Abstract
Background
Glioblastoma, the most aggressive primary brain tumor, is genetically heterogeneous. Alternative splicing (AS) plays a key role in numerous pathologies, including cancer. The objectives of our study were to determine whether aberrant AS could play a role in the malignant phenotype of glioma and to understand the mechanism underlying its aberrant regulation.
Methods
We obtained surgical samples from patients with glioblastoma who underwent 5-aminolevulinic fluorescence-guided surgery. Biopsies were taken from the tumor center as well as from adjacent normal-appearing tissue. We used a global splicing array to identify candidate genes aberrantly spliced in these glioblastoma samples. Mechanistic and functional studies were performed to elucidate the role of our top candidate splice variant, BAF45d, in glioblastoma.
Results
BAF45d is part of the switch/sucrose nonfermentable complex and plays a key role in the development of the CNS. The BAF45d/6A isoform is present in 85% of over 200 glioma samples that have been analyzed and contributes to the malignant glioma phenotype through the maintenance of an undifferentiated cellular state. We demonstrate that BAF45d splicing is mediated by polypyrimidine tract-binding protein 1 (PTBP1) and that BAF45d regulates PTBP1, uncovering a reciprocal interplay between RNA splicing regulation and transcription.
Conclusions
Our data indicate that AS is a mechanism that contributes to the malignant phenotype of glioblastoma. Understanding the consequences of this biological process will uncover new therapeutic targets for this devastating disease.
Keywords: glioma, alternative splicing, BAF45d, PTBP1
Importance of the study
Glioblastoma, the most aggressive primary brain tumor, has an overall survival of less than 14 months. Glioblastoma is a genetically and histologically heterogeneous tumor, with few mutations occurring in more than 50% of the cases. The study of AS in this tumor is therefore highly relevant. Our results suggest that aberrant AS in BAF45d contributes to the pathogenesis of glioblastoma through the maintenance of an undifferentiated cellular state. We show that the aberrant splicing of BAF45d is mediated by PTBP1, an RNA-binding protein known to regulate splicing. Our results also identify a novel BAF45d/PTBP1 feedback mechanism in glioblastoma splicing. Of importance, our data implicate AS of BAF45d as a central mechanism in glioblastoma development, thereby introducing new avenues of study into pathways and potential therapeutic targets for this devastating tumor.
Glioblastoma, which is the most aggressive of all primary brain tumors, is genetically and histologically heterogeneous: even the most frequent mutations are found in <50% of cases.1 The study of alternative splicing (AS) therefore holds great promise for understanding this disease, as it plays a key role in the pathogenesis of many cancers.2 A previous global analysis of AS in high-grade gliomas, however, did not identify any genes with a consistent pattern of AS.3,4 In recent years, AS of 2 genes, pyruvate kinase muscle (PKM) and annexin 7 (ANXA7),5,6 has been shown to be important in the initiation and maintenance of glioblastomas. AS differences were not identified via global analysis, however, reducing the ability to assess the true relevance of AS in glioma pathogenesis. The introduction of 5-aminolevulinic acid (5-ALA) fluorescence-guided surgery (FGS)7 has greatly increased the extent of resection in glioblastoma, however, and the residual fluorescent tissue has prognostic value in patients with gross total resection.8 In addition, FGS facilitates the analysis of paired samples of tumor and adjacent normal tissue.
In this study, we conducted a global analysis of AS in paired samples obtained via FGS from the same patient. Our data reveal the aberrant AS of 45-KDa Brahma (BRM) and Brahma/SWI2-related gene 1 (BRG)–associated factor (BAF45d) in glioblastoma. BAF45d is a member of the switch/sucrose nonfermentable (SWI/SNF) complex,9 which is known to play a role in the development of the CNS.10 BAF45d withdraws from this complex as a necessary step in allowing neural progenitors to exit mitosis and initiate differentiation.11 Functionally, we demonstrate that BAF45d participates in the maintenance of an undifferentiated phenotype and in in vivo tumorigenesis. We show that the aberrant splicing of BAF45d is mediated by polypyrimidine tract-binding protein 1 (PTBP1), an RNA-binding protein known to regulate splicing.12,13 PTBP1 also contributes to neurogenesis14 and the malignant progression of glioblastomas.5,6 Most importantly, our results uncover that BAF45d also positively regulates the expression of PTBP1 and its participation in the regulation of AS.
Materials and Methods
Detailed materials and methods data can be found in the Supplementary materials.
Clinical Samples, RNA Extraction, and RNA Array Analysis
Samples were obtained from 20 glioblastoma patients at Clínica Universidad de Navarra and 30 glioblastoma patients from La Fe Hospital (Fig. 1A and Supplementary Tables S1 and S2) who underwent 5-ALA FGS. The study protocol was approved by our institutions’ ethical committees, and all patients gave written informed consent.
Fig. 1.
Fluorescence-guided surgery (FGS) and glioblastoma alternative splicing (AS) candidate discovery. (A) Image of 5-ALA FGS during glioblastoma resection. a: Image of the surgical field using a white-light module; b: image of the surgical field under 5-ALA fluorescence, where red/pink and blue correspond to the tumor and macroscopically normal tissue, respectively. (B) Schematic representation and representative agarose PCR gels of the AS of the following candidates obtained from global splicing analysis: BAF45d, RALGAPA, ITAG6, ARHGEF7, BBX, LRRFIP2, and NDE1. Normal brain (N); tumor (T). (C) Validation of the 7 candidate genes using conventional PCR in 10 additional patient samples. Quantification of the Percent Splicing Index (PSI) provides the inclusion level of each exon. Error bars represent the mean ± SD.
