TERT expression is susceptible to BRAF and ETS-factor inhibition in BRAFV600E/TERT promoter double-mutated glioma

The BRAF gene and the TERT promoter are among the most frequently altered genomic loci in low-grade (LGG) and high-grade-glioma (HGG), respectively. The coexistence of BRAF and TERT promoter aberrations characterizes a subset of aggressive glioma. Therefore, we investigated interactions between those alterations in malignant glioma. We analyzed co-occurrence of BRAFV600E and TERT promoter mutations in our clinical data (n = 8) in addition to published datasets (n = 103) and established a BRAFV600E-positive glioma cell panel (n = 9) for in vitro analyses. We investigated altered gene expression, signaling events and TERT promoter activity upon BRAF- and E-twenty-six (ETS)-factor inhibition by qRT-PCR, chromatin immunoprecipitation (ChIP), Western blots and luciferase reporter assays. TERT promoter mutations were significantly enriched in BRAFV600E-mutated HGG as compared to BRAFV600E-mutated LGG. In vitro, BRAFV600E/TERT promoter double-mutant glioma cells showed exceptional sensitivity towards BRAF-targeting agents. Remarkably, BRAF-inhibition attenuated TERT expression and TERT promoter activity exclusively in double-mutant models, while TERT expression was undetectable in BRAFV600E-only cells. Various ETS-factors were broadly expressed, however, only ETS1 expression and phosphorylation were consistently downregulated following BRAF-inhibition. Knock-down experiments and ChIP corroborated the notion of a functional role for ETS1 and, accordingly, all double-mutant tumor cells were highly sensitive towards the ETS-factor inhibitor YK-4-279. In conclusion, our data suggest that concomitant BRAFV600E and TERT promoter mutations synergistically support cancer cell proliferation and immortalization. ETS1 links these two driver alterations functionally and may represent a promising therapeutic target in this aggressive glioma subgroup. Electronic supplementary material The online version of this article (10.1186/s40478-019-0775-6) contains supplementary material, which is available to authorized users.


Introduction
Glioma represents the most common tumor type in the central nervous system (CNS) across all age groups [37]. The biology and clinical behavior of glioma are highly heterogeneous as reflected by WHO grades ranging from I to IV [27]. Generally, they are divided into low-grade glioma (LGG), comprised of WHO grades I/II, and WHO grade III/IV tumors which are referred to as high-grade glioma (HGG). Moreover, glioma encompasses a variety of histologic subtypes some of which can present either as LGG or HGG [27].
BRAF is a serine/threonine kinase and central mediator in the well-described oncogenic mitogen-activated protein kinase (MAPK) signaling pathway [14]. Various alterations such as activating mutations of BRAF are commonly found in cancerous tissues [14]. In the pediatric patient population, more than half of LGG are characterized by genetic alterations of the BRAF gene resulting in increased cellular proliferation due to hyperactivation of downstream signaling [16,39]. Moreover, the missense mutation BRAF V600E is present in a considerable amount of LGG namely pleomorphic xanthoastrocyma (PXA) and ganglioglioma (GG), but also other subtypes of astrocytoma [43]. With respect to HGG, BRAF V600E has been described in anaplastic PXA or anaplastic GG [43], as well as pediatric (6-12%) [8,43] and adult (7.7%) glioblastoma (GBM), often accompanied by an epithelioid phenotype [8,20]. The biological differences between BRAF-mutant LGG and HGG remain poorly understood. To date, only concomitant deletion of the CDKN2A locus has been described to synergistically promote glioma development [15] and to define inferior outcome in BRAF V600E -positive glioma [21,34]. Smallmolecule inhibitors of BRAF and its downstream-target MEK have already been approved for other BRAF-driven cancer types, such as melanoma [14] and have been shown to effectively inhibit glioma growth both in preclinical models [5,9,22,36] and small patient cohorts [7,17,21]. Consequently, phase I/II trials with BRAF-or MEK-inhibitors either as single agent (NCT01677741, NCT01748149, NCT 03363217, NCT01089101, NCT02285439, NCT 03213691) or in combination (NCT02684058, NCT 03340506, NCT02034110) have already been initiated. First analyses show promising results in both the pediatric [3,7,21,42] and the adult patient population [17].
The telomerase reverse transcriptase (TERT) gene codes for the core catalytic subunit of telomerase, an enzyme which is responsible for elongating the telomeric ends of chromosomes, thereby enabling cancer cells to bypass senescence. Hence, telomerase re-activation is a frequent mechanism, used in malignant tissues to render replicative immortality and is associated with worse prognosis in various types of brain tumors [11,26]. Specific mutations within the TERT promoter, C250T (−146C > T), C228T (−124C > T) and A161C (−57A > C), have been identified to play an important role in telomerase re-activation in multiple tumor types including HGG [13,46]. C228T represents the most frequent of either mutation in both LGG as well as HGG [18]. Functionally, all three non-coding mutations open new binding-sites for e-twenty-six (ETS/TCF) family transcription factors involved in TERT promoter hyperactivation [4,13]. In addition to a major role of GABPA [4], contribution of MAPK-activated ETS-factors have been reported in BRAF-mutant melanoma and thyroid cancer [45,50].
Pathologic activation of the MAPK signaling pathway in cancer cells is well-known to cause oncogene induced senescence (OIS), a tumor suppressing mechanism [38] which has also been described in BRAF-altered glioma [2]. Interestingly, re-expression of TERT has been shown to promote escape from OIS in BRAF-mutant cancer cells [38]. Moreover, TERT promoter and BRAF doublemutant papillary thyroid cancer exhibits a particularly aggressive course of disease, suggesting an important interaction of these two prominent oncogenic genomic aberrations [35,52]. In brain tumors, cases with concurrent mutations of BRAF and the TERT promoter have been identified and appear to be associated with an aggressive tumor biology [29,33,34,40,54].
Hence, in this study we sought to elucidate the role of concomitant BRAF V600E and TERT promoter mutations in the malignant phenotype of glioma, to dissect the involvement of different ETS-factors and investigate potential therapeutic implications.

