Upregulation of the Mevalonate Pathway through EWSR1-FLI1/EGR2 Regulatory Axis Confers Ewing Cells Exquisite Sensitivity to Statins

Simple Summary The objective of this project was to search for new dependencies in Ewing sarcoma, a deadly disease for which new therapeutic approaches are urgently needed. A pharmacological screening of off-patent approved drugs (FDA agency) and the investigation of downstream targets of EGR2 were performed. The two approaches showed the MVA pathway as a major dependency in Ewing sarcoma and statin, an inhibitor of this pathway, as a potential new therapeutic agent for the treatment of Ewing sarcoma. Abstract Ewing sarcoma (EwS) is an aggressive primary bone cancer in children and young adults characterized by oncogenic fusions between genes encoding FET-RNA-binding proteins and ETS transcription factors, the most frequent fusion being EWSR1-FLI1. We show that EGR2, an Ewing-susceptibility gene and an essential direct target of EWSR1-FLI1, directly regulates the transcription of genes encoding key enzymes of the mevalonate (MVA) pathway. Consequently, Ewing sarcoma is one of the tumors that expresses the highest levels of mevalonate pathway genes. Moreover, genome-wide screens indicate that MVA pathway genes constitute major dependencies of Ewing cells. Accordingly, the statin inhibitors of HMG-CoA-reductase, a rate-limiting enzyme of the MVA pathway, demonstrate cytotoxicity in EwS. Statins induce increased ROS and lipid peroxidation levels, as well as decreased membrane localization of prenylated proteins, such as small GTP proteins. These metabolic effects lead to an alteration in the dynamics of S-phase progression and to apoptosis. Statin-induced effects can be rescued by downstream products of the MVA pathway. Finally, we further show that statins impair tumor growth in different Ewing PDX models. Altogether, the data show that statins, which are off-patent, well-tolerated, and inexpensive compounds, should be strongly considered in the therapeutic arsenal against this deadly childhood disease.


Gene Expression Profile
The differential analysis of genes modulated by EWSR1-FLI1 and EGR2 was performed with data obtained with Affymetrix Human Gene 2.1 ST arrays as described by [13]. The data were deposited at the Gene Expression Omnibus (GEO; GSE62090). The differential expression of mevalonate pathway enzymes and prenyltransferases (in Green) in 6 Ewing cell lines was performed with RNAseq data obtained after invalidation of EWSR1-FLI1 by siEF1 at different time points. The data were deposited at the Gene Expression Omnibus (GSE 132966, GSE 156653, GSE 150777, and GSE 164373).

High-Throughput Drug Screening and Hit Calling
The high-throughput drug screening was performed using the PCL on the BioPhenics platform of Institut Curie, Paris, France (Elaine Del-Nery). The PCL V3 (1280 small molecules) and V2 (1200 small molecules) off-patent approved drugs were selected for their bioavailability and safety for humans. For the screening, the A673 cell line was seeded in 384-well plates the day prior to compound addition. The next day, a single dilution of each compound of PCL V2 and V3 at 10 µM was added to the cells for 24 h and 72 h, respectively. After fixation with 4% formaldehyde solution, cell nuclei were stained with 0.2 µg/mL of DAPI and image acquisition was performed using the IN Cell 2200 automated wide-field microscopy system (GE Healthcare) at 10× magnification. Nucleus counts were performed using a specific module implemented in IN Cell Analyzer Workstation software (GE Healthcare). Raw cell-count values were obtained by the average of 4 image fields per well and transformed in log function to obtain a robust Z-score-standardized robust estimator value, as previously described [20]. The compounds with an RZ-score < −2 correspond to decreased cell proliferation. Two biological replicates were obtained for the 24 h time-point experiment and three for the 72 h time-point experiment. Data analysis and hit-calling details are provided in Figure S3.

ChIP-Seq
Chromatin Immunoprecipitation (ChIP) experiments were performed with the POE shEGR2_4#22 clone stably expressing the EGR2 protein fused to the HA tag. Previously, endogenous EGR2 protein expression was inhibited by doxycycline-inducible shRNA against EGR2 for 5 days. ChIP was conducted following the manufacturer's protocol using the iDeal ChIP-seq kit for transcription factors and histones (Diagenode; Denville, NJ, USA) with anti-HA (2 µg; Y11; Santa-Cruz), rabbit polyclonal anti-Fli (2 µg; 15289; Abcam), rabbit polyclonal anti-H3K4me3 (1 µg; C15410003; Diagenode), and rabbit polyclonal anti-H3K27ac (1 µg; ab4729; Abcam). ChIP sequencing was performed on Illumina HiSeq 2500 using 100 bp single-end sequencing; libraries were generated using the TruSeq ChIP library preparation kit (Illumina, San Diego, CA, USA). The obtained reads were aligned in the human reference genome (GRCh37/hg19) using bowtie2-2.2.9 [21]. Uninformative reads (multimapped reads, duplicated reads, and reads with low mapping score) were filtered out with samtools 1.3 [22]. Peaks were called with MACS2 2.1.1 [23] with the narrow option for EGR2 ChIP-seq and the broad option for histone marks. ChIP-seq were normalized according to the input DNA sample. The ChIP-seq signal tracks were generated by macs2 with the bdgcmp option (and -m FE to compute fold enrichment between the ChIP and the control). Then, we ran bedGraphToBigWig to convert the file to a binary format (BigWig).