Total RNA was extracted from samples by Trizol. Three paired samples of tumor and normal-appearing cells from each patient and one sample of pooled RNA from healthy donors, as a control, were subjected to a global splicing array (Human Junction Array; Affymetrix). The results were analyzed using the ExonPointer algorithm.15 The microarray data from this study have been submitted to the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo) under accession number GSE76070.
Glioma Cell Lines
The U87MG cell line was obtained from American Type Culture Collection and cultured in Dulbecco’s modified Eagle’s medium (DMEM)/F12 (1:1) + GlutaMAX™ (ThermoFisher Scientific)-I supplemented with 10% fetal bovine serum and 1% penicillin streptomycin at 37°C with a 5% CO2 atmosphere. The glioma stem cell lines GSC23 and GSC11 (which were generated from patients at the MD Anderson Cancer Center in Houston, Texas and kindly provided by Dr Frederick Lang) were cultured in serum-free DMEM/F12 (1:1) (1X) + GlutaMAX™-I supplemented with 1xB27, 20 ng/mL human fibroblast growth factor, 20 ng/mL epidermal growth factor, and 1% penicillin streptomycin, and incubated at 37°C with a 5% CO2 atmosphere. All the cell lines were tested and authenticated at the Centro de Investigación Medica Aplicada (CIMA) Genomic Core Facility using short tandem repeat DNA profiling.
Small Interfering RNA
Three small interfering RNAs (siRNAs) targeting the junction sequence between exons 6 and 7 of the tumor isoform (siBAF45d/6A−) were designed using the I-designer tool.
Overexpression of BAF45d/6A+ in GSC23 and GSC11 Cells
GSC23 and GSC11 cells were stable-transfected with a pcDNA3.1(+) plasmid TOPO TA system (Invitrogen/Life Technologies) containing the BAF45d/6A+ cDNA sequence amplified from normal brain total RNA by polymerase chain reaction (PCR) using 5ʹ-AGGCAGAGG AACAGGGAAGATG-3ʹ as the forward primer and 5ʹ-CTAAGG CTGTTTCTCTCCTCCACTT-3ʹ as the reverse primer.
Cell Viability Assays
U87MG and GSC23 cells treated with temozolomide were seeded at a density of 1 × 103 cells per well in 96-well plates. Cell viability was measured with a CellTiter 96 Aqueous Assay Kit using MTS [(3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium), inner salt; Promega] at the indicated times.
In Vivo Tumorigenicity Study
Ethical approval for animal studies was granted by the Animal Ethical Committee of the University of Navarra (CEEA) under protocol number CEEA/093-14. Athymic mice (Harlan Laboratories) were maintained at CIMA in specific pathogen-free conditions and fed standard laboratory chow. U87MG human glioma cells (5 × 105) were engrafted into the caudate nucleus using a guide-screw system as previously described.16
Immunohistochemical Analysis
Paraffin-embedded sections of mice brain were immunostained with specific antibodies for Ki67 (Chemicon International) using conventional procedures. For immunohistochemical staining, Vectastain ABC Kits (Vector Laboratories) were used according to the manufacturer’s instructions.
RNA Immunoprecipitation
To assess the association of PTBP1 with BAF45d mRNA, RNA immunoprecipitation (RIP) of ribonucleoprotein complexes was performed as described previously.17 A list of primers is found in the Supplementary materials.
Chromatin Immunoprecipitation
Chromatin immunoprecipitation (ChIP) was conducting using a ChIP Assay Kit (Upstate Biotechnology) according to the manufacturer’s instructions. A list of antibodies and primers is provided in the Supplementary materials.
Statistical Analysis
Differences in RNA isoform expression and in the cell viability assay were analyzed by Student’s t-tests. The survival rate was performed using the Kaplan–Meier method. For statistical comparisons across groups, the log-rank test was used. All statistical analyses were conducted in GraphPad.
Results
Aberrant Splicing in Glioblastoma Samples
We obtained 3 paired samples of RNA from tumor and normal-appearing tissue from each patient with high specificity18 (Fig. 1A). From the global splicing array, we obtained 51 events from 41 candidate genes with AS (Supplementary Table S1). We proceeded to validate 7 of these candidates based on their significance and structure: BAF45d, Ral GTPase-activating protein subunit alpha-1-like protein (RALGAPA), ITGA6, ARHGEF7, BBX, LRRFIP2, and NDE1 (Fig. 1B). We confirmed the AS of BAF45d and RALGAPA both in Human Junction Array and in 7 patient samples (Fig. 1C).
