Concurrent CIC mutations, IDH mutations, and 1p/19q loss distinguish oligodendrogliomas from other cancers

Authors


Abstract

Oligodendroglioma is characterized by unique clinical, pathological, and genetic features. Recurrent losses of chromosomes 1p and 19q are strongly associated with this brain cancer but knowledge of the identity and function of the genes affected by these alterations is limited. We performed exome sequencing on a discovery set of 16 oligodendrogliomas with 1p/19q co-deletion to identify new molecular features at base-pair resolution. As anticipated, there was a high rate of IDH mutations: all cases had mutations in either IDH1 (14/16) or IDH2 (2/16). In addition, we discovered somatic mutations and insertions/deletions in the CIC gene on chromosome 19q13.2 in 13/16 tumours. These discovery set mutations were validated by deep sequencing of 13 additional tumours, which revealed seven others with CIC mutations, thus bringing the overall mutation rate in oligodendrogliomas in this study to 20/29 (69%). In contrast, deep sequencing of astrocytomas and oligoastrocytomas without 1p/19q loss revealed that CIC alterations were otherwise rare (1/60; 2%). Of the 21 non-synonymous somatic mutations in 20 CIC-mutant oligodendrogliomas, nine were in exon 5 within an annotated DNA-interacting domain and three were in exon 20 within an annotated protein-interacting domain. The remaining nine were found in other exons and frequently included truncations. CIC mutations were highly associated with oligodendroglioma histology, 1p/19q co-deletion, and IDH1/2 mutation (p < 0.001). Although we observed no differences in the clinical outcomes of CIC mutant versus wild-type tumours, in a background of 1p/19q co-deletion, hemizygous CIC mutations are likely important. We hypothesize that the mutant CIC on the single retained 19q allele is linked to the pathogenesis of oligodendrogliomas with IDH mutation. Our detailed study of genetic aberrations in oligodendroglioma suggests a functional interaction between CIC mutation, IDH1/2 mutation, and 1p/19q co-deletion. Copyright © 2011 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Introduction

Oligodendroglioma, a glioma subtype marked by unique clinical, pathological, and genetic characteristics, is composed of neoplastic cellular elements that resemble oligodendrocytes. Unlike other gliomas such as astrocytomas and ependymomas, oligodendrogliomas are chemosensitive and often progress in a slow and predictable manner 1. Oligodendrogliomas display a classical appearance of cells, with round, regular nuclei associated with clearing of the cytoplasm and in close proximity to fine branching vasculature 2. Most importantly, oligodendrogliomas are strongly linked to co-deletions of the short arm of chromosome 1 (1p) and the long arm of chromosome 19 (19q). These recurrent genetic alterations in turn have been strongly linked to favourable clinical behaviour and classic oligodendroglioma histology such that 1p/19q co-deletion is considered a defining feature 2–4.

Recent developments in brain tumour genomics have highlighted the uniqueness of oligodendrogliomas 5–7. In particular, it has been shown that the proneural gene expression signature is enriched in oligodendrogliomas, particularly those with 1p/19q co-deletion 5, 8. In addition to displaying the proneural signature, 1p/19q co-deleted oligodendrogliomas are associated with a constellation of positive prognostic markers including methylation of the MGMT promoter, IDH1 mutations, and the recently described CpG island methylator phenotype (G-CIMP) 9. Although these markers are also present in glioblastomas that arise from low-grade astrocytomas, an important divergence in the molecular pathogenesis of low-grade oligodendrogliomas and astrocytomas is 1p/19q co-deletion in the former and TP53 mutations in the latter 10. The mutual exclusivity of these events underscores the distinct molecular characteristics of oligodendrogliomas.

Despite these advances, our understanding of the genetic underpinnings of oligodendrogliomas remains unclear. Deletions of 1p and 19q, which result in loss of heterozygosity in those regions, may unmask mutations leading to the oligodendroglioma phenotype. This could arise strictly within 1p and 19q or could be the result of global genomic or epigenomic changes. However, efforts to find candidate tumour-associated genes in 1p and 19q, and more globally, have met with limited success 11–14. Until recently, our progress in understanding oligodendrogliomas has also been hampered by a dearth of robust model systems with which to conduct functional studies 15.

