https://orcid.org/0000-0003-4600-3648
Figures
Abstract
The long non-coding RNA (lncRNA), HAR1A is emerging as a putative tumour suppressor. In non-neoplastic brain cells, REST suppresses HAR1A expression. In gliomas REST acts as an oncogene and is a potential therapeutic target. It is therefore conceivable that REST promotes glioma progression by down-regulating HAR1A. To test this hypothesis, glioma clinical databases were analysed to study: (I) HAR1A/REST correlation; (II) HAR1A and REST prognostic role; (III) molecular pathways associated with these genes. HAR1A expression and subcellular localization were studied in glioblastoma and paediatric glioma cells. REST function was also studied in these cells, by observing the effects of gene silencing on: (I) HAR1A expression; (II) cancer cell proliferation, apoptosis, migration; (III) expression of neural differentiation genes. The same phenotypes (and cell morphology) were studied in HAR1A overexpressing cells. Our results show that REST and HAR1A are negatively correlated in gliomas. Higher REST expression predicts worse prognosis in low-grade gliomas (the opposite is true for HAR1A). REST-silencing induces HAR1A upregulation. HAR1A is primarily detected in the nucleus. REST-silencing dramatically reduces cell proliferation and induces apoptosis, but HAR1A overexpression has no major effect on investigated cell phenotypes. We also show that REST regulates the expression of neural differentiation genes and that its oncogenic function is primarily HAR1A-independent.
Citation: Waters E, Pucci P, Rahman R, Yatsyshyna AP, Porter H, Hirst M, et al. (2024) REST-dependent glioma progression occurs independently of the repression of the long non-coding RNA HAR1A. PLoS ONE 19(11): e0312237. https://doi.org/10.1371/journal.pone.0312237
Editor: Peyman Tabnak, Tabriz University of Medical Sciences, ISLAMIC REPUBLIC OF IRAN
Received: May 8, 2024; Accepted: October 3, 2024; Published: November 27, 2024
Copyright: © 2024 Waters et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the article and its Supporting Information files.
Funding: This work was supported by the European Union’s NextGenerationEU (grant number CN00000041). The grant was awarded to Istituto Italiano di Tecnologia (IIT), with Dr. Benedetto Grimaldi as the recipient of the funding.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Gliomas are the most common primary brain tumours with very high mortality rates [1]. Their morphology can resemble that of different glial cells, including astrocytes, oligodendrocytes, and ependymal cells. This results in differing histologies, including lower grade I, II or III astrocytoma, oligodendroglioma, or the mixed cell oligoastrocytoma. Eighty percent of astrocytomas progress to the more aggressive grade 4 glioblastoma (GBM), which is scarcely differentiated. Patients diagnosed with GBM have a 5-year survival rate of less than 5%, which has not improved in decades [2]. Lower grade gliomas are the most frequent brain tumour in children, whereas GBM is a common brain tumour in adults [3]. Gliomas are highly infiltrative; therefore, surgery can rarely remove the whole tumour, and is not always an option depending on the location of the neoplasm (i.e., proximity to eloquent anatomical regions) and age of the patient. For example, Diffuse Midline Glioma (DMG), one of the most aggressive pediatric gliomas [4], is almost invariably inoperable. The characteristic inter- and intra-heterogeneity of gliomas makes developing therapeutic targets difficult [5]. Further research is required to identify earlier diagnostic markers, prognostic biomarkers for tumour progression and timing of recurrence and to identify more efficacious patient-tailored targeted therapeutics.
Long non-coding RNAs (lncRNAs) are transcripts longer than 200 nucleotides (nts), which do not encode proteins. LncRNAs have been implicated in multiple diseases, including cancer. Their expression is often specific to cellular location, tissue, and disease [6]. As lncRNAs are functional units, their subcellular localization is important to determine their function. Two understudied lncRNAs are the transcripts encoded by human accelerated region 1 A (HAR1A) and HAR1B. These lncRNAs overlap on human chromosome 13 and their sequences are human-specific [7]. HAR1A was shown to have a tumour suppressor role in oral cancer via interaction with Alpha Kinase-1 (ALPK1), and in non-small cell lung cancer via modulation of the Signal Transducers and Activator of Transcription 3 (STAT3) pathways [8, 9]. However, the function of these lncRNAs in gliomas has not been elucidated.
HAR1A and HAR1B transcription was shown to be repressed by RE1-silencing transcription factor (REST) in Huntington’s disease [10]. Intriguingly, REST is known to have an oncogenic role in gliomas, through the repression of tumour-suppressor genes. For example, REST downregulates synapsin-1 (SYN1), contributing to glioma pathogenesis [11, 12]. It is currently unknown whether the REST-HAR1 axis is functionally relevant for glioma pathogenesis, or whether REST exerts its oncogenic function in a HAR-independent manner (e.g., primarily via the repression of protein-coding genes).
Based on this background, we hypothesize that REST exerts its oncogenic function by repressing HAR1A and HAR1B. Hence, the objectives of this study are to investigate the clinical significance of REST, HAR1A, and HAR1B expression in pediatric and adult gliomas and their functional roles in key cancer hallmarks (including cell survival, proliferation and migration).
Materials and methods
Bioinformatic analysis
CBioPortal.
RNA-sequencing expression data of HAR1A, HAR1B and REST was analyzed on cBioPortal (https://www.cbioportal.org/, accessed on 4th April 2023), to determine correlations in lower grade gliomas and glioblastoma. Data were obtained from Brain Lower Grade Glioma (TCGA, Firehose Legacy) and Glioblastoma Multiforme (TCGA, Firehose Legacy). Linear regression analysis was performed on GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA). Sample size: the CBioPortal LGG study had 530 samples– 512 of these contained HAR1A data and 514 of these with HAR1B data; he GMB study had 619 samples– 157 of these had both HAR1A and HAR1B.
