EWSR1-WT1 Target Genes and Therapeutic Options Identified in a Novel DSRCT In Vitro Model

Simple Summary Desmoplastic small round cell tumor (DSRCT) is an extremely rare soft tissue sarcoma arising in the abdomen of adolescents and young adults. This sarcoma is driven by a single genomic rearrangement, resulting in the expression of the EWSR1-WT1 fusion gene. No effective treatment exists for DSRCT patients, which highlights the need for preclinical models to study disease progression and drug sensitivity. The aim of this study is to develop a pre-clinical DSRCT in vitro model, which enables investigating the molecular target genes of the EWSR1-WT1 fusion gene and allows for medium-throughput drug screening to discover new therapeutic options. Abstract Desmoplastic small round cell tumor (DSRCT) is a rare and aggressive soft tissue sarcoma with a lack of effective treatment options and a poor prognosis. DSRCT is characterized by a chromosomal translocation, resulting in the EWSR1-WT1 gene fusion. The molecular mechanisms driving DSRCT are poorly understood, and a paucity of preclinical models hampers DSRCT research. Here, we establish a novel primary patient-derived DSRCT in vitro model, recapitulating the original tumor. We find that EWSR1-WT1 expression affects cell shape and cell survival, and we identify downstream target genes of the EWSR1-WT1 fusion. Additionally, this preclinical in vitro model allows for medium-throughput drug screening. We discover sensitivity to several drugs, including compounds targeting RTKs. MERTK, which has been described as a therapeutic target for several malignancies, correlates with EWSR1-WT1 expression. Inhibition of MERTK with the small-molecule inhibitor UNC2025 results in reduced proliferation of DSRCT cells in vitro, suggesting MERTK as a therapeutic target in DSRCT. This study underscores the usefulness of preclinical in vitro models for studying molecular mechanisms and potential therapeutic options.


Introduction
Desmoplastic small round cell tumor (DSRCT) is a highly aggressive and rare soft tissue sarcoma [1][2][3]. This sarcoma was first characterized as a separate entity by Gerald and Rosai in 1989, who described the histologic appearance of DSRCT as nests of small round blue cells, separated by desmoplastic stroma [4]. DSRCT exhibits features of multi-phenotypic differentiation, since mesenchymal, epithelial, and neural markers are expressed within this sarcoma [3,4]. Genetically, DSRCT is characterized by a chromosomal translocation, resulting in a gene fusion involving Ewing sarcoma region 1 (EWSR1) and Wilms tumor 1 (WT1). The chimeric protein contains the N-terminal domain of EWSR1

Single-Cell RNA Sequencing
OV-054 DSRCT cells were digested using TrypLE™ Express Enzyme (Gibco™, Thermo Fisher Scientific, #12605010, Life Technologies, Merelbeke, Belgium) and mechanically dissociated by pipetting up and down. The 2D and 3D OV-054 DSRCT cells were subsequently sorted for DAPI negative cells using the FACSJazz (BD bioscience, Franklin Lakes, NJ, USA) and SH800S Cell Sorter (SONY Europe B.V., Weybridge, Surrey, UK), respectively. Single DSRCT cells were sorted into a 384-well plate for SORT-seq performed by Single Cell Discoveries B.V. [15]. Gene counts per gene per well are provided in Tables S1-S4. Cluster analysis was performed using RaceID3 [16]. WT1 reads represent EWSR1-WT1 fusion expression, since wild-type WT1 is not expressed in OV-054 DSRCT cells.

Comparison R2 Dataset
The DSRCT patient described in this paper is included in the INFORM study [17] and available microarray data were analyzed in R2 (www.r2.amc.nl, accessed on 14 January 2020). Two gene sets were created from the gene expression list of the original DSRCT patient: high-expressed genes (Z-score > 1) and low-expressed genes (Z-score < −1). Average read counts of the genes in these gene sets were collected from the here generated scRNAseq data of the 2D and 3D cultured DSRCT cells.