Next-generation sequencing technology and advancements in bioinformatics have transformed the field of cancer genomics. Together, these developments afford rapid, cost-effective ‘deep-sequencing’ of cancer genomes, thus enabling genome-wide searches for cancer-associated mutations 16–23. Within the last month, two groups have independently reported their findings from next-generation sequencing of oligodendrogliomas with markedly different results. One group sequenced all exons, microRNA, splice sites, and promoter regions on 1p and 19q in seven oligodendrogliomas and found no recurrent alterations 14. The other sequenced the exomes of seven oligodendrogliomas, and performed directed sequencing in an additional 27 oligodendrogliomas, and reported recurrent mutations in CIC in 53% of their cases and FUBP1 in 15% 24.

Here, we report the results of our concurrent effort to identify somatic mutations potentially contributing to 1p/19q co-deleted oligodendroglioma through a combination of exome, transcriptome, and whole genome shotgun sequencing. We found frequent mutations in the CIC gene, noting that these cluster in the DNA-interacting HMG domain and the protein–protein interacting GRO-L domain. Thirteen out of 16 tumours in our discovery set harboured protein-altering somatic mutations, nonsense mutations, or insertions/deletions (INDELs) of CIC. Subsequent deep amplicon sequencing of CIC from 73 gliomas with matched normal tissues identified a further eight somatic CIC mutations and INDELs, and seven out of eight of these were found in 1p/19q co-deleted oligodendrogliomas with IDH1 mutation. Our study highlights a unique relationship between recurrent chromosomal aberrations and mutations in CIC, and suggests that mutations in CIC are key events in the development of oligodendrogliomas.

Materials and methods

Sample acquisition and preparation

Fresh frozen tumour tissue, matched normal blood, and formalin-fixed, paraffin embedded (FFPE) tissue from patients undergoing surgery for oligodendroglioma, oligoastrocytoma, and astrocytoma were obtained from the Calgary Brain Tumour Bank. H&E-stained sections of the samples were examined for sample quality. Tumour samples with estimated greater than 60% tumour cell content and less than 20% necrosis were used for nucleic acid extraction. DNA was extracted from fresh frozen tissue, FFPE tissue, and frozen blood buffy coat using the QIAamp DNA Mini Kit (Qiagen, Valencia, CA, USA). RNA was extracted from tumour tissue using Trizol (Invitrogen, Burlington, Ontario, Canada) and was evaluated for quality using an Agilent 2100 Bioanalyzer (Agilent, Mississauga, Ontario, Canada). RNA samples with a RIN ≥ 8 were used for transcriptional analyses. Additional tumour and matched normal specimens were provided by the Cancer Genetics Laboratory at BCCA and from the Massachusetts General Hospital Brain Tumour Repository. All samples were collected, de-identified, and used in accordance with approval by the respective local research ethics boards.

Clinicopathological information

Demographic, diagnostic, treatment, and follow-up information from Calgary cases was retrieved from medical records. 1p and 19q status was determined previously by routine PCR microsatellite analysis or FISH. IDH1 and IDH2 genotyping was performed as described previously 25. Progression-free survival (PFS) was calculated from the date of diagnosis to the date of recurrence or the date last seen without recurrence. Overall survival (OS) was calculated to the date of death or the date last seen. Comparisons of mutation status with tumour histology or grade were performed using Fisher's exact test. Stratified analyses were performed using the Mantel–Haenszel statistic. PFS and OS according to mutation status were examined using Kaplan–Meier survival curves with statistical testing by log-rank test. p values were two-tailed and p < 0.05 was considered significant. SAS version 9.2 (SAS Institute, Cary, NC, USA) was used for statistical analyses.

Sequencing of matched tumour and blood specimens

Whole genome shotgun sequencing (WGSS)

WGSS was performed on the SOLiD 4.0 platform as previously described 26. Paired end reads were aligned to the reference human genome (NCBI Build 36.1, hg18) using diBayes. Single nucleotide variant (SNV) detection was performed using the Samtools package (0.1.5) 27, and candidate somatic mutations were identified by comparison of tumour and matched normal DNA sequence data. Candidate somatic mutations were disqualified if one or more reads from the matched normal sample presented evidence for the SNV.

