Transcriptome analysis of desmoplastic small round cell tumors identifies actionable therapeutic targets: a report from the Children's Oncology Group

Abstract

To further understand the molecular pathogenesis of desmoplastic small round cell tumor (DSRCT), a fatal malignancy occurring primarily in adolescent/young adult males, we used next-generation RNA sequencing to investigate the gene expression profiles intrinsic to this disease. RNA from DSRCT specimens obtained from the Children's Oncology Group was sequenced using the Illumina HiSeq 2000 system and subjected to bioinformatic analyses. Validation and functional studies included WT1 ChIP-seq, EWS-WT1 knockdown using JN-DSRCT-1 cells and immunohistochemistry. A panel of immune signature genes was also evaluated to identify possible immune therapeutic targets. Twelve of 14 tumor samples demonstrated presence of the diagnostic EWSR1-WT1 translocation and these 12 samples were used for the remainder of the analysis. RNA sequencing confirmed the lack of full-length WT1 in all fusion positive samples as well as the JN-DSRCT-1 cell line. ChIP-seq for WT1 showed significant overlap with genes found to be highly expressed, including IGF2 and FGFR4, which were both highly expressed and targets of the EWS-WT1 fusion protein. In addition, we identified CD200 and CD276 as potentially targetable immune checkpoints whose expression is independent of the EWS-WT1 fusion gene in cultured DSCRT cells. In conclusion, we identified IGF2, FGFR4, CD200, and CD276 as potential therapeutic targets with clinical relevance for patients with DSRCT.

Figure 1

Molecular validation of the EWSR1-WT1 fusion status in the study samples. (A) PCR amplification of the EWSR1-WT1 fusion gene in DSRCT samples using primers that flank the presumptive fusion junction. The JN-DSRCT-1 cell line serves as the positive control for the EWSR1-WT1 amplicon, HELA cells are the negative control. (B) RNA-sequencing reveals splice-site alterations in DSRCT samples expressing alternative EWSR1-WT1 fusion transcripts. Altered splice sites (red boxes) flanking the intron7/exon8 boundary (black dotted line).

Figure 2

Clustering analysis of DSRCT specimens. (A) Consensus clustering and Spearman rank correlation analysis of the DSRCT specimens after ssGSEA analysis demonstrating that DSRCT specimens segregate into two groups. Color bar indicates correlation strength (blue = weaker, red = stronger). (B) Unsupervised hierarchical clustering using average linkage and Spearman correlation illustrating the 20 most differentially expressed gene set enrichment terms between the two subgroups of DSRCT patient specimens. Color bar indicates row Z-score. (C, D) Principal component analysis of gene expression data from alveolar rhabdomyosarcoma (ARMS; red), alveolar softpart sarcoma (ASPS; green), DSRCT (blue), Ewing sarcoma (ES; purple), and synovial sarcoma (SS; gold) displayed as (C) 3-dimensional and (D) 2-dimensional projections.

Figure 3

EWS-WT1 ChIP-seq using the JN-DSRCT-1 cell line. (A) 2036 Peaks were found to be significantly enriched and a majority of the peaks were found within the intergenic and intronic regions of the genome. (B) 1,284 peaks were associated with a protein-coding gene including ROCK1, PEX5, CTCFL, FGFR4, IGF2 and TSPAN7. (C) Co-occupancy analysis of EWS-WT1 with RNA polymerase. (D) Analysis of the motifs associated with EWS-WT1 peaks showed enrichment for known WT1 binding sequence. (E) Pathway analysis demonstrated enrichment in several potentially targetable pathways.

Figure 4

IGF2 is a highly expressed direct target of EWS-WT1 in DSRCT cells. (A) IGF2 gene expression in patient samples (gray). (B) Characterization of IGF2 expression in patient samples (gray) and the JN-DSRCT-1 cell line (red) using ranked gene list of all genes with TPM (log2) values > 2. (C) Comparative analysis of IGF2 expression across multiple fusion-positive pediatric sarcomas using gene expression array data. ASPS (alveolar soft part sarcoma), EWS (ewing sarcoma), DSRCT (desmoplastic small round cell tumor), ARMS (alveolar rhabdomyosarcoma), SS (synovial sarcoma). (D) EWS-WT1 (blue), input control (red), ChIP-seq peaks and RNA-seq coverage (black) corresponding to the IGF2 (right) and H19 (left) genomic loci. (E) Western blot of EWS-WT1, GAPDH, and IGF2 (conditioned media) protein expression in JN-DSRCT cells transduced with inducible control shRNA or inducible WT1 shRNA with and without 48-h doxycycline treatment. Independent biological triplicates of the western blotting experiment were performed and a single representative example is shown.

Figure 5

EWS-WT1 regulates FGFR4 expression in DSRCT cells and exhibits variable expression in patient samples. (A) EWS-WT1 (blue), input control (red), ChIP-seq peaks and RNA-seq coverage (black) corresponding to the FGFR4 genomic locus. (B) Western blot of FGFR4, EWS-WT1, and GAPDH protein levels in JN-DSRCT cells transduced with inducible control shRNA or inducible WT1 shRNA with and without 48-h doxycycline treatment. Independent biological triplicates of the western blotting experiment were performed and a single representative example is shown. (C) FGFR4 gene expression in patient samples (gray) and the JN-DSRCT-1 cell line (red). (D) Characterization of FGFR4 expression in patient samples (gray) and the JN-DSRCT-1 cell line (red) using ranked gene list of all genes with TPM (log2) values > 2. (E) Comparative analysis of FGFR4 expression across multiple fusion-positive pediatric sarcomas using gene expression array data. (F) Low level (top panel) and high level (bottom panel) expression of FGFR4 in DSRCT patient tissues via immunohistochemistry (representative images); each panel represents a different patient.

Figure 6

DSRCTs express CD276/B7H3 and CD200 in a EWS-WT1 independent manner. (A) CD200 and CD276/B7H3 gene expression in patient samples (gray) and the JN-DSRCT-1 cell line (red). (B) Western blot of CD200, CD276/B7H3, EWS-WT1, and GAPDH protein levels in JN-DSRCT cells transduced with inducible control shRNA or inducible WT1 shRNA with and without 48-h doxycycline treatment. Independent biological triplicates of the western blotting experiment were performed and a single representative example is shown. (C–H) Immunohistochemical analysis of CD276/B7H3 expression in DSRCT patient samples (representative images); each panel represents a different patient.