Characterization of the Aberrant Alternative Splicing of BAF45d in Glioblastoma
We focused on BAF45d due to its high degree of validation and its involvement in the SWI/SNF complex, which is known to play a key role in the development of the CNS and in the maintenance of neural precursors.11 The tumor samples harbored a single BAF45d isoform with 11 exons, whereas the normal and normal-appearing tissue harbored an additional isoform that retains exon 6a (BAF45d/6A+) (Fig. 1A and Supplementary Figure S1A, B). The loss of exon 6a in the tumor-predominant spliced isoform of BAF45d (BAF45d/6A−) leads to the formation of a zinc finger domain that enables DNA binding19 (Fig. 2B), whereas the retained exon 6a in the predominant isoform in the normal brain (BAF45d/6A+) prevents the formation of this structure (Fig. 2B). RNA-sequencing data from the University of California–Santa Cruz Genome Browser confirms that the BAF45d/6A+ isoform is found in normal brain and heart tissue only (Supplementary Figure S2A). Zhang and colleagues20 reported that in the mouse brain, BAF45d/6A+ expression was preferentially restricted to neurons and oligodendrocytes and generally absent from astrocytes (Supplementary Figure S2B, C).
Fig. 2.
Characterization of BAF45d alternative splicing in glioblastoma. (A) Left: A schematic representation of the BAF45d exon structure, showing that exon 6a is present in BAF45d/6A+ (in normal and normal-appearing tissue) but not in BAF45d/6A− (in the tumor). Lower right: Representative photograph of a conventional PCR gel showing the presence of the 2 isoforms (6A+ and 6A−) in the normal brain and the single isoform (6A−) in the tumor. Upper right: Quantification of the Percent Splicing Index (PSI) provides the inclusion level of each exon. (B) BAF45d protein isoform structures obtained from ensembl (www.ensembl.org). Inclusion of exon 6a in the BAF45d/6A+ isoform disrupts the protein domains formed by exons 6 and 7. (C) mRNA expression of BAF45d in normal tissue (n = 9), normal-appearing tissue (normal-like [LK], n = 20), and glioblastoma (tumor, n = 50) samples. (D) Expression of BAF45d/6A− and BAF45d/6A+ mRNA represented as the ratio (BAF45d/6A−/BAF45d/6A+) in glioblastoma, normal-appearing, and normal tissue samples. (E) BAF45d expression heatmap generated from The Cancer Genome Atlas RNA-Seq data corresponding to 166 glioblastomas and 116 oligodendrogliomas.
Interestingly, we found no significant difference in the total expression of BAF45d in normal, normal-appearing, or tumor samples (Fig. 2C). This finding was corroborated by a meta-analysis21 that reported no difference in the expression of BAF45d between nontumor and astrocytoma grade II, anaplastic astrocytoma grade III, and oligodendroglioma grade II and anaplastic oligodendroglioma grade III tumors (Supplementary Figure S3A). Of importance, the ratio of the expression of BAF45d/6A− to BAF45d/6A+ increased significantly in tumor samples compared with normal and normal-appearing samples (Fig. 2D), a trend that has also been observed in a previously published study on the expression of exon 6a in normal, glioblastoma, and oligoastrocytoma samples (Supplementary Figure S3B).22 This result is further corroborated by RNA-seq data from The Cancer Genome Atlas; exon 6a was nearly absent from glioblastomas (n = 166) and oligodendrogliomas (n = 116), indicating nearly null expression of the BAF45d/6A+ isoform (Fig. 2F). Together, these data indicate that the BAF45d/6A− splicing isoform predominates over the BAF45d/6A+ isoform in tumor samples.
BAF45d/6A− Contributes to the Maintenance of an Undifferentiated Phenotype in Glioma Cell Lines
Analyses of the expression of BAF45d isoforms in a panel of glioblastoma cell lines revealed that BAF45d/6A+ was nearly absent (Supplementary Figure S4A). Additionally, BAF45d/6A+ levels were low in normal human astrocytes, neural precursors, and astrocyte progenitors A2B5+ and A2B5− (Supplementary Figure S4B). Next, we attempted to elucidate the role of BAF45d/6A− in the malignant phenotype of gliomas. From a functional point of view, the inhibition of BAF45d/6A− or total BAF45d in GSC23 and U87MG cell lines (Supplementary Figure S5A) resulted in a significant reduction in their in vitro proliferation (Fig. 3A and Supplementary Figure S5B). Overexpression of BAF45d/6A+ alone or combined overexpression of BAF45d/6A+ and inhibition of BAF45d/6A− (Supplementary Figure S5A) also resulted in a significant reduction in cell proliferation (Fig. 3B and Supplementary Figure S5C). Surprisingly, GSC23 and GSC11 glioma cell lines in which we previously modulated the expression of the different BAF45d isoforms did not exhibit an altered chemosensitivity to temozolomide (Supplementary Figure S5D). Importantly, the inhibition of BAF45d significantly decreased in vivo tumorigenicity in mice with orthotopic intracranial tumors: 5 out of 5 short hairpin (sh)Scramble mice displayed tumors, whereas only 1 out of 5 mice in which BAF45d was stably depleted displayed a tumor. These mice displayed a significant increase in overall survival (P = 0.006; median survival could not be assessed because they survived beyond the study period) compared with shScramble mice (median survival = 40 days) (Fig. 3C). Mice bearing cells overexpressing BAF45d/6A+ also exhibited a significant increase in overall survival (median survival = 32 days; P = 0.0003) and long-term disease-free survival (Fig. 3D). Moreover, analysis of BAF45d isoforms in longer-surviving mice confirmed the reexpression of BAF45d/6A+ (Supplementary Figure S5E, F). Cells that simultaneously overexpressed BAF45d/6A+ and lacked BAF45d/6A− did not survive long enough in culture to perform in vivo experiments. The Ki67-proliferation index was significantly higher in the shScramble or pcDNA empty control mice tumors compared with tumors in which the shBAF45d gene was stably depleted or the BAF45d/6A+ isoform was overexpressed (pcDNA_6A+; Fig. 3E, F).