Additional analyses on the Pediatric CbioPortal (https://pedcbioportal.kidsfirstdrc.org/, dataset name: Pediatric Brain Tumor Atlas, accessed on the 25th of July 2023) were performed on the PBTA study (2182 samples).
Metascape.
Gene ontology analysis of HAR1A and HAR1B was conducted on Metascape (https://metascape.org/, accessed on 4th April 2023). Genes co-expressed with HAR1A, HAR1B, and REST, with an R2 threshold of ≥0.7 were selected from the cBioPortal lower grade glioma study, were queried on Metascape, and analyzed with ontology sources. The most significant regulators and most significant downstream pathways were analyzed on GraphPad Prism 8.
Kaplan-Meier plotter for correlation of prognosis (clinical datasets).
HAR1A, HAR1B, and REST expression correlation with prognosis in lower grade gliomas and glioblastoma were analyzed in the KM-plotter platform (https://kmplot.com), using the PAN-cancer Glioblastoma (N = 153) dataset and the PAN-cancer Lower Grade Glioma (N = 510) datasets. Genes co-expressed with HAR1A were selected from the Metascape analysis (R2 threshold of ≥0.7), and then analysed using the PAN-cancer Glioblastoma (N = 153) dataset on KM-plotter Private Edition (https://kmplot.com/private/). Genes significantly correlated with glioblastoma prognosis were further analyzed in vitro.
Cell culture
U-373 MG (Uppsala) cells were obtained from the European Collection of Authenticated Cell Cultures (ECACC) and were cultured in EMEM media (ATCC) supplemented with a 1% (v/v) antibiotic-antimycotic solution and a 10% heat-inactivated fetal bovine serum (Thermo Fisher Scientific, Loughborough, UK). VUMC-DIPG-A (H3.3 K27M) cells were kindly provided by Dr. Esther Hulleman (VUMC Cancer Center, Amsterdam, the Netherlands), and were cultured in 1:1 DMEM-F12 and Neurobasal-A cell media and supplemented with a 10% heat-inactivated fetal bovine serum 1% (v/v) glutamax supplement, a 1% (v/v) antibiotic–antimycotic solution, 10 mM HEPES, a 1% (v/v) MEM non-essential amino acid solution, and 1 mM sodium pyruvate (Thermo Fisher Scientific), hereafter referred to as TBM medium. GCE28 and GIN28 cells were kindly provided by Dr. Ruman Rahman (University of Nottingham, Nottingham, UK), and were cultured with DMEM low glucose, supplemented with a 1% (v/v) antibiotic-antimycotic solution and a 10% heat-inactivated fetal bovine serum (Thermo Fisher Scientific). Cell lines were grown adherent and passaged using 0.25% (v/v) trypsin-EDTA and washed using HBSS (Thermo Fisher Scientific). LNCaP cells were obtained from ATCC and cultured in RPMI 1640 (GIBCO, Thermo Fisher) supplemented with 10% FBS and 1% penicillin-Streptomycin (Gibco, Thermo Fisher). Cells were incubated at 37°C and 5% CO2 in a humidified incubator.
SiRNA reverse transfection
SiRNA knockdown of REST was performed using the reverse transfection method. Cells were seeded in 6-well plates (1.2 × 105 cells/well) with a transfection mix containing RNAiMAX lipofectamine (Invitrogen, Thermo Fisher Scientific), siRNAs (IDT, Leuven, Belgium) and opti-MEM (Thermo Fisher Scientific). The final concentration of siRNA doses was 2 nM. The duplexes used were: anti-REST DsiRNA hs.Ri.REST.13.1 and hs.Ri.REST.13.2, and scrambled negative control DS NC1.
Lentiviral overexpression
U-373 cells were stably transfected with a lentivirus-derived particle that induced HAR1A overexpression (Genecopoeia USA, Cat# LPP-CUST-GVO-GC). 7.0 × 104 U-373 cells were seeded in 24-well plates and incubated overnight. The next day, fresh media was supplemented with a final concentration of 8 μg/ml polybrene (Sigma Aldrich, Gillingham, UK) to increase the efficiency of transduction. 5 μl of the purified human HAR1A lentiviral particles (Titer: 1.37 × 108 TU/ml where 1 TU = 100 copies of viral genomic RNA) were added to the wells and incubated overnight. The next day, the wells were washed with media and incubated until confluent. 72 hours post-transduction, U-373 cells were split 1 in 2 into 6-well plates and were incubated to adhere. 0.3 μg/ml of puromycin (Sigma Aldrich) was used for antibiotic selection for 2 weeks, with media change every 3 days. These cells were referred to as U-373-HAR1A and results were normalized to cells transduced with the negative control vector, U-373-NC.
Analysis of gene expression
Total RNA was isolated from cultured cells using the RNeasy Plus Mini Kit (Qiagen, Manchester, UK), according to the manufacturer’s instructions. Reverse transcription of 1 μg of RNA was performed using the High‐Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Loughborough, UK), according to kit instructions. This cDNA was diluted 1 in 10 for RT-qPCR analysis. TaqMan gene expression assays were obtained from Thermo Fisher: HAR1A (Hs05038333_s1), HAR1B (Hs03299152_m1), REST (Hs05028212_s1), GAPDH (Hs02786624_g1), MALAT1 (Hs00273907_s1), HPRT1 (HS02800695_M1), SNAP25 (Hs00938957_m1), CPLX1 (Hs00362510_m1), SYN1 (Hs00199577_m1), DLG4 (Hs01555373_m1), GABRG2 (Hs00168093_m1). GAPDH was used as the housekeeping gene, unless otherwise stated in the figure legends.