Clinical samples and patient data
Tumor tissues for analyses and establishment of patientderived cell models were derived from patients treated at the General Hospital of Vienna or the Department of Neurosurgery at the Neuromed Campus, Kepler University Hospital in Linz. The histopathological diagnoses were assessed by experienced neuropathologist teams according to the 2016 WHO classification. Clinical histories and characteristics were obtained from patient charts available at the respective hospitals.

Cell culture
All cell models were kept under humidified conditions containing 5% CO2 at 37°C (normal cell culture conditions) and were regularly checked for mycoplasma contamination. Cell authentication was performed by short tandem repeat (STR) analysis. All primary glioma cell lines originating from the Department of Neurosurgery, Neuromed Campus, Kepler University Hospital, Linz (BTL53, BTL1333, BTL1304, BTL2231, BTL2176) and from the Medical University of Vienna (VBT4, VBT92, VBT125, VBT150, VBT172) were cultured in RPMI-1640 medium (Sigma-Aldrich, Missouri, USA) supplemented with 10% fetal calf serum (FCS, Gibco, Thermo Fisher Scientific, MA, USA).
NMC-G1, and AM38 cells were purchased from the Japanese Collection of Research Bioresources Cell Bank (Japan) and were cultured according to the distributor's recommendations. DBTRG-05MG was purchased from the "Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH" (Braunschweig, Germany) and cultured in RPMI-1640 medium supplemented with 10% FCS. Neither antibiotics nor any other anti-microbial substances were used during this study. All experiments with both primary and stable cell models were performed between passages 5 and 15.

Molecular characterization
DNA of tumor tissues and cell cultures was extracted and characterized for the respective BRAF V600E and TERT promoter mutation status by direct sequencing as previously published [47]. For international cell models, available genetic information was extracted from the COSMIC database [48]. CDKN2A status was assessed by Ion Torrent sequencing and qRT-PCR, whilst activation of CDK4/6-signaling was estimated by detecting the phosphorylation of the Retinoblastoma-associated protein (Rb) on immunoblots. Copy number variants of the CDNK2A and TERT locus were confirmed using array comparative genome hybridization data either derived from COSMIC database (NMC-G1, DBTRG-05MG, AM38) [48] or analyzed in house as previously published (BTL1333, BTL53, BTL2176, VBT92, VBT125) [32]. The p53-pathway was evaluated through expression analysis of total p53 and the downstream target p21 by Western blot as well as sequencing data from COSMIC database (NMC-G1, DBTRG-05MG, AM38) [48] or established by Ion Torrent sequencing (BTL53, BTL1333, BTL2176, BTL2231, VBT92, VBT125, BTL1304). The Ion Torrent PGM System, the "Ion AmpliSeq Cancer Hotspot Panel v2 with 207 Amplicons" library and "Ion Torrent Suite Software (Version 5.10.1)" software were used to sequence tumor hotspot mutations. Sequencing data from NMC-G1, DBTRG-05MG and AM38 were derived from COSMIC, however, only previously reported pathogenic mutations were included in the manuscript.

In silico analyses
RNA sequencing data derived from the cancer genome atlas (TCGA) from GBM (n = 166), skin cutaneous melanoma (n = 469) and bladder urothelial carcinoma (n = 408) were stratified according to their BRAF mutation status. Average logarithmic relative expression values of TERT and different ETS-factors in the respective subgroups were calculated by RNA sequencing algorithms (DESeq2). A dataset of glioma cases including information on BRAF and TERT promoter status, WHO grade and histologic subtype was compiled from COSMIC [48] database and a previously published dataset [18].

