Tumor-suppressive role of miR-216b impacts cell cycle regulation in high-grade glioma

Abstract

High-grade gliomas (HGGs) are the most aggressive adult brain tumors, with a dismal median survival of approximately 15 months, highlighting the need for novel therapeutic strategies. In a prior immunotherapy trial using dendritic cells against glioblastoma, miR-216b emerged as a potential predictive biomarker. Thus, we hypothesize that miR-216b impacts glioma aggressiveness and thereby therapeutic success. Here, we demonstrate that miR-216b is significantly downregulated in the majority of Isocitrate dehydrogenase 1/2 (IDH) wild-type HGG tissue samples (n = 42) and cell models (n = 18). Functional assays revealed that miR-216b overexpression impairs glioma cell proliferation, migration, and stemness characteristics. Transcriptomic and target prediction analyses identified CDK4, a key cell cycle regulator, as a direct target of miR-216b, confirmed via luciferase reporter assays. Correspondingly, upregulating miR-216b (mimic) via transfection decreased CDK4 mRNA and protein levels accompanied by a p21-dependent increase of cells in G0/G1 phase. In addition, miR-216b expression correlated with increased sensitivity to the CDK4/6 inhibitor Abemaciclib. Notably, miR-216b levels were significantly higher in less aggressive IDH-mutant gliomas (n = 21), linking its downregulation to malignancy grade. Collectively, our findings discovered miR-216b as a tumor suppressor in HGGs, modulating CDK4 expression and affecting the responsiveness to CDK4/6 inhibitors. The observed results support the potential of miR-216b as both a prognostic and predictive indicator in HGGs.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40478-025-02215-5.

Introduction

High grade gliomas (HGGs) are the most common and aggressive primary brain tumor entity in adults. The characteristic diffusely-infiltrating growth pattern, fast progression and limited therapeutic success lead to an average survival rate of 15 months after diagnosis [31]. The standard of care treatment involves maximum safe resection followed by radiotherapy and chemotherapy with Temozolomide [51]. In a previous experimental immunotherapy study (phase II) by our group, glioblastoma patients were treated with a personalized, autologous dendritic cell vaccine but the approach showed no improvement in overall survival [16]. Nevertheless, five miRNAs were found to be upregulated in a cohort of long-term survivors with this immunotherapy regime (miR-216a, miR-216b, miR-let7i, miR-200a, miR-708) [17].

Micro RNAs (miRNAs) are short non-coding RNAs in the lengths of 20–22 nucleotides, which bind to the 3′-UTR of target genes and have the ability to silence genes. One miRNA can regulate multiple genes at once and one gene can be regulated by multiple miRNAs. Of note, given this target multiplicity, the precise function a single miRNA fulfils, is often manifold and rather than single genes, miRNAs often regulate whole biological processes including differentiation, proliferation and apoptosis [29, 38]. All this renders them important molecules of interest for cancer therapy development.

In an earlier study we observed a positive correlation of e.g. upregulated miR-216b and prolonged survival in patients treated with dendritic cells immunotherapy, therefore we assumed an impact of this and other miRNAs on tumor aggressiveness and on the tumor immune microenvironment [17]. Following this hypotheses, this project comprehensively explored the presence of the five above-mentioned miRNAs in HGGs via analyzing FFPE material, snap frozen HGG tissue and glioma cell lines. Among these miRNAs, so far there is limited data about the impact of miR-216b on glioma aggressiveness and progression. First found in nasopharyngeal carcinoma, miR-216b is generally downregulated in various cancer types including breast cancer, liver cancer, lung cancer, colorectal cancer, pancreatic cancer and melanoma as well as HGG [10, 35]. Apart from being involved in tumorigenic processes, miR-216b has also been found upregulated in acute pancreatitis and reported to be a sensitizer to chemotherapy in ovarian cancer, colorectal cancer and non-small cell lung cancer [29].

While miR-216b has been observed as a modulator in various cancer types, the actual targets vary. Among the most prominent proven targets are FOXM1 (cervical, liver, lung cancer and melanoma), SOX9 (lung cancer) and Beclin-1 (colorectal and lung cancer, melanoma) [29]. With respect to glioma, only AEG1 and MTDH have been identified as targets of miR-216b in tissue and cell culture samples, inhibiting cell proliferation, invasion, tumor growth and migration [10, 30, 35].

To close the current gap of knowledge, in this study, we performed a comprehensive investigation on the targets of miR-216b in glioma. Via identifying CDK4 as a direct target of miR-216b, we establish a link between this miRNA and key histopathological features of HGG. Our findings further highlight the role of miR-216b in modulating glioma cell aggressiveness and the response to cell cycle inhibitors.

Methods

FFPE tissue

For this study a cohort of formalin-fixed, paraffin-embedded (FFPE) tissue blocks of 67 patients who underwent surgery at the Department of Neurosurgery, Medical University of Vienna was used. The patients were diagnosed with a high-grade glioma (WHO grade 3 and 4) as observed by the neuropathologists of the Division of Neuropathology and Neurochemistry, Department of Neurology, Medical University of Vienna (EK 1454/2018). Detailed information about patient characteristics are outlined in Supplementary Table 1.

Immunohistochemistry

Blocks of FFPE tissue of the patient cohort were cut in 3 µm sections and stained with a CDK4 antibody (1:25, Santa Cruz Biotechnology, Dallas Texas USA, sc-23896). Deparaffinization was done with a standard in-house procedure and pre-treated for 92 min with CC1 buffer (Ventana). The staining was performed on a Ventana BenchMark staining system and visualized with the UltraView Universal DAB detection kit (760–500). Consecutive slides were stained with hematoxylin to visualize nuclei, followed by differentiation and bluing according to standard protocols. Subsequently, sections were counterstained with eosin to visualize cytoplasmic and extracellular components. After staining, sections were dehydrated through graded ethanol solutions, cleared in xylene, and covered using a permanent mounting medium.