Analysis of protein levels
Cell lysates were extracted using RIPA buffer. The protein was quantified using the Pierce BCA assay, according to the manufacturer’s protocol. Protein was resolved by gel electrophoresis on reducing SDS/PAGE. The proteins were transferred onto a nitrocellulose membrane. The membrane was then blocked in 5% non-fat milk dissolved in Tris-buffered saline (TBS). The blots were incubated overnight at 4°C with protein-specific antibodies. REST and GAPDH were the primary antibodies used. After the overnight incubation, the blots were washed in TBS with 0.1% Tween® 20 Detergent. The blots were then incubated with HRP-conjugated anti-rabbit secondary antibody for 1 hour at room temperature. The blot was then washed in TBS with 0.1% Tween® 20. Finally, the blot was revealed by ECL substrate and visualized using Syngene Gbox with GeneTools software.
Analysis of publicly available RNA-seq datasets
RNA-seq datasets from GIN31 and GCE31 cell lines (GSE233380) and MRI localised glioblastoma samples (GSE59612) were retrieved from the Gene Expression Omnibus (GEO). Data from primary and recurrent glioblastoma samples was retrieved from the TCGA-GBM cohort using the TCGAbiolinks (v2.31.3) package in R. Raw gene count data was processed in R (version 4.1.1) using RStudio (2021.09.0+351 Release). Differential expression analysis and variance stabilizing transformation was conducted using DESeq2 (v1.34.0). Differentially expressed genes were defined as those with an absolute log2 fold change greater than 1 and an adjusted p value less than 0.05.
Localisation of lncRNAs
RNA was extracted from cells and fractionation was performed using the PARIS kit (Invitrogen), according to the manufacturer’s instructions. For RT-qPCR validation, the probes MALAT1 was used as a nuclear marker, GAPDH as the cytoplasmic marker, and HPRT1 as the housekeeping gene.
Proliferation assays
U-373 and DIPGA cells were transfected with siRNAs, or the overexpression model plated and incubated for 2, 4, 6, and 8 days. On these days, the cells were trypsinised and a pellet was obtained to count the number of cells using the LUNA cell counter (Logos Biosystems). The number of cells were plotted using GraphPad Prism 8.
Caspase 3/7 assays
1.0 × 104 U-373 cells were plated on white 96-well pates and treated with siRNAs. On day 4 post-transfection, Caspase-Glo reagent (Promega, Southampton, UK) was added to each well, according to the manufacturer’s instructions. Luminescence was then quantified using the BMG POLARstar plate reader (BMG Labtech, Aylesbury, UK).
Wound healing assay (migration).
2.5 × 105 HAR1A-overexpressed and control cells were plated in 24-well plates. Once confluent, a scratch was made in the cell monolayer using a sterile P20 pipette tip. The cells were imaged twice a day until the wound was closed. Images were analyzed using the MRI wound-healing tool on ImageJ.
Statistical analysis
All data were obtained from two or three independent experiments, indicated in Figure legends by N, and analyzed using GraphPad Prism 8 software. Values presented as mean ± standard error of the mean (SEM). Significant differences between the groups were analyzed using one-way ANOVA with Dunnett’s multiple comparison test or two-way ANOVA with Šídák’s multiple comparison test. A p < 0.05 was set as threshold for statistical significance.
Results
Expression and clinical significance of REST, HAR1A and HAR1B in human gliomas
To analyze the relationship between HAR1A and HAR1B with REST in gliomas, we began by correlating their RNA expression in lower grade gliomas and glioblastoma, using the Cbio Portal. Linear regressions were analyzed in the following datasets: Brain Lower Grade Glioma (TCGA, Firehose Legacy) and Glioblastoma Multiforme (TCGA, Firehose Legacy) (Fig 1). We found that HAR1A and HAR1B have a significant positive correlation in both datasets (Fig 1A and 1D), HAR1A and REST have a significant negative correlation in both datasets (Fig 1B and 1E), and HAR1B and REST have a significant negative correlation in lower grade gliomas (Fig 1C), but unexpectedly, do not correlate in the GBM dataset (Fig 1F).
Linear regressions in lower grade gliomas of a) HAR1A and HAR1B (p<0.001, R2 0.65), b) HAR1A and REST (p<0.001, R2 0.2), c) HAR1B and REST (p<0.001, R2 0.18). Linear regressions in glioblastoma of d) HAR1A and HAR1B (p<0.001, R2 0.29), e) HAR1A and REST (p<0.001, R2 0.07), f) HAR1B and REST. Expression z-score relative to all samples.
Following the linear regression analysis of HAR1A, HAR1B, and REST, we determined whether aberrant expression of these lncRNAs and REST impacted glioma patient prognosis, using a Kaplan Meier analysis (Fig 2). In the lower grade glioma dataset, we found that a poorer prognosis of patients correlated with low expression of HAR1A and HAR1B (Fig 2A and 2B), and with high expression of REST (Fig 2C). In the GBM dataset, we found no significant correlation between patient overall survival for the lncRNAs or REST (Fig 2D–2F).
Kaplan-Meier plots from KMP-Plotter show high expression (red) or low expression (grey), relative to the median. Lower grade glioma: a) low HAR1A expression correlates with significantly worse lower grade glioma patient survival; b) low HAR1B expression correlates with significantly worse lower grade glioma patient survival; c) high REST expression correlates with worse lower grade glioma patient survival. Glioblastoma: d) HAR1A; e) HAR1B; f) REST expression does not correlate with glioblastoma patient prognosis.
To test whether our results were also valid for pediatric malignancies, we performed correlation and survival analyses on a large pediatric glioma dataset (Pediatric CbioPortal, PBTA study). Our results confirmed that HAR1A and HAR1B are negatively correlated with REST expression in pediatric gliomas (S1A–S1C Fig). Interestingly, we found that none of the investigated genes had a prognostic significance in this dataset (S1D–S1F Fig).
Taken together, these results suggest that HAR1A and REST are negatively correlated in all glioma subtypes analyzed, and that the expression of these molecules is of prognostic relevance in adult lower grade gliomas, but not in the most aggressive gliomas (DIPG and GBM).