Time-Lapse Imaging and Particle Analysis
OV-054 DSRCT cells with shRNA 3 and NT shRNA were induced with DOX, 24 h prior to live imaging. Live imaging was performed using the Leica DMi8, while incubated at 37 • C and 5% CO 2 . Images were taken every 2 min from representative areas in each well for 16 h in total (24-40 h post-DOX). Cell counts and cell area from pictures of DOX-induced and non-induced shRNAs in OV-054 DSRCT cells were analyzed using particle analysis in ImageJ (National Institutes of Health and the Laboratory for Optical and Comutational Instrumentation (LOCI, University of Wisconsin), Madison, WI, USA). The number of segments represent the number of cells, and the segment areas represent the area of each cell. Cell migration and cell adhesion genes were identified using the DAVID gene annotation tool [19,20]. Cell-matrix interactions were quantified using the cell counter tool in ImageJ (National Institutes of Health and the Laboratory for Optical and Comutational Instrumentation (LOCI, University of Wisconsin), Madison, WI, USA).

Medium-Throughput Compound Screen
Two days prior to the addition of compounds, 5 × 10 3 OV-054 DSRCT cells were seeded, per well, in 384-well plates. A panel of 201 drugs and DMSO controls were added to the 384-well plates in different concentrations using the Beckman Coulter Biomek i7 Hybrid liquid handling workstation. After 5 days of exposure of the compounds, cell viability was measured with CellTiter-Glo ® assay (Promega, #G9681, Madison, WI, USA) using the Spectramax i3x luminescence microplate reader. Values were normalized to DMSO. AUC Z-scores were calculated by (AUC-AUC avg )/st.dev for OV-054 DSRCT cultures and two Ewing cultures (ES-046 and ES-041). Relevant plasma concentrations from previous phase I and II trials (Table S8) were plotted to compare the viability curves.

Establishment and Characterization of a DSRCT In Vitro Preclinical Model
Surgically resected tissue was obtained from a DSRCT patient and successfully cultured in vitro under various conditions, including as spheroids in 3D basement membrane extract type 2 (BME) droplets, as a 2D layer in 0.1% BME, and in suspension without BME ( Figure 1A). The DSRCT in vitro cultures were established and maintained for over two years in serum-free basal medium (AdDMEM/F12+), supplemented with the specific growth factors EGF, FGF2, IGF1, RKI, and BMP4. While these growth factors were essential for initial growth of DSRCT cells, the established OV-054 DSRCT cultures were able to be maintained without the additional growth factors EGF, FGF2, IGF1, RKI, and BMP4 for over 10 passages. The addition of the growth factors independently did not affect DSRCT cell expansion in vitro, compared to the control medium without the additional growth factors ( Figure S1), suggesting that these OV-054 DSRCT cells are able to adapt fast and are able to propagate independently of their microenvironment. For consistency, all further experiments were conducted using the complete medium, in which the OV-054 DSRCT model was established. H&E staining on OV-054 DSRCT cultures revealed nests of small round cells with large nuclei ( Figure 1A), which is typically seen in DSRCT tissue [4]. Generally, patient-derived tumor tissue grown in 3D BME droplets can form organized structures, so-called organoids, that can capture disease heterogeneity [21]. To investigate whether our 2D and 3D DSRCT in vitro models also consist of distinct cell populations, single-cell RNA sequencing (scRNAseq) was performed. scRNAseq analysis of 2D and 3D OV-054 DSRCT cultures revealed a homogeneous, tight population of DSRCT cells ( Figure 1B,C). The slight difference between clustering of 2D versus 3D cultured cells was likely due to sequencing depth variation ( Figure 1D). Approximately 0-2 reads, per million (RPM) of WT1, were picked up per cell in the scRNAseq analysis ( Figure 1G), indicating low EWSR1-WT1 expression. Importantly, the expression of EWSR1-WT1 was similar between the 2D and 3D cultures ( Figure 1E,G). The DSRCT patient, described in this study, was enrolled in the INFORM (individualized therapy for relapsed malignancies in childhood) pilot study [17]. The INFORM registry applied comprehensive molecular profiling to provide information on actionable gene variants, which may be used for subsequent therapeutic approaches. Microarray data of the original tumor of this DSRCT patient were obtained from the INFORM study and used to create two gene sets from the gene expression list: high (Z-score > 1) and low genes (Z-score < −1). Average read counts of the two gene sets were collected from the scRNAseq data of the 2D and 3D cultured DSRCT cells, and a similar trend was found between expression of the gene sets in the in vitro cultures, compared to the original tumor ( Figure 1F). This indicated that both the 2D and 3D cultured DSRCT cells closely resemble features of the original tumor. For consistency, subsequent experiments were The DSRCT patient, described in this study, was enrolled in the INFORM (individualized therapy for relapsed malignancies in childhood) pilot study [17]. The INFORM registry applied comprehensive molecular profiling to provide information on actionable gene variants, which may be used for subsequent therapeutic approaches. Microarray data of the original tumor of this DSRCT patient were obtained from the INFORM study and used to create two gene sets from the gene expression list: high (Z-score > 1) and low genes (Z-score < −1). Average read counts of the two gene sets were collected from the scRNAseq data of the 2D and 3D cultured DSRCT cells, and a similar trend was found between expression of the gene sets in the in vitro cultures, compared to the original tumor ( Figure 1F). This indicated that both the 2D and 3D cultured DSRCT cells closely resemble features of the original tumor. For consistency, subsequent experiments were conducted with 2D cultured OV-054 DSRCT cells. To check whether wild-type WT1 is expressed in OV-054 DSRCT cells, the EWSR1-WT1 fusion breakpoint from cDNA, as well as several regions of WT1, were amplified. In our in vitro OV-054 DSRCT model, only the 3 -most exons of WT1 were expressed ( Figure 1H,I), while the expression of WT1 exon 4-7 was absent. This indicated that exons 9-10 from WT1 were derived from the EWSR1-WT1 fusion cDNA and that wild-type WT1 was not expressed in these cells ( Figure 1H,I). As a control, cDNA of a WT1-expressing malignant rhabdoid tumor of the kidney (MRTK) model was taken along, which, indeed, showed expression of all regions of the WT1 gene ( Figure 1H).