Cell culture

Two immortalized glioma (U251MG, LN229) and 16 primo-cell culture models were established and provided by our collaboration partners (Center for Cancer Research Vienna and Kepler University Hospital Linz). The study was approved by the ethics committee of Upper Austria, Linz (E-39–15) and ethics committee of Medical University of Vienna (EK 1616/2020 and EK419/2008). Patient-derived primo-cell models (VBT72, VBT136, VBT185, VBT186, VBT195, VBT208, VBT230, VBT262, VBT279, VBT580, VBT584, BTL53, BTL1528, BTL1529, BTL2175, BTL2176) were cultivated in RPMI-1640 medium (Sigma-Aldrich, St. Louis, Missouri, USA) supplemented with 10% fetal calf serum (FCS, Gibco Thermo Fisher Scientific, Waltham, MA, USA) and 2% Glutamin (Gibco_, #25030-024)_. U251 and LN229 were held in Dulbecco’s modified Eagle’s medium (DMEM) supplied with 10% FCS. The cells were incubated at 37 °C in a 5% CO₂ humidified incubator. All cell lines were re-cultivated from frozen stocks after 9 weeks to avoid genetic changes in high passages. Subsequently we focused on seven primary cell models (VBT186, VBT262, VBT 580, BTL2176, BTL1528, BTL1529, BTL53) due to similar growth properties and their most promising effects on miRNA transfection. Genomic alterations of primo-cell models were analyzed by targeted next generation sequencing using the Ion AmpliSeq™ Cancer Hotspot Panel v2 or the Oncology Cancer Analysis v3 panel on the Ion Torrent™ platform (Thermo Scientific). Details are shown in Supplementary Fig. 2.

Spheroid models were held in Nunclon Sphera 24well plates (#174930, Thermo Scientific) with Neurobasal Medium (Gibco, Thermo Fisher Scientific), 20 ng/ml human epidermal growth factor (PeproTech, Thermo Fisher, #AF-100-15), 20 ng/ml human FGF2 (PeproTech, #100-18B), 1% B27 and 1% N2 (Gibco, Thermo Fisher Scientific) and 2 µM Glutamine.

Total RNA and miRNA extraction; quantitative real-time PCR (RT-qPCR)

Total RNA including miRNA was extracted from FFPE tissue slides with the miRNeasy FFPE Kit (#217504, Qiagen, Venlo, Netherlands) following the provided protocol. RNA from cell models was isolated using TRI-Reagent® (Sigma-Aldrich, T9424). RNA concentration was measured with a spectrophotometer (Thermo fisher Scientific, Nanodrop). 20 ng RNA was reverse transcribed into cDNA with the miRCURY LNA RT Kit (Qiagen, #339340) or with High-Capacity cDNA Reverse Transcription Kit (applied biosystems, Thermo Fisher #4368814). qPCR was performed using the miRCURY LNA miRNA PCR Assay Kit (Qiagen, 339306), U6 snRNA was used as housekeeping gene (Supplementary Table 2). For CDK4 and CDK6 gene expression the GoTaq® qPCR Kit from Promega (Madison, Wisconsin, USA) was used (A6001) and GAPDH was used as housekeeping gene. Detailed information can be found in the Supplementary Table 3. As positive controls Human Brain Total RNA (636530, Clonetech, Takara, Kusatsu, Shiga, Japan), Human Astrocytes (1800-5, ScienCell, San Diego, California, USA) and in-house established cerebral organoids derived from human embryonic stem cells (hESCs) were used.

Transfection of miRNA

Adherent cell models were seeded in 24-well plates (3526, Corning Incorporated, costar®, Corning, New York, USA; 3 × 10⁴–1 × 10⁵ per well), in their respective medium and kept in the incubator overnight. On the next day, cells were transfected with 10 nM and 30 nM of the following probes (mirVana, Invitrogen, Thermo Fisher): MiR-216b-5p mimic (#4464066), which overexpresses the naturally occurring miRNA by being loaded into the RNA-induced silencing complex (RISC) and miR-216b-5p Inhibitor (#4464084) which are antisense oligonucleotides, once introduced to the cell, it will be the preferred binding partner of the endogenous miRNA rather than its physiological targets. As negative controls mimic negative control (#4464058) and Inhibitor negative control (#4464076) were induced to the cells. The negative controls contain sequences with no homology to any known miRNA or other human, mouse or rat genome sequence. They therefore might not affect any cellular processes. The oligonucleotides were introduced to the cell by using Lipofectamin RNAiMAX (Invitrogen, Thermo Fisher, #13778150). The procedure was performed according to the manufacturer’s protocol (mirVana, Invitrogen, Thermo Fisher). The transfection reagents were left on the cells overnight, a total medium change back to the respective medium of the cells was performed on the following day. Cells were kept for another 2 days or up to 1 week, depending on the assay. Successful transfection was confirmed via qRT-PCR and the predicted miRNA targets by Western Blotting (detailed information given below).

For the establishment of three-dimensional (3D) anchorage-independent growth, cells were seeded (2 × 10⁴ per well) in Neurobasal medium in Nunclon Sphera 24-well plates and transfected, as described above.

Protein expression analyses

1 × 10⁵ to 4 × 10⁵ cells/well were seeded into 6-well plates and cultivated in their respective medium overnight. Transfection was performed as described earlier after the cells reached 80 to 90% confluence. Protein isolation was accomplished after scraping cells and washing with PBS. The cells were lysed using a lysis buffer (50 mM Tris/HCl, pH 7.6); 300 mM NaCl, 0.5% Triton X-100), protease inhibitors PMSF (Sigma, P7626-1G), cOmplete (Merck, Rahway, New Jersey, USA, 11697498001) and phosphatase inhibitor PhosSTOP (Roche, Basel, Switzerland, 04906837001), following by ultrasound sonication for at least five minutes. Protein concentrations were detected with the BCA Protein Assay Kit (Pierce, Thermo Scientific #23221) following the company’s protocol. For western blotting 15 µg proteins were loaded on 10% polyacrylamide gels and electrophoresis was run at 90 V (BioRad, Hercules, California, USA,). Semi-dry plotting onto polyvinylidenfluorid membranes (Amersham™, Cytiva, Marlborough, Massachusetts, USA, Hybond® Membranen, PVDF, GE10600023) was achieved with the Trans-Blot® Turbo™ Transfer System from Bio Rad (1704150EDU). All antibodies were diluted in a 3% BSA in Tris-buffer with 0.1% Tween20 (BioRad, #1706531) and β-actin was used as loading control (Supplementary Table 4).