Fig. 3.
BAF45d/6A− contributes to the tumorigenicity and maintenance of an undifferentiated phenotype in glioma cell lines. (A and B) The proliferation index quantified by MTS in GSC23 cell lines upon (A) inhibition of the BAF45d/6A− isoform (siBAF45d/6A−), the whole gene (siBAF45d), or the control siScramble (siScrbl); (B) overexpression of the empty vector pcDNA-3.1 (pcDNA), pcDNA-3.1_6A+ concomitant with the inhibition of pLKO1_scramble (pcDNA_6A+/pLKO_Scrbl), or pCDNA-3.1_6A+ concomitant with the inhibition of pLKO1_BAF45d/6A− (pcDNA_6A+/pLKO_6A-). (C) Kaplan–Meier graph of the overall survival of mice bearing orthotopic cells depleted of BAF45d (shBAF45d) compared with shScramble mice (shScbrl). (D) Kaplan–Meier graph of the overall survival of mice bearing orthotopic cells with stable overexpression of BAF45d/6A+ (pcDNA_6A−) compared with mice with empty pcDNA-3.1 (pcDNA). (E) Left: The Ki67 index, which represents the percentage of the total labeled nuclei in 10 high-power fields (40x), in the control group (shScrbl) and the BAF45d-inhibited group (shBAF45d). Right: Microscopic (20x) Ki67 immunohistochemical images corresponding to the control group (shScrbl) and the BAF45d-inhibited group (shBAF45d). (F) Left: Ki67 index quantified in the control group (pcDNA) and the group overexpressing the BAF45d/6A+ isoform (pcDNA_6A+). Right: Microscopic (20x) Ki67 immunohistochemical images corresponding to the control group (shScrbl) and the BAF45d-inhibited group (shBAF45d). (G) Self-renewal capacity assay in GSC23 cells transfected with a scramble siRNA (siScrbl) or siBAF45d/6A−. (H) Self-renewal capacity assay in GSC23 cells transfected with an empty pcDNA-3.1 (pcDNA) or overexpressing BAF45d/6A+ concomitant with transfection with pLKO1 carrying a scramble short hairpin (sh) (pcDNA_6A+/pLKO_Scrbl) or overexpressing BAF45d/6A+ concomitant with inhibition of the BAF45d/6A− isoform with an sh specific for that isoform (pcDNA_6A+/pLKO_6A−). (I) Representative western blot of markers associated with an undifferentiated cell phenotype after the indicated treatments in GSC23. Representative PCR image of the different BAF45d isoforms after transfection with an empty pcDNA-3.1 (pcDNA) or overexpressing BAF45d/6A+ concomitant with transfection with pLKO1 carrying a scramble sh (pcDNA_6A+/pLKO_Scrbl) or overexpressing BAF45d/6A+ concomitant with inhibition of the BAF45d/6A− isoform with an sh specific for that isoform (pcDNA_6A+/pLKO_6A−). Quantification of the Percent Splicing Index (PSI) is reported as the mean ± SD of 3 independent experiments. (J) Forced differentiation of GSC23 in 10% fetal bovine serum medium for 10 days. Sex determining region Y–box 2 representative western blot. Representative PCR image of the different BAF45d isoforms after forced differentiation. Quantification of the PSI is reported as the mean ± SD. (K) BAF45d/6A− and BAF45d/6A+ mRNA expression in SH-SY5Y neuroblastoma cells instructed with staurosporin to differentiate toward neurons or stably transfected with pcDNA_6A+/pLKO_6A− as control. In all panels, error bars represent the mean ± SD.
Inhibition of the BAF45d/6A− isoform in GSC23 and GSC11 cells significantly reduced cell self-renewal capacity, resulting in changes in both the expression profile and cell morphology, such that cells had a more differentiated phenotype (Fig. 3G and Supplementary Figure S5G, H), supporting the notion of a possible role of BAF45d/6A− in self-renewal. Overexpression of BAF45d/6A+ or its isoform together with the inhibition of BAF45d/6A− led to a significant reduction in the self-renewal potential of these cells and reduced expression of differentiation markers such as sex determining region Y–box 2 and nestin (Fig. 3H, I and Supplementary Figure S5H). In addition, the expression of the stem cell marker CD133 was significantly decreased in GSC23 and GSC11 cells that overexpressed BAF45d/6A+ and/or inhibited BAF45d/6A (Supplementary Figure S5I), suggesting a role for BAF45d/6A− in the maintenance of stem cell status. Forced differentiation of GSC23 using 10% fetal bovine serum also resulted in a switch toward the BAF45d/6A+ isoform (Fig. 3J). SH-SY5Y neuroblastoma cells treated with staurosporin to trigger neuronal differentiation exhibited reduced levels of BAF45d/6A− compared with parental cells (Fig. 3K). These findings suggest that BAF45d/6A− is the predominant transcript in early postnatal murine neural precursors but ceases to be expressed as cells are instructed to start differentiating into neurons (Supplementary Figure S5J).