Exomic and transcriptomic sequencing

An Agilent SureSelect target enrichment kit was used for exome library construction according to the manufacturer's protocol (Agilent). Library construction for RNA-Seq was performed as previously described 16, 28. Sequencing was performed on the Illumina GAIIx platform. RNA-Seq reads were processed as previously described 16, 19, and paired end reads were aligned to the reference human genome assembly using BWA (0.5.7) 29. SNV detection was performed using SNVMix2 with somatic non-synonymous SNVs identified by referring to the matched normal pileup 29, with filtering to include SNVs such that the combined probability of either heterozygous or homozygous SNV was greater than 0.99 30.

CIC deep amplicon sequencing

Deep sequencing of the entire CIC gene (chr19: 47,467,725-47,480,658; hg18), including the unannotated exon 0 identified in RNA-Seq, was performed using the Illumina HiSeq2000 platform (see Supporting information, Supplementary materials and methods for detailed methods).

Validation of mutations

All somatic SNVs identified in WGSS and exome sequencing were examined and confirmed manually using IGV Browser 2.0 (Broad Institute, Cambridge, MA, USA). PCR and Sanger sequencing were used to validate all CIC mutations and INDELs in tumour and matched normal samples (Supplementary Figures 1–5 and Supplementary Tables 1–3). Lastly, exons 5 and 20 were Sanger-sequenced in an extended validation set of 111 gliomas.

Results

Next-generation sequencing of 1p/19q co-deleted oligodendrogliomas identifies known and novel somatic mutations

To survey the spectrum of somatic aberrations in the coding sequence present in oligodendrogliomas, we conducted Illumina exome sequencing on a discovery set of 16 pairs of 1p/19q co-deleted oligodendrogliomas and matched normal bloods. Sequencing was performed to more than 100 times average redundancy of the on-target region from the Agilent SureSelect kit (v1.0). After identification of somatic SNVs, non-synonymous mutations were identified and grouped by gene (Table 1). This analysis revealed 3212 candidate somatic non-synonymous mutations in 2696 genes. Of the 3212 mutations, 340 SNVs were manually verified in IGV.

Table 1. Recurrent mutations and recurrently mutated genes seen in three or more libraries. List of genes with recurrent or more than one non-synonymous change. Sixteen tumour and matched normal exomes were sequenced and SNV calls were annotated. Novel non-synonymous mutations were identified by filtering against the normal pileup and dbSNP129. All mutations were manually curated using the IGV browser
GeneChrPositionSample(s)Nucleotide (cDNA)Amino acid
IDH12208821357HS1893, HS2040, HS2042, HS2805, HS2806, HS2807, HS2808, HS2809, HS2810, HS2811, HS2812, HS2813, HS2815, HS3124c. 395 C > Tp.R132H
CIC1947483558HS2040c.604 C > Tp.R202W
CIC1947483100HS2814c.142 G > Ap.G107E
CIC1947488171HS2813c.2980 C > Tp.Q994X
CIC1947483598HS2815c.644 G > Ap.R215Q
CIC1947483690HS2810c.736 A > Gp.K246E
CIC1947490900HS2045c.4544 G > Tp.R1515L
CIC1947490900HS3124c.4544 G > Ap.R1515H
CIC1947483580HS2805c.626 T > Gp.I209S
CIC1947483653–47483655HS2807DEL : CCT 
CIC1947486863–47486864HS2806DEL : AA 
CIC1947490858–47490858HS2814INS : A 
NOTCH19138528889HS2805, HS2811c.2101 T > Gp.T701P
NOTCH19138532470HS2811c.1195 T > Gp.T399P
NOTCH19138522244HS3124c.3494 C > Tp.G1165D
NOTCH19138532105–138532107HS2814DEL : TTG 
IDH21588432842HS2045, HS2814c.515 C > Tp.R172K
PCDH81352318741HS2805, HS2811c.1832 T > Gp.H611P

Three genes were mutated in three or more cases (IDH1, CIC, and NOTCH1). The canonical mutation resulting in p.Arg132His in IDH1 was seen in 14 cases, and the remaining two tumours were mutated in IDH2, resulting in p.Arg172His. Out of a total set of 340 IGV validated non-synonymous mutations, 54 more mutations were seen in the same gene in more than one case—these included alterations of SOX11, DACH1, PTPRZ1, and ATM. Two post-treatment tumours displayed significantly elevated rates of somatic mutation approximately nine-fold greater than the rest of the tumour exomes (Supplementary Figures 1–5 and Supplementary Tables 1–3). Similar to a subgroup of recurrent glioblastomas, somatic mutations in the mismatch repair gene MSH6 were present in both of these cases 31, 32. TP53 mutations were present only in these two hypermutator tumours.