Clonogenicity assay

1 × 10⁴ to 5 × 10⁴ cells were seeded in a 24-well plate and transfected with the mirVana Kit from Qiagen as described previously. After medium exchange, cells were kept for 3 days and microphotographs were either taken with the Incucyte ZOOM™ (Sartorius, Incucyte S3, Essen Bioscience) and percentage of confluence in the well was calculated by the Incucyte program and analyzed with the GraphPad Prism software. Or cells were fixed with methanol following crystal violet staining (VWR, 87811.180). The wells were photographed and analyzed using Fiji software. Therefore images were transformed into binary images and the black particles were analyzed with the “Measure” function of Fiji. Integrated density was computed and analyzed with the GraphPad Prism software.

Wound healing assay

4 × 10³ cells were seeded in a 96-well plate in triplets and transfected with the mirVana Kit from Qiagen as described previously. After 72 h medium was refreshed, and a scratch was made with the Incucyte® WoundMaker on the next day. Migration of the cells was followed in an incubator and observed with the Incucyte ZOOM-program until the wound was closed again in at least one condition. The area of the scratch confluence was measured daily with the Incucyte ZOOM-program.

Three-dimensional anchorage-independent growth

5 to 7.5 × 10³ cells were cultivated in Neurobasal medium with previous mentioned supplements in 24-well plates, following transfection with mimic, inhibitor or the respective controls as already described. 12 h later, Neurobasal medium was refreshed and the formation of spheres was followed over time in an inverse light microscope and photomicrographs were taken daily for 7 days. The size of the spheres (area in µm²) was measured using Fiji.

Drug sensitivity assay

4 to 5 × 10³ cells were seeded in triplets in 96-well plates and transfected after 24 h as described previously. After the medium change on day three, Abemaciclib, Palbociclib and Ribociclib (Selleck Chemicals (Houston, TX, USA) was added in a concentration gradient from 0.5 to 25 µM and cells were incubated for another 72 h. Subsequently the viability of the remaining cells was measured via an ATP assay (CellTiter-Glo®, Promega, Madison, Wisconsin). Luminescence signals were analyzed with the Tecan plate reader (Tecan Infinite 200Pro, Zurich, Switzerland).

Library preparation and sequencing

Libraries were prepared with the QuantSeq 3′ mRNA-Seq V2 Library Prep Kit with UDI (Lexogen, Austria) according to the manufacturer’s instructions. 160 ng total RNA was used as input and final amplification of libraries was performed with 16 PCR cyles. Libraries were pooled in equimolar ratio and sequenced on Illumina NovaSeq SP Flowcell in SR100 mode.

Data analysis

1. Trimming, mapping, quantification of reads and differential expression analysis

Overall quality of the next-generation sequencing data was evaluated automatically and manually with fastQC v0.11.8 [4] and multiQC v1.7 [18]. Reads from all passing samples were adapter trimmed and quality filtered using bbduk from the bbmap package v38.69 [7] and filtered for a minimum length of 17nt and phred quality of 30. Alignment steps were performed with STAR v2.7 [13] using samtools v1.9 [34] for indexing, whereas reads were mapped against the genomic reference GRCm38.p6 provided by Ensembl [59]. Assignment of features to the mapped reads was done with htseq-count v0.13 [3]. Differential expression analysis with edgeR vNA [49] used the quasi-likelihood negative binomial generalized log-linear model functions provided by the package. The independent filtering method of DESeq2 [37] was modified for operating with edgeR to remove low abundant genes and thus improve the false discovery rate (FDR) correction.

• 2.Target prediction for miR-216b-5p

Target miRNAs are used to derive potential target genes. The tool miRNAtap [43] allows to combine the results of 5 different miRNA-mRNA interaction databases (Targets are aggregated from 5 most commonly cited prediction algorithms: DIANA [40], Miranda [15], PicTar [33], TargetScan [19] and miRDB [55]. Targets were only included if they were found in at least 2 databases. The aggregated rank was calculated by using the geometric mean rank of all databases.

• 3.Gene ontology analysis

Differentially expressed mRNAs based on an FDR cutoff (FDR < = 0.05) were used in a GO-term enrichment analysis. Enriched biological processes (BP) were identified by using the Kolmogorov Smirnov (KS) test [39, 41], with the unadjusted p-values from the differential expression analysis as rank information with the tool topGO [1].

Furthermore, GO enrichment analysis was performed for predicted targets of miR-216b-5p and overlap of gene sets regulated in the data set and of the predicted targets were calculated and visualized.

Secrete-pair dual luminescence assay

VBT186 and BTL53 were seeded in their respective medium in 24-well plates (1 × 10⁵/well). After 24 h they were co-transfected with miR-216b (mimics, inhibitor and negative controls) and a plasmid containing the CDK4 3′UTR target sequence of miR-216b cloned into a pEZX-MT05 vector (GeneCopoeia, Rockville, MD, USA) using DharmaFect (DH-T-2006-01). This vector harbors the bioluminescent reporter protein Gaussia luciferase (GLuc) linked to the miR-216b target (CDK4) and the reporter gene Secreted Alkaline Phosphatase (SeAP) which was used as an internal control. Supernatant of the samples was measured in quadruplicates after 48 h and 72 h with the Secrete-pair dual luminescence kit (GeneCopoeia, GC-LF032) and normalized to the SeAP value (GLuc/SeAP ratio). Luminescence signals were detected with the Tecan Infinite 200Pro.

