Figures and data

Multiomic analyses reveal distinct signatures between non-EwS and EwS cells.
(A) UMAP embedding of single cell RNA sequencing and ATAC sequencing of EwS and non EwS cells. (B) ChromVAR deviation scores of EWS::FLI motif enrichment across EwS and non-EwS cells (C) ChromVAR deviation scores of EWS::FLI binding site enrichment, separated by regions with or without GGAA microsatellites (µSAT; GGAA repeat ≥ 4), grouped by cell line. (D) Fraction of cells in each cell line with detection of EWS::FLI fusion transcript. (E) Differentially expressed genes (FDR ≤ 0.01 & Log2FC ≥ 1.25) across EwS/non-EwS landscape. (F) Differentially accessible peaks (FDR ≤ 0.01 & Log2FC ≥ 1.25) of EwS and non EwS cells with row dendrogram showing phenotypic clusters. Row dendrogram shows EwS cell clusters highlighted in bold and red.

EwS cells utilize distinct gene regulatory modules.
(A) Heatmap of unsupervised kmean clusters of peak-to-gene links (PGLs) across 9 EwS cell lines. Left-peaks, right-genes, top-colors denote cell lines. Right of heatmap-UMAP of cell lines labeled by subgroup. Group 1 labelled with navy/dashed lines, group 2 in orange and group 3 in light blue. (B) Violin plot of modules 1-3 gene expression and peak accessibility across EwS cell lines. (C) Universal enrichment analysis of module gene (top 300 filtered on peak-to-gene correlation) sets using both MSigDB Hallmark pathways and Gene Ontologies: Biological Process. The top 7 terms filtered by lowest p-value (those greater than 0.05 are not plotted) and count are shown. More significant terms are darker blue. The number of genes in each module that overlap with a gene set term is indicated by the count metric—represented as dot size. (D) HOMER de novo motif enrichment for each module peak set. Results depict the top significantly enriched motifs in each module peak set, and enrichments in other modules. (E) TFs with high expression and motif enrichment in entire single-cell data set of EwS cell lines. Statistically significant TFs correlation to gene expression > 0.5 and a maximum inter EwS subgroup difference in motif deviation z-score (denoted max TF motif delta) > 1.9 highlighted in red. (F) ChromVAR motif deviation scores and gene expression of highly enriched TFs, parsed by EwS subgroup. (G) Western blot of selected TFs across EwS cell lines. GAPDH = loading control.

EwS gene regulatory modules are differentially regulated by EWS::FLI1.
(A) HOMER enrichment of EWS/ETS motifs (EWS::FLI, pink; FLI::ETS, gray; EWS::ERG, black) in module peak sets. (B) HOMER enrichment of EWS::FLI binding sites, separated by regions with (orange) or without (dark purple) GGAA microsatellites. (C) Universal enrichment analysis of MSigDB EWS::FLI1 and Ewing sarcoma terms in module gene sets. (D) UMAP embeddings of shNS and shFLI1 samples. EwS subgroup identity of cell lines represented by dashed boxes. Group 1 (A673) labelled with navy/dashed lines, group 2 (CHLA10, PDX305) in orange and group 3 (TC71) in light blue. (E) Expression of EwS target gene expression across shNS vs. shFLI1 samples. (F-H) Module gene set expression across knockdown samples.

CHLA10 is a microcosm of EWS::FLI1 mediated heterogeneity.
(A) UMAP of CHLA10 colored by cluster, C22 dark purple, C23 light purple. (B) UMAP of module peak accessibility and gene set expression in CHLA10 cells. (C) Differential motif enrichment between C22 and C23, TFs with -log10(p-value) > 250 highlighted. (D) Universal enrichment of GO: Biological Process terms in differentially expressed genes between C22 and C23, terms associated with TGF-β in bold. (E) Universal enrichment of MSigDB Ewing Sarcoma terms in differentially expressed genes between C22 and C23. (F) Pathway activity scores across C22 and C23, TGF-β highlighted in bold.

Sub-populations of EwS cell lines differentially respond to TGF-β stimulus.
(A-C) UMAP of A673, CHLA10, and TC71 cells colored by condition (vehicle control or TGF-β1 treatment) – column 1; and cluster – column 2; Fraction of treated versus control cells in each cluster with highest and lowest indicated by arrow - column 3; Expression of fusion regulated module 2 gene set, highest and lowest TGF-β clusters highlighted with dashed lines – column 4. (D-F) Correlation between TF expression and chromVAR TF deviation scores of EwS TFs across clusters, highest and lowest TGF-β clusters highlighted by arrows. (G) Western blot of active (pS465/467) and total SMAD2 and FOSL2 protein in TC71 and CHL10 cells after short (30 minute) and long term (24 hr) TGFβ1 treatment durations. (H) H&E staining of 3 CHLA10 xenografts in NSG mice generated by subcutaneous or tail vein injection (upper left). Those regions of interest (ROIs) subjected to whole human transcriptome digital spatial profiling are highlighted. Heatmap of module gene expression across ROIs (bottom left). (I) Scatter plot of average expression of Gene Ontology: Biological Process (GO:BP) TGF-β response genes and extracellular matrix (ECM) organization genes compared to FOSL2 expression in ROIs. Pearson R and p-value highlighted (top left).

Human tumor data show usage of EWS::FLI1 gene regulatory modules.
(A) UMAP embedding of five EwS PDX samples (left) UMAP of fusion regulated module gene sets in PDX samples (right) (B) UMAP embedding of single nuclei sequencing of 6 patient tumor samples (left). Patient tumor scRNA expression of fusion regulated modules (right). (C) Schematic depicting deconvolution method (D) Heatmap EwS subgroup signal in bulk EwS patient tumor RNA-seq. The intercept row represents remaining signal (bottom). Patient tumor samples annotated by GEO BioProject. (E) Scatter plot of bulk RNA patient samples by group 2 signature and GO: BP TGF-β response genes and colored log10 FOSL2 expression. Pearson R and p-value highlighted (top left). (F) Scatter plot of microarray patient samples by module 2 expression and GO: BP TGF-β response genes and colored FOSL2 expression. Pearson R and p-value highlighted (top left). (G) Survival curves of patient microarray data51 (n = 46), high expression of module 2 gene set in red (top 15% expressing samples, n = 7), low expression in blue (bottom 85%, n = 39) (H) Survival curves of high expression of FOSL2 samples (top 15%, n = 7, bottom 50%, n=39).