shRNAs Targeting the EWSR1-WT Breakpoint Create an Effective Knock-Down of the Fusion mRNA
The OV-054 DSRCT in vitro model enabled us to study the molecular target genes of the fusion driving DSRCT. By generating a knock-down model of the EWSR1-WT1 fusion mRNA, target genes could be identified that were affected by the expression level of EWSR1-WT1. To this end, we sequenced the breakpoint of EWSR1-WT1 cDNA, in order to identify the exact location of the gene-fusion and found that EWSR1 exon 7 was fused to WT1 exon 8. This enabled us to design short hairpin RNAs (shRNAs), targeting the exact EWSR1-WT1 breakpoint (shRNA 3, 5) or exon 8-10 of WT1 (shRNA 2, 4, 6). With these, we created a DOX-inducible knock-down of the EWSR1-WT1 fusion mRNA ( Figure 2A). As a control, a non-targeting (NT), DOX-inducible scrambled shRNA sequence was taken along. An effective knock-down of the EWSR1-WT1 fusion was observed with shRNA 2, 3, and 5 after 24 h of DOX induction ( Figure 2B). When cells were harvested 8, 16, 24, 32, and 40 h after DOX induction, a gradual decrease of EWSR1-WT1 transcripts was observed ( Figure 2C), where shRNA 3 resulted in the most efficient knock-down. To identify genes that are affected upon EWSR1-WT1 knock-down, RNA sequencing was performed. RNA sequencing reads were mapped to a reference transcriptome, in which EWSR1 and WT1 were split into a 3 end and a 5 end, in order to distinguish between the EWSR1-WT1 fusion and wild-type EWSR1 and WT1 transcripts. 5 WT1 was not expressed ( Figure 2D), again confirming that wild-type WT1 was not expressed in these DSRCT cells; 5 EWSR1 and 3 WT1 showed a gradual decrease upon DOX induction ( Figure 2D), indicative of efficient knock-down of the EWSR1-WT1 fusion transcripts. A slight decrease was seen in 3 EWSR1 transcripts ( Figure 2D), suggesting that shRNA 3 could also target wild-type EWSR1. However, the RPKM did not drop below the 3 EWSR1 levels in the NT shRNA control, indicating that this effect was minimal. Sequencing reads were picked-up from both +KTS and −KTS isoforms, which were decreased to a similar extent after induction of shRNA 3 ( Figure 2E). Hence, knock-down of the EWSR1-WT1 fusion with shRNA 3 did not result in an isoform bias. Together, these data showed that induction of shRNA 3 (targeting the EWSR1-WT1 breakpoint in DSRCT cells) resulted in an efficient knock-down of the fusion gene. Wild-type EWSR1 and WT1 were not affected, and no bias was observed for the +KTS or −KTS isoform. This EWSR1-WT1 knock-down model was used to further investigate genes affected by levels of EWSR1-WT1 expression.