Upstream and downstream pathways associated with HAR1A and HAR1B
After showing that HAR1A and HAR1B have positive prognostic roles in lower grade gliomas, we determined possible downstream pathways and upstream regulators of the lncRNAs, using gene ontology (Fig 3). The genes most significantly coexpressed with HAR1A and HAR1B in lower grade gliomas were analyzed in the publicly available ontology resource Metascape. As expected from the previous analysis, we identified REST as the most significant regulator of both lncRNAs (Fig 3A and 3B). Further to this analysis, we also determined the most likely downstream pathways of HAR1A and HAR1B (Fig 3C and 3D): chemical synaptic transmission and other synapse-related pathways.
Analysis of a) HAR1A upstream regulators, b) HAR1B upstream regulators, c) HAR1A downstream pathways (enriched terms), d) HAR1B downstream pathways. Analysis conducted on Metascape. -Log10(P) is p-value in log base 10.
Downstream pathways associated with REST
Using the same dataset for REST-correlated genes, we found that REST expression is associated with several cancer-relevant pathways (Fig 4), including GTPase activity, TP53 regulation, MAPK cascade, cell division, and wound healing.
Gene ontology analysis with REST co-expressed genes, associating downstream pathways (enriched terms). Analysis conducted on Metascape. -Log10(P) is p-value in log base 10(P) is p-value in log base 10.
Expression of REST, HAR1A, and HAR1B in glioma cells
Following the bioinformatic analysis, we observed low expression of HAR1A and HAR1B in U-373 (GBM) and DIPGA (DMG) cell cultures (Ct>35). To understand whether these expression levels could be functionally relevant, we compared HAR1A expression in glioma vs other cell types, using the human Expression Atlas [13]. Among 16 organs, the brain and the prostate showed the highest levels of HAR1A (S2A Fig). We therefore measured HAR1A levels in prostate cancer cells (LNCaP), finding that HAR1A has a higher expression in the prostate cancer cells, compared to GBM and DMG cells (S2B Fig). To the contrary, REST expression was highly detectable in both glioma cells, as shown in previous studies [14]. Therefore, we silenced REST using siRNAs, as shown previously [10] which resulted in increased expression of HAR1A and HAR1B in both DIPGA and U-373 cell lines (Fig 5A, 5B and 5D–5G). Moreover, we measured protein levels by western blot, confirming that REST protein expression was reduced in both cell lines (Fig 5C). For the siRNA that we used, siREST-2 significantly increased HAR1A and HAR1B in DIPGA cells, likely because this siRNA had the largest silencing effect on REST (Fig 5D and 5E). Next, in U-373 cells we found that REST silencing significantly increased the expression of HAR1A (Fig 5F), however, HAR1B was not expressed in this cell line.
REST silencing using two independent siRNAs in a) DIPGA and b) U-373. c) Western blot of REST levels when gene silenced in U-373 and DIPGA cells. d) HAR1A expression upon REST silencing in DIPGA, e) HAR1B expression upon REST silencing in DIPGA, f) HAR1A expression upon REST silencing in U-373. g) HAR1A and REST expression upon HAR1A overexpression and control in U-373. In a,b,d-g) bars represent the mean values of 3 independent experiments with error bars denoting SEM. GAPDH was used as the housekeeping gene. One-way ANOVA with Dunnett’s multiple comparison tests used to compare the siRNAs silenced or HAR1A overexpressing cells to the negative control. ** p<0.01, ***p<0.005, ****p<0.001. N = 3.
In parallel with the REST silencing model, we also stably overexpressed HAR1A in the U-373 cell line (Fig 5G) and observed that the expression of REST did not change between the control and HAR1A overexpression.
After determining the expression of HAR1A and HAR1B in the glioma cell lines, we identified their subcellular localization (Fig 6). The lncRNA expression was analyzed by cell fractionation and RT-qPCR in DIPGA (Fig 6A), U-373 (Fig 6B) and HAR1A-overexpressed U-373 (Fig 6C) cell lines. We found HAR1A to be primarily localized in the organelle enriched component in all tested cell lines (which is mostly nuclear, as independently validated in S3 Fig), whereas HAR1B to be primarily cytoplasmic in DIPGA cells. Using specific markers of nuclear and cytoplasmic fractions, we independently validated our cell fractionation results. Localization of HAR1B was not determined in cells U-373 (Fig 6B) and HAR1A-overexpressed U-373, as HAR1B was not expressed in these cell lines.
Expression of a) HAR1A and HAR1B in DIPGA cells, b) HAR1A in U-373 cells, c) HAR1A in HAR1A overexpressed U-373 cells. MALAT1 is nuclear control, and GAPDH is cytoplasmic control. HPRT1 was used as the housekeeping gene. N = 3.
Effects of REST and HAR1A on glioma cell proliferation
Having identified the subcellular localization of the lncRNAs in the cell, we then investigated whether REST silencing and HAR1A overexpression affect the cell proliferation and apoptosis of cancer cells (Fig 7). REST was silenced with siRNAs in DIPGA cells and U-373 cells, then cell counts were taken every 2 days over 6 days. There was no significant difference between the control and REST-silenced cells for DIPGA cells (Fig 7A). For U-373 cells, we found REST silencing significantly reduced cell proliferation, compared to the control (Fig 7B; 68–86% decrease vs siRNA control). A caspase 3/7 apoptosis assay was performed on these cells on day 4 of REST-silencing (Fig 7X). This showed that REST silencing resulted in an increase in caspase activity in U-373 cells. To identify whether this difference in U-373 cell proliferation and survival was mediated by the REST-dependent increase in HAR1A expression, we repeated the cell count using a HAR1A overexpression and negative control models (Fig 7D). Here, we found no difference between the control vector and HAR1A overexpressing cells in 2 out of 3 timepoints (2 and 4 days). HAR1A overexpression seemed to induce a slight (albeit significant) increase in cell proliferation after 6 days (22% increase vs control vector).
a) Cell count assay with REST-silencing in DIPGA cells. b) Cell count assay with REST-silencing in U-373 cells, c) apoptosis assay in U-373 cells, d) cell count assay in U-373 HAR1A_overexpression cell line. Two-way ANOVA with Dunnett’s multiple comparison test used to compare the siRNAs silenced or HAR1A overexpressed cells to the negative control in the cell count assays. One-way ANOVA with Dunnett’s multiple comparison test used to compare the siRNAs silenced cells to the negative control in the apoptosis assay. * p<0.05, ** p<0.01, ***p<0.005, ****p<0.001. N = 3. Two-way ANOVA with post-hoc multiple comparison test for the HAR1A OE experiment in d).