Although somatic coding mutations of FUBP1 were recently reported in a subset of oligodendrogliomas 24, we did not find any non-synonymous somatic mutations in this gene in the 16 oligodendroglioma cases, despite having five-fold redundant coverage or better of all the FUBP1 exonic regions except in the first and last exons. We did, however, identify two FUBP1 candidate somatic splice site variants at chr1 : 78205319 and chr1 : 78193604.

CIC mutations are common in oligodendrogliomas but rare in other cancers

After IDH1, the gene next most frequently mutated in our discovery cohort was CIC (Source: UniProtKB/Swiss-Prot; Acc: Q96RK0), located on chromosome 19q within the commonly deleted region 33. In our exome libraries, we found 13 of 16 tumours mutated in CIC, one of which carried two distinct mutations. To validate the exome findings and to determine the frequency of CIC mutations in oligodendrogliomas, we performed CIC-focused deep sequencing on the original 16 tumour/normal pairs in the discovery set as well on tumour/normal pairs from 13 additional known 1p/19q co-deleted tumours. All mutations identified in the 13/16 discovery samples by exome sequencing were confirmed by deep amplicon sequencing. Seven of the 13 additional cases were also identified as having CIC mutations, yielding a CIC mutation rate in oligodendrogliomas of 20/29 (69%).

In contrast, deep sequencing of 60 tumour/normal pairs from a variety of astrocytomas and mixed oligoastrocytomas without 1p/19q co-deletion revealed only one mutant tumour (2%). This set contained four cases with 1p loss only and nine cases with 19q loss only. Interestingly, the single somatic non-coding mutation was found in a tumour with 19q loss only.

To further determine whether CIC mutations were commonly found in other cancer types, including those outside the brain, we searched for CIC alterations in the Genome Sciences Centre (GSC) database containing SNVs from next-generation sequencing of 2315 cancer and normal libraries 34. Whereas CIC mutations were present in 13/16 (81%) of our oligodendroglioma exome libraries, the GSC database search revealed that CIC alterations in general are rare—0% of normal libraries and only 7.3% of libraries from other non-oligodendroglial cancers.

Preferential distribution of CIC mutations in oligodendrogliomas

Of the 21 mutations detected in the 20 CIC-mutant oligodendrogliomas (one carrying two distinct mutations), nine were clustered in exon 5 within an annotated DNA-interacting domain, three were in exon 20 within an annotated protein-interacting domain, and the remaining nine were in other exons (Figure 1). We also detected recurrent somatic mutations resulting in p.Arg201Trp and p.Arg215Trp changes and a recurrent out-of-frame deletion at chr19 : 47483653delTCC in exon 5. Another recurrent somatic mutation in codon 1515 of exon 20, leading to change of Arg > His and Arg > Leu, was also identified. When the rare CIC mutations were identified in the GSC human variant database, exons 5 and 20 mutations were represented in 8.6% of non-oligodendroglial tumours but disproportionally in 69% of oligodendrogliomas. Similarly, a survey of the Catalogue of Somatic Mutations in Cancer (COSMIC v54) database showed ten reported mutations in CIC, none occurring in exon 5 or 20.

Figure 1.

Mutations are identified along a schematic of the human CIC gene with the highly conserved DNA-interacting HMG domain (exon 5) and protein–protein interacting GRO-L homology domain (exon 20) identified. Somatic mutations are identified with arrows and amino acid changes. Recurrent mutations are outlined. The vast majority of mutations from 1p/19q co-deleted oligodendroglioma (IDH1/2-mutated) are concentrated within exons 5 and 20

To assess the significance of the clustering of somatic mutations in CIC exons 5 and 20, we performed multiple sequence alignments using CLUSTALW 35. This analysis revealed that CIC exon 5 encoded a conserved annotated DNA-binding region, while exon 20 encoded a conserved domain lacking annotation in humans (Supporting information, Supplementary materials and methods). Side-by-side alignments of human and fruit fly CIC protein sequences revealed that the second conserved region mapped to a protein–protein interaction region, annotated in the fly as the ‘GRO-L interaction region’ (not shown).

