YAP1 is a key regulator of EWS::FLI1-dependent malignant transformation upon IGF-1-mediated reprogramming of bone mesenchymal stem cells

Abstract

Ewing sarcoma (EwS) is an aggressive cancer of adolescents in need of effective treatment. Insulin-like growth factor (IGF)-1 is an autocrine growth factor for EwS, but only 10% of patients respond to IGF-1 receptor (IGF-1R) blockade. Although EwS is presumed to originate from mesenchymal progenitors during bone development, targeting of the EwS driver oncogene EWS::FLI1 to the mesenchymal lineage in a mouse model does not result in tumor formation but in skeletal malformations and perinatal death. We report that transient exposure to IGF-1 concentrations mimicking serum levels during puberty reprograms limb-derived mesenchymal cells of EWS::FLI1-mutant mice to stable transformation and tumorigenicity. We identify a modular mechanism of IGF-1-driven tumor promotion in the early steps of EwS pathogenesis, in which Yap1 plays a central role. Pharmacologic Yap1/Tead inhibition reverses the transformed phenotype of EWS::FLI1-expressing cells. Our data provide a rationale for combined IGF-1R and YAP/TEAD inhibition in the treatment of EwS patients.

Keywords: CP: Stem cell research; EWS::FLI1; Ewing sarcoma; IGF-1/insulin signaling; YAP1/TAZ signaling; endochondral bone development; mesenchymal stem cells; puberty.

Graphical abstract

Figure 1

Chromatin accessibility is altered in EF^Prx1 MSCLCs (A) Volcano plot of differentially accessible chromatin regions between EF^Prx1 MSCLCs and WT MSCLCs identified by DEseq2. The x axis indicates the logarithms of the fold changes of individual genes. The y axis indicates the negative logarithm of their p value to base 10 (DESeq2; p_adj < 0.05, |log₂FC| > log₂(1.5); n [EF^Prx1] = 4, n [WT] = 3). See Tables S4 and S5. (B) Distribution of differentially accessible peaks (from A) with respect to the distance from the transcription start site of the nearest gene compared to all regions without significant change (EF^Prx1 MSCLCs ≈ WT MSCLCs). (C) Fast gene set enrichment analysis (fGSEA) of GGAA microsatellite repeats in peaks that are more accessible in EF^Prx1 MSCLCs than in WT MSCLCs. The barcode indicates peaks with at least eight repeats of the GGAA motif with a variable spacer of 0–3 bp. Peaks were preranked by the DESeq2 test statistic for the comparison of EF^Prx1 MSCLCs and WT MSCLCs. (D) Distribution of log₂ fold changes (between EF^Prx1 MSCLCs and WT MSCLCs) in peaks separated by the number of repeats of the GGAA microsatellite within these peaks. The plot indicates the kernel density estimate per fold change value, which is similar to a histogram for continuous values.

Figure 2

IGF-1 reprogramming of EF^Prx1 MSCLC to full malignant transformation (A) Approach assessing in vitro anchorage-independent growth of limb-derived MSCLCs from EF mutant mice as a surrogate of malignant transformation. EF^Prx1 MSCLCs were grown in soft agar in the absence or presence of IGF-1 (500 ng/mL) and insulin (INS; 100 ng/mL) either alone or in combination according to the indicated scheme. Representative examples of soft-agar plates illustrating the appearance of transformed colonies after 4 weeks of incubation are shown. Cells derived from these colonies are referred to as hormone-activated (ha-) EF^Prx1 MSCLCs. Colonies >0.5 mm were counted using ImageJ software. Results are presented as the mean ± SD of triplicate samples (hr-EF^Prx1 MSCLC #2) from representative data of three independent experiments. Statistical significance was determined using one-way ANOVA (^∗∗∗p < 0.0005). Unt, untreated (absence of IGF-1/INS). (B) Same as in (A) but using ha-EF^Prx1 cells as the starting material and culturing cells exclusively in the absence of hormone supplementation. Cells from colonies arising under these conditions are referred to as hormone-reprogrammed (hr-) EF^Prx1 MSCLCs. Violin plots as in (A). n = 3. Statistical significance was determined using one-way ANOVA (^∗∗∗p < 0.0005). (C) Tumor formation upon subcutaneous injection of parental EF^Prx1 #2 and derived hr-EF^Prx1 MSCLCs in SCID mice (n = 4 mice per group). The scheme, representative pictures, and mean tumor size increase over 31 days are shown. ^∗p < 0.1, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. (D) Hematoxylin and eosin stain of a frozen section of a representative tumor arising from the subcutaneous injection of hr-EF^Prx1 MSCLCs in SCID mice. Scale bar: 100 μm.