Taken together, these data and the results of the previous section show that REST inhibits the expression of HAR1A in glioma cells, and that REST appears to be essential for the proliferation of GBM cells. However, HAR1A modulation alone does not seem to have an inhibitory effect on the proliferation of GBM cells.
Effects of HAR1A on glioma cell migration, nuclear structure, and differentiation
In addition to its role in cell proliferation, REST has been previously shown to be a key driver of glioma cell migration [14]. For this reason, we analyzed the expression of REST and HAR1A in primary GBM cells derived from the tumour core and from the invasive margin of the same patient (patient 28) [15]. Our data showed a significant up-regulation of REST, but not HAR1A in the invasive margin, relative to the tumour core (Fig 8A). To validate against paired gene expression data of tumour core and invasive margin cell lines from an independent patient (patient 31) we analysed untreated controls from published RNA-seq data from our previous study of GIN31 and GCE31 response to electric field stimulation [16]. In contrast, REST was not significantly upregulated in GIN31 compared to GCE31 (Fig 9A). To further investigate the clinical relevance of REST and HAR1A/HAR1B expression, we compared the expression of these genes in a publicly available dataset of glioblastoma samples taken from MRI contrast enhancing (CE) tumour core and non-enhancing (NE) tumour periphery regions [17]. REST, HAR1A, and HAR1B were all differentially expressed between the tumour core and periphery samples (Fig 9B). However, as the non-enhancing region contains both invasive glioblastoma cells and surrounding parenchyma and this change is consistent with comparison to healthy brain controls, we are unable to comment on expression of REST in invasive glioblastoma relative to core in these samples. Finally, as glioblastoma invasive cells are known to drive tumour recurrence, we analysed expression of these genes in primary and recurrent glioblastoma samples from TCGA. We identified 595 differentially expressed genes (padj < 0.05 and absolute log2 fold change > 1), however there was no significant difference in REST, HAR1A, or HAR1B gene expression between these unmatched primary and recurrent samples (Fig 9C). Conversely, relative to normal solid tissue samples, REST was upregulated in primary and recurrent tumour samples and HAR1A / HAR1B were downregulated in both primary and recurrent tumours (Fig 9C).
a) HAR1A, HAR1B and REST expression in GIN28 (invasive margin) and GCE28 (tumour core), data obtained with RT-qPCR; b) Images of the wound closing over 72 hours, c) percentage of the scratch is open for HAR1A U-373 control and overexpression. GAPDH was used as the housekeeping gene. For a), two-way ANOVA with Šídák’s multiple comparison test was used to compare the gene expression between the two cell lines. **p<0.01. N = 3.
a) VST normalised gene counts were not differentially expressed between GIN31 and GCE31 (n = 3) (DESeq2 DEA adjusted p value > 0.05), mean + SEM. b) VST normalised gene read counts from MRI localised tissue samples from the tumour core (CE, contrast enhancing, n = 39) tumour margin (NE, non-enhancing, n = 36), and non-neoplastic (NN, from access biopsy from patients undergoing ventriculoperitoneal shunt placement, n = 17) controls. DESeq2 DEA showed that REST (Log2 Fold change -0.58, adjusted p value 1.708477e-05), HAR1A (Log2 Fold Change 3.74, adjusted p value 2.310446e-17), and HAR1B (Log2 Fold Change 3.24, adjusted p value 1.059286e-11) were differentially expressed between NE and CE tumour regions. c) VST normalised gene read counts from unmatched Primary (n = 158) and Recurrent (n = 13) tumour samples and non-tumour brain tissue (n = 5) controls from the TCGA-GBM cohort. DESeq2 DEA showed no difference in REST, HAR1A, and HAR1B expression between recurrent and primary tumours (adjusted p value > 0.05).
In keeping with this observation, we did not find any changes in cell migration between the HAR1A overexpression model and control in U-373 cells, using a scratch assay (Fig 8B, 8C).
Having not yet found sufficient evidence of a functional role of HAR1A, we next investigated the correlations between the expression of HAR1A and synaptic genes. This is based upon the previous gene ontology results indicating that HAR1A may have a role in synaptic transmission (Fig 3C). Based on our pathway analysis, 15 synaptic genes were significantly associated with HAR1A expression. We investigated the prognostic role of these genes through the Kaplan Meier Plotter dataset (Fig 10). Five synaptic genes were selected based on their prognosis status in gliomas (higher expression = better prognosis): SNAP25 (encoding Synapsis Associated Protein 25), CPLX1 (encoding Complexin 1), SYN1, DLG4 (encoding Postsynaptic Density Protein 95), and GABRG2 (encoding Gamma-Aminobutyric Acid Type A Receptor Subunit Gamma2).
Kaplan-Meier plots from KMP-Plotter show high expression (red) or low expression (grey), relative to the median. a) High expression of SNAP25 results in worse patient survival. b) High expression of CPLX1 results in worse patient survival. c) High expression of SYN1 results in worse patient survival. d) High expression of DLG4 results in worse patient survival. e) High expression of GABRG2 results in worse patient survival.