EWSR1-WT1 Expression Affects Cell Shape, Cell Propagation, and the Transcriptome
To investigate the phenotypic effects of EWSR1-WT1 in OV-054 DSRCT cells, timelapse imaging was performed. Knock-down of the EWSR1-WT1 fusion mRNA, in a time lapse between 24 h and 40 h after DOX induction, resulted in a decrease of cell-cell adhesion and increase of cell-matrix adhesion (Figures 3A and S2). Additionally, after 4 days of DOX-induction of shRNA 3, OV-054 DSRCT cells appeared more stretched, creating more cell-matrix adhesion, while non-induced cells, as well as cells with DOX-induction of the NT shRNA control, remained in clusters of inter-adhesive cells ( Figure 3B). Upon knock-down of EWSR1-WT1, the number of cells decreased (Figure 3C), while the area of the cells increased ( Figure 3D), showing that decreased expression of EWSR1-WT1 quickly affects shape and expansion of DSRCT cells.

EWSR1-WT1 Expression Affects Cell Shape, Cell Propagation, and the Transcriptome
To investigate the phenotypic effects of EWSR1-WT1 in OV-054 DSRCT cells, timelapse imaging was performed. Knock-down of the EWSR1-WT1 fusion mRNA, in a time lapse between 24 h and 40 h after DOX induction, resulted in a decrease of cell-cell adhesion and increase of cell-matrix adhesion (Figures 3A and S2). Additionally, after 4 days of DOX-induction of shRNA 3, OV-054 DSRCT cells appeared more stretched, creating more cell-matrix adhesion, while non-induced cells, as well as cells with DOX-induction of the NT shRNA control, remained in clusters of inter-adhesive cells ( Figure 3B). Upon knockdown of EWSR1-WT1, the number of cells decreased (Figure 3C), while the area of the cells increased ( Figure 3D), showing that decreased expression of EWSR1-WT1 quickly affects shape and expansion of DSRCT cells. To investigate the effect of EWSR1-WT1 expression on the transcriptome, RNA sequencing was performed 8, 24, and 32 h after induction of shRNA 2, 3, and NT. To control for off-target effects, both shRNA 2 and shRNA 3 were used, which target WT1 exon 10 and the EWSR1-WT1 breakpoint, respectively. For both shRNA 2 and shRNA 3, genes that were at least 25% up-or downregulated were selected, compared to the NT shRNA control. A cut-off was set at 5 RPKM, in order to extract truly expressed genes. Genes that were affected by both shRNA 2-and shRNA 3-mediated knock-down of the EWSR1-WT1 fusion were called true target genes of EWSR1-WT1 ( Figure 4A). After 32 h of DOX-induction, we identified 75 genes that were downregulated upon EWSR1-WT1 knock-down and 174 genes that were upregulated, upon EWSR1-WT1 knock-down ( Figure 4B). The expression levels of the 50 most upregulated genes and most downregulated genes, upon EWSR1-WT1 knock-down, are shown in heatmaps ( Figure 4C,D). The phenotypic effects that we observed were accompanied by an increased expression of several cell migration and cell adhesion genes ( Figure S3C). While phenotypically the DSRCT cells appeared more mesenchymal upon knock-down of the EWSR1-WT1 fusion, no evidence was found for epithelial to mesenchymal transition (EMT) ( Figure S3D) [22]. Together, these data showed that cell migration and adhesion genes are upregulated upon knock-down of EWSR1-WT1, resulting in phenotypic changes, including a decreased cell expansion and To investigate the effect of EWSR1-WT1 expression on the transcriptome, RNA sequencing was performed 8, 24, and 32 h after induction of shRNA 2, 3, and NT. To control for off-target effects, both shRNA 2 and shRNA 3 were used, which target WT1 exon 10 and the EWSR1-WT1 breakpoint, respectively. For both shRNA 2 and shRNA 3, genes that were at least 25% up-or downregulated were selected, compared to the NT shRNA control. A cut-off was set at 5 RPKM, in