Somatic gene delivery faithfully recapitulates a molecular spectrum of high-risk sarcomas

Abstract

A major challenge hampering therapeutic advancements for high-risk sarcoma patients is the broad spectrum of molecularly distinct sarcoma types and the corresponding lack of suitable model systems. Here we describe the development of a genetically-controlled, yet versatile mouse modeling platform allowing delivery of different genetic lesions by muscle electroporation (EPO) in wildtype mice. This EPO-GEMM (EPO-based genetically engineered mouse model) platform allows the generation of ten genetically distinct sarcomas on an isogenic background, including the first model of ETV6::NTRK3-driven sarcoma. Comprehensive histological and molecular profiling reveals that this mouse sarcoma cohort recapitulates a spectrum of molecularly diverse sarcomas with gene fusions acting as major determinants of sarcoma biology. Integrative cross-species analyses show faithful recapitulation of human sarcoma subtypes, including expression of relevant immunotherapy targets. Comparison of syngeneic allografting methods enables reliable preservation and scalability of sarcoma-EPO-GEMMs for preclinical treatment trials, such as NTRK inhibitor therapy in an immunocompetent background.

Fig. 1. Modeling diverse subtypes of fusion-driven sarcoma.

a Scheme of versatile vector combination for in situ sarcoma induction. Reporter genes were not utilized here. Created in BioRender. Banito, A. (2025) https://BioRender.com/kpnm994. b Matrix of vector combinations chosen based on human sarcoma profiling studies. c Kaplan–Meier curves of murine PAX3::FOXO1-driven tumor induction and representative histographs compared to human aRMS (d). e Kaplan–Meier curves of murine SS18::SSX1/2 tumor induction and representative histographs compared to human SS (f). g Kaplan–Meier curves of murine ASPSCR1::TFE3-driven tumor induction and representative histographs compared to human ASPS (h). i Kaplan–Meier curves of murine ETV6::NTRK3-driven tumor induction and representative histographs compared to human IFS and the emerging WHO entity of NTRK-rearranged spindle cell neoplasm (NRSCN) (j). Murine tumors exhibited two main histological types, named ETV6::NTRK3-Monomorphic and ETV6::NTRK3-Pleomorphic. If not specified otherwise, histographs were stained by H&E. All Kaplan–Meier curves correspond to n ≥ 6 mice per group electroporated bilaterally. P values for comparing Kaplan–Meier curves were determined by two-sided log-rank tests and corrected for multiple testing with the Benjamini-Hochberg method. Scale bars = 50 µm. Source data are provided as a Source data file.

Fig. 2. Modeling sarcomas driven by oncogenic RAS and gene inactivation.

a Kaplan–Meier curves of RAS-driven mouse sarcomas and representative H&E histographs compared to human eRMS (b). c Kaplan–Meier curves of mouse sarcomas driven by Trp53-inactivation and secondary mutations and representative histographs compared to human MPNST, MRT and UPS (d). All Kaplan–Meier curves correspond to n ≥ 6 mice per group electroporated bilaterally. P values for comparing Kaplan–Meier curves were determined by two-sided log-rank tests and corrected for multiple testing with the Benjamini-Hochberg method. e Integrated score for jH2AX IHC stainings, depicted as boxplots ordered by median from low to high. Boxplots display individual values, median, and interquartile range (IQR). Whiskers extend to the most extreme data points within 1.5 times the IQR from the lower and upper quartiles. P values were determined by unpaired two-sided Wilcoxon tests and corrected for multiple testing by the Bonferroni-Holm method. n ≥ 3 tumors per group. f CNV profiles derived from DNA methylation data, condensed as Percent altered (≤/≥. 1) and Genomic Index. n ≥ 3 tumors per group. Scale bars = 50 µm. Source data are provided as a Source data file.