The expression of these genes was measured by RT-qPCR in siREST-silenced DIPGA and U-373 cells and HAR1A-overexpressed cells (Fig 11). Our results showed that REST silencing caused the overexpression of CPLX1, SYN1 and SNAP25 in DIPGA (Fig 11A), and all synaptic genes in U-373 (Fig 11B). However, we did not find any effects of HAR1A overexpression on the expression of the same genes (Fig 11B).
a) Expression of synaptic genes in REST-silenced DIPGA cells, b) Expression of synaptic genes in REST-silenced U-373 cells, c) expression of synaptic genes in HAR1A-overexpressed U-373 cells. * p<0.05. N = 3.
Overall, our results show that REST exerts pleiotropic pro-oncogenic functions in gliomas, as shown by its pro-survival and anti-apoptotic effects, and by its higher expression in invasive glioma cells. As expected, REST suppresses the expression of HAR1A in both pediatric and adult glioma cells. However, none of the oncogenic roles of REST seem to be mediated by HAR1A.
Discussion
In this study, we have shown that HAR1A is negatively regulated by REST in glioma cells, and that the expression of these two molecules has an opposite prognostic significance in glioma patients. Our results are in line with previous publications [18] but expand these findings on two key points: (1) we show that the prognostic role of these molecules is more relevant in lower grade gliomas, compared to GBM. This may be due to the extremely short survival time of GBM patients, which makes prognostic stratification more challenging; (2) we show that REST exerts its oncogenic function through different mechanisms that include enhancing GBM cell proliferation and survival, as well as reducing the expression of neural differentiation genes (but largely/mostly not via HAR1A repression). Loss of neural differentiation is a key step in gliomagenesis [19]. REST is a known negative regulator of neural differentiation, with repressive effects on the expression of SNAP25 [20] and SYN1 [21]. However, our results show that REST also inhibits the expression of CPLX1 and DLG4. CPLX1 is a protein required for the calcium-dependent exocytosis of synaptic vesicles [22]. DLG4 is a post-synaptic scaffold protein [23], which has been implicated in glioma pathogenesis by previous bioinformatic analysis [24]. To the best of our knowledge, the function of the CPLX1 protein has never been described in gliomas. Here we show that the CPLX1 gene is suppressed by REST and that higher CPLX1 expression predicts better prognosis in gliomas. These findings warrant further exploration of the REST-CPLX1 axis.
To broaden the significance of our findings, we have employed two glioma cell lines with different characteristics: one deriving from an adult GBM and the other from a pediatric glioma. The results obtained with the two cell lines are only partially overlapping, probably because of the heterogeneous nature of gliomas. Whilst the regulation of HAR1A by REST seems to be a cell type-independent mechanism, the effects of REST silencing on proliferation were much more marked in GBM than in DMG cells. Furthermore, we confirmed that HAR1A and HAR1B expression is comparable intra-tumour cell lines derived from both the GBM core and clinically relevant invasive margin, the latter region reflective of residual disease post-surgery.
Due to the strong negative correlation between REST and HAR1A, we have hypothesized that the oncogenic activity of REST in glioblastoma was at least in part mediated by HAR1A silencing. For this reason, we have overexpressed HAR1A, and confirmed that this transcript is probably localized in the nucleus, where it has been shown to have onco-suppressive functions [18]. Our overexpression studies did not corroborate a tumour suppressive role for HAR1A in glioma. The overexpression of this lncRNA had no inhibitory effect on cell proliferation and migration. If anything, we found that the overexpression of HAR1A could slightly increase the proliferation of GBM cells. This appears to be in contrast with our prognostic findings and with other studies describing a tumour suppressive role for HAR1A. However, the magnitude of HAR1A effect on proliferation is modest (22% increase; compared to a 68–86% reduction in proliferation upon REST silencing). Since REST silencing induces both cell death and HAR1A over-expression, it is likely that the slight pro-survival effects associated with HAR1A upregulation are counterbalanced by the inactivation of several other REST-dependent pro-survival pathways. In keeping with this hypothesis, a recent study showed that REST silencing down-regulates several pathways involved in glioma cell mitosis and cell proliferation [25]. This hypothesis is corroborated by our GO Pathway analysis (Fig 4), showing that REST expression is correlated with key oncogenic pathways (e.g., MAP kinase, cell division, wound healing). Synaptic reprogramming is emerging as a key driver of oncogenesis in gliomas [26]. We were therefore intrigued to identify a strong bioinformatic signal linking synaptogenesis with HAR1A expression. However, our RT-qPCR experiments (Fig 11) shows that this lncRNA does not regulate the expression of synaptogenesis genes in GBM. Nuclear structure and nuclear/cytoplasm ratio alterations are another hallmark of cancerogenesis [27], which may be controlled by lncRNAs. Hence it may be conceivable that HAR1A affects this aspect of carcinogenesis. Notably, HAR1A-downstream pathways in gliomas have not been explored so far. This potential function should be therefore investigated by future studies.
Our study provides new insights into the REST-HAR1A relationship in gliomas; however, we recognize that our investigation has some limitations. First, we studied the role of HAR1A in a limited collection of cell types (glioma and glioblastoma). HAR1A is also expressed in neurons [7] and in other cell types, including the prostate (S2 Fig), lung epithelium [9] and peripheral blood [28]. It is therefore conceivable that HAR1A could have a functional role in these cell types, which are not the object of our investigation. In addition, our overexpression system induced very high levels of HAR1A expression, which are likely non-physiological. The lack of measurable phenotypes in these conditions further suggests a minor role of this lncRNA in GBM cells. Our GO term analyses (Figs 3 and 4) are based on co-expression between REST/HAR1 and protein coding genes. These correlations are not a definitive mechanistical explanation of the pathways activated by these genes, but they have provided valuable insights into REST-dependent regulatory pathways, some of which we have experimentally confirmed (Fig 11). Finally, we would like to clarify that our overexpression experiments rule out a trans role for this lncRNA, but do not completely exclude a locus-specific function (cis) for this transcript. Also, given the known role of REST in glial migration [29], and its up-regulation in invasive glioma cells, it would have been interesting to test whether silencing this gene affected the migratory properties of invasive glioma cells. However, since REST silencing caused extensive cell death in a relatively short timeframe, the migration experiment could not be performed.