Colony formation assay
Dabrafenib, vemurafenib, and YK-4-279 were purchased from Selleck Chemicals (Houston, TX, USA). Low densities of cells ranging from 0.5 × 10 3 -3 × 10 3 cells/well depending on the respective cell proliferation time were seeded in 500 μl growth media in duplicates in 24-well plates and settled for 24 h under normal cell culture conditions. Upon the recovery time, the indicated drug concentration was added in 100 μl growth medium. Following drug exposure time of 7 days, cells were washed with phosphate-buffered saline (PBS) and fixed with icecold methanol for 30 min at 4°C before cells were stained using crystal violet. Digital photographs were taken using a Nikon D3200 camera and processed with ImageJ software. For quantification, crystal violet was eluted using 2% sodium dodecyl sulfate and color absorbance was measured at 560 nm at the Tecan infinite 200Pro (Zurich, Switzerland). Values were analyzed using GraphPad Prism software 5.0 and are given in arbitrary units (AU) as mean +/− standard deviation (SD) normalized to untreated control.
Cell viability assay -ATP assay Dabrafenib, vemurafenib, and YK-4-279 were purchased from Selleck Chemicals (Houston, TX, USA). Cells were seeded in 100 μl of the respective growth medium in triplicates of 96-well plates at cell densities ranging from 2 × 10 3 -4 × 10 3 /well cells depending on the respective cell proliferation time. Following a 24 h recovery time under normal cell culture conditions, cells were treated with different drug concentrations in 100 μl growth medium. Upon 72 h, cell viability was analyzed based on the cellular ATP content following manufacturer's instructions ("CellTiter-Glo® Luminescent Cell Viability Assay", Promega, Madison, WI, USA). Luminescence was measured at 1000 nm at the Tecan infinite 200Pro. Raw data were analyzed using GraphPad Prism software 5.0. Results are given as mean +/− SD and were normalized to untreated control cells.
Protein isolation and Western blotting 4 × 10 5 -6 × 10 5 cells/well were seeded in 2 ml of growth medium in 6-well plates and left under normal cell culture conditions for recovery. On the next day when 80-90% confluence was reached, cells were treated with 1 μM dabrafenib for 6 h. Upon scraping and washing the cells in PBS, cells were mechanically (ultrasound) and chemically lysed (lysis buffer: 50 mM Tris/HCl (pH 7.6), 300 mM NaCl, 0.5% Triton X-100, supplemented with protease inhibitors PMSF and complete and phosphatase inhibitor PhosSTOP; all supplements from Roche, Rotkreuz, Switzerland).
Total protein concentrations were determined following manufacturer's instructions ("Pierce™ BCA Protein Assay Kit", Rockford, IL, USA). 15 μg of proteins were loaded onto 10% polyacrylamide-gels and polyacrylamide gel electrophoresis was run at 90 V. Proteins were blotted onto polyvinylidenfluorid membranes via semidry Western blotting. Blotting efficiency was checked with Ponceau protein staining. Antibodies detecting target proteins (Additional file 1: Table S1) were diluted 1:1000 in 3% BSA in Tris-buffered saline with 0.1% Tween20.

RNA-extraction, reverse transcription and qRT-PCR
4 × 10 5 -6 × 10 5 cells/well were seeded in 6-well plates in 2 ml of the respective growth medium. Upon 1 day, cells were exposed to 1 μM dabrafenib for 16 h. Total RNA was isolated using TRIzol reagent (Thermo Fisher Scientific, Waltham, MS, USA) and chloroform isolation according to standard protocols and checked for purity (260/280 ratio > 1.8) and concentration (100-500 ng/μl) using Nanodrop 1000 (Thermo Fisher Scientific). 1 μg of RNA was reverse transcribed into cDNA using Revert aid reverse transcriptase (Thermo Fisher Scientific). cDNA was diluted 1:25 and mixed to the same parts with 2x GoTaq Green Master Mix (Promega) and 10 nM of both forward and reverse primers (Eurofins Scientific, Luxembourg, Luxembourg; primer table in Additional file 1: Table S2). CFX Connect Real-Time PCR Detection System and analysis software (BioRad, Hercules, CA, USA) was used for running quantitative PCR. Raw data were normalized to internal control RPL-41 (dCT) and converted to a linear form using 2 -dCT (mean from triplicates). In treatment experiments, expression values were normalized to the housekeeping gene RPL-41 as well as to the respective untreated control (ΔΔCT), set as 1, and were converted to a linear form using 2 -ΔΔCT (mean, +/− SEM from triplicates).

siRNA-mediated knock-down of ETS1
2 × 10 5 -3 × 10 5 cells/ml were seeded in 500 μl or 2 ml growth medium into 24-well or 6-well plates, respectively, and incubated for 24 h under standard cell culture conditions in order to recover. On the following day, knock-down was performed using 50 nM ETS1-targeting SMARTpool siRNA (UCAUUAGCUAUGGUAUUGA, GUCUCAAGCAUUAAAAGCU, CCCCAAGGUUUA AAUACAA, GGUUGGACUCUGAAUUUUG) or 50 nM Accell Green non-targeting siRNA (GE Healthcare Little Chalfont, UK). Transfection was performed using Xfect RNA transfection reagent (Takara Bio, Kyoto, Japan) according to company's recommendations. Upon 48 h of incubation under normal cell culture conditions, total RNA was isolated and qRT-PCR was performed as described above.
Luciferase reporter assay 4 × 10 5 -6 × 10 5 cells were seeded in 2 ml of growth medium into 6-well plates and incubated under standard cell culture conditions for 24 h. Upon recovery, cells were transfected with the indicated plasmids as described previously [47] using Lipofectamine 3000 according to manufacturer's recommendations. After incubation for 24 h, cells were treated with 1 μM dabrafenib. Following 16 h drug exposure, proteins were isolated and luciferase signals were analyzed using the Dual-Glo Luciferase Assay System (Promega) according to manufacturer's instructions.

