RNA-sequencing of IDH-wild-type glioblastoma with chromothripsis identifies novel gene fusions with potential oncogenic properties

Abstract

Glioblastoma (GBM) is the most frequent and most aggressive form of glioma. It is characterized by marked genomic instability, which suggests that chromothripsis (CT) might be involved in GBM initiation. Recently, CT has emerged as an alternative mechanism of cancer development, involving massive chromosome rearrangements in a one-step catastrophic event. The aim of the study was to detect CT in GBM and identify novel gene fusions in CT regions. One hundred and seventy IDH-wild-type GBM were screened for CT patterns using whole-genome single nucleotide polymorphism (SNP) arrays. RNA sequencing was performed in 52 GBM with CT features to identify gene fusions within CT regions. Forty tumors (40/52, 77%) harbored at least one gene fusion within CT regions. We identified 120 candidate gene fusions, 30 of which with potential oncogenic activities. We validated 11 gene fusions, which involved the most recurrent fusion partners (EGFR, SEPT14, VOPP1 and CPM), by RT-PCR and Sanger sequencing. The occurrence of CT points to underlying gene fusions in IDH-wild-type GBM. CT provides exciting new research avenues in this highly aggressive cancer.

Fig. 1

Landscape of copy number alterations in 170 IDH-wild-type GBM. Whole chr losses are in dark green, partial losses in light green, chr gains in red (which were mostly whole chr gains) and copy neutral loss of heterozygosity (LOH) in light blue. Amplifications and homozygous deletions of key cancer genes in GBM are shown in orange and CT is shown in dark blue. Chr: chromosome; CT: chromothripsis. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2

Chromothripsis occurrence across the genome in 170 IDH-wild-type GBM A. CT mostly involved chr 7, 9 and 12. B.EGFR amplification was more frequent in cases with CT on chr 7 compared to cases without CT on chr 7 (22/26, 84.6% vs 49/144, 34.0%, p < 0.001). C.CDKN2A homozygous deletion was more frequent in cases with CT on chr 9 compared to cases without CT on chr 9 (18/19, 94.7% vs 79/151, 52.3%, p < 0.001). D. Amplification of MDM2 and/or CDK4 were more frequent in cases with CT on chr 12 compared to cases without CT on chr 12 (12/12, 100.0% vs 14/158, 8.9%, p < 0.001). Exact Fisher test; p-value: 0.12 (ns), 0.033 (*), 0.002 (**), <0.001 (***). CT: chromothripsis, chr: chromosome, HD: homozygous deletion.

Fig. 3

Selection of the fusion partners with potential oncogenic properties. The association of genes to cancer was estimated with Oncoscore, a bioinformatic tool that ranks cancer-related genes based on citation frequencies in the literature (OncoScore cut-off threshold = 21.09 (horizontal dashed line), according to the developer's recommendations and publications). The relevance of the putative oncogenes (triangle) and tumor suppressor genes (square) were manually checked from the available Pubmed literature. IF: in-frame.

Fig. 4

Identification of 30 putative in-frame oncogenic gene fusions within CT regions in 22 IDH-wild-type GBM. Chromosomes are represented in blocks in the inner ring. The outer ring indicates the names of the genes. Putative in-frame oncogenic fusions are represented by arcs joining the two fusion partners. Extensive inter- or intra-chromosomal rearrangements may lead to the formation of oncogenic fusions. Putative in-frame gene fusions within CT regions that involved at least one partner with potential or well-known oncogenic properties were selected.

Fig. 5

Candidate gene fusions identified within CT regions of IDH-wild type GBM. Each figure shows 1) chr ideograms (top), with a vertical line indicating the location of the partner gene within each chr, 2) transcript portions (greyed-out) of each partner gene whose fusion is represented by a connecting line (middle), and 3) predicted chimeric protein (bottom) with the protein domain annotations. A.CPM-MDM2 gene fusion (case 38). B.CPA6-CPM gene fusion (case 59). C.LEMD3-CPM gene fusion (case 144). D.VOPP1-SEPT14 gene fusion (case 95). E.BLVRA-SEPT14 gene fusion (case 109). F.VOPP1-ABCA13 gene fusion (case 105). Chr: chromosome; TMhelix: transmembrane helix; LEM: LAP2, emerin, MAN1.

Supplementary file 1

Chromothripsis is a one-step catastrophic event leading to massive chromosome rearrangements. A. Schematic example of CT, by which one chr is shattered into pieces randomly reassembled by the DNA repair machinery. The derivative chr might harbor fusion genes with potential oncogenic properties. Loss of DNA fragments may lead to the inactivation of tumor suppressor genes and (onco)gene amplification may lead to the formation of double minutes. B. Copy number profile from the same chr as shown in A. There is an oscillating pattern of copy number states with interspersed loss and retention of heterozygosity and focal amplification (fragment F). Chr: chromosome.

Supplementary file 3

Detection of 120 gene fusions within CT regions in 52 IDH-wild type GBM. Putative in-frame oncogenic fusions are represented by arcs joining the two fusion partners. RNA-seq allowed the identification of 120 gene fusions, mostly resulting from intra-chromosomal rearrangements (109/120, 90.8%).

Supplementary file 4

Characterization of EGFR status in 52 IDH-wild-type GBM with CT. A.EGFR amplification was detected in 51.9% (27/52) of IDH-wild-type GBM with CT (irrespective of the chr involved). All five cases harboring an EGFR gene fusion also showed an amplification of EGFR (detected by SNP arrays) but consistently lacked the EGFRvIII rearrangement (identified by RNA-seq). EGFRvIII rearrangement was observed in 18.5% (5/27) of EGFR-amplified GBM.B. FPKM values reflect expression levels of mRNA. When EGFR gene was amplified, EGFR transcript levels were not significantly different whether there was an additional EGFR gene fusion or an additional EGFRvIII rearrangement.

Supplementary file 5

Examples of EGFR gene fusions identified within CT regions in IDH-wild-type GBM. Each figure shows 1) chr ideograms (top), with a red line indicating the location of the two partner genes within each chr, 2) transcript portions (greyed-out) of each partner gene whose fusion is represented by a connecting red line (middle), and 3) predicted chimeric protein (bottom) with the protein domain annotations. A.EGFR-SEPT14 gene fusion (sample 123). B.EGFR-VOPP1 gene fusion (sample 15). c.EGFR-VSTM2A gene fusion (sample 51). Chr: chromosome; Rec_L_domain: Receptor L domain; TMhelix: Transmembrane Helix; I-set V-Set: Immunoglobulin V-set domain.