Conclusions
We show that REST can impact the prognosis and progression of gliomas via different mechanisms, including enhanced cell proliferation, repression of neural-differentiation genes and potentially increased migration. Our results seem to exclude an in vitro function for HAR1A in gliomas. This does not exclude an in vivo function for this lncRNA, perhaps mediated by the interaction between the microenvironment and malignant cells. This function could be studied in the future using animal models or 3D co-cultures. Considering the ongoing development of REST small molecule inhibitors for glioma treatment [30], our results have therapeutic relevance. We show that some REST-downstream effects (e.g. pro-survival signals, silencing of neural differentiation genes) are more therapeutically relevant than others (e.g. HAR1A silencing). This evidence may inform the selection of more promising inhibitors and the identification of biomarkers of efficacy in pre-clinical and clinical studies.
Supporting information
S1 Fig. Clinical significance of HAR1A, HAR1B and REST in paediatric gliomas.
(A-C): Linear regressions in paediatric gliomas of A) HAR1A and REST, B) HAR1B and REST, C) HAR1B and HAR1A. (D-F): Overall survival of paediatric glioma patients according to the expression of D) REST, E) HAR1A, F) HAR1B. Data from the Paediatric Cbioportal (PTBA-provisional, 2182 samples). Gene expression (mRNA) is measured as Z scores, log2 scale.
https://doi.org/10.1371/journal.pone.0312237.s001
(TIF)
S2 Fig. HAR1A expression in a panel of organs and cell lines.
(A) Expression plot of HAR1A from the Illumina Body Map (https://www.ebi.ac.uk/). (B) RT-qPCR (Ct values) of HAR1A expression in LNCaP (prostate), U-373 (brain) and DIPGA (brain) cells. GAPDH was used as reference.
https://doi.org/10.1371/journal.pone.0312237.s002
(TIF)
S3 Fig. Validation of nuclear and cytoplasmic fractionation.
Immunoblotting analysis of preparation of separate nuclear and cytoplasmic lysates from cancer cells trough PARIS Kit protocol (Invitrogen, Cat #AM1921). Cytoplasmic beta-Tubulin (Cell Signaling Cat #2128S) and GAPDH (Cell Signaling Cat #5174T) proteins were adopted to evaluate potential cytosolic contaminations in the nuclear fraction. Lysosomal LAMP1 (Cell Signaling Cat # 9091) protein was used to evaluate potential contamination of organelles in the nuclear fraction. Histone H3 (Invitrogen Cat # 710282) protein was used to confirm that the fraction separated from cytosolic fraction was nuclear.
https://doi.org/10.1371/journal.pone.0312237.s003
(TIF)
S1 Raw images. Raw blot images.
This Supporting Information file has raw blots showing expression of REST and GAPDH in U-373 cells from Fig 5c; REST and GAPDH in DIPGA cells from Fig 5c; as well raw blots showing expression of LAMP1, beta-tubulin, GAPDH, histone H3 from S3 Fig.
https://doi.org/10.1371/journal.pone.0312237.s004
(PDF)
References
- 1. Bondy ML, Scheurer ME, Malmer B, Barnholtz-Sloan JS, Davis FG, Il’yasova D, et al. Brain Tumour Epidemiology: Consensus from the Brain Tumour Epidemiology Consortium (BTEC). Cancer. 2008 Oct 1;113(7 Suppl):1953–68.
- 2. Ostrom QT, Gittleman H, Liao P, Rouse C, Chen Y, Dowling J, et al. CBTRUS Statistical Report: Primary Brain and Central Nervous System Tumours Diagnosed in the United States in 2007–2011. Neuro Oncol. 2014 Oct;16(Suppl 4):iv1–63.
- 3. Lin Z, Yang R, Li K, Yi G, Li Z, Guo J, et al. Establishment of age group classification for risk stratification in glioma patients. BMC Neurol. 2020 Aug 20;20:310. pmid:32819307
- 4. Jones C, Karajannis MA, Jones DTW, Kieran MW, Monje M, Baker SJ, et al. Pediatric high-grade glioma: biologically and clinically in need of new thinking. Neuro Oncol. 2017 Feb;19(2):153–61. pmid:27282398
- 5. Hernández Martínez A, Madurga R, García-Romero N, Ayuso-Sacido Á. Unravelling glioblastoma heterogeneity by means of single-cell RNA sequencing. Cancer Letters. 2022 Feb 28;527:66–79. pmid:34902524
- 6. Bridges MC, Daulagala AC, Kourtidis A. LNCcation: lncRNA localization and function. J Cell Biol. 2021 Jan 19;220(2):e202009045. pmid:33464299
- 7. Pollard KS, Salama SR, Lambert N, Lambot MA, Coppens S, Pedersen JS, et al. An RNA gene expressed during cortical development evolved rapidly in humans. Nature. 2006 Sep;443(7108):167–72. pmid:16915236
- 8. Lee CP, Ko AMS, Nithiyanantham S, Lai CH, Ko YC. Long noncoding RNA HAR1A regulates oral cancer progression through the alpha-kinase 1, bromodomain 7, and myosin IIA axis. J Mol Med (Berl). 2021 Sep;99(9):1323–34. pmid:34097087
- 9. Ma J, Cao K, Ling X, Zhang P, Zhu J. LncRNA HAR1A Suppresses the Development of Non-Small Cell Lung Cancer by Inactivating the STAT3 Pathway. Cancers (Basel). 2022 Jun 8;14(12):2845. pmid:35740511
- 10. Johnson R, Richter N, Jauch R, Gaughwin PM, Zuccato C, Cattaneo E, et al. Human accelerated region 1 noncoding RNA is repressed by REST in Huntington’s disease. Physiological Genomics. 2010 Feb 23;41(3):269–74. pmid:20179156
- 11. Wang G, Yang X, Qi M, Li M, Dong M, Xu R, et al. Systematic analysis identifies REST as an oncogenic and immunological biomarker in glioma. Sci Rep. 2023 Feb 21;13:3023. pmid:36810892
- 12. Zhang D, Li Y, Wang R, Li Y, Shi P, Kan Z, et al. Inhibition of REST Suppresses Proliferation and Migration in Glioblastoma Cells. Int J Mol Sci. 2016 May 3;17(5):664. pmid:27153061
- 13. Conti L, Crisafulli L, Caldera V, Tortoreto M, Brilli E, Conforti P, et al. REST controls self-renewal and tumourigenic competence of human glioblastoma cells. PLoS One. 2012;7(6):e38486.