order to extract truly expressed genes. Genes that were affected by both shRNA 2-and shRNA 3-mediated knock-down of the EWSR1-WT1 fusion were called true target genes of EWSR1-WT1 ( Figure 4A). After 32 h of DOX-induction, we identified 75 genes that were downregulated upon EWSR1-WT1 knock-down and 174 genes that were upregulated, upon EWSR1-WT1 knock-down ( Figure 4B). The expression levels of the 50 most upregulated genes and most downregulated genes, upon EWSR1-WT1 knock-down, are shown in heatmaps ( Figure 4C,D). The phenotypic effects that we observed were accompanied by an increased expression of several cell migration and cell adhesion genes ( Figure S3C). While phenotypically the DSRCT cells appeared more mesenchymal upon knock-down of the EWSR1-WT1 fusion, no evidence was found for epithelial to mesenchymal transition (EMT) ( Figure S3D) [22]. Together, these data showed that cell migration and adhesion genes are upregulated upon knock-down of EWSR1-WT1, resulting in phenotypic changes, including a decreased cell expansion and increase in cell surface area. Subsequently, OV-054 DSRCT cells, with decreased EWSR1-WT1 levels, create more cell-matrix interactions ( Figure S3A,B).  The genes that were affected upon knock-down of EWSR1-WT1 were next compared to a recent study, in which a siRNA-mediated knock-down of the EWSR1-WT1 fusion gene was performed in the JN-DSRCT-1 and BER cell lines [10]. From the 174 upregulated genes upon EWS-WT1 knock-down in our OV-054 DSRCT cells, 6% (11/174) showed an overlap with both the JN-DSRCT-1 and BER lines: CST1, FILIP1L, CHST1, ALCAM, EPHA4, KRT8, ELF3, KRT18, SLC1A1, PCP4, and ASS1 ( Figure 5A). We found that 9% (7/75) of the downregulated genes, upon EWSR1-WT1 knock-down, in our OV-054 DSRCT cells were shared with the other two in vitro models: LOXL3, FOXL1, RRAD, ST6GAL-NAC5, CCL25, IGFBP3, and C1QL4 ( Figure 5B). ShRNA3 resulted in a more efficient knock-down of EWSR1-WT1, compared to shRNA 2-mediated knock-down. When only The genes that were affected upon knock-down of EWSR1-WT1 were next compared to a recent study, in which a siRNA-mediated knock-down of the EWSR1-WT1 fusion gene was performed in the JN-DSRCT-1 and BER cell lines [10]. From the 174 upregulated genes upon EWS-WT1 knock-down in our OV-054 DSRCT cells, 6% (11/174) showed an overlap with both the JN-DSRCT-1 and BER lines: CST1, FILIP1L, CHST1, ALCAM, EPHA4, KRT8, ELF3, KRT18, SLC1A1, PCP4, and ASS1 ( Figure 5A). We found that 9% (7/75) of the downregulated genes, upon EWSR1-WT1 knock-down, in our OV-054 DSRCT cells were shared with the other two in vitro models: LOXL3, FOXL1, RRAD, ST6GALNAC5, CCL25, IGFBP3, and C1QL4 ( Figure 5B). ShRNA3 resulted in a more efficient knock-down of EWSR1-WT1, compared to shRNA 2-mediated knock-down. When only the genes affected upon shRNA 3-mediated knock-down were compared with the JN-DSRCT-1 and BER cell lines, we indeed discovered additional genes to overlap between the three in vitro models ( Figure S4). The overlap of the transcriptomic effects, upon knock-down of EWSR1-WT1 in three independent DSRCT cell lines, confirmed that these genes are direct or indirect targets of the EWSR1-WT1 fusion and potentially play an important role in tumor development and progression.
the genes affected upon shRNA 3-mediated knock-down were compared with the JN-DSRCT-1 and BER cell lines, we indeed discovered additional genes to overlap between the three in vitro models ( Figure S4). The overlap of the transcriptomic effects, upon knock-down of EWSR1-WT1 in three independent DSRCT cell lines, confirmed that these genes are direct or indirect targets of the EWSR1-WT1 fusion and potentially play an important role in tumor development and progression.

