Article Figures & Data
Figures
- Figure 1.
Mutational landscape of Ewing sarcoma tumors. A, sample data, sequencing method, and recurrent aberrancies detected by massively parallel sequencing are indicated for 92 Ewing sarcoma tumors and 11 Ewing sarcoma cell lines. Box colors for each panel are indicated in the key. SNV, single-nucleotide variant. B, arm-level SCNAs from 92 Ewing sarcoma tumors determined by segmentation analysis of WES. C, focal SCNAs for 18 Ewing sarcoma tumors determined from SNP-array analysis. B and C, plotted are the q values for copy-number losses (blue) and gains (red) for each chromosomal arm or focal region calculated with GISTIC2.0 (33). Gains of chromosome 8 were the most significant arm-level SCNA, and CDKN2A is located at the most significant focal loss. RNASeq data demonstrated that ABCC12 was likely not expressed in any sample regardless of copy-number status. Focal deletions at 11q24.3 and 22q12.2 involve the FLI and EWS genes in the EWS–FLI rearranged samples. D, the locations of variants in STAG2 and TP53 indicated in linear protein domain models. Domain annotations are taken from recent publications (15, 73). Numbers below each protein name indicate the number of amino acids in the full-length protein. Variant details are listed in Supplementary Table S11. Solid boxes, mutations identified in tumors with normal pairs (including multiple tumors from the same patient); open triangles, variants from tumors without normal pairs after germline filtering; and open circles, variants from cell lines after filtering. Mutations: MS, missense; FS, frameshift; NS, nonsense; SS, splice site. Protein domains: STAG, STAG superfamily; SCD, stromalin conservative domain; TADI/II, transcriptional-activation I and II; DNA, DNA binding; OD, tetramerization; CTD, carboxyl terminus.
- Figure 2.
Loss of STAG2 expression in Ewing sarcoma. A, IHC staining for STAG2 in Ewing sarcoma cell lines and tumors demonstrates a loss of STAG2 expression in mutated samples (red asterisk). The A673 Ewing cell line, expressing wild-type STAG2, stained positive for STAG2 expression, whereas TC32, mutated at STAG2, does not express STAG2 protein. Also shown are two primary Ewing sarcoma tumor samples stained for STAG2 expression. The tumor with a STAG2 mutation does not stain for STAG2 expression, whereas the tumor with wild-type STAG2 stains positive for STAG2 expression. B, Western immunoblot analysis of STAG2 protein expression in Ewing sarcoma cell lines with mutated (black box), rearranged (green box), and wild-type (white box) STAG2. Mutations and rearrangements of STAG2 result in loss of expression of protein, but some cell lines with wild-type STAG2 also have loss of protein expression. C, top, schematic illustrating novel STAG2–MAP7D3 fusion transcripts detected by RNASeq in the Ewing sarcoma cell line TTC466. Bottom, RT-PCR detected the presence of both STAG2–MAP7D3 fusion forms in TTC466 but not in CADO-ES1 (negative control) Ewing sarcoma cell lines. D, scatter plot of tumor ploidy for tumors staining positive for STAG2 versus STAG2 loss (mean: 2.12 vs. 2.02; P = 0.585 by Mann–Whitney test). E, scatter plot of the fraction of the genome affected by SCNAs in all Ewing tumors staining positive for STAG2 versus STAG2 loss by IHC (mean: 0.15 vs. 0.36, respectively; P = 0.009 by Mann–Whitney test). F, the fraction of genome affected by SCNAs in TP53 wild-type diagnostic tumors staining positive for STAG2 versus STAG2 loss (mean: 0.14 vs. 0.27; P = 0.108 by Mann–Whitney test). G, 88% of patients with STAG2 loss by IHC at diagnosis presented with metastatic disease compared with 27% of patients whose diagnostic tumors expressed STAG2 (P = 0.002 by Fisher exact test). H, heatmap of ssGSEA enrichment z scores for the significant gene sets (SNR ≥ 1.5 and FDR q value <0.05) positively correlated with STAG2 loss (top) or STAG2 expression (bottom). Geneset names are listed in Supplementary Table S13. I and J, GSEA plots of running enrichment scores for two metastatic signatures (j and k) significantly enriched in STAG2 loss samples. **, P < 0.01.
- Figure 3.