ChIP
Protein crosslinking was performed using 1% methanolfree paraformaldehyde (Thermo Fisher Scientific) and reaction was stopped with glycine. Dynabeads Protein A (Thermo Fisher Scientific) were precleared, subsequently blocked using bovine serum albumin and finally loaded with antibodies targeting ETS1, GABPA, IgG and AcH3K27 described in Additional file 1: Table S1. Chromatin was sonicated and validated for suitable fragment size via agarose gels. Crosslinked protein-chromatin suspension was added to the antibody pre-loaded beads and incubated over night at 4°C on an overhead rotator. The next day, beads were washed to remove unbound fragments and DNA was eluted from the beads upon heatinduced reverse-crosslinking. DNA was isolated by phenol-chloroform (Sigma Aldrich) purification and genomic fragments were quantified with qRT-PCR as described above using primers adjacent to the prominent TERT promoter mutations C228T and C250T. The used primer sequences are listed in Additional file 1: Table  S2.

Ectopic TERT expression using adenoviral constructs
The HA-tagged TERT adenoviral construct (HA-TERT, human, #349917A) was purchased from ABM (Richmond, BC, CAN) and multiplied by several rounds of HEK-293 cell amplification. GFP adenovirus was constructed using AdEasy Adenoviral Vector System (Agilent, La Jolla, CA, USA) and served as infection control. Cells were cracked by three freeze and thaw cycles in Tris/HCl (pH 8). DBTRG-05MG cells were infected using 30 moi of the respective viruses. RNA for qRT-PCR (primer sequences are listed in Additional file 1:

Results
TERT promoter mutations are associated with TERT expression and enhanced aggressiveness in BRAF V600Emutated glioma We analyzed the TERT promoter mutation status in a small cohort of pediatric cases with BRAF V600E -mutated glioma (n = 8, Additional file 1: Table S3) treated at the General Hospital of Vienna. TERT promoter mutation status of these BRAF V600E -positive glioma patients was correlated to clinical parameters including gender, age, WHO grade and overall survival. Interestingly, the single patient harboring a tumor with additional TERT promoter mutation showed the most aggressive course of disease (Additional file 1: Table S3). To further analyze this clinical finding on a broader basis, we curated a dataset of 103 BRAF V600E -mutated glioma with information on tumor grade and TERT promoter mutation from publicly available datasets (Additional file 2: Table S4). Corroboratively, double-mutant tumors were significantly enriched (Fisher's exact test, p = 0.003) in HGG (WHO grade III/IV; 19/69, 28%) as compared to LGG (WHO grade I/II, 1/34, 3%).
To investigate the underlying oncogenic mechanisms of concomitant TERT promoter and BRAF V600E mutations in glioma, we analyzed the expression of different ETS-factors (ETS1, GABPA, GABPB-1S, GABPB-1 L, GABPB-2) and their downstream targets cyclin D1 and TERT in tissues of BRAF V600E -positive glioma (n = 8) (Additional file 3: Figure S1). Strikingly, TERT mRNA was only expressed in the tumor with concomitant TERT promoter and BRAF V600E mutations, whereas neither differences in expression of ETS-factors nor the downstream target cyclin D1 were observed (Additional file 3: Figure S1). This was confirmed by analysis of in silico RNA sequencing data, additionally including the ETS-factors ETV1, ETV4 and ETV5 (Additional file 3: Figure S2A). Interestingly, investigation of in silico data rather showed a trend towards lower TERT mRNA expression in BRAF-mutant GBM, but not other BRAFmutant tumor types (Additional file 3: Figure S2B).
Based on these data, we sought to elucidate the interplay of BRAF V600E signaling and downstream effects on the TERT promoter in more detail. Therefore, we established a panel of twelve glioma-derived cell lines with different BRAF and TERT promoter status, containing nine BRAF V600E -mutant and three BRAF wild-type cell models ( Table 1). The latter were considered as references to dissect the effect of oncogenic BRAF activation in the background of both a mutated (BTL2176) and a wild-type (BTL1333, BTL53) TERT promoter. Consistent with our previous tissue analyses, only BRAF V600E /TERT promoter-mutant cell lines, but not the models with an isolated BRAF V600E mutation, expressed TERT mRNA (Table 1). Notably, all available BRAF V600E -positive stable cell lines derived from our neurosurgical departments or from commercial sources turned out to be TERT promoter-mutated. In contrast, primo-cell cultures from three BRAF V600E -positive glioma specimens not expressing TERT mRNA and not developing into stable cell models were TERT promoter-wild-type (Table 1). Consequently, only one of these BRAF V600E /TERT promoter wild-type primo-cell models could be propagated sufficiently for further in vitro experiments. Moreover, none of the double-mutant cell lines showed a gain of the TERT gene locus (Additional file 3: Figure S3), supporting our hypothesis that promoter mutation is the primary driver of telomerase re-activation in these tumors.
We further characterized the panel for the respective CDKN2A and TP53 status. All analyzed BRAF V600E cell models lacked TP53 mutations and, in contrast to BRAF wild-type cells, homogenously expressed the p53 downstream target p21 (Additional file 3: Figure S3). Notably,  Figure S3). In line with CDKN2A loss-of-function and consecutive activation of the cyclin D1/cyclin dependent kinase (CDK) 4/6 complex [15], Rb was hyperphosphorylated in the majority of double-mutant glioma cells, indicating functional inhibition (Additional file 3: Figure S3).