Validation of CIC exon 5 and 20 mutations in an expanded number of gliomas

Focused PCR and Sanger sequencing of CIC exons 5 and 20 was then performed on an extension set of 111 primary glioma specimens including 68 oligodendrogliomas (WHO II and III), 11 mixed oligoastrocytomas (WHO II and III), 32 astrocytomas (WHO II–IV), and four normal brain specimens. The association of CIC mutations with 1p/19q co-deleted oligodendrogliomas was seen again in this cohort. In this extension set, 15 mutations were detected, of which 14 (93%) were found in IDH1-mutated, 1p/19q co-deleted oligodendrogliomas (see the Supporting information, Supplementary materials and methods for a detailed list of samples and mutations).

Association of CIC mutations with oligodendroglioma histology, IDH1/2 mutations, and 1p/19q status

Of the 204 tumour samples studied (16 discovery set exome and CIC deep sequenced, 77 CIC deep sequenced, and 111 CIC ex5/20 Sanger sequenced), 191 were informative for CIC, IDH1/2, and 1p/19q. CIC mutations were closely associated with classic oligodendroglioma histology (p < 0.001), and therefore also associated with other oligodendroglioma-related mutations in IDH1/2 mutations (p < 0.001), and 1p/19q loss (p < 0.001). No exon 5 CIC mutations were noted in tumours that were both IDH1/2 wild-type and 1p/19q intact. The single instance of a CIC mutation outside of classic oligodendrogliomas occurred in exon 20 in a case of WHO IV astrocytoma/gliomatosis cerebri, raising the possibility that exon 20 mutations may be less specific for oligodendrogliomas than other mutations. No CIC mutations were identified in four normal brain tissue samples. Thus, CIC mutations appear to be specific to prototypical oligodendrogliomas wherein we frequently observed co-alteration of IDH, 1p/19q, and CIC abnormalities.

Clinicopathological correlation of oligodendrogliomas with and without mutations in CIC

To determine whether the presence of CIC mutation has prognostic significance, we examined relationships with tumour grade, progression-free survival, and overall survival. Among the 84 oligodendrogliomas with 1p/19q co-deletion (out of 99 oligodendrogliomas total), CIC mutations were detected in 15/44 (34.1%) low-grade tumours and 19/40 (47.5%) high-grade tumours (p = 0.27). Controlling for 1p/19q status in a stratified analysis, there was a trend towards an association between the presence of a CIC mutation and high-grade histology (OR 2.05, 95% CI 0.86–4.66, p = 0.10). Clinical follow-up was available for 43 patients with oligodendrogliomas of known CIC status (23 wild-type and 20 mutant). In this cohort, CIC mutation was not associated with recurrence (13/22 wild-type versus 10/21 mutant, p = 0.55), progression-free survival (p = 0.89), or overall survival (p = 0.99) (Figure 2).

Figure 2.

Kaplan–Meier analyses show no differences in progression-free survival (PFS) or overall survival (OS) in CIC-mutant versus wild-type oligodendroglioma patients

Validation of CIC mutations in oligodendroglioma cell lines

The initiation and maintenance of different types of brain tumours have been linked to the presence of ‘tumour-initiating cells’ (TICs) 36, 37. These cells have the ability to undergo asymmetric and symmetric division to propagate and to generate various types of neoplastic progeny. TIC lines propagated under stem cell-promoting conditions have become valuable reagents in the study of cancer biology as they more reliably maintain phenotypic and genotypic characteristics of their parent cancers. Recently, two oligodendroglioma-derived TIC lines (BT54 and BT88) were established that display the oligodendrogliomatous phenotype and co-deletions of 1p and 19q 15. To determine whether these cell lines were similar at the DNA sequence level to our patient cohorts and to assist in the use of the lines for functional studies of oligodendroglioma, we sequenced the genomes of the two TIC lines using SOLiD technology to an average of 30× haploid coverage (Supporting information, Supplementary materials and methods). We confirmed that line BT54 carried the mutation chr19 : 47483801G > C affecting the CIC exon 6 splice acceptor site and that line BT88 has a missense mutation leading to p.ARg1515His, both of which were identified in the respective parental tumour tissues from which the lines had been derived.