PI staining and fluorescence activated cell sorting (FACS)

2.5 × 10⁵ to 1 × 10⁶ cells were seeded in duplicates in 6-well plates. On the next day transfection of miRNA 216b-5p mimic and mimic negative control was done as already described. Cells were cultivated for 48 and 72 h and then detached with Trypsin. Duplicates were pooled and centrifuged. Cell pellets were resuspended in 0.9% NaCl and diluted in 70% ethanol. After a centrifugation step, the pellet was resuspended in 500 µl FACS-PBS with added RNase, stained with Propidium Iodide (PI) and measured by flow cytometry (BD LSR Fortessa X-20 Flow Cytometer). Results were analyzed with the FlowJo Software (BD Bioscience).

Statistical analyses

Statistical analyses were conducted using GraphPad Prism (version 8). Data are shown as mean ± SEM unless indicated otherwise. Normality was assessed prior to statistical testing. For comparisons between two groups, unpaired two-tailed Student’s t-tests were used for normally distributed data, whereas non-parametric data were analyzed using Mann–Whitney U-test. Comparisons among multiple groups or involving more than one experimental factor were performed using two-way ANOVA, as appropriate, with post hoc correction for multiple comparisons. A p value < 0.05 was considered statistically significant.

Results

Expression levels of miR-216b correlate with histo-pathological parameters in high-grade gliomas

To characterize miRNA expression patterns in glioma tissue, we analyzed a set of 58 miRNAs in a previously published cohort of FFPE glioblastoma samples (n = 38). The analysis revealed distinct expression levels across the miRNA panel, with several established oncomiRs showing consistently high expression, while others were present at comparatively low levels [9, 23, 56]. Notably, miR-216b exhibited one of the lowest expression levels across the reference cohort, ranking second lowest among all analyzed miRNAs, which provided a strong rationale for its subsequent in-depth analysis (Fig. 1a, Supplementary Table 5) [17]. In a new cohort (n = 39) of HGG patients were retrospectively recruited and their tumor tissue was examined with respect to the expression levels of the five miRNAs (miR-200a, miR-216a, miR-708-, miR-216b, miR-let-7i) that had been attributed with a predictive impact on patient survival in the aforementioned previous study investigating dendritic cell vaccination immunotherapy in HGG (Fig. 1b) [17]. qPCR results demonstrated generally low expression levels of the investigated miRNAs in the majority of samples, while only a small subgroup was characterized by increased miRNA levels (Fig. 1b). To confirm these first results seen in the FFPE samples, we additionally analyzed the expression of miR-216b in a subset of our glioma patient cohort where snap-frozen tissue was available, in comparison to non-cancerous brain control samples. The frozen tissue collection (patient 1–12) was divided in three groups according to their miRNA expression levels as detected in the matching FFPE tissue block (high, low, average) and compared to a control group consisting of epileptic tissue, brain organoid and whole brain lysates (blue). Again, only a small glioma subgroup was characterized by high miR-216b levels, exceeding even the expression levels in the non-malignant control group (Fig. 1c), while generally expression levels were low. Since miR-216b had been among the most predictive miRNA in our previous study impacting long-term survival in immunotherapy-treated glioblastoma patients, we continued focusing on an in-depth analysis of this miRNA. With respect to clinico-pathological parameters, we did not observe any association between miRNA expression and age, Karnofsky Performance Status (KPS), MGMT promoter methylation or TERT promoter mutation (Supplementary Fig. 1a–d). However, our cohort exhibited a strong significant association between high-levels of miR-216b and the presence of an IDH1/2 mutation, characterizing this less aggressive glioma subtype (Supplementary Table 1 and Fig. 1d). As our clinical data had pointed towards an impact of particular miRNAs on glioma cell aggressiveness, we next screened the expression of the five miRNAs in 18 different glioma cell culture models (Fig. 1e). Here, miRNA levels showed comparable expression patterns as observed in the patient tissue samples (Fig. 1b and c). The generally low levels of miR-216b in all investigated cell models held the potential for a straightforward and meaningful manipulation of upregulating this miRNA in order to decipher its biological effects. Thus, based on their properties with respect to growth, transfection-tolerance, and passaging, we chose seven cell lines for further cell-biological and molecular investigations (marked with asterisks in Fig. 1f). Molecular characteristics of our in-house established patient-derived cell models (Vienna Brain Tumor, VBT and Brain Tumor Linz, BTL, model) are provided in Supplementary Fig. 2.

Fig. 1.

miRNA expression levels in glioma tissue and cell lines. Data showing miRNA expression levels of 58 miRNAs analyzed in the cohort of FFPE glioblastoma samples (n = 38) as previously published [17]. Values are expressed as 2^−dCT and log₁₀-transformed for visualization. miRNA expression was normalized to the mean of 5S rRNA, SNORD44 and SNORD48. Well-known oncomiRs are shown as filled, tumor suppressor miRNAs as checkered boxplots. MiR-216b is emphasized with a green triangular marker (a). MiRNA expression levels of five different miRNAs were investigated in FFPE tissue of an independent glioma cohort (n = 39) using RT-qPCR. IDH1/2-mutated tumors are highlighted as squares (b). MiR-216b expression levels were analyzed in frozen samples of glioma tissue and grouped as “high”, “low”, and “average” according to the values determined in the matching FFPE sample (compare b). The control group includes one epileptic foci sample, one cerebral organoid and a whole brain sample. 2^−dCT values were log transformed and plotted as bar graphs indicating each single sample (upper panel) or as scatter plots showing the mean expression in each group (lower panel) (c). MiR-216b expression is depicted in FFPE tissue of IDHmut (n = 17) and IDHwt (n = 43) gliomas. Statistical significance was assessed using the Mann–Whitney-U test (d). Expression levels of the five indicated miRNAs were determined by RT-qPCR in 18 glioma cell lines (e), and levels of miR-216b are separately plotted for each individual cell line (f). Models used for further investigations are marked with asterisks. Values were normalized to the negative control (U6 snRNA), subsequently expressed as linear or logarithmic 2^−dCT. All measurements were done with RT-qPCR