- 14. Derrien T, Johnson R, Bussotti G, Tanzer A, Djebali S, Tilgner H, et al. The GENCODE v7 catalog of human long noncoding RNAs: analysis of their gene structure, evolution, and expression. Genome Res. 2012 Sep 1;22(9):1775–89. pmid:22955988
- 15. Smith SJ, Rowlinson J, Estevez-Cebrero M, Onion D, Ritchie A, Clarke P, et al. Metabolism-based isolation of invasive glioblastoma cells with specific gene signatures and tumourigenic potential. Neurooncol Adv. 2020 Jul 13;2(1):vdaa087.
- 16. Jain A, Gosling J, Liu S, Wang H, Stone EM, Chakraborty S, et al. Wireless electrical-molecular quantum signalling for cancer cell apoptosis. Nat Nanotechnol. 2024 Jan;19(1):106–114. pmid:37709951
- 17. Gill BJ, Pisapia DJ, Malone HR, Goldstein H, Lei L, Sonabend A, et al. MRI-localized biopsies reveal subtype-specific differences in molecular and cellular composition at the margins of glioblastoma. Proc Natl Acad Sci U S A. 2014 Aug 26;111(34):12550–5. pmid:25114226
- 18. Zou H, Wu LX, Mei Y, Li S, Mei Y, Liu YB, et al. lncRNAs PVT1 and HAR1A are prognosis biomarkers and indicate therapy outcome for diffuse glioma patients. Oncotarget. 2017 Aug 13;8(45):78767–80. pmid:29108264
- 19. Friedmann-Morvinski D, Bushong EA, Ke E, Soda Y, Marumoto T, Singer O, et al. Dedifferentiation of Neurons and Astrocytes by Oncogenes Can Induce Gliomas in Mice. Science. 2012 Nov 23;338(6110):1080–4. pmid:23087000
- 20. Duan P, Sun S, Li B, Huang C, Xu Y, Han X, et al. miR-29a modulates neuronal differentiation through targeting REST in mesenchymal stem cells. PLoS One. 2014;9(5):e97684. pmid:24841827
- 21. Paonessa F, Latifi S, Scarongella H, Cesca F, Benfenati F. Specificity protein 1 (Sp1)-dependent activation of the synapsin I gene (SYN1) is modulated by RE1-silencing transcription factor (REST) and 5’-cytosine-phosphoguanine (CpG) methylation. J Biol Chem. 2013 Feb 1;288(5):3227–39. pmid:23250796
- 22. Chang S, Reim K, Pedersen M, Neher E, Brose N, Taschenberger H. Complexin Stabilizes Newly Primed Synaptic Vesicles and Prevents Their Premature Fusion at the Mouse Calyx of Held Synapse. J Neurosci. 2015 May 27;35(21):8272–90. pmid:26019341
- 23. Levy AM, Gomez-Puertas P, Tümer Z. Neurodevelopmental Disorders Associated with PSD-95 and Its Interaction Partners. Int J Mol Sci. 2022 Apr 15;23(8):4390. pmid:35457207
- 24. Li W, Li K, Zhao L, Zou H. Bioinformatics analysis reveals disturbance mechanism of MAPK signaling pathway and cell cycle in Glioblastoma multiforme. Gene. 2014 Sep 1;547(2):346–50. pmid:24967941
- 25. Perycz M, Dabrowski MJ, Jardanowska-Kotuniak M, Roura AJ, Gielniewski B, Stepniak K, et al. Comprehensive analysis of the REST transcription factor regulatory networks in IDH mutant and IDH wild-type glioma cell lines and tumors. Acta Neuropathol Commun. 2024 May 6;12(1):72. pmid:38711090
- 26. Krishna S, Choudhury A, Keough MB, Seo K, Ni L, Kakaizada S, et al. Glioblastoma remodelling of human neural circuits decreases survival. Nature. 2023 May;617(7961):599–607. pmid:37138086
- 27. Jevtić P, Edens LJ, Vuković LD, Levy DL. Sizing and shaping the nucleus: mechanisms and significance. Curr Opin Cell Biol. 2014 Jun;28:16–27. pmid:24503411
- 28. Akbarzadeh S, Tayefeh-Gholami S, Najari P, Rajabi A, Ghasemzadeh T, Hosseinpour Feizi M, et al. The expression profile of HAR1A and HAR1B in the peripheral blood cells of multiple sclerosis patients. Mol Biol Rep. 2023 Mar;50(3):2391–8. pmid:36583781
- 29. Yu T, Lin Y, Xu Y, Dou Y, Wang F, Quan H, et al. Repressor Element 1 Silencing Transcription Factor (REST) Governs Microglia-Like BV2 Cell Migration via Progranulin (PGRN). Neural Plast. 2020 Nov 25;2020:8855822. pmid:33299399
- 30. Charbord J, Poydenot P, Bonnefond C, Feyeux M, Casagrande F, Brinon B, et al. High throughput screening for inhibitors of REST in neural derivatives of human embryonic stem cells reveals a chemical compound that promotes expression of neuronal genes. Stem Cells. 2013 Sep;31(9):1816–28. pmid:23712629