Expression of mutant CIC transcripts and protein

Analysis of RNA-Seq data confirmed that the CIC mutations identified at the DNA level in the exomes resulted in the expression of mutation-bearing CIC transcripts. Additional gene expression analyses of 16 oligodendrogliomas, including four wild-type and 12 with mutations in CIC, also showed no significant difference in transcript expression levels between wild-type and mutated CIC (data not shown). The assembly of RNA-Seq data using ABySS 38, 39 revealed the presence of a novel alternative transcript. A previously unknown exon (exon 0) was spliced to exon 2. Thus, we predict two predominant transcripts of CIC in oligodendrogliomas: a short isoform predicted to encode a protein of 1608 amino acids (160 kD), and a long isoform, incorporating exon 0 and predicted to encode 2517 amino acids (250 kD). We confirmed the specificity of our antibody via CIC sequence-specific siRNA knockdown, which also confirmed the presence of both short and long form proteins in protein blot and immunoprecipitation–MRM assay (Figures 3A, 3B1, and 3B2). Next, we confirmed that CIC protein expression in lysates from the control human embryonic kidney cell line (HEK293), BT54 (somatic mutation affecting splice acceptor in exon 6), and BT88 (p.Arg1515His) were equivalent, indicating that CIC mutant protein is expressed and that both predicted transcripts encode CIC protein (Figure 3C).

Figure 3.

(A) Antibody specificity for CIC confirmed by siRNA. (B1-2) MRM show long isoform expression. (C) A representative western blot showing high levels of CIC mutant protein expression both in nuclear and in cytoplasmic protein fractions from oligodendroglioma cell lines, line 54 (L54), and line 88 (L88). CE = mouse cerebellum; HEK = HEK293 embryonic kidney line

Discussion

The data presented in this study, generated from systematic sequencing of tumour exomes, genomes, and transcriptomes of oligodendroglioma, demonstrate the power of massively parallel next-generation sequencing in uncovering novel genetic changes in cancers. We sequenced a total of 628 096 883 783 bp of tumour and 499 887 339 249 bp of normal DNA from genome, RNA-Seq, and exome libraries. From this, we confirmed that IDH mutations are a universal feature of oligodendrogliomas. We also found novel potential driver mutations in CIC in a majority of the tumours.

On a broad level, the data show that oligodendrogliomas carry a relatively modest and stable somatic mutational load compared with many other cancers—much lower than that present in glioblastoma, melanoma, lymphoma, and colorectal cancer (Broad Institute, unpublished data). In addition to known positive prognostic factors such as the proneural gene signature and MGMT promoter methylation, the finding that oligodendrogliomas bear mutations relatively infrequently might also explain the relatively benign clinical course of this tumour and its initial chemosensitivity, a hypothesis that has yet to be formally tested. A particularly striking finding from our analysis, however, was the identification of recurrent mutations in CIC and the apparent relationship between mutations in the CIC and IDH genes, and the 1p/19q anomaly, which appear to be distinguishing features of oligodendrogliomas.

CIC, found on chromosome 19q13.2, encodes the mammalian homolog of the Drosophila transcriptional repressor Capicua. In Drosophila, Capicua is a transcriptional repressor with crucial roles in development—it represses genes downstream of the RAS/MAPK pathway that signal the differentiation of wing veins, imaginal eye discs, and head and tail polarity development 40–42. Mammalian Capicua is highly expressed in the brain, particularly in the external granular cells of the developing cerebellum 43, and its overexpression has been identified in medulloblastomas, an aggressive primitive neuroectodermal tumour of the central nervous system 44. However, CIC mutations in medulloblastoma have not been reported. Association of CIC alterations with human disease is restricted to two diseases affecting cells of the neural crest lineage. Spinocerebellar ataxia I (SCAI), a progressive neurodegenerative disease, and Ewing's family tumours, a cohort of aggressive soft tissue tumours affecting children and young adults, both involve defects of the CIC gene which in turn lead to dysregulated interactions of the CIC protein with downstream signalling partners 45, 46. With respect to the Ewing's family of tumours, rare cases of Ewing's sarcoma have a recurrent chromosomal translocation, t(4;19)(q35;q13), that generates a novel CIC–DUX4 fusion protein. The fusion oncoprotein exhibits enhanced transcriptional activation of the ETS family genes ERM/ETV5 and ETV145, 47, whose expression levels are also elevated in ‘traditional’ Ewing's sarcoma with the EWS–FLI1 fusion 48.