PTBP1 Mediates the Aberrant Splicing of BAF45d
To elucidate the mechanism underlying the splicing of BAF45d, we analyzed the BAF45d sequence, which revealed several binding sites for PTBP1 flanking exons 6, 6a, and 7 (Fig. 4A and Supplementary Figure S6A). Moreover, analyses of these binding sites for PTBP1 using the Tomtom algorithm23 revealed that they were significantly conserved across species (Supplementary Figure S6B). PTBP1 is a ribonucleoprotein that regulates splicing, is overexpressed in gliomas, and contributes to the malignant phenotype of this disease.5,6,13,24 Radioimmunoprecipitation assays demonstrated the binding of PTBP1 to BAF45d mRNA in different positions. Interestingly, the location interrogated with primers O3F/R, which had the most putative binding sites, was the most enriched in this protein (Fig. 4A), suggesting that PTBP1 may regulate BAF45d splicing. Inhibition of PTBP1 in glioma cells (U87MG and GSC23) led to a splicing switch from the BAF45d/6A− isoform to an increase in the BAF45d/6A+ isoform (Fig. 4B and Supplementary Figure S6C). This result further supports the hypothesis that PTBP1 binding to the BAF45d pre-mRNA regulates this splicing event. The reduction in BAF45d/6A− splicing after inhibition of the PTBP1 protein was accompanied by a significant reduction in the prevalence of the proliferation phenotype (Fig. 4C and Supplementary Figure S6D). Expression of PTBP1 is high in glioblastomas5,6 due to the loss of miR-124, which promotes neuronal differentiation14; therefore, we evaluated whether reexpression of miR-124 impacts BAF45d splicing. We expressed miR-124 in GSC23 and GSC11 (Supplementary Figure S6E) and observed that PTBP1 mRNA expression significantly decreased in both cell lines (Supplementary Figure S6F; P < 0.001). Moreover, overexpression of miR-124 also resulted in the inclusion of exon 6A+, favoring the expression of the BAF45d/6A+ isoform (Supplementary Figure S6G).
Fig. 4.
PTBP1 mediates exon 6a exclusion in BAF45d pre-RNA. (A) Cross-linking immunoprecipitation (CLIP) of PTBP1-bound BAF45d RNA. Reverse transcription quantitative (RT-q)PCR identified the BAF45d region bound by PTBP1 in vivo. The locations of primer pairs (OXF/R) along the BAF45d transcript are indicated in the diagram above. BS: number of binding sites. Glyceraldehyde 3-phosphate dehydrogenase was used as a negative control. *P = 0.01 compared with an immunoglobulin G control. (B) Left: Expression of mRNA of BAF45d/6A−, BAF45d/6A+, and PTBP1 in U87MG cells after inhibition of PTBP1 (siPTBP1) compared with expression with scramble control (siScrbl). RNA expression was quantified by RT-qPCR. Lower right: Representative photograph of a conventional PCR gel showing the expression of BAF45d/6A− and BAF45d/6A+ isoforms after siPTBP1 or siScramble (siScrbl). Upper right: Quantification of the Percent Splicing Index (PSI) provides the inclusion level of each exon. (C) Upper panel: Representative western blot depicting the inhibition of PTBP1. Lower panel: Proliferation index quantified by MTS in the U87MG cell line upon inhibition of PTBP1 (siPTBP1) or the control siScramble (siScrbl). (D) A proposed model for exon 6a splicing of BAF45d by PTBP1. The diagram shows the exon 6a sequence (underlined) and part of the flanking introns. Green bases correspond to the sequence of the 3 regions (branch point, 3ʹ, and 5ʹ splice sites) that are initially recognized by the splicing machinery (spliceosome). SnRNP proteins that specifically recognize these regions are included. Depicted in red are the binding motifs for specific recognition by PTBP1. PTBP1 is represented in its monomeric form with 4 RNA recognition motif (RRM) domains. In the model, the binding of PTBP1 leads to the formation of an mRNA exon loop, and then exclusion of the exon from the spliced mRNA. In all panels, error bars represent the mean ± SD.
Based on these data, we propose a hypothetical model in which PTBP1 binds to BAF45d pre-mRNA and mediates its splicing (Fig. 4D).
BAF45d/6A− Transcriptionally Regulates PTBP1 and Controls Splicing Events
The inhibition of BAF45d/6A− resulted in a decrease in PTBP1 expression (Fig. 5A and Supplementary Figure S7A). Moreover, whereas stable overexpression of BAF45d/6A+ did not result in PTBP1 inhibition, BAF45d/6A+ overexpression concomitant with BAF45d/6A− inhibition led to a significant decrease in PTBP1 mRNA and protein expression (Fig. 5B). This finding implicates BAF45d/6A− in the regulation of this pivotal splicing gene and, therefore, in the regulation of the splicing of other genes regulated by PTBP1, some of which, such as epidermal growth factor receptor (EGFR) and PKM, are of stark relevance to the glioblastoma phenotype.1,5 To confirm that BAF45d plays a key role in the regulation of PTBP1, we studied the PTBP1 promoter and identified several binding sites for this protein (Supplementary Figure S7B). ChIP analyses revealed the direct binding of BAF45d to the PTBP1 promoter at both binding sites analyzed in the GSC23 and U87MG cell lines (Fig. 5C and Supplementary Figure S7C). Interestingly, we could not detect the binding of BAF45d to the PTBP1 promoter in cells with concomitant stable overexpression of BAF45d/6A+ and depletion of BAF45d/6A− (Fig. 4D), indicating direct regulation of PTBP1 by BAF45d and revealing a link between the product of splicing regulation and its own regulation. Finally, we confirmed that the inhibition of BAF45d/6A− resulted in the inversion of gene splicing, such as protein phosphatase 3 catalytic subunit gamma, EGFR, enhancer of zeste homolog 2, or paired immunoglobin-like type 2 receptor alpha (Fig. 5E), all of which have been described previously as direct splicing targets for PTBP1.24 These findings are summarized in Fig. 5F.