Drug Screen on OV-054 Cells Reveals Effective Compounds Targeting RTKs and Downstream Pathways
The established preclinical OV-054 DSRCT in vitro model enabled us to perform a medium-throughput drug screen for DSRCT. The screen involved a panel of 201 different compounds. Several of these compounds affected DSRCT cell viability. The area under the curve (AUC) Z-scores of OV-054 DSRCT cells were compared to the Z-scores of two Ewing sarcoma in vitro models, ES-041 and ES-046, to unravel compounds that showed a higher sensitivity in DSRCT cells ( Figure 6A). As DSRCT is closely related to Ewing sarcoma, the compounds showing AUC Z-score possibly present sarcoma specific sensitivity. From these drugs, the compounds were selected that presented with an IC50 value, which was below or around the relevant plasma concentrations ( Figure 6B). Several of the compounds that showed an effective response were targeting apoptosis regulators (such as XIAP and BCL2L), including AZD5582, Birinapant, and Navitoclax ( Figure 6B), suggesting that the regulation of apoptosis is relevant in DSRCT cells. Interestingly, many other compounds that showed an effective response, including regorafenib, lapatinib, entrectinib, linsitinib, crizotinib, dovitinib, sorafenib, vandetanib, and brigatinib ( Figure 6B), target RTKs. Additionally, compounds targeting downstream targets of RTK signaling pathways ( Figure 6B), such as the PI3K-AKT and the mTOR signaling pathways, showed effective cell killing, including AT7519 and AZD8055. Together, the drug screening data suggested that the regulation of apoptosis-and RTK-driven signaling pathways are important for DSRCT tumor cell survival.

Drug Screen on OV-054 Cells Reveals Effective Compounds Targeting RTKs and Downstream Pathways
The established preclinical OV-054 DSRCT in vitro model enabled us to perform a medium-throughput drug screen for DSRCT. The screen involved a panel of 201 different compounds. Several of these compounds affected DSRCT cell viability. The area under the curve (AUC) Z-scores of OV-054 DSRCT cells were compared to the Z-scores of two Ewing sarcoma in vitro models, ES-041 and ES-046, to unravel compounds that showed a higher sensitivity in DSRCT cells ( Figure 6A). As DSRCT is closely related to Ewing sarcoma, the compounds showing AUC Z-score possibly present sarcoma specific sensitivity. From these drugs, the compounds were selected that presented with an IC50 value, which was below or around the relevant plasma concentrations ( Figure 6B). Several of the compounds that showed an effective response were targeting apoptosis regulators (such as XIAP and BCL2L), including AZD5582, Birinapant, and Navitoclax ( Figure 6B), suggesting that the regulation of apoptosis is relevant in DSRCT cells. Interestingly, many other compounds that showed an effective response, including regorafenib, lapatinib, entrectinib, linsitinib, crizotinib, dovitinib, sorafenib, vandetanib, and brigatinib ( Figure 6B), target RTKs. Additionally, compounds targeting downstream targets of RTK signaling pathways ( Figure 6B), such as the PI3K-AKT and the mTOR signaling pathways, showed effective cell killing, including AT7519 and AZD8055. Together, the drug screening data suggested that the regulation of apoptosis-and RTK-driven signaling pathways are important for DSRCT tumor cell survival.