Figure 3

Transcriptional and chromatin changes associated with IGF-1 activation and reprogramming (A) Volcano plot of differentially expressed genes between EF^Prx1 MSCLCs and ha-EF^Prx1 (left) or hr- EF^Prx1 (right) MSCLCs identified by DEseq2. The x axis indicates the logarithm of the fold changes of individual genes. The y axis indicates the negative logarithm of their p value to base 10 (DESeq2; p_adj < 0.05, |log₂FC| > log₂(2); n [EF^Prx1] = 2, n [ha-EF^Prx1] = 2, and n [hr-EF^Prx1] = 4). See Tables S6 and S7. (B) Venn diagrams showing the overlap of differentially expressed genes between ha-EF^Prx1 and hr-EF^Prx1 MSCLCs for up- and downregulated genes in comparison to EF^Prx1 MSCLCs (from A). (C) Volcano plot of differentially accessible chromatin regions (from ATAC-seq) between hr-EF^Prx1 MSCLCs and EF^Prx1 MSCLCs. Each point represents an ATAC-seq peak, and significantly up- and downregulated regions are highlighted in blue and red, respectively (DESeq2; p_adj < 0.05, |log₂FC| > log₂(1.5); n [hr-EF^Prx1] = 2 and n [EF^Prx1] = 4). See Table S5. (D) Fast gene set enrichment analysis (fGSEA⁴³) of GGAA microsatellite repeats in peaks that are more accessible in hr-EF^Prx1 MSCLCs compared to EF^Prx1 MSCLCs. The barcode indicates peaks with at least eight repeats of the GGAA sequences with a variable spacer of 0–3 bp. Peaks were preranked by the DESeq2 test statistic for the comparison of hr-EF^Prx1 MSCLCs and EF^Prx1 MSCLCs.

Figure 4

Comparative analysis of differentially accessible chromatin modules in MSCLCs (A) Heatmap showing row-scaled normalized read counts for all peaks that were considered differentially accessible in at least one comparison (EF^Prx1 MSCLCs/WT MSCLCs, hr-EF^Prx1 MSCLCs/EF^Prx1 MSCLCs, or MSCLC + IGF-1/WT MSCLCs; n_total = 2,038). Peaks are grouped into five “modules” (M1–M5). GGAA motif density in differentially open regions is indicated to the right of the heatmap. See Table S4. (B) Scatterplot comparing changes in the chromatin accessibility of ATAC-seq peaks (x axis) with the corresponding changes in gene expression of the nearest genes (y axis). One point is indicated for each combination of differentially accessible peak (from A) and differentially expressed gene. Color indicates the mean of both fold changes. (C) The protein levels of YAP1, LAMA5, and β-actin were detected by western blot for WT MSCLCs, EF^Prx1 MSCLCs, and hr-EF^Prx1 MSCLCs (top) and those of Yap1 by immunofluorescence staining (bottom). Scale bars: 50 μm. (D) Bar plots showing DNA sequence motifs (mouse and human TF motifs from JASPAR 2022⁵⁰) overrepresented in peaks belonging to each of the five modules from (A). Each plot chart lists the top three motifs per module and each bar indicates the percentage of peaks with at least one match to the given motif. Enrichment was calculated using Fisher’s exact test (one-tailed). ^∗p_adj < 0.05, ^∗∗p_adj ≤ 0.01, and ^∗∗∗p_adj ≤ 0.005. See Table S9. (E) Left: Western blot analysis of YAP1 levels in hr-EF^Prx1 MSCLCs #2 upon knockout of Yap1 using three CRISPR single-guide (sg)RNAs (sg-Yap1). Middle: Representative soft-agar assay for hr-EF^Prx1 cells transduced with sg-Ctrl versus sg-Yap1. Right: The number of cell colonies was counted on days 21 after plating. Data are presented as the mean ± SE (n = 3), ^∗∗∗p < 0.001. Statistics were calculated by one-tailed, paired Student’s t test. (F) Quantitative analysis by RT-qPCR of relative mRNA expression levels of Yap1, Igf2bp1, Hpf1, Cenpq, and Ccnd1 after knockout of Yap1. Data are presented as the mean ± SE (n = 3), ^∗∗∗p < 0.001 and ^∗∗p < 0.001. Statistics were calculated by one-tailed, paired Student’s t test.

Figure 5

YAP/TEAD blockade reduces colony-forming ability and synergizes with IGF-1R inhibition in hr-EF^Prx1 MSCLCs and human EwS (A) The effects of K-975 on soft-agar colony formation of hr-EF^Prx1 MSCLCs upon 12 days incubation. Soft-agar assay shows a concentration-dependent decrease in the number of hr-EF^Prx1 MSCLC colonies. Colonies >0.5 mm were counted using ImageJ software. Data are expressed as the mean ± SD (n = 3); two-way ANOVA was used to determine statistical significance. ^∗∗p < 0.01 and ^∗p < 0.05. (B) Overview of synergy scores of NVP-AEW541 and K-975 drug combinations across hr-EF^Prx1 MSCLCs #1 and #2 in monolayer culture conditions. (C) Overview of synergy scores of NVP-AEW541 and K-975 drug combinations for STA-ET-1 and SK-N-MC human EwS cell lines in spheroid culture conditions. The heatmaps show Bliss excess across the hr-EF^Prx1 MSCLCs (additive if =0 and non-additive if ≠0; within non-additive cases, it is synergistic if >0 or antagonistic if <0).