Fig. 5.
BAF45d positively regulates PTBP1 and splicing events. (A) Assessment of PTBP1 mRNA (upper panel) and protein expression (lower panel) after siBAF45d/6A− or SiScramble (siScrbl) in the GSC23 cell line. (B) Assessment of PTBP1 mRNA (upper panel) and protein expression (lower panel) after pcDNA (control empty vector), pcDNA_BAF45d/6A+/pLKO_Scrbl, or pcDNA_6A+/pLKO_6A− in the GSC23 cell line. (C) Enrichment (relative to input) of BAF45d in the GSC23 PTBP1 promoter as determined by ChIP-qPCR. (D) Enrichment (relative to input) of BAF45d in the GSC23 PTBP1 promoter after transfection with pcDNA_BAF45d/6A+/pLKO_Scrbl or pcDNA_6A+/pLKO_6A− in the GSC23 cell line as determined by ChIP-qPCR. (E) Schematic representation and representative pictures of agarose PCR gels depicting the inversion in the splicing pattern of genes regulated by PTBP1 after treatment with siBAF45d/6A− or siScramble (siScrbl). (F) Scheme depicting a proposed model for the role and mechanism of action of BAF45d during gliomagenesis and neurogenesis. In all panels, error bars represent the mean ± SD.
Together our results reveal BAF45d/6A− as a product of AS in glioblastoma and a central element in the regulation of this process through the positive control of the expression of PTBP1, which suggests a possible link between splicing regulation and chromatin remodeling.
Discussion
Two aspects of this study distinguish it from previous studies and provide validity to the results: first, the use of FGS to obtain samples and, second, the use of a global splicing array. FGS, the specificity of which has been demonstrated by previous studies of histological correlation with 5-ALA,18 enables the extraction of paired samples of tumor and nontumor tissue. The heterogeneity of glioblastoma highlights the importance of using paired samples, as the dependability of the results is compromised if tumor samples are not compared with tumor-free samples from the same individual.25 The other distinctive aspect of this study is that this is the first glioblastoma study based on a global splicing array as opposed to an exon array.3,4
AS is more highly conserved in the brain than in other tissues, suggesting a role in conserved functions.26 Previous studies have reported fundamental roles for AS in neural development and in the establishment and function of neural networks.27 For example, a previous study reported the importance of coordinated AS, including of BAF45d, in the control of key functions in the mouse CNS, although the significance of BAF45d was not investigated.28 Our previous detailed study of the murine CNS, which produced both a transcriptome and a splicing database, reported that the BAF45d/6A+ isoform was predominantly expressed in neurons,20 suggesting a lineage-specific exon.
The high frequency of AS of BAF45d and the role of BAF45d in maintaining an undifferentiated phenotype suggest that AS associated with gliomagenesis occurs at an early stage in glioma stem cells. These observations are consistent with an established model involving a specific neural precursor of the SWI/SNF complex, the neural progenitor npBAF, whose composition changes as cells are instructed toward a more differentiated phenotype, the nBAF postmitotic neuron.11
BAF45d is not the only example of lineage-specific splicing in a protein that participates in the initiation or maintenance of a tumor phenotype. In fact, Ferrarese et al demonstrated that the lineage-specific splicing of ANXA7 diminished endosomal targeting of the EGFR protein, thereby enhancing its signaling during glioblastoma progression.6 Moreover, these authors identified PTBP1 as the main regulator of ANXA7 exon-skipping and showed that PTBP1 can repress miR-124, which participates in the conversion of fibroblasts to neurons.29 Of note, both miR-124 and miR-9 have been shown to repress the npBAF complex to facilitate the instruction for a cell to differentiate toward a neuron.30 These results illustrate a complex network of interacting proteins that may provide critical checkpoints during CNS development.
PTBP1 is a key protein implicated in glioblastoma splicing regulation.3,5,31–33 Our study found this RNA-binding protein to be the main mediator of AS of BAF45d. An in silico model based on a genome-wide mapping of PTB–RNA interactions predicted binding between PTBP1 and BAF45d, a prediction that is corroborated by the results of the current study. It also has been shown that PTBP1 represses miR-124, which is consistent with the hypothesis that the PTBP1–BAF45d axis is related to an undifferentiated phenotype.14 Moreover, repression of PTBP1 in fibroblasts elicits cellular reprogramming toward a neuronal lineage.29 Furthermore, a decrease in PTBP1 levels has previously been shown to be associated with the switch from neuronal precursor proliferation to differentiation. This decrease in PTBP1 also is associated with global remodeling of the AS landscape.34 In the context of glioblastoma, PTBP1 is overexpressed and amplified in glioblastoma cells6; knockdown of PTBP1 expression is reported to reduce glioblastoma malignancy, inhibit angiogenesis, and drive cells to adopt a neuron-like phenotype.6 The data therefore suggest that PTBP1 not only plays a pivotal role in the development of CNS but also contributes to gliomagenesis when deregulated in the neural precursor compartment.
