Gliomas are the most common primary brain tumors in children and adults. We recently identified frequent alterations in chromatin remodelling pathways including recurrent mutations in H3F3A and mutations in ATRX (α-thalassemia/mental-retardation-syndrome-X-linked) in pediatric and young adult glioblastoma (GBM, WHO grade IV astrocytoma). H3F3A mutations were specific to pediatric high-grade gliomas and identified in only 3.4 % of adult GBM. Using sequencing and/or immunohistochemical analyses, we investigated ATRX alterations (mutation/loss of expression) and their association with TP53 and IDH1 or IDH2 mutations in 140 adult WHO grade II, III and IV gliomas, 17 pediatric WHO grade II and III astrocytomas and 34 pilocytic astrocytomas. In adults, ATRX aberrations were detected in 33 % of grade II and 46 % of grade III gliomas, as well as in 80 % of secondary and 7 % of primary GBMs. They were absent in the 17 grade II and III astrocytomas in children, and the 34 pilocytic astrocytomas. ATRX alterations closely overlapped with mutations in IDH1/2 (p < 0.0001) and TP53 (p < 0.0001) in samples across all WHO grades. They were prevalent in astrocytomas and oligoastrocytomas, but were absent in oligodendrogliomas (p < 0.0001). No significant association of ATRX mutation/loss of expression and alternative lengthening of telomeres was identified in our cohort. In summary, our data show that ATRX alterations are frequent in adult diffuse gliomas and are specific to astrocytic tumors carrying IDH1/2 and TP53 mutations. Combined alteration of these genes may contribute to drive the neoplastic growth in a major subset of diffuse astrocytomas in adults.

Recurrent mutations affecting the histone H3.3 residues Lys27 or indirectly Lys36 are frequent drivers of pediatric high-grade gliomas (over 30 % of HGGs). To identify additional driver mutations in HGGs, we investigated a cohort of 60 pediatric HGGs using whole-exome sequencing (WES) and compared them to 543 exomes from non-cancer control samples. We identified mutations in SETD2, a H3K36 trimethyltransferase, in 15 % of pediatric HGGs, a result that was genome-wide significant (FDR = 0.029). Most SETD2 alterations were truncating mutations. Sequencing the gene in this cohort and another validation cohort (123 gliomas from all ages and grades) showed SETD2 mutations to be specific to high-grade tumors affecting 15 % of pediatric HGGs (11/73) and 8 % of adult HGGs (5/65) while no SETD2 mutations were identified in low-grade diffuse gliomas (0/45). Furthermore, SETD2 mutations were mutually exclusive with H3F3A mutations in HGGs (P = 0.0492) while they partly overlapped with IDH1 mutations (4/14), and SETD2-mutant tumors were found exclusively in the cerebral hemispheres (P = 0.0055). SETD2 is the only H3K36 trimethyltransferase in humans, and SETD2-mutant tumors showed a substantial decrease in H3K36me3 levels (P < 0.001), indicating that the mutations are loss-of-function. These data suggest that loss-of-function SETD2 mutations occur in older children and young adults and are specific to HGG of the cerebral cortex, similar to the H3.3 G34R/V and IDH mutations. Taken together, our results suggest that mutations disrupting the histone code at H3K36, including H3.3 G34R/V, IDH1 and/or SETD2 mutations, are central to the genesis of hemispheric HGGs in older children and young adults.Electronic supplementary materialThe online version of this article (doi:10.1007/s00401-013-1095-8) contains supplementary material, which is available to authorized users.

No abstract

Pediatric high-grade gliomas (pHGGs), encompassing hemispheric and diffuse midline gliomas (DMGs), remain a devastating disease. The last decade has revealed oncogenic drivers including single nucleotide variants (SNVs) in histones. However, the contribution of structural variants (SVs) to gliomagenesis has not been systematically explored due to limitations in early SV analysis approaches. Using SV algorithms, we recently created, we analyzed SVs in whole-genome sequences of 179 pHGGs including a novel cohort of treatment naïve samples-the largest WGS cohort assembled in adult or pediatric glioma. The most recurrent SVs targeted MYC isoforms and receptor tyrosine kinases, including a novel SV amplifying a MYC enhancer in the lncRNA CCDC26 in 12% of DMGs and revealing a more central role for MYC in these cancers than previously known. Applying de novo SV signature discovery, we identified five signatures including three (SVsig1-3) involving primarily simple SVs, and two (SVsig4-5) involving complex, clustered SVs. These SV signatures associated with genetic variants that differed from what was observed for SV signatures in other cancers, suggesting different links to underlying biology. Tumors with simple SV signatures were TP53 wild-type but were enriched with alterations in TP53 pathway members PPM1D and MDM4. Complex signatures were associated with direct aberrations in TP53, CDKN2A, and RB1 early in tumor evolution, and with extrachromosomal amplicons that likely occurred later. All pHGGs exhibited at least one simple SV signature but complex SV signatures were primarily restricted to subsets of H3.3 K27M DMGs and hemispheric pHGGs. Importantly, DMGs with the complex SV signatures SVsig4-5 were associated with shorter overall survival independent of histone type and TP53 status. These data inform the role and impact of SVs in gliomagenesis and mechanisms that shape them.

During the S-phase, the DNA replication process is finely orchestrated and regulated by two programs: the spatial program that determines where replication will start in the genome (Cadoret et al. (2008 Oct 14), Cayrou et al. (2011 Sep), Picard et al. (2014 May 1) [1], [2], [3]), and the temporal program that determines when during the S phase different parts of the genome are replicated and when origins are activated. The temporal program is so well conserved for each cell type from independent individuals [4] that it is possible to identify a cell type from an unknown sample just by determining its replication timing program. Moreover, replicative domains are strongly correlated with the partition of the genome into topological domains (determined by the Hi-C method, Lieberman-Aiden et al. (2009 Oct 9), Pope et al. (2014 Nov 20) [5], [6]). On the one hand, replicative areas are well defined and participate in shaping the spatial organization of the genome for a given cell type. On the other hand, studies on the timing program during cell differentiation showed a certain plasticity of this program according to the stage of cell differentiation Hiratani et al. (2008 Oct 7, 2010 Feb) [7], [8]. Domains where a replication timing change was observed went through a nuclear re-localization. Thus the temporal program of replication can be considered as an epigenetic mark Hiratani and Gilbert (2009 Feb 16) [9]. We present the genomic data of replication timing in 6 human model cell lines: U2OS (GSM2111308), RKO (GSM2111309), HEK 293T (GSM2111310), HeLa (GSM2111311), MRC5-SV (GSM2111312) and K562 (GSM2111313). A short comparative analysis was performed that allowed us to define regions common to the 6 cell lines. These replication timing data can be taken into account when performing studies that use these model cell lines.