ETS-factors are hyperactivated in BRAF V600E -mutated glioma cells
Based on the well-described activation of ETS-factors via MAPK-signaling [51] and having confirmed ETS-factor expression in the respective tumor tissues, we hypothesized that BRAF V600E mutant glioma cells were characterized by hyperactivated ETS-signaling. Analyses of ETS-factor mRNA expression confirmed that the transcription factors ETS1, GABPA, GABPB, ETV1, ETV4 and ETV5 were widely expressed throughout the entire cell panel, however, no differences between the genotypes were observed. Similarly, the distinct splicevariants of GABPB (GABPB-1S, GABPB-1 L, GABPB-2) showed no differences in expression between the genotypes. In contrast, cyclin D1 displayed significantly higher expression in BRAF V600E positive cell models as compared to wild-type cells. No differences between the investigated genotypes were observed for TERT mRNA expression (Fig. 1a). Next, we investigated expression and signaling activation of MAPK pathway members on protein level. Consistent with oncogenic BRAF signaling, phosphorylation levels of MEK were markedly enhanced in BRAF V600E -mutant as compared to BRAF wild-type cell models. In contrast, no distinct differences were observed for activation of S6 and ERK indicating activation of the respective pathways by alternative mechanisms in the BRAF wild-type models, an effect which has previously been described in both glioma [36] and melanoma [55]. With respect to ETS-factors, ETS1 was variably expressed throughout the cell panel and showed phosphorylation in all genotypes (Fig. 1b, Additional file 3: Figure S4). GABPA was widely and ETV1 constitutively expressed in the investigated cell line panel. In line with our qRT-PCR results, cyclin D1 was predominantly detectable in cell models harboring BRAF alterations (Fig. 1b).

BRAF V600E /TERT promoter double-mutant glioma cells are highly sensitive towards BRAF-inhibitors
In order to validate the expected dependency of BRAF V600E -mutant glioma on hyperactivated MAPK-signaling, we tested the anti-proliferative effects of the BRAF-inhibitor dabrafenib. As predicted, dabrafenib was only effective in BRAF-mutated cell models (Additional file 3: Figure S5a, Fig. 2a). Interestingly, sensitivity was highest in those cell lines harboring additional TERT promoter mutations (Fig. 2a). In contrast, cell proliferation of cell models lacking BRAF V600E mutation was not inhibited but rather increased (Fig. 2a). In addition, these results were confirmed with vemurafenib, another BRAF-inhibitor (Additional file 3: Figure S5b). In addition to basal expression levels, we further investigated downstream effects of BRAF-inhibition in the respective genotypes. Dabrafenib effectively decreased both MEK-as well as ERK-phosphorylation levels in all BRAF V600E -mutated cell lines. In contrast, despite inhibition of MEK-phosphorylation, ERK-phosphorylation was stable or even increased in BRAF wild-type cell models (Fig. 2b). This finding is well in agreement with the so-called RAF-paradox by BRAF-inhibition under wild-type conditions [14]. Moreover, dabrafenib treatment resulted in downregulation of cyclin D1 in doublemutant cell models (Fig. 2b). No effect on cell proliferation was observed upon short term BRAF-inhibitor treatment (data not shown).

Oncogenic MAPK-signaling mediates TERT expression in BRAF V600E /TERT promoter double-mutant glioma
After confirming the respective impact of BRAF-inhibition on downstream signaling in the different genotypes, we tested the effect of dabrafenib on TERT expression. TERT mRNA was constitutively downregulated by dabrafenib (Fig. 3a) or vemurafenib (Additional file 3: Figure S6) in double-mutant glioma models only. Conversely, dabrafenib rather stimulated TERT expression in BRAF wild-type/TERT promoter-mutant cells and had no effect in the double wild-type background. No effect on cell proliferation was observed upon short term BRAF-inhibitor treatment (data not shown). In order to clarify the role of TERT promoter mutation in telomerase re-activation in the respective genotypes, we analyzed activation of the wild-type and mutant promoter sequences in each genotype via luciferase reporter assays. As expected, the C228T-mutated TERT promoter construct was significantly more active as compared to the wild-type promoter in most of the cell models. This high promoter activity could be suppressed by treatment with dabrafenib solely in the BRAF V600E -mutated cell models. In contrast, BRAF-inhibition increased the activity of the mutated TERT promoter in a double wild-type background (Fig. 3b).

