Samples and RNA Extraction

Supplementary Methods

Samples and RNA extraction.

Tumor samples were derived from thetissue collection (frozen material) and the archives (FFPE material) of the Department of Neuropathology, University of Bonn Medical Center and NeuropathologicalBrainTumorReferenceCenter located at this department. Informed consent was obtained from the parents of the patients. The collection represents a retrospectice, inhomogenously treated patient cohort. Each tissue was checked for histological composition by frozen section before RNA extraction to avoid contamination by normal tissue, hemorrhage or necrosis. Total RNA was extracted after DNA digestion using the RNeasy kit (Qiagen, Hilden, Germany) according to the manufacturers’ instructions. RNA integrity was checked using a Bioanalyzer 2100 (Agilent Technologies, Waldbronn, Germany). Only samples with tumor content > 80% and sufficient RNA quality were used in this study.

RNA sequencing.

DNA libraries suitable for mRNA sequencing were prepared from 500 ng of total RNA samples on a Sciclone NGS platform (PerkinElmer, Waltham, U.S.A.) in accordance with the mRNA sample preparation protocol using Illumina TruSeq RNA and DNA sample preparation kits (v2). The adaptor-ligated libraries were enriched by PCR for 15 cycles and purified. The libraries were quality controlled on Agilent 2100 Bioanalyzer chips and showed average sizes of approximately 280 bp. Sequencing on the HiSeq2500 platform (Illumina) involved 101 cycles for read1, 7 cycles for the barcode read and 101 cycles for read2. Bcl files were converted into fastq format using the configureBcltoFastq script in CASAVA1.8.2.

Fusion detection.

If a sample was run on several lanes or in several sequencing runs, all reads from that sample were merged. For de-novo identification of gene fusions, FusionCatcher, version 0.99 ( was applied on the initial, merged FASTQ files using default parameters. FusionCatcher's pipeline includes various pre-processing steps such as adapter trimming.

Expression analysis of RNA sequencing data.

Reads were trimmed using cutadapt [6] in order to remove the Illumina TruSeq adapter sequences and bases with low Phred quality of less than 20 at the end of a read. Reads with length less than 20 base pairs after trimming were discarded. Reads were then mapped against Ensembl GRCh37 transcriptome and genome using TopHat [3] version 2.0.9 with default parameters which internally uses Bowtie [4] version 2.1.0. Fastqc ( was used for quality control after every described pipeline step.For obtaining read counts per gene, mapped reads were first sorted according to read name using SAMtools [5], which is a requirement of the counting tool. Then, for counting HTSeq-count ( was used with Ensembl annotation and parameter 'stranded' set to no. With HTSeq-count's otherwise default settings, multiply mapped reads and reads mapping to introns or to more than one gene were disregarded. For inspecting the quality of counts, we used the corresponding tool of QualiMap [2]. RPKM(reads per kilobase exon per million mapped reads) values were computed using custom R scripts. Gene counts were further analyzed using DESeq [1] version 1.14 which initially computes library-size normalization and per-condition dispersion estimation. DeSeq assumes that dispersion is a function of mean and estimates it by using regression. We used DeSeq's local regression fit type for this task.

For unsupervised clustering, library-size normalized and variance-stabilized gene expressions computed by DESeq were used for hierarchical clustering in R. Genes on the gonosomes were removed. Hierarchical clustering was performed using 1-Pearson correlation as distance, the 100 most variable genes and complete clustering for agglomeration. Gene scores of genes with a base mean of at least 30 read counts in compared conditions were analyzed for enriched gene sets [7] using the R package SeqGSEA [8], also with fit type local used for dispersion estimation and with parameter alpha set to one for identifying enriched gene sets using only scores from differential expression.Gene sets with FDR at most 0.05 are reported as significantly enriched. Case #18 with a predicted C11orf95 -NCOA1 fusion was not included in this analysis.

Reverse transcription and PCR.

250 ng of total RNA of the ependymoma samples were reverse transcribed using the TaKaRa Prime Script RT reagent Kit (Perfect Real Time; Takara Bio Inc., Saint-Germain-en-Laye, France) and random primers. PCR was performed with the following primers located in exon 2 ofC11orf95(5’ aggaagtcatcagcaacagc 3’), and in exon 3 of RELA (5’ tcttggtggtatctgtgctc 3’). The PCR conditions were as follows: annealing temperature 60°C, elongation at 72°C for 30 sec for a total of 45 cycles. The generated 321 bpPCR fragments were analysed on a 2 % agarose gel.PCR products were visualized anddocumented on a Geldoc 1000 system (Biorad, Munich, Germany). The PCR products were purifiedusing a PCR purification kit (Qiagen). Direct Sanger sequencing reactions were performed in duplicate(forward and reverse) as custom service (Eurofins MWG Operon, Ebersberg, Germany) using 30 ngof the PCR product.

Immunohistochemistry for RelA.

FFPE tissue slides were stained on an immunostaining system (BenchMark XT, Ventana-Roche, Mannheim, Germany) using the rabbit anti-NF-κB p65 antibody D14E12(1:400, Cell Signaling, Danvers, U.S.A.) after antigen retrieval (buffer CC1, Ventana-Roche, 30 min).

References for Supplementary Methods

1)Anders S, Huber W (2010) Differential expression analysis for sequence count data. Genome Biol. 11: R106.

2)García-AlcaldeF, Okonechnikov K, Carbonell J, Cruz LM, Götz S, Tarazona S, DopazoJ, Meyer TF, Conesa A (2012) Qualimap: evaluating next-generation sequencing alignment data. Bioinforma. Oxf Engl 28: 2678–2679.

3)Kim D, Pertea G, Trapnell C, Pimentel H, Kelley R, Salzberg SL (2013)TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol 14: R36.

4)Langmead B, Salzberg SL (2012) Fast gapped-read alignment with Bowtie 2. Nat Methods 9: 357–359.

5)Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin R, 1000 Genome Project Data Processing Subgroup (2009) The Sequence Alignment/Map format and SAMtools. Bioinforma Oxf Engl 25: 2078–2079.

6)Martin M (2011) Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal 17: 10–12.

7)Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Paulovich A, Pomeroy SL, Golub TR, Lander ES, Mesirov JP (2005) Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad SciUSA 102: 15545–15550.

8)Wang X, Cairns MJ (2013) Gene set enrichment analysis of RNA-Seq data: integrating differential expression and splicing. BMC Bioinformatics 14, Suppl 5: S16.