MERTK, Regulated by EWSR1-WT1, Is a Potential Therapeutic Target in DSRCT
Since RTKs and downstream pathways of RTKs appeared to be important for DSRCT tumor progression, we looked for RTKs that are regulated by EWSR1-WT1 expression. MERTK is one of the most downregulated genes, upon knock-down of EWSR1-WT1, in OV-054 DSRCT cells ( Figure 4C) and the JN-DSRCT-1 cell line ( Figure 5B), suggesting that high expression of MERTK ( Figure 7A) is regulated by the EWSR1-WT1 fusion. Interestingly, MERTK has been described as a therapeutic target in several cancers, including melanoma, leukemia, glioblastoma, and gastric cancer [34][35][36]. To investigate the function of MERTK in DSRCT cells, different concentrations of a MERTK/FLT3 small-molecule inhibitor UNC2025 were added to the culture medium, and live cells were counted after 4 and 7 days. Inhibition of MERTK resulted in reduced cell expansion of OV-054 DSRCT cells in vitro in a dose dependent manner ( Figure 7B,D), and a similar effect was found in the JN-DSRCT-1 cell line ( Figure 7B,D). In both DSRCT cell lines, the IC50 of UNC2025 was around 104 nM ( Figure 7C), in line with a previous study of UNC2025-mediated MERTK inhibition in leukemia [34]. Normal human small intestinal (huSI) organoids were exposed to UNC2025 as a control and were only affected by UNC2025 at the highest concentration of 400 nM ( Figure 7B,D). This shows that the effect of UNC2025, seen in both DSRCT in vitro models, is a targeted effect, rather than a non-specific cytotoxic effect. Together, this suggested that MERTK is an important driver for cell proliferation in DSRCT and a potential therapeutic target.

Discussion
DSRCT is a highly aggressive and rare soft tissue sarcoma, characterized by a chromosomal translocation, resulting in the EWSR1-WT1 gene fusion. Expression of EWSR1-WT1 presumably regulates the genetic targets responsible for oncogenesis in DSRCT. The 5-year, event-free survival rate is 18% [37], showing the urgence of novel therapies to improve the outcomes for DSRCT patients.