miR-216b inhibits proliferation and migration of glioma cell models

To determine the effects of upregulated and inhibited miR-216b, glioma cell models (VBT186, VBT262, BTL53, BTL2176 and BTL1529) were transfected with miRCURY LNA miRNA mimics and inhibitors, and their respective negative controls. MiRNA mimic represents double-stranded RNA mimicking the endogenous miR-216b by entering the RNA-induced silencing complex (RISC), enabling to study gain-of-function mechanisms. The inhibitor on the other hand is a single-stranded antisense oligonucleotide which binds directly to the endogenous miR-216b, causing an inhibition of its function. After up to 7 days of observation, manipulated cells with miRNA mimics showed reduced cell proliferation compared to the respective non-targeting negative control (mimic negative). Surprisingly, inhibition of functional miR-216b using a small molecule inhibitor did not further enhance proliferation capacity compared to the respective controls (inhibitor negative). Quantitative analysis confirmed the observations by showing decreased survival of glioma cells upon transfection with miRNA (mimic) compared to cells with non-functional miRNA (inhibitor) and their negative controls (Fig. 2 and Supplementary Fig. 3).

Fig. 2.

Impact on cell proliferation upon miR-216b manipulation in GBM cell models. Photomicrographs of clonogenic assays in VBT186 and BTL53 taken over the course of 7 days (a and b). Cell confluence was monitored in the IncuCyte system and quantitative analysis is shown at the experimental endpoint for both cell models (c and d). Values are expressed relative to the respective (resp.) negative controls (mimic negative or inhibitor negative), which were set to 1. Bar = 100 µm

Besides proliferation, we investigated the impact of miR-216b overexpression (mimic) and inhibition (inhibitor) on migratory properties by performing scratch assays. Upon blockade of functional miR-216b in BTL53 cells (inhibitor), wound closure was distinctly pronounced after 72 h (Fig. 3a and b), while cells with upregulated levels of miR-216b (mimic) showed decreased migration abilities compared to the respective controls (mimic negative) (Fig. 3a and b). These findings indicate impaired migration of glioma cells with high miR-216b levels. However, the migratory potential of VBT186 cells was not altered upon miR-216b manipulation, pointing towards intertumoral molecular heterogeneity among HGGs (Supplementary Fig. 4).

Fig. 3.

Wound healing ability of BTL53 upon miR-216b manipulation. Cells were transfected with miR-216b mimic (upregulating) and miR-216b inhibitor (blocking). The respective negative controls were included. After 72 h, when cells reached confluence, a scratch was made with a wound maker and cells were observed in the IncuCyte cell monitoring system for another 72 h (a). Quantitative analysis is depicted in b. Statistical significance was determined using a two-way ANOVA

3D anchorage-independent growth indicated a loss of stemness characteristics in cells with high levels of miR-216b

In addition to proliferation and migration capacities, we then explored whether miR-216b has an impact on stemness characteristics in glioma cells. Accordingly, we transfected VBT186, VBT262 and BTL53 cells with the indicated miRNA oligonucleotides mimicking or inhibiting miR-216b plus the respective controls and transferred the manipulated cells to ultra-low attachment plates to observe their ability to grow anchorage-independent as spheres, which is an indicator for stemness characteristics. After 3–7 days, a distinct difference was observed in cells with upregulated miR-216b compared to its negative control. The spheres of cells with induced expression of miR-216b (mimic) were distinctly smaller and more dissociated while those with blocked miR-216b (inhibitor) and the respective negative control (inhibitor negative) formed larger spheres (Fig. 4a–c). For an objective quantification, the area of each individual sphere was subsequently measured resulting in statistically significant smaller spheres upon treatment with miR-216b mimic compared to the respective negative control (mimic negative) and to the group transduced with the miR-216b inhibitor (Fig. 4d–f).

Fig. 4.

Three-dimensional (3D) anchorage-independent growth upon miR-216b manipulation. Glioma cells were seeded in ultra-low attachment plates in Neurobasal medium following transfection with miR-216b mimic and miR-216b inhibitor, as well as the respective negative controls. 3D growth of the indicated cell models (VBT186, VBT262, BTL53) was monitored over time and photomicrographs were taken (a-c). At the experimental endpoint the area of each sphere was quantified using ImageJ software and results are presented as violin plots. Statistical significance was evaluated via student’s t-test, adjusted p-values are shown (d–f). Bar = 100 µM

mRNA sequencing predicted miR-216b target genes involved in cell division

To further investigate which targets are affected by miR-216b, not only untreated glioma cell lines were submitted for RNA sequencing but also five glioma models which were transfected with miR-216b mimic and inhibitor. Five different miRNA-mRNA interaction databases were then used to predict target genes of miR-216b. Amongst the genes identified were e.g. CDGSH Iron Sulfur Domain 2 (CISD2), which is involved in calcium homeostasis, ageing and is known to be involved in cancer development [36]. Another affected gene was Guanylylcyclase domain containing 1 (GUCD1) which is associated with liver tumorigenesis [5]. CDK4 was listed as well under the top twelve downregulated targets in glioma cells transduced with miRNA oligonucleotides mimicking miR-216b, affirming the cell cycle as a possible target (Fig. 5a). This data was further strengthened by GO enrichment analyses, showing mitotic cell cycle pathways (including CDK4, Supplementary Table 6) among the overlapping top hits (blue dot) according to miR-216b target gene prediction (green dots) but also based on differentially regulated genes upon miRNA manipulation in our cell models (purple dots) (Fig. 5b).

Fig. 5.