ETS1 activation is impaired upon BRAF-inhibition in BRAF V600E -mutant glioma
In order to investigate the impact of BRAF-mediated downstream signaling on ETS-factors in more detail, we selected ETS1, which is well described for being transcriptionally and post-translationally activated by MAPK signaling, for further analyses [30,44,50]. In correspondence to the expression patterns observed for TERT, ETS1 was more effectively downregulated by dabrafenib in all BRAF V600E /TERT promoter-mutant cell models (Fig. 4a). Moreover, ETS1 activation via phosphorylation was blocked in response to the BRAF-inhibitors dabrafenib (Fig. 4b) and vemurafenib (Additional file 3: Figure S7) in all BRAF V600E -mutant cells hence paralleling phosphorylation of ERK. In contrast, the levels of ETS1 expression and phosphorylation in BRAF wild-type cell lines were not efficiently blocked, obviously based on the above described paradoxical ERK-activation by the BRAF-inhibitors under wild-type conditions. GABPA protein expression, however, was not affected by BRAF-inhibition in double-mutant Ratios between phosphorylated and total proteins as indicated were calculated after normalization to β-actin. wt = wild-type, mut = mutated glioma cells (Fig. 4b). To exclude whether this was a side-effect of cell growth inhibition we tested short-term BRAF-inhibitor treatment and observed no effect on cell proliferation (data not shown).
With respect to other ETS-factors, only GABPB-1S expression was also predominately inhibited by dabrafenib in a BRAF V600E -mutated background whereas GABPA, GABPB-1 L and GABPB-2 showed variable response patterns (Fig. 4c-f ). In line with the data from our cell models, a residual tumor of an anaplastic PXA case operated during combination treatment of dabrafenib and the MEK-inhibitor trametinib had lost expression of ETS1, cyclin D1 and TERT (Additional file 3: Figure S8).

ETS1 mediates TERT expression in BRAF V600E and TERT promoter-mutated glioma cells
To more specifically dissect the link between MAPK-signaling and telomerase re-activation, we applied siRNA mediated knock-down of ETS1. Knock-down consistently decreased TERT mRNA expression across BRAF V600E /TERT promoter-mutated glioma cells as well Fig. 2 Anti-proliferative effects and altered downstream-signaling upon BRAF-inhibition. a Clone formation assays of with different BRAF and TERT promoter status as indicated are shown. Cells were seeded at low density and treated with 1 μM dabrafenib for 7 days. The upper panel depicts one representative well per condition. The lower panel shows the quantitative results represented as mean +/− SD of the respective untreated control. ***p < 0.001 (unpaired student's t-tests) (b) Western blot analyses of cell models with different BRAF and TERT promoter status are depicted. Cell models were treated with 1 μM dabrafenib for 6 h. Expression and phosphorylation of the indicated MAPK pathway mediators as well as cyclin D1 are shown. Fold values are given as normalized expression to β-actin followed by activated kinase/total kinase and are normalized to the respective control. wt = wild-type, mut = mutated, dabra = dabrafenib as in a TERT promoter-mutated cell model (Fig. 5a). Notably, no effect on cell proliferation was observed upon siRNA-mediated knock-down (data not shown). Moreover, by ChIP-qRT-PCR we confirmed strong binding of ETS1 and GABPA selectively to the mutant TERT promoter locus, paralleled by activating H3K27-acetylation. In one double mutant cell model ETS1 even was the dominant of the investigated ETS-factors bound to the mutant promoter site (Fig. 5b). To clarify, whether ETS-factors play a central role in the malignant phenotype of BRAF V600E -mutant glioma, we assessed sensitivity to the ETS-inhibitor YK-4-279 [10] across our cell panel. Indeed, BRAF-mutant cell models were hypersensitive (IC 50 < 5 μM) towards this compound suggesting a wide dependency of cell viability/proliferation on ETS-mediated signals (Fig. 5c). Accordingly, in contrast to TERT promoter wild-type glioma, ETS-factor inhibition by YK-4-279 reduced TERT mRNA expression in double-mutant cell models whereas ETS1 expression was more variable (Fig. 5d). Moreover, TERT re- Fig. 3 Regulation of TERT expression and TERT promoter activity upon BRAF-inhibition. a TERT mRNA expression following dabrafenib treatment (1 μM, 16 h) of the indicated cell models is shown. Mean +/− SD; unpaired student's t-tests (b) Luciferase reporter assays were performed in cell lines with different BRAF and TERT promoter status as indicated using wild-type or mutated (C228T) TERT promoter sequences. Cells were treated with 1 μM dabrafenib for 16 h. Results are given as ratio of firefly to renilla luciferase (internal control) and were normalized to a promoter-less construct (−Co, set to 1). Values are given as mean +/− SD from duplicates. One representative experiment out of three, delivering comparable results, is shown. Tukey's multi-comparison one-way ANOVA was applied for statistical analysis. *p < 0.05, **p < 0.01, ***p < 0.001; wt = wild-type, mut = mutated, n.d. = not detected, dab = dabrafenib  (1 μM, 16 h). *p < 0.05, **p < 0.01, ***p < 0.001 (unpaired students' t-tests); All values are given as mean +/− SD; wt = wild-type, mut = mutated, dabra = dabrafenib  Figure S9). Combination of the BRAF inhibitor dabrafenib and the ETS-factor inhibitor YK-4-279 revealed additive to rather antagonistic effects, especially in the double-mutant glioma cell models (Additional file 3: Figure S10).