Perhaps the most striking and original finding of our study concerns the regulation of PTBP1 by BAF45d. There are few published reports on the mechanisms by which PTBP1 is regulated and, among those reports, the role of PTBP1 in splicing together with any demonstration of regulation is generally limited to the gene under study.5,14 We have demonstrated that PTBP1 is regulated by a product of its own splicing, BAF45d, in a feedback mechanism that has not been described previously. In addition, we have shown the effect of BAF45d on other genes with a key role in gliomagenesis and whose splicing is regulated by PTBP1, such as PKM2 and EGFR, suggesting that BAF45d controls several processes at the transcriptional and posttranscriptional levels and providing a possible link between chromatin remodeling and splicing.
In summary, our data indicate that BAF45d AS contributes to glioblastoma pathogenesis in 2 ways. First, the BAF45d/6A− splicing product contributes to the maintenance of cells in an undifferentiated state. Second, the BAF45d/6A− splicing product regulates the splicing of BAF45d; this is the first observation of such a feedback mechanism in glioblastoma splicing. In addition, our data suggest that AS is a mechanism that may be central to glioblastoma development, introducing new avenues of study for the treatment of this devastating tumor.
Supplementary material
Supplementary material is available online at Neuro-Oncology (http://neuro-oncology.oxfordjournals.org/).
Funding
This work was supported by Marie Curie (IRG270459 to M.M.A.), the Instituto de Salud Carlos III and Fondos Feder Europeos (PI13/125 and PI16/00066 to M.M.A. and MS11/00147 and CP14/00077 to A.A.), the Spanish Ministry of Science and Innovation (Ramón y Cajal contract RYC-2009-05571 and IEDI-2015-00638 to M.M.A.), the L’Oreal-Unesco Foundation for Women in Science (to M.M.A.), the Department of Health of the Government of Navarra (to M.M.A.), the Basque Foundation for Health Research (BIOEF, BIO13/CI/005), Asociación Pablo Ugarte-Fuerza, Julen, Colegio La Milagrosa-Lodosa and the Fundación Caja Navarra (2015 to M.M.A.).
Supplementary Material
Acknowledgments
We thank Drs Howard Weiner (Department of Neurosurgery, Texas Children’s Hospital–Baylor College of Medicine, Houston), Fernando De Miguel, Daniel Ajona, and Rubén Pío (Program of Solid Tumors, CIMA, Pamplona) for helpful discussions. We thank Kristine Eco (Texas Children’s Hospital) for editing the manuscript.
Conflict of interest statement
There are no potential conflicts of interest to disclose by any of the authors.
References
- 1. Cancer Genome Atlas Research Network. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature. 2008;455(7216):1061–1068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Zhang J, Manley JL. Misregulation of pre-mRNA alternative splicing in cancer. Cancer Discov. 2013;3(11):1228–1237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Cheung HC, Baggerly KA, Tsavachidis S, et al. Global analysis of aberrant pre-mRNA splicing in glioblastoma using exon expression arrays. BMC Genomics. 2008;9:216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Sadeque A, Serão NV, Southey BR, Delfino KR, Rodriguez-Zas SL. Identification and characterization of alternative exon usage linked glioblastoma multiforme survival. BMC Med Genomics. 2012;5:59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. David CJ, Chen M, Assanah M, Canoll P, Manley JL. HnRNP proteins controlled by c-Myc deregulate pyruvate kinase mRNA splicing in cancer. Nature. 2010;463(7279):364–368. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ferrarese R, Harsh GR 4th, Yadav AK, et al. Lineage-specific splicing of a brain-enriched alternative exon promotes glioblastoma progression. J Clin Invest. 2014;124(7):2861–2876. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Stummer W, Stocker S, Wagner S, et al. Intraoperative detection of malignant gliomas by 5-aminolevulinic acid-induced porphyrin fluorescence. Neurosurgery. 1998;42(3):518–525; discussion 525. [DOI] [PubMed] [Google Scholar]
- 8. Aldave G, Tejada S, Pay E, et al. Prognostic value of residual fluorescent tissue in glioblastoma patients after gross total resection in 5-aminolevulinic acid-guided surgery. Neurosurgery. 2013;72(6):915–920; discussion 920. [DOI] [PubMed] [Google Scholar]
- 9. Wilson BG, Roberts CW. SWI/SNF nucleosome remodellers and cancer. Nat Rev Cancer. 2011;11(7):481–492. [DOI] [PubMed] [Google Scholar]
- 10. Lessard J, Wu JI, Ranish JA, et al. An essential switch in subunit composition of a chromatin remodeling complex during neural development. Neuron. 2007;55(2):201–215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Ronan JL, Wu W, Crabtree GR. From neural development to cognition: unexpected roles for chromatin. Nat Rev Genet. 2013;14(5):347–359. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Patton JG, Mayer SA, Tempst P, Nadal-Ginard B. Characterization and molecular cloning of polypyrimidine tract-binding protein: a component of a complex necessary for pre-mRNA splicing. Genes Dev. 1991;5(7):1237–1251. [DOI] [PubMed] [Google Scholar]