Wild-type ETS genes are not expressed in Ewing sarcoma. A, schematic depicts the generation of gene-level and pre/postfusion RPKMs for EWS and ETS genes (FLI and ERG) from Ewing sarcoma tumor samples. Gene-level RPKMs for ETS genes (green) are derived from all exons for that gene. Reads generated from the wild-type allele or translocated allele cannot be distinguished. Prefusion EWS RPKMs (blue) are generated from all reads that map to the exons before the site of the EWS–ETS breakpoint (reads from rearranged and wild-type alleles cannot be distinguished). Postfusion EWS RPKMs (top, black) map to exons after the breakpoint. Prefusion ETS RPKMs (bottom, black) are derived from alleles before the EWS–ETS breakpoint. Postfusion ETS RPKMs (red) are derived from exons after the breakpoint (reads from the rearranged and wild-type alleles cannot be distinguished). B, left, gene-level RPKMs for FLI are plotted for each tumor with available RNASeq data. Tumors are grouped by whether they express a EWS–FLI or EWS–ERG fusion. Right, gene-level RPKMs for ERG are plotted for each tumor and grouped by fusion type. C, left, prefusion and postfusion EWS RPKMs for all tumors with RNASeq data. Because postfusion exons are not involved in the EWS–ETS rearrangement, expression of these exons indicates the expression of the wild-type EWS allele. Middle, prefusion and postfusion FLI RPKMs for EWS–FLI rearranged tumors. Right, prefusion and postfusion ERG RPKMs for EWS–ERG rearranged tumors. Expression of prefusion ETS exons depends on the wild-type promoter. Thus, prefusion RPKMs reflect expression of wild-type ETS alleles.
- Figure 4.
Ewing sarcoma tumors acquire somatic aberrancies with treatment. A, coding mutations in 24 Ewing sarcoma tumor samples identified from WES and WGS of patient tumors and paired normal tissue with available clinical data. Bars indicate the number of coding-region mutations in each tumor and are grouped by disease state (diagnostic or treated) and ordered by number of mutations (high to low) within each group. B, the rate of coding mutations is higher in tumors sequenced after treatment than in tumors from diagnostic samples. Plotted are the average mutation rates for diagnostic and treated samples ± the standard deviation (SD), P = 0.0017 by Mann–Whitney test. C, treated tumors have a lower fraction of *CpG to T transitions and a higher fraction of substitutions at cytosines in TpC dinucleotides. The NpN nomenclature indicates the dinucleotide sequence involved in the indicated mutation and * indicates the mutated base. Plotted are the fractions of SNVs for each mutational category ± the 95% CI; *, P < 0.05; **, P < 0.01 by z test. D, shown is the rate of SNVs by 3-base context from tumors acquired at diagnosis (left) or after treatment (right). Colors indicate the single-nucleotide change, and the location of each bar in the matrix indicates the flanking bases. E and F, circos plots of seven tumors sequenced by WGS. Chromosomes displaying cytobands are arranged end-to-end in the outer ring with the inner ring showing segmented copy-number variations generated from exome sequencing. Interchromosomal rearrangements are displayed as purple arcs and intrachromosomal as green arcs. Rearrangements at t(21;22)(q22;q12) in SJDES010 and SJDES018-R indicate the EWS–ERG fusions in these samples. All other tumors demonstrate a translocation at t(11;22)(q24;q12), consistent with the EWS–FLI fusion.
- Figure 5.
Characteristics of Ewing sarcoma cell lines. A, frequency of molecular aberrancies and gender in Ewing sarcoma cell lines (red) compared with tumor samples (black). STAG2 loss refers to absence of protein expression by Western blot analysis or IHC. There was a statistically significant increase in the percentage of cell lines with STAG2 mutation (P = 0.016), STAG2 loss (P = 0.0006), TP53 mutations (P < 0.0001), CDKN2A loss (P = 0.028), and chromosome 1q gain (P = 0.011) compared with tumor samples by Fisher exact test. *, P < 0.05; ***, P < 0.001; ****, P < 0.0001. B, projection on the first two components of a principal component analysis applied to RNASeq data from Ewing sarcoma tumors and cell lines. Principal components are ordered on the basis of the percentage of explained variance in the data. Black dots indicate tumor samples and red indicate cell lines. Cell lines and tumors separate into distinct clusters. C, plot of the normalized P value (by −log10) of Student t tests comparing the principal component scores for tumors versus cell lines. The red line indicates a P = 0.05. Only PC1 has a significant P value. D, plot of FDR q value (normalized by −log10) versus normalized enrichment score (NES) of ssGSEA applied to genes (ranked by PC1-weight) to identify Kyoto Encyclopedia of Genes and Genomes (KEGG) canonical pathways enriched by differences in Ewing sarcoma tumors versus cell lines. A negative NES indicates a pathway enriched in tumors and a positive NES indicates a pathway enriched in cell lines. Black dots indicate significant pathways (FDR q value < 0.05) associated with environmental interactions with the extracellular membrane, transmembrane signaling, and the immune system. Red dots indicate significant pathways associated with cellular proliferation including cell division, growth, and metabolism. Significantly enriched signatures are listed in Supplementary Table S17.
- Figure 6.