Fig. 3. Fusion gene and TP53 mutation status determine sarcoma biology and microenvironment.

a Heatmap view of mouse sarcomas, quantified for six morphological and based on H&E staining and integrated expression scores for 10 antigens determined by immunohistochemistry. Asymmetric color scale for combined visualization of low (CD45, CD3, cCas3) and high-scoring antigens. Quantification by blinded expert pathology review. b Pearson correlation matrix based on data from a. Insignificant p values (≥0.05) are indicated. Blue indicates positive, red indicates negative correlation, and point size reflects effect size. c, d IHC scores of Ki-67 and CD45 from a visualized as boxplots ordered by median from low to high. Boxplots display individual values, median, and interquartile range (IQR). Whiskers extend to the most extreme data points within 1.5 times the IQR from the lower and upper quartiles. P values were determined by unpaired two-sided Wilcoxon tests and corrected for multiple testing by Bonferroni–Holm method. n ≥ 3 tumors per group. Source data are provided as a Source data file.

Fig. 4. Mouse sarcomas exhibit distinct genotype-dependent molecular signatures.

a tSNE clustering based on the top 10,000 differentially methylated CpG sites. b Correlation clustering based on the top 2000 differentially expressed genes. c k-means clustering based on the top 2000 differentially expressed genes. The top five most significantly enriched GO terms in the category GOTERM_BP_DIRECT (Biological processes) are shown alongside examples of genes in each k-means group. Source data are provided as a Source data file.

Fig. 5. Murine sarcomas resemble a spectrum of human sarcomas.

a Analysis scheme of cross-species sarcoma analysis based on RNA sequencing. b tSNE visualization based on cross-species transcriptome analysis of mouse (n = 63) and human (n = 299) sarcoma specimens. SS synovial sarcoma, aRMS alveolar sarcoma, IFS infantile fbrosarcoma, ASPS alveolar soft part sarcoma, RMS rhabdomyosarcoma, MPNST malignant peripheral nerve sheath tumor, MRT malignant rhabdoid tumor, UPS Undifferentiated pleomorphic sarcoma, MFS myofibroblastic tumor, LMS leiomyosarcoma, LPS liposarcoma, GIST gastrointestinal stromal tumor. Source data are provided as a Source data file.

Fig. 6. Syngeneic allograft models enable scalability for in vivo testing.

a Schematics of syngeneic allograft modeling (SAM) procedure, systematically comparing four different allograft types. Created in BioRender. Banito, A. (2025) https://BioRender.com/abys0b6. b Tumor-free survival of SAMs compared to corresponding GEMMs. n ≥ 8 tumors per condition, including tumors with different genotypes. c tSNE analysis based on the top 10,000 differentially methylated CpG sites of SAMs, mouse-tumor-derived cell lines and corresponding GEMMs. d Median-ordered boxplots of Pearson correlation coefficients based on DNA methylation data from GEMMs and corresponding SAMs. n = 5 tumors per condition with 5 different genotypes. e Representative H&E histographs of GEMMs and corresponding SAMs after orthotopic engraftment of 2D-cultured mouse tumor lines. Scale bars equal 100 µm. f NTRKi in vitro sensitivity testing. Data points correspond to asymmetric drug sensitivity scores (DSS, as explained in the methods). DDS = 0 indicates resistance, and values > 10 indicate sensitivity. n ≥ 2 cell lines grouped by entity, in two independent experiments (triangles and squares). For the IMT condition, only one cell line was available. IMT refers to a human tumor cell line derived from an Inflammatory myofibroblastic tumor driven by ETV6::NTRK3 used to clone the mouse transposon vectors for electroporation. Three eRMS (KRAS^G12V/sgTrp53) tumor lines were used as fusion negative controls for comparison. Boxplots in b, d, and f display individual values, median, interquartile range (IQR). Whiskers extend to the most extreme data points within 1.5 times the IQR from the lower and upper quartiles. P values refer to unpaired two-sided t-tests, corrected for multiple testing using the Bonferroni-Holm method. g In vivo treatment of FD-IFS tumor line re-engrafted orthotopically into wildtype C57BL6/J host mice. Treatment started when mice developed palpable tumors (days 8–10 post engraftment) by oral gavage of NTRK inhibitors or vehicle twice a day. Mean, standard deviation and individual values are depicted. n = 8 mice per group. Source data are provided as a Source data file.