Recurrent internal tandem duplications of BCOR in clear cell sarcoma of the kidney

Abstract

The X-linked BCL-6 co-repressor (BCOR) gene encodes a key constituent of a variant polycomb repressive complex (PRC) that is mutated or translocated in human cancers. Here we report on the identification of somatic internal tandem duplications (ITDs) clustering in the C terminus of BCOR in 23 of 27 (85%) pediatric clear cell sarcomas of the kidney (CCSK) from two independent cohorts. We profile CCSK tumours using a combination of whole-exome, transcriptome and targeted sequencing. Identical ITD mutations are found in primary and relapsed tumour pairs but not in adjacent normal kidney or blood. Mutant BCOR transcripts and proteins are markedly upregulated in ITD-positive tumours. Transcriptome analysis of ITD-positive CCSKs reveals enrichment for PRC2-regulated genes and similarity to undifferentiated sarcomas harbouring BCOR-CCNB3 fusions. The discovery of recurrent BCOR ITDs defines a major oncogenic event in this childhood sarcoma with significant implications for diagnostic and therapeutic approaches to this tumour.

Figure 1. Recurrent somatic ITDs in the BCOR gene in CCSKs.

(a) View of aligned whole-transcriptome sequencing reads from a single ITD-positive CCSK (347T) demonstrating a marked focal increase in read coverage corresponding to the ITD in exon 15 of BCOR on Xp11.4. Only unpaired reads of discordant mate pairs were used for local realignment. A representation of the ITD within the BCOR gene is shown beneath. The parental segment (P) that is duplicated is depicted in green and the tandem duplicated segment (ITD) is shown in red. (b) Targeted PCR and gel electrophoresis of BCOR exon 15 in samples from four representative male and two female subjects showed the expected wild-type products (288 bp) in the peripheral blood (C) and adjacent normal kidney (N) tissues, and larger products corresponding to the ITDs (87–114 bp) in the primary CCSK tumours (T) and relapsed metastatic tumours (R). Nearly undetectable levels of the wild-type products were observed in tumour samples from males; in females, both ITD-positive and wild-type products were evident. (c) Sanger sequence trace from case 347T showing the immediate sequence context surrounding the proximal genomic breakpoint in BCOR exon 15 (hg19 coordinates, negative strand). The wild-type genomic sequence around the breakpoint including the termination codon and 3′-UTR are shown above and the parental and duplicated segments of the ITD are below. The proximal breakpoint at the second base of the stop codon (TGA) alters it to a TTA (leucine). (d) Schematic of predicted BCOR protein sequences from ITD-positive CCSKs demonstrating the clustering of all ITDs within the C-terminal PUFD domain. ITD types I–V were numbered based on genomic breakpoints and ITD sequence (Table 1). The BCOR wild-type protein sequence (amino acids (aa) 1,701–1,755) is shown on top with the predicted protein sequence of each ITD-positive case below. Parental segments that have been duplicated are shown in green and the ITDs in red. Novel junctional amino acids (bold black font, underlined) were introduced by the ITDs in cases 385T and 504T. A stretch of 14 residues (aa 1,724–1,737) is common to every ITD type. ANK, ankyrin repeats; BBD, BCL6-binding domain.

Figure 2. BCOR expression in CCSKs.

(a) Targeted RT–PCR of a segment of the BCOR transcript (exons 14 and 15) in CCSKs from female patients demonstrated expression of both the wild-type product (491 bp) and a larger product corresponding to the mutant (ITD) allele, confirming expression of the ITD from the active X chromosome. (b) Box-and-whisker plot of estimated BCOR transcript abundance from RNA-seq data in ITD-positive CCSKs (ITD+) demonstrating high expression of BCOR in comparison to an ITD-negative CCSK (ITD−), Wilms tumours (Wilms) and assorted soft-tissue sarcomas (STS). UDS with BCOR–CCNB3 fusions also had upregulated BCOR expression. Bar representing 25th–75th percentile, line representing the maximum and minimum values. (c) Immunoblot using an antibody to full-length BCOR protein demonstrated a 192-kDa product corresponding to the predicted size of BCOR in ITD-positive CCSK tumours (T) but not in matched normal kidney samples (N). ACTB, beta-actin. (d) Haematoxylin-and-eosin-stained section of CCSK showing classic histologic pattern. Immunohistochemistry with BCOR antibody demonstrated strong nuclear staining in the tumour cells in (e) CCSKs but not in (f) Wilms tumours.

Figure 3. Transcriptome profiling of renal tumours and soft-tissue sarcomas.

(a) Unsupervised hierarchical clustering revealed ITD-positive CCSKs (n=6) to cluster separately from Wilms tumours (n=11), the lone ITD-negative CCSK (case 381) and other soft-tissue sarcomas (STS; n=31). The transcriptomes of BCOR–CCNB3 fusion-positive UDS (n=4) cluster together and are closely related to those of the ITD-positive CCSKs. The heatmap reflects the variance in gene expression relative to the mean across all samples (n=53) with upregulated genes in yellow and downregulated genes in blue. (b) GSEA of CCSKs revealed significant upregulation of PRC2 targets. Enrichment score computed by the GSEA algorithm (y axis) is plotted against the PRC2 target genes (x axis) rank ordered by the degree of differential expression in ITD+ CCSKs compared with Wilms tumours. High-ranking PRC2 target genes are upregulated in CCSKs relative to Wilms tumours (c) Heatmap showing expression values of PRC2 targets in individual ITD-positive CCSK cases as a deviation from the mean in Wilms tumours in units of s.d. The PRC2 target gene set (BENPORATH_PRC2_TARGETS) includes genes with promoter regions bound to H3K27me3, SUZ12 and EED in human H9 ES cells. p, nominal p-value; FDR, false discovery rate; FWER, family-wise error rate.