Clonal evolution in Ewing sarcoma. A and B, Venn diagrams indicate the number of SNVs and indels detected in paired tumor samples by WES for mutations occurring at bases with adequate coverage to power mutation detection in both tumors. Purple circles include mutations identified in the tumor sequenced from the 26 tumor–normal cohort, and the blue circles include mutations identified in the subsequent relapse. “Cloud plots” indicate the cancer cell fraction (CCF) of these mutations by a central dot surrounded by a cloud. Light dots indicate single mutations. Black dots indicate multiple mutations with the same CCFs in both tumors. Clouds depict the 90% CI of each CCF estimate (smaller clouds represent higher certainty). A, left and middle, cloud plots demonstrate the overlap of mutations identified in 2 patients from whom we sequenced a diagnostic tumor and relapsed metastatic tumor. Right top, MRI and CT images of primary pelvic tumor at diagnosis and lung metastasis at relapse for patient SJDES004. Right bottom, a model depicting the proposed evolution of a metastatic relapsed tumor from a primary tumor. Blue circles represent Ewing sarcoma cells. Symbols within cells represent clusters of mutations (circles) or specific mutations. Time moves from left to right with tumor sampling indicated by vertical lines. The large teardrop indicates the evolving relapsed cell population. Smaller, multicolored teardrops indicate subclonal populations of cells. The absence of some mutations at relapse that are estimated to be clonal at diagnosis indicates that metastatic (met) spread likely occurs after tumor initiation but before the emergence of the dominant clone in the primary tumor. B, left and middle, cloud plots demonstrate mutations identified in two separate locally relapsed tumors from patients with Ewing sarcoma. Right top, MRI and PET images (left to right) from diagnosis, relapse 1 (R1), and relapse 2 (R2) from patient SJDES007 with Ewing sarcoma involving the scapula. Right bottom, schematic depicts the proposed mechanism of evolution from diagnosis to each relapse for tumors from patient SJDES007. At diagnosis, tumors likely contain a mutation in STAG2 (extrapolated from IHC demonstrating loss of STAG2 expression) that cannot be 100% clonal due to the presence of different STAG2 mutations at R1 and R2. The presence of clonal mutations in R1 that are not present in R2 (and vice versa) suggest that local relapses emerge from cells that have diverged from the primary tumor.
Tables
- Table 1.
Demographics
Demographic Percentage (number/total) Male 55% (51/92) Mean age (min–max) 11 y (4 mo–21 y) White 96% (25/26) Black 4% (1/26) Axial primary (chest, vertebrate, pelvis) 42% (25/59) Tumors with EWS–FLI (vs. EWS–ERG) 85% (22/26) Survival (patients sampled at diagnosis) 69% (11/16) Survival (patients sampled after treatment failure) 25% (2/8) NOTE: Summary of available clinical data (details are given in Supplementary Table S1).
Supplementary Data
Files in this Data Supplement:
- Supplementary Methods, Figure Legends, Figures S1 - S6, Tables S1, S3 - S6, S8 - S15, S17 - This file contains supplementary methods, supplementary references, supplementary figures S1-S6, supplementary tables S1, S3-S6, S8-S15, and S17, and legends for all supplementary figures and tables. Supplementary Figure S1. Summary of Ewing sarcoma tumor and cell line cohorts. Supplementary Figure S2. RNASeq validation of mutations identified by WES and WGS. Supplementary Figure S3. Power calculations for the detection of significantly mutated genes. Supplementary Figure S4. Copy number variants in Ewing sarcoma tumors. Supplementary Figure S5. ETS1 deletions associated with EWS/FLI rearrangements. Supplementary Figure S6. Copy number variants in diagnostic vs. treated Ewing sarcoma tumors. Supplementary Figure S7. SCNAs in paired Ewing sarcoma tumor samples. Supplementary Table S1. Tissue samples. Supplementary Table S3. Multiply mutated genes. Supplementary Table S4. Rearrangements and fusions detected by WGS and RNASeq. Supplementary Table S5. Mutational significance for tumor/normal pairs. Supplementary Table S6. Rate of coding mutations in Ewing sarcoma tumors. Supplementary Table S8. Mutational significance for all tumors. Supplementary Table S9. Gene set enrichment analyses of mutated genes. Supplementary Table S10. Significance of arm-level SCNAs. Supplementary Table S11. Multiply mutated genes of interest in all sample cohorts. Supplementary Table S12. Immunohistochemical staining of STAG2 in Ewing sarcoma tumors. Supplementary Table S13. Gene set enrichment analyses of STAG2 expressing vs. STAG2 loss. Supplementary Table S14. Gene and gene-fragment RPKM Values for EWS and ETS genes. Supplementary Table S15. ETS1 copy-number changes. Supplementary Table S17. Gene set enrichment analyses of tumors vs. cell lines.
- Supplementary Table S2 - Mutations identified from WES and WGS of 26 tumors with paired normals.
- Supplementary Table S7 - Variants identified from WES of 92 tumors after filtering.
- Supplementary Table S16 - Variants identified from WES of 11 cell lines after filtering.
- Supplementary Table S18 - Mutations identified from WES of 4 subsequent relapses.
Volume 4, Issue 11