- 13. Han SP, Tang YH, Smith R. Functional diversity of the hnRNPs: past, present and perspectives. Biochem J. 2010;430(3):379–392. [DOI] [PubMed] [Google Scholar]
- 14. Makeyev EV, Zhang J, Carrasco MA, Maniatis T. The microRNA miR-124 promotes neuronal differentiation by triggering brain-specific alternative pre-mRNA splicing. Mol Cell. 2007;27(3):435–448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. de Miguel FJ, Sharma RD, Pajares MJ, Montuenga LM, Rubio A, Pio R. Identification of alternative splicing events regulated by the oncogenic factor SRSF1 in lung cancer. Cancer Res. 2014;74(4):1105–1115. [DOI] [PubMed] [Google Scholar]
- 16. Lal S, Lacroix M, Tofilon P, Fuller GN, Sawaya R, Lang FF. An implantable guide-screw system for brain tumor studies in small animals. J Neurosurg. 2000;92(2):326–333. [DOI] [PubMed] [Google Scholar]
- 17. Lopez de Silanes I, Zhan M, Lal A, Yang X, Gorospe M. Identification of a target RNA motif for RNA-binding protein HuR. Proc Natl Acad Sci U S A. 2004;101(9):2987–2992. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Idoate MA, Díez Valle R, Echeveste J, Tejada S. Pathological characterization of the glioblastoma border as shown during surgery using 5-aminolevulinic acid-induced fluorescence. Neuropathology. 2011;31(6):575–582. [DOI] [PubMed] [Google Scholar]
- 19. Mertsalov IB, Ninkina NN, Wanless JS, Buchman VL, Korochkin LI, Kulikova DA. Generation of mutant mice with targeted disruption of two members of the d4 gene family: neuro-d4 and cer-d4. Dokl Biochem Biophys. 2008;419:65–68. [DOI] [PubMed] [Google Scholar]
- 20. Zhang Y, Chen K, Sloan SA, et al. An RNA-sequencing transcriptome and splicing database of glia, neurons, and vascular cells of the cerebral cortex. J Neurosci. 2014;34(36):11929–11947. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Sun L, Hui AM, Su Q, et al. Neuronal and glioma-derived stem cell factor induces angiogenesis within the brain. Cancer Cell. 2006;9(4):287–300. [DOI] [PubMed] [Google Scholar]
- 22. French PJ, Peeters J, Horsman S, et al. Identification of differentially regulated splice variants and novel exons in glial brain tumors using exon expression arrays. Cancer Res. 2007;67(12):5635–5642. [DOI] [PubMed] [Google Scholar]
- 23. Gupta S, Stamatoyannopoulos JA, Bailey TL, Noble WS. Quantifying similarity between motifs. Genome Biol. 2007;8(2):R24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Xue Y, Zhou Y, Wu T, et al. Genome-wide analysis of PTB-RNA interactions reveals a strategy used by the general splicing repressor to modulate exon inclusion or skipping. Mol Cell. 2009;36(6):996–1006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Rampazzo E, Della Puppa A, Frasson C, et al. Phenotypic and functional characterization of glioblastoma cancer stem cells identified through 5-aminolevulinic acid-assisted surgery [corrected]. J Neurooncol. 2014;116(3):505–513. [DOI] [PubMed] [Google Scholar]
- 26. Barbosa-Morais NL, Irimia M, Pan Q, et al. The evolutionary landscape of alternative splicing in vertebrate species. Science. 2012;338(6114):1587–1593. [DOI] [PubMed] [Google Scholar]
- 27. Raj B, Blencowe BJ. Alternative splicing in the mammalian nervous system: recent insights into mechanisms and functional roles. Neuron. 2015;87(1):14–27. [DOI] [PubMed] [Google Scholar]
- 28. Fagnani M, Barash Y, Ip JY, et al. Functional coordination of alternative splicing in the mammalian central nervous system. Genome Biol. 2007;8(6):R108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Xue Y, Ouyang K, Huang J, et al. Direct conversion of fibroblasts to neurons by reprogramming PTB-regulated microRNA circuits. Cell. 2013;152(1-2):82–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Yoo AS, Staahl BT, Chen L, Crabtree GR. MicroRNA-mediated switching of chromatin-remodelling complexes in neural development. Nature. 2009;460(7255):642–646. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Izaguirre DI, Zhu W, Hai T, Cheung HC, Krahe R, Cote GJ. PTBP1-dependent regulation of USP5 alternative RNA splicing plays a role in glioblastoma tumorigenesis. Mol Carcinog. 2012;51(11):895–906. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Jin W, Bruno IG, Xie TX, Sanger LJ, Cote GJ. Polypyrimidine tract-binding protein down-regulates fibroblast growth factor receptor 1 alpha-exon inclusion. Cancer Res. 2003;63(19):6154–6157. [PubMed] [Google Scholar]
- 33. Cheung HC, Hai T, Zhu W, et al. Splicing factors PTBP1 and PTBP2 promote proliferation and migration of glioma cell lines. Brain. 2009;132(Pt 8):2277–2288. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Linares AJ, Lin CH, Damianov A, Adams KL, Novitch BG, Black DL. The splicing regulator PTBP1 controls the activity of the transcription factor Pbx1 during neuronal differentiation. Elife. 2015;4:e09268. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.