Discussion
The discovery of genomic aberrations in BRAF as drivers of certain glioma subtypes has resulted in a magnificent extension of the therapeutic repertoire for these patients [17,39,43]. Recent studies, however, indicate that additional factors such as loss of CDKN2A or telomerase reactivation may significantly influence the clinical outcome of glioma patients with oncogenic BRAF, suggesting a biological heterogeneity within this subgroup [34,40]. In this context, OIS and replicative senescence have been described as central obstacles for proliferation in BRAF-altered glioma cells [2]. Furthermore, TERT derepression has been demonstrated to promote escape of cancer cells from OIS [38]. Previous clinical observations and case-report studies delivered preliminary evidence that TERT promoter mutations indicate tumors with higher aggressiveness within BRAF V600E -mutated glioma [33,34,40,54]. Consequently, we systematically investigated the cellular factors contributing to the interplay of BRAF V600E and TERT promoter mutations in glioma.
To begin with, we analyzed the clinical impact of telomerase re-activation in a small cohort of BRAF V600Emutated glioma. In accordance with previous reports, we observed a highly aggressive course in one anaplastic PXA with both BRAF V600E and TERT promoter mutations in our patient collective [18,28]. Moreover, only this double-mutant tumor expressed detectable levels of TERT mRNA. Consistent with this observation, an in silico analysis of 103 BRAF V600E -mutated glioma showed that TERT promoter mutations are significantly enriched in WHO grade III/IV tumors. Within this dataset, 28% of BRAF V600E -mutated HGG harbored additional TERT promoter mutations, which is consistent with earlier reports of secondary HGG and anaplastic PXA [18,20,28,33,34,40]. Particular aggressiveness of double-mutant tumors has already been described for other tumor-types such as melanoma and thyroid cancer [23,35,52].
In order to elucidate the potential role of TERT promoter mutations in the malignant phenotype of certain BRAF V600E -mutated glioma, we curated a unique set of cell models containing nine BRAF V600E -mutant gliomas, the largest panel reported to date. Notably, all stable BRAF V600E -positive cell lines harbored TERT promoter mutations corroborating the aggressive biology of double-mutant tumors. Moreover, double-mutant models were further characterized by loss of the CDKN2A locus and BRAF-induced cyclin D1 expression. In a previous reported study, depletion of CDKN2A, CDKN1A, and TP53 has been demonstrated as essential factor to overcome OIS in a patient-derived model of pilocytic astrocytoma [2]. These findings support our data as all stable BRAF V600E -mutant cell models harbor a loss at the CDKN2A locus. Moreover, the same study showed that TERT re-expression in this BRAF-driven cell model enabled escape from replicative senescence [2]. This supports the notion that the aggressive behavior of these tumors is fueled by a synergistic activation of MAPK-signaling in addition to CDK4/6-activation and telomerase re-expression.
The pathogenic mechanism of TERT promoter mutations is widely mediated by transcriptional activation which involves the binding of ETS-factors [13]. Specifically, the ETS-factors ETS1, GABPA, GABPB, and the splice variant GABPB-1 L, have been linked to activation of the mutated TERT promoter [4,13,31,53]. Interestingly, both overall expression (e.g. GABPB, ETS1) as well as posttranscriptional activation (e.g. ETS1) can be stimulated via the MAPK-pathway [6,25,41,51]. Accordingly, with respect to BRAF V600E -mutated melanoma, both phosphorylation of ETS1 and upregulation of GABPB via Fos have been shown to link oncogenic BRAF signaling to activation of the mutant TERT promoter [25,50]. We analyzed the expression of the major ETS-factors ETS1, GABPA, and GABPB with its splice variants, ETV1, ETV4 and ETV5 all of which were widely expressed across our glioma cell panel. In contrast, TERT mRNA was only detectable in BRAF V600E cells with additional TERT promoter mutations. Cell proliferation assays with BRAF-inhibitors confirmed the (See figure on previous page.) Fig. 5 Activation of the mutant TERT promoter by ETS1. a An siRNA approach was applied to ETS1 knock-down in cell models with different BRAF and TERT promoter status as indicated. TERT mRNA expression was detected using qRT-PCR. Significance levels were evaluated by unpaired students' t-tests (mean +/− SEM). b Binding of ETS1 and GABPA to the different TERT promoter variants and presence of the activating histone mark H3K27-Ac was analyzed by ChIP-qRT-PCR. IgG served as negative control. Values are given as % Input, depicted mean +/− SD from duplicates. c Cytotoxicity assay were performed in cell models of different BRAF and TERT promoter background as depicted. Half-maximal inhibitory concentration (IC50) after YK-4-279 treatment for 72 h was calculated. Tukey's multi-comparison one-way ANOVA was applied (mean +/− SD). d TERT and ETS1 mRNA expression levels were analyzed