RNA sequencing for target gene prediction and target testing upon miR-216b manipulation. Volcano plot showing differential gene expression between the two groups. Log fold change (logFC) is displayed on the x-axis, and statistical significance is shown on the y-axis as − log10 (adjusted P value/FDR). Genes exceeding the predefined significance threshold (FDR < cutoff) were considered significantly differentially expressed. Accordingly, the most strongly downregulated genes, located furthest to the left on the x-axis, are highlighted and labeled in green (a). GO enrichment analyses were performed and enrichment scores are plotted as dot plots. Pathways predicted to be regulated by miR-216b are shown in green, differentially regulated genes upon miRNA manipulation are depicted as purple and overlapping top hits as blue dot (b)

miR-216b targets proteins involved in cell cycle regulation in glioma cells

To test if CDK4 is a direct target of miR-216b in glioma cells, a dual-luciferase reporter assay was performed using the Secrete-pair Dual Luminescence Assay Kit (GeneCopoeia). After co-transfecting the cells with miR-216b (mimics and inhibitors) and the plasmid containing the CDK4 3′UTR, GLuc-SeAP-ratios were investigated. Since the GLuc-reporter protein is linked to the miR-216b target 3′UTR, the detected luminescence level correlates with the expression level of CDK4. We found reduced luciferase signals in cells with upregulated miR-216b levels (via mimics transfection) (Fig. 6a and c) and upregulated luminescence in cells with blocked miR-216b activity (via inhibitor transfection) as shown in Fig. 6b and d. These findings confirm the direct interaction of miR-216b and CDK4.

Fig. 6.

Impact of miR-216b manipulation on the induction of the predicted target gene CDK4. CDK4 promoter activity was measured using a Secrete-pair dual luminescence assay in two cell models (as indicated). Results are depicted as luciferase activity relative to the respective negative controls, set to 1. Statistical differences were analyzed by student´s t-test. (a-d). CDK4 protein expression was investigated by immunohistochemical staining in HGG patients. Representative CDK4 and hematoxylin-eosin (H&E) stainings of FFPE tissue sections (total n=8) are shown, bar = 250µm (e). Quantitative analysis of CDK4-positive cells (f, %) and of miR-216b expression levels (g) in matched FFPE samples (n=8). Samples were stratified into low- and high-miR-216b expression groups and displayed as graphs. Statistical differences were analyzed by student’s t-test. CDK4 mRNA expression was analyzed by RT-PCR in FFPE samples where sufficient material was available (n=32) and stratified according to the mean miR-216b status. Statistical differences were assessed using the Mann-Whitney-U test (h). Endogenous CDK4 mRNA levels were measured by qRT-PCR in cell models (n=12) and stratified according to mean miR-216b expression into high- and low-miRNA groups (i). Changes in CDK4 (j) and CDK6 (k) mRNA expression following miRNA manipulation were evaluated in VBT186, VBT580, BTL2176, BTL53 and BTL1529. Gene expression levels were normalized to the respective negative controls (set to 1) and statistical analysis performed by student´s t-tests. Western blot analyses depict protein levels of CDK4 and of CDK4-dependent cell cycle regulators (p21 and RB) in BTL53 and VBT186 (l)

CDK4 protein expression was next evaluated in FFPE tissue sections from HGG patients (n = 8). Representative CDK4 and hematoxylin–eosin stainings demonstrated lower CDK4 protein levels in slides with high miRNA expression and vice versa (Fig. 6e). Quantitative analysis confirmed a significant negative correlation between CDK4 (% of stained cells) and miR-216b levels (Fig. 6f). In the same cohort miR-216b expression was quantified and stratified into low and high expression groups based on the mean. Analysis revealed a significant difference between these two groups (Fig. 6g). An association between miR-216b and CDK4 was further confirmed in the broader FFPE cohort (n = 32), where CDK4 mRNA levels were significantly higher in tumors with low miR-216b levels (Fig. 6h). Similarly, in our investigated HGG cell panel, endogenous CDK4 gene expression inversely correlated with miR-216b expression (Fig. 6i).

To verify and functionally investigate the predicted target (CDK4 of the RNA sequencing), we performed RT-qPCR and western blot analyses in our cell models (VBT186, VBT580, BTL2176, BTL53, BTL1529). Indeed, induction of miR-216b by miRNA mimics led to consistent downregulation of CDK4 transcripts in cells transfected with the miRNA mimics (Fig. 6j). Given the close functional relationship between CDK4 and CDK6 in cell cycle control, we further examined the regulatory effect of miR-216b on CDK6. Besides BTL53 and VBT186, no impact on CDK6 expression was observed among the investigated cell models (Fig. 6k). In contrast to the consistent suppression of CDK4, the effects on CDK6 mRNA expression following miR-216b manipulation appears to be cell line-dependent. At the protein level, Western blot analyses in BTL53 (RB-deficient), VBT186 and VBT262 confirmed that miR-216b regulation of CDK4 also modulated downstream cell cycle regulators (Fig. 6l, Supplementary Fig. 5). Corroborating our findings on CDK4 transcription, induction of miR-216b (mimic) resulted in decreased CDK4 protein expression accompanied by induced levels of p21 (CDK-inhibitor 1) in the investigated cell model. In addition, phosphorylation of RB was impaired upon overexpression of miR-216b in the RB-proficient VBT186 model, while BTL53 cells lack RB expression altogether. Vice versa, when miR-216b was blocked by transduction of the cells with the respective oligonucleotides (inhibitor), reduction of p21 and an increase of pRB was observed. In VBT262 cells, characterized by CDKN2A loss, the impact on RB phosphorylation was less pronounced (Supplementary Fig. 5).

As p21 can act as a tumor suppressor and is able to halt the cell cycle, our results suggest a tumor-suppressive potential of miR-216b (Fig. 6l), which is further confirmed in cell cycle analyses. In detail, upregulated miR-216b levels (mimic) resulted in an increase of cells in the G0/G1 phase (Supplementary Fig. 6, blue) and a decreasing percentage of cells in the G2 phase (Supplementary Fig. 6, yellow) in both cell models. All together our observations strengthen the tumor-suppressing properties of miR-216b on cell proliferation (compare Fig. 2) and cell cycle progression.