Discussion
DSRCT is a highly aggressive and rare soft tissue sarcoma, characterized by a chromosomal translocation, resulting in the EWSR1-WT1 gene fusion. Expression of EWSR1-WT1 presumably regulates the genetic targets responsible for oncogenesis in DSRCT. The 5-year, event-free survival rate is 18% [37], showing the urgence of novel therapies to improve the outcomes for DSRCT patients.
OV-054 DSRCT cells in vitro grow as nests of small round cells with large nuclei, typically also seen in DSRCT tissue [4]. Both 2D and 3D culture conditions of OV-054 DSRCT consist of a homogeneous cell population, exhibiting similarities to the transcriptional profile of the original tumor. The EWSR1-WT1 fusion is expressed similarly under 2D and 3D culture conditions. This novel DSRCT in vitro model can be used to investigate molecular pathways driving this rare sarcoma type and further explore therapeutic options, which is urgently needed to improve the poor prognosis of this sarcoma.
To explore the molecular mechanisms of EWSR1-WT1, we used an shRNA knockdown approach on our primary DSRCT in vitro model. When the genes that were affected upon shRNA-mediated knock-down of EWSR1-WT1 were compared with similar gene sets previously described for the JN-DSRCT-1 and BER cell lines [10], we indeed found overlap in up-and downregulated genes. Of note, ASCL1 was downregulated upon EWSR1-WT1 knock-down, which was previously described as a direct target of the gene fusion [7]. This confirms that these genes are (in)direct targets of the EWSR1-WT1 fusion and likely play a role in tumor development and progression.
Gedminas and colleagues show similarities between the molecular mechanisms of EWSR1-FLI1 in Ewing sarcoma and EWSR1-WT1 in DSRCT, despite their different DNAbinding domains [10]. These common features include the dysregulation of the DNA damage response, an alteration in the E2F transcription factor family members and modulation of other pathways, including TGFb and IGF/mTOR signaling. While the mechanism behind these common features is not clear, DSRCT cells show a striking dependence on ERG expression. This is a close family member of FLI1, and it is upregulated by EWSR1-WT1 in DSRCT [10]. Interestingly, Franzetti and colleagues found that a knock-down of EWSR1-FLI1 expression in Ewing sarcoma affected cell dynamics [38]. Major changes were observed in the dynamics of the actin cytoskeleton and cell-to-cell adhesions shifted to cell-matrix adhesion, associated with an increase of cell migration and invasion potential in vivo. The dynamical changes of the actin cytoskeleton and a shift from cell-cell adhesion to cell-matrix adhesion, shown in the Ewing sarcoma model, are similar to the observations we found in DSRCT, upon knock-down of EWSR1-WT1 in vitro. In our study, knock-down of EWSR1-WT1 in DSRCT cells affected cell shape and propagation. We found several cell migration genes and cell-cell adhesion genes to be affected upon EWSR1-WT1 knock-down. Thus, despite the differences in DNA binding motifs of EWSR1-FLI1 and EWSR1-WT1, the underlying mechanisms driving Ewing sarcoma and DSRCT are possibly similar.
Because of the rarity of DSRCT, this sarcoma is often excluded from clinical trials. Currently, there are 10 clinical trials ongoing or completed that involve DSRCT patients and just one of these clinical trials included an RTK-targeting compound, i.e., sorafenib (ClinicalTrials.gov identifier: NCT01946529). Unfortunately, the interim analysis of this clinical trial determined that the therapy did not meet the anticipated response; therefore, the trial was stopped. To discover novel therapeutic entities, we performed a mediumthroughput drug screen on the established preclinical OV-054 DSRCT in vitro model. A total of 201 different compounds were included in the drug screen for DSRCT cell viability. The compounds to which the OV-054 DSRCT cells were sensitive included several drugs targeting the regulators of apoptosis and many drugs targeting RTKs and downstream pathways of RTKs, including the PI3K-AKT and mTOR signaling pathways. Our data suggests that the regulation of apoptosis is important in DSRCT cells and that RTK-driven pathways are key players in DSRCT tumor progression. Therefore, other compounds that target RTKs or downstream pathways of RTKs might show a more effective response, either alone or in combination with current standard chemotherapy.
MERTK levels in OV-054 DSRCT are decreased upon knock-down of EWSR1-WT1, showing that this gene is likely regulated by the fusion protein. Interestingly, MERTK expression is higher in the OV-054 DSRCT cells, compared to the JN-DSRCT-1 cells. This inconsistency can be the result of the different fusion breakpoints of the EWSR1-WT1 fusion genes in the OV-054 DSRCT and JN-DSRCT1 cells. However, we cannot exclude the possibility that this effect is a result of differences in the culture media. Despite these different expression levels of MERTK in the different DSRCT in vitro models, the effect of small-molecule inhibitor UNC2025 was similar between the OV-054 DSRCT model and the JN-DSRCT-1 cell line. MERTK has several downstream signaling pathways, including MAPK/ERK, PI3K/AKT, and JAK/STAT, regulating multiple biological processes [35,39,40]. MERTK is involved in multiple malignancies, including leukemia, glioma, melanoma, and rhabdomyosarcoma [35,36], while it has been described as novel therapeutic target in several of these malignancies [36]. Here, we show that MERTK inhibition, with small-molecule inhibitor UNC2025, affects propagation of OV-054 DSRCT cells in vitro, providing a rationale for investigating MERTK as a therapeutic target in DSRCT by small molecule inhibitors, such as UNC2025.

Conclusions
Using a primary DSRCT patient-derived 2D and 3D cell culture system, we were able to characterize the molecular mechanisms that are driven by the DSRCT-specific EWSR1-WT1 fusion protein. The preclinical DSRCT in vitro model also enables us to perform a medium-throughput drug screen. This screen reveals compounds that affect the cellular pathways that are important for DSRCT cell viability, including RTK-driven signaling pathways. Interestingly, we show that expression the RTK family member MERTK correlates with the expression of the EWSR1-WT1 fusion gene. To our knowledge, this is the first study that reveals effective therapeutic compounds that likely target EWSR1-WT1-driven mechanisms.