by qRT-PCR after 16 h treatment with the indicated concentrations of YK-4-279. Ordinary one-way ANOVAs for every cell line were calculated (Dunnet correction, 0.05 (95% confidence interval) as controls vs. treatments (mean +/− SEM). *p < 0.05, **p < 0.01, ***p < 0.001, wt = wild-type, mut = mutated, n.d. = not detected selective activity against BRAF V600E -mutated glioma cells, as already reported by previous preclinical studies and case series [7,12,19,21,22]. Strikingly, BRAF-inhibition also blocked activating phosphorylation of ETS1 selectively in BRAF V600E -mutant glioma cells. This effect has previously been described in other BRAF-mutated tumor types like melanoma [50], but not yet in glioma. Additionally, we could detect an inhibitory effect on cyclin D1 expression, confirming previous data, suggesting it as a downstream target of oncogenic MAPK-signaling in BRAF-mutated glioma [39].
In a following step, we demonstrated that activation of TERT transcription is dependent on BRAF signaling solely in the background of BRAF V600E and TERT promoter double-mutation. Moreover, the clinical relevance of this finding is supported by investigation of the tumor material derived from a single patient treated with a combination of BRAF-and MEK-inhibitors resulting in undetectable levels of both ETS1 and TERT during therapy. Oncogenic BRAF signaling has been shown to bridge telomerase re-activation via the mutated TERT promoter sequence. With respect to ETS-factor activation, GABPB expression has been described to be upregulated via Fos, a well described downstream effector of MAPK and ETS1 signals [25,41], and ETS1 is directly phosphorylated via the MAPK-pathway [50]. Next, we tested the impact of BRAF-inhibition on the expression of the respective ETS-factors and found that only ETS1 expression, but not expression of GABPA or the different GABPB splice variants, were significantly and consistently downregulated throughout all BRAF V600Emutated glioma. These findings are well in agreement with a feed-forward loop on ETS1 expression exerted by MAPK-signaling and ETS1 as predominant effector molecule [30]. Additionally, we show that knock-down of ETS1 reduces TERT expression confirming it as mediator of BRAF V600E -driven TERT promoter activation. Accordingly, ChIP-analysis revealed ETS1 and GABPAbinding specifically to the mutant TERT promoter site. Previous studies in BRAF wild-type glioblastoma models have already demonstrated a central role of GABPA in activation of the mutant TERT promoter [4]. Our data point towards cooperation of GABPA with ETS1, especially in a BRAF V600E -mutant glioma background. Accordingly, a recent study in melanoma has shown that the C228T variant, the predominant mutation detected in our panel, is more efficiently activated by ETS1 as compared to the other TERT promoter mutations [1]. The potential role of other MAPK-induced ETS-factors is of high interest to be investigated in further studies.
BRAF-inhibition has become the standard of care in BRAF V600E -mutated melanoma [14]. However, acquired insensitivity plays an important role in therapy failure [24]. Additionally, first results of clinical application in BRAF V600E -positive glioma indicate that also a proportion of these tumors exhibit intrinsic BRAF-inhibitor resistance [17]. Consequently, inhibition of ETS-factors as downstream MAPK-signal transmitters have been suggested to provide a novel therapeutic opportunity to overcome upstream resistance development [49]. Therefore, we tested the effect of YK-4-279, a well-described ETS-factor inhibitor [10], in our cell line panel. In line with previous reports from our group concerning meningioma [47], also glioma cell models harboring mutant TERT promoters were hypersensitive towards YK-4-279 treatment [47] and YK-4-279 distinctly reduced TERT mRNA expression. Strikingly, BRAF V600E -mutant models were even more sensitive towards YK-4-279, irrespective of the underlying TERT promoter status. TERT re-expression from a viral promoter partially rescued doublemutated glioma cells from the cytotoxic effect of YK-4-279. As we found that both dabrafenib as well as YK-4-279 were highly active against BRAF V600E mutated glioma, we aimed to investigate interactions between the two drugs. No synergistic, but rather antagonistic effects were identified particularly in the double-mutant glioma cell models. These results correspond to our findings that dabrafenib alone already reduced the expression levels of ETS1, the target of YK-4-279. Taken together, our data suggest that apart from the demonstrated TERT promoter activating properties, ETS-factors appear to play an important role in tumor biology of BRAF V600Emutated glioma.

Conclusions
Summarizing, we prove that telomerase re-activation based on a mutant TERT promoter sequence in BRAF V600E -mutant glioma is driven by oncogenic BRAF signaling predominantly via downstream activation of ETS-factors. Accordingly, ETS-factor inhibition is a promising therapeutic option for therapy-resistant BRAF-mutant glioma.