Manipulation of miR-216b levels impacts the sensitivity towards CDK4/6 inhibitors

Finally, to further confirm the correlation between miR-216b and CDK4/6 and to analyze the role of these kinases as possible targetable vulnerabilities for gliomas, we tested the sensitivity towards the current FDA approved CDK4/6 inhibitors, Abemaciclib, Ribociclib and Palbociclib, in our panel of patient-derived HGG cell models. Dose–response data for BTL53, VBT186 and VBT262 are depicted as bar graphs (Supplementary Fig. 7a–c). As treatment with Ribociclib did not reach an IC₅₀ in any of the tested cell lines, this compound was excluded from subsequent analyses (Supplementary Fig. 7a–c). The response to Abemaciclib and Ribociclib was further investigated in our extended cell panel including VBT580, BTL2176, BTL1528 and BTL1529. Stratification of IC₅₀ values for Palbociclib and Abemaciclib by miR-216b expression levels revealed increased sensitivity to Abemaciclib in miR-216b–low cells, whereas no such trend was observed for Ribociclib (Supplementary Fig. 7d and e). Inhibition of functional miR-216b with corresponding synthetic oligonucleotides (inhibitor) significantly enhanced the effects of Abemaciclib at low micromolar concentrations (1 or 2.5 µM) in miR-216b-low BTL53 cells (Fig. 7a). Conversely, in the VBT186 model, which exhibits high endogenous miR-216b levels, further upregulation by miRNA mimics significantly increased sensitivity to Abemaciclib compared to the negative control. (Fig. 7b). With respect to VBT262, miR-216b inhibition or overexpression had no impact on the sensitivity towards the investigated CDK4/6 inhibitors (Supplementary Fig. 8). In addition, Palbociclib efficacy was unaffected by miR-216b manipulation in the investigated cell models (Fig. 7c and d). Our findings suggest that, depending on the endogenous miR-216b levels, manipulation of this miRNA can increase the responsiveness of HGG cells to the CDK4/6 inhibitor Abemaciclib.

Fig. 7.

Sensitivity towards cell cycle inhibitors and impact on miR-216b manipulation. MiR-216b was induced (mimic) or blocked (inhibitor) in BTL53 and VBT186 cells following a treatment with Abemaciclib (a and b) or Palbociclib (c and d) at the indicated concentrations. After 72 h the viability of the remaining cells was examined using CellTiter-Glo® Luminescent Cell Viability Assay. Depicted are the luminescence signals, normalized to the respective untreated controls. Statistical significance was determined using a two-way ANOVA

Discussion

In this study, we investigated the role of miR-216b on glioma cell aggressiveness and how this miRNA impacts target genes. miRNA-216b is generally low expressed in glioma tissue samples. Accordingly, induction of miR-216b expression by using oligonucleotide mimics, resulted in tumor-suppressive effects in vitro, indicated by decreased glioma cell proliferation and migration as well as impaired stem cell features. Additionally, RNA sequencing suggested the cell cycle protein CDK4 as a target of miR-216b. Protein and RNA analysis assays confirmed these findings and allowed—for the first time—to state CDK4 as a direct target of miR-216b. CDK4 overexpression is known to induce cell proliferation in gliomas [28, 47, 48, 50]. Therefore, it might act as target for glioma therapy, which could be silenced by miRNA application in the future.

In recent years, the role of miRNA in cancers has been a well-discussed topic. The potential of regulating multiple target genes with one miRNA is a promising tool in experimental cancer therapies. MiRNAs are known for their modulatory effects on cell proliferation, apoptosis, differentiation, stem cell maintenance, metabolism, autophagy, and immune suppression [45, 54]. While in many cancer types so-called oncogenic miRNAs (oncomiRs) have been found to be upregulated and to silence tumor suppressor genes, it is also known that certain miRNA are downregulated which target oncogenic mRNAs [32]. A downregulation of miR-216b has been observed in nasopharyngeal carcinoma, breast, liver, colorectal cancer, pancreatic ductal adenocarcinoma and glioma [12, 14, 21, 35, 60].

Accordingly, our data confirmed low expression of miR-216b in HGG tissue samples and cell models. Interestingly, a small subgroup of HGG specimens depicted increased expression of the investigated miRNA. This might be explained by the significant association found between miR-216b levels and presence of an IDH1/2-mutation, defining a more favorable, less aggressive glioma subtype [58].

Based on these results and on a previous work of our group demonstrating higher expression of miR-216b in long-term glioma survivors in an immunotherapy trial with dendritic cells, we continued to investigate the effects of this miRNA on cell aggressiveness [6, 16, 17]. In line with that, upon miR-216b induction we observed a decreased proliferative, migratory and stem-cell capacity in glioma cells. These findings match the results found in non-CNS tumors, including nasopharyngeal carcinoma, melanoma and cervical cancer cells [52] and in one study investigating glioma cells [35].

Based on the fact that one miRNA can target multiple mRNAs we analyzed potential targets by performing mRNA sequencing with the aim to predict novel or at least confirm already published targets of miR-216b described for different cancer types, including FoxM1, PKC-α, KRAS or AEG [11, 12, 20, 24]. This had the intention to bring us one step closer to understanding the biological processes behind gene silencing through miR-216b in HGG. Although we did not find any of the aforementioned targets, our data still demonstrated that miR-216b is directly modulating the expression of CDK4. Indeed, we could confirm the direct interaction between the investigated miRNA and CDK4 using cell-based assays, which has not been described in the literature yet.

Since the CDK4/6-Cyclin D-RB pathway is frequently dysregulated in cancer cells including HGG, we assume a direct impact of miR-216b on cell cycle progression in HGG cells [25, 57]. Accordingly, overexpression of miR-216b via oligonucleotides decreased the transition of cells from G1 to S phase and therefore indicated cell cycle arrest. Western blot analysis underpinned these observations, demonstrating induction of p21 and subsequently reduction of RB phosphorylation.