It is intriguing that recurrent mutation in CIC, located on chromosome 19q, is found almost exclusively in 1p/19q co-deleted oligodendrogliomas with IDH1 mutation, yet loss of chromosome 1p is more strongly associated with the oligodendrogliomatous phenotype and clinical behaviour than 19q loss 3. In addition, 19q loss, which is occasionally found in astrocytomas, has traditionally been considered less specific to oligodendroglioma. Thus, it was unexpected that our sequencing did not yield more recurrent candidate coding point mutations in 1p, including mutations in FUBP1 as reported by others 24. Since our study focused on the detection of recurrent point mutations, other types of recurrent alterations affecting the 1p region have not been excluded. Larger-scale sequencing of panels of whole tumour and normal genomes would address this possibility.

High levels of cross-species sequence conservation and identification of functional domains encoded by the human CIC gene support possible functional consequences of mutations in CIC. Two highly conserved regions in CIC include the high mobility group (HMG) DNA-binding domain in exon 5 and the protein–protein interaction GRO-L domain in exon 20 at the C-terminus of the protein—both of these regions contained recurrent mutations in our patient series. We have confirmed that mutant CIC mRNA and mutant CIC protein are expressed. Thus, based on the known structure/function of CIC, it is likely that the oligodendroglioma mutations play a key role in the biology of the disease.

Two-thirds of CIC somatic mutations in oligodendroglioma occurred in exon 5 within the HMG-box domain. A majority of these are clustered around several hotspots in the following order of frequency: p.Arg215 > p.Arg201 > p.Arg202, and the affected residues display a high degree of evolution conservation in all examined vertebrate homologs. Furthermore, 33% of the unique exon 5 mutations were predicted to affect protein function using POLYPHEN2 analysis which included the recurrent p.Arg201Trp and p.Arg215Trp changes. Interestingly, five out of 20 exon 5 mutations affected amino acid position 215 and three of these caused p.Arg215Trp changes, which corresponds to the Drosophila cicArg505Trp mutation associated with loss of CIC function. These mutations result in aberrant Drosophila eye development 42. After exon 5, the next most frequently altered CIC region was in the C-terminal portion of exon 20. We found four mutations causing amino acid changes at p.Arg1515 to His, Leu, and Cys. All were predicted to be deleterious to protein function by POLYPHEN2 analysis. It is undetermined, however, whether the hemizygous CIC mutations in oligodendroglioma result in loss or gain of protein function. Whatever the consequence, the recurrent nature of CIC mutations in such a high percentage of tumours supports the view that they are driver mutations central to the genesis and/or identity of the tumours. It is clear that carefully designed functional studies taking into account genomic context (such as 1p/19q loss and IDH mutations) are necessary to delineate the role of CIC in the pathogenesis of oligodendroglioma.

Further investigation of this complex relationship and of the downstream molecular consequences of CIC mutation will necessitate the development of novel biological reagents and appropriate model systems. To this end, the two oligodendroglioma TIC lines analysed here represent a unique resource to study the relationship between 1p/19q co-deletions and the special biology and clinical behaviour of this tumour. Future studies should also concentrate on sequencing of CIC in larger cohorts of oligodendrogliomas with clinical follow-up data to better distil potential influences on outcome. This will most likely involve international multicentre collaboration similar to a recent study 49.

In the minority of oligodendrogliomas where we do not find CIC mutations, it is reasonable to speculate that other mutations in the RAS/MAPK pathway or in CIC interacting proteins may result in a phenocopy of CIC mutants. One potential candidate is ATXN2, which we found mutated in two oligodendrogliomas in our exome dataset. CIC is known to interact with ATXN1 in human disease 46, and ATXN1 is modulated by ATXN2 50. Similarly, we and others 24 have found a low frequency of NOTCH mutations in oligodendrogliomas. Recent work suggests direct involvement of ATXN1 in the NOTCH signalling pathway that has critical roles in neural development and tumourigenesis 51, 52. Lastly, FUBP1 was reported to be mutated in oligodendroglioma 24, and although we did not find non-synonymous coding mutations in our tumours, we did detect infrequent somatic splice site variants. The possible relationship between FUBP1 and CIC, however, remains to be elucidated. Our working model is that mutations in CIC (in the majority of tumours) or a number of related genes such as ATXN2, NOTCH1, and FUBP1 (in a minority of tumours) may be functionally equivalent in the context of 1p/19q loss, IDH1/2 mutation, and the genesis of oligodendroglioma (Figure 4). In our exome discovery cohort, we noted various patterns of co-occurrence of somatic mutations affecting CIC, NOTCH genes, and ATXN2. However, such a model has yet to be formally tested in a larger cohort of tumours and also in the context of the two predominant patterns of mutations in CIC affecting the two functional domains.

