulator for mir-137-mediated TRAIL sensitivity in GBM cells in vitro.

To determine the therapeutic benefit of combined utilization of miR-137 and TRAIL in vivo, we first investigated the inhibitory effect of miR-137 on the expression of XIAP in U-87 MG cell-derived xenografts. The results of IHC staining showed that overexpression of miR-137 significantly inhibited endogenous expression of XIAP in vivo (Figure 7A). Furthermore, in the established xenograft mouse model, we used liposome-capsuled TRAIL-expressing plasmid to perform a multipoint injection on tumor tissues in miR-137-overexpressed or control xenografts. Each mouse was administrated liposome-capsuled TRAIL-expressed or control plasmids for five times every four days. Although monotherapy with miR-137 overexpression or TRAIL administration led to anti-tumor effect in vivo, combined utilization of miR-137 and TRAIL had a more significant effect on regression of tumor growth (Figures 7B, C). Subsequently, the combined effect was also confirmed by the number of apoptotic cancer cells using TUNEL staining assay (Figures 7D, E). Taken together, the above results further demonstrated that overexpression of miR-137 could be a potential strategy to improve the efficacy of TRAIL-based therapeutics in GBM (Figure 8).