As we confirmed a direct impact of miR-216b on cell cycle regulatory processes in glioma cells, we continued with testing the sensitivity of cell cycle inhibitors in glioma cells upon miR-216 manipulation. CDK4/6 inhibitors are currently approved for HER2-negative advanced or metastatic breast cancer and are further tested for other cancer types [42]. There are subtle differences in the mechanisms of Palbociclib, Abemaciclib and Ribociclib. Abemaciclib is apparently the most potent drug of them and targets also CDK9, PIM1, HIPK2, DYRK2, CK2 and GSK3b. This is in accordance with our data, as Abemaciclib showed the most promising effects in our investigated cell models [26]. Besides being a multi-target compound, it shows also a greater selectivity for CDK4 than CDK6 and inhibits the phosphorylation of RB [44].

Abemaciclib has also been tested in xenograft models of the glioma cell line U87MG, where it exhibited distribution across the BBB, thus resulting in a significantly increased survival rate [46]. In accordance with that, a clinical phase I study confirmed detectable concentrations of Abemaciclib in cerebrospinal fluid and plasma. Even more, two of the 17 enrolled HGG patients showed significantly increased progression free survival [44]. With respect to Palbociclib, no antitumor activity was observed when administered as monotherapy in heavily pretreated RB-proficient glioblastoma patients [53].

Our preclinical data highlight differences in the responsiveness towards CDK4/6 inhibitors based on endogenous miR-216b levels. Strikingly, in the cell model with high endogenous miR-216b expression, additionally characterized by a CDK4 amplification (VBT186), further upregulation by miRNA mimic enhanced sensitivity to Abemaciclib. We propose that residual CDK4/6 activity becomes critical for maintaining proliferation in this context, rendering these cells particularly vulnerable to pharmacological inhibition. On the opposite, in miR-216b–low cell model (BTL53), reduced miRNA-mediated repression of CDK4 likely results in increased dependence on CDK4 signaling, thereby enhancing sensitivity to Abemaciclib-mediated CDK4/6 inhibition.

Nevertheless, our findings do not rule out a potential impact of the molecular background on cellular responsiveness to CDK4/6 inhibitors. In detail, the presence of a CDK4 amplification or a CDKN2A loss might act as cofounder factor modulating the anti-cancer effects of CDK4/6 inhibitors in HGG. Accordingly, a glioblastoma subgroup demonstrating stable disease or tumor response to Abemaciclib treatment in the phase I study from Patnaik et al. was characterized by a non-altered CDK4 gene locus [44]. Vice versa, in the group of patients showing progressive disease under Abemaciclib therapy, several tumors harbored a CDK4 amplification. Similar findings were reported in cell-based assays, where RB-wild-type glioma cells overexpressing CDK4 but lacking CDKN2A/B were completely resistant to CDK4/6 inhibitors [8]. In this context, our data support the potential of ectopic miR-216b induction to overcome resistance to Abemaciclib in CDK4-amplified HGG. Loss of CDKN2A may further predict the response to CDK4/6 inhibitors. In line with the literature, where low levels of CDKN2 are described as predictors for CDK4/6-targeting compounds [2, 22], cell models characterized by a homozygous CDKN2A loss (VBT262, BTL1528 and BTL1529) displayed the lowest IC₅₀ values for Abemaciclib. Besides the CDKN2 status, RB was considered to be one of the most important biomarkers of sensitivity to CDK4/6 therapy [2, 27]. In our study, only BTL53 cells presented with a RB-deletion, thus defined to be intrinsically resistant to CDK4/6 inhibitors. Nevertheless, inhibition of miR-216b re-sensitized these cells to Abemaciclib. To validate how molecular alterations and their interaction with miR-216b predict response to CDK4/6-targeting therapies, it is advisable to expand the number of patient-derived HGG cell models. In addition, we aim to include more complex preclinical models like brain-tumor-organoids and xenograft models in future studies.

In summary, this study once more corroborates the importance of molecular tumor profiling since that approach can generate crucial insights about the mechanisms of and response to novel treatment strategies. Our findings reveal a robust correlation between miR-216b and cell cycle alterations, which makes miR-216 a suitable predictive marker for experimental CDK4/6 inhibitor treatment in glioblastoma.

Author contributions

AL, DLG and FE designed the study. AL and DLG wrote the main manuscript. AL, SB, LM, CNJ, FCR, LK, LP, AN, KB and AL2 did the experiments. TRP, AW and RH helped with the analysis of histological slides. BK, GR, SSK, MM, DS, GW, KR, SM and JAH helped with sample acquisition. BN, LM, LGP, SM helped with data analysis and kindly provided access to their facility and technical support. ASB, DLG, FE, KR, DL, WB, JAH and RH supervised the study, corrected the results and proof read the manuscript. LE carried out the figure design. MH and MP executed mRNA sequencing and data analysis. JR and DK conducted cell data and FACS analysis. DLG and FE, WB and SM revised the initial study concept, supervised the experiments and proof read the manuscript.

Funding

This work was supported by Innovative Interdisciplinary Cancer Research grant of the Medical-Scientific Fund of the Mayor of the federal capital Vienna (Fonds des Bürgermeisters der Stadt Wien), via a Grant to FE (Number 19071) and DLG (Number 21165), the Austrian Science Fund—FWF (T 906-B28 to DLG) and the Comprehensive Cancer Center- Research Grant (to DLG).

Data availability

All data relevant to the study is accessible in the article or as supplementary figures and tables. Sequencing or raw data will be provided upon request.

Declarations

Ethics approval and consent to participate

This study was approved by the ethics committee of Upper Austria, Linz (E-39-15) and ethics committee of Medical University of Vienna (EK 1616/2020 and EK419/2008).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.