Figure 4.

Proposed CIC interaction network in the context of observed mutations

Our discovery that CIC mutations are a core abnormality in the majority of oligodendrogliomas has translational and clinical implications in neuro-oncology. Improved understanding of the role of CIC in a background of IDH1/2 mutations in co-deleted oligodendroglioma will significantly enhance our understanding of this disease and may pave the way for the development of more targeted therapies for this common type of glioma.

Acknowledgements

We would like to acknowledge Dr John Kelly for cell lines, Dr Peter Forsyth and the Calgary Brain Tumour Bank for tissues, Colleen Anderson and Susan Hui for technical assistance, Alberta Innovates Health Solutions (JAC, EES), the Brain Tumour Foundation of Canada (JGC), the Alberta Cancer Foundation (JGC), and the Clark Smith Family. We also acknowledge Dr Robert Jenkins for performing mFISH on the brain tumour-initiating cell lines BT054 and BT088 and sharing the data. MAM acknowledges the support of the BC Cancer Foundation, Western Economic Diversification, Genome Canada, and Genome British Columbia. MAM is also grateful to Dr Brian Thiessen, BCCA, for early conversations that helped shape the project. SY acknowledges the generous support of BrainCare BC and the BCCA Hershey and Yvette Porte Neuro-Oncology Endowment Fund. We would also like to acknowledge Maria Mendez-Lago, Diane Trinh, Kane Tse, and Thomas Zeng at the BCGSC for invaluable scientific advice and technical support.

Author contribution statement

JGC and MAM funded and co-directed the project, and designed the study. SY planned and executed the validation experiments, and performed data analysis. YSB conducted next-generation sequencing (NGS) bioinformatic analysis (WGSS, exome, amplicon). JAC identified tumours and performed the histopathology and analysis, and provided tumour and blood samples for NGS. MDB collected and identified tumour samples for sequencing. AK and S Young contributed additional tumour and blood samples for validation. DNL contributed additional tumour samples for validation. CC and MP assisted in the collection and identification of tumour samples. OM conducted NGS bioinformatic analysis (WTSS, exome, hypermutation, CIC homology). SM conducted NGS bioinformatic analysis (WTSS). SC coordinated and led the amplicon deep sequencing and functional/proteomic validation, and assembled the data. GBM coordinated and performed proteomic analysis. AN performed proteomic analysis. AJM coordinated NGS. YZ performed NGS. SJ coordinated NGS bioinformatic analysis. RC conducted NGS bioinformatic analysis (WGSS). KLM conducted NGS bioinformatic analysis (ABySS pipeline). IB, RC, EC, RD, MH, SJ, HL, JQ, NT, and RV assisted in NGS bioinformatic analysis. MF and JTW performed deep amplicon sequencing and biochemical analysis. RM coordinated sequencing and NGS bioinformatic analysis. AM conducted biochemical analysis (mass spectrometry). GR and EEE contributed clinical information and performed survival analysis. SW generated TIC lines from tumours. WW performed statistical analysis. JGC, MAM, SY, YSB, JAC, OM, and SC wrote the manuscript.

SUPPORTING INFORMATION ON THE INTERNET

The following supporting information may be found in the online version of this article.

Supplementary materials and methods.

Figure S1. Next-generation sequencing analysis of loss of heterozygosity status and gene expression of oligodendroglioma.

Figure S2. Mutational load of oligodendrogliomas in discovery set.

Figure S3. Long-range PCR of the CIC gene for deep amplicon sequencing.

Figure S4. Identification of previously unannotated exon 0 in human CIC.

Figure S5. Analysis of CIC sequence homology across different species.

Table S1. List of IGV validated somatic single nucleotide variants identified in the oligodendroglioma exome discovery set.

Table S2. List of PCR primers for focused Sanger sequencing of CIC, IDH1, and IDH2.

Table S3. Summary of tumour diagnosis, 1p19q status, IDH1/2 status, and CIC mutational status.

Ancillary