Cell Proliferation in 2D and 3D Cultures (Spheroid Assay) and IC50 Determination
The assessment of the effect of Atorvastatin and Simvastatin on cell proliferation and the IC50 determination were conducted by seeding, respectively, 5 × 10 4 A673, TC71, MCF7, A549, SKNSH, and KD, 7.5 × 10 4 POE, and 10 × 10 4 SKNMC in each well of 24-well plates. For PDX-dissociated cells, 14 × 10 4 cells were seeded per well. After 24 h, a serial dilution of statins (30 µM->0.12 µM) was added for 72 h for the treatment of the cell lines and for 6 days for PDX-dissociated cells. Cell counting was performed by the trypan blue exclusion method using a Vi-cell XR Viability Analyzer (Beckman Coulter). IC50 values were determined by nonlinear fit dose-response curves using prism graphpad 8 software, (San Diego, CA, USA) and normalized to cells treated with DMSO.
For the 3D culture, single spheroids per well were formed from 800 dissociated cells derived from a PDX tumor in a low ultra-attachment 96-well plate (ULA plate; 7007; Corning). After 4 days of spheroid formation, the serial dilution of statins (30 µM->0.12 µM) was added for 7 days. Images of spheroids were taken on a Leica microscope with a DFC 3000G camera at 10× magnification. The growth of spheroids was measured on day 7 by area calculation using imageJ software. All experiments were performed in duplicate. The measure of the surface area of spheroids was normalized to spheroids treated with DMSO.

Cell-Cycle Analysis and Annexin V Staining
A673 (3.5 × 10 5 cells), TC71 (4 × 10 5 cells), and POE (6 × 10 5 cells) were seeded in flasks of 25 cm 3 . After 24 h, the cells were treated with, respectively, 1 µM, 2.6 µM, and 2.4 µM Atorvastatin for 72 h. Cell-cycle analysis was performed by EdU staining (5 ethylnyl-2 -deoxyuridine) with 10 µM EdU solution for 30 min at room temperature. The cells were harvested, washed with PBS, resuspended in 0.5% BSA-PBS, and fixed with 70% ethanol overnight at 4 • C. Then, the cells were centrifuged and permeabilized with 0.2% Triton X-100 for 20 min. The detection of EdU labeling was performed with azide coupled with 1.8 µL of Alexa Fluor 647 dye (C10419; ThermoFisher Scientific) in PBS buffer containing 2 µM CuSO4 and 10 µM Sodium Ascorbate for 30 min at room temperature in the dark. The cells were stained with PI solution (0.5 µg/mL Pi and 50 µg/mL RNAseA) and analyzed on an LSRII flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA). Cell-cycle progression was studied using an EdU pulse-chase experiment on asynchronized cells, and the Double Thymidine block (DT) method was used to synchronize cells at the G1/S boundary, with the A673 and TC71 cell lines having been treated for 72 h with Atorvastatin at, respectively, 1 µM and 4.6 µM, corresponding to 80% of cell proliferation inhibition. For the EdU pulse-chase, cells were pulsed with EdU (10 µM) for 30 min, washed, and released in fresh media. EdU-positive cells and unstained cells were followed overtime (2 h, 4 h, 6 h, 8 h, and 10 h) through the phases of the cell cycle by flow cytometry. For the DT analysis, 2 mM thymidine was added to the culture medium for 14 h; the cells were washed twice with culture medium and covered with medium for 9 h. Then, 2 mM thymidine was added to the culture for 14 h. At the end of the second round of thymidine treatment, the cells were washed and collected at 0, 2 h, and 4 h after release. Cells were stained with EdU (10 µM) for 30 min each time they were collected.

Rescue Experiments on Cell Proliferation
The implication of the mevalonate pathway in the cytotoxic effects of statins on Ewing cell lines and PDX-dissociated cells was assessed by rescue experiments performed by the addition of intermediate products of lipid/cholesterol synthesis. Three Ewing cell lines (A673, TC71, and POE) were seeded in a 24-well plate as described previously. After 24 h, Atorvastatin (1 µM A673, 2.6 µM TC71, and 2.4 µM POE) was added to the culture medium in the absence or presence of 100 µM MEV, 10 µM Fpp, or GGpp for 72 h. The experiments with NAC, an inhibitor of intracellular ROS (NAC), were performed in flasks of 25 cm 3 with 4.5 × 10 5 of A673 cells. A total of 24 h after being seeded, cells were pre-treated for 2 h with NAC at 3 mM before the addition of Atorvastatin (1 µM) and collected 48 h post treatment. Cell viability was measured with the trypan blue exclusion method using a Vi-cell XR Viability Analyzer (Beckman Coulter). All analyses were conducted in duplicate in 3 independent experiments and normalized to cells treated with DMSO.

Intracellular Cell ROS and Lipid Peroxidation Measurement
The intracellular ROS level was studied using a Probes CellROX Deep Red kit (C10422; ThermoFisher Scientific) and lipid peroxidation using Bodipy 580/591 C11 (Lipid peroxidation sensor) (D3861; ThermoFisher Scientific). Briefly, the A673, TC71, and POE cell lines were seeded at 1 × For the study of the mevalonate pathway's role in the modulation of intracellular ROS levels and lipid peroxidation induced by statins, the A673 cell line was treated with 500 nM Atorvastatin and the rescue experiments were performed with MEV (100 µM), Fpp (10 µM), and GGpp (10 µM) for 72 h. For the measurement of ROS upon NAC treatment, the A673 cell line was seeded at 1.5 × 10 5 and pretreated with 3 mM NAC, followed by the addition of Atorvastatin (1 µM).
All experiments were performed in duplicate, in 3 independent experiments. Mean fluorescence intensity (MFI) was performed with flowjo 9.8.2 software (BD Biosciences) and relative specific MFI was calculated by normalizing to cells treated with DMSO.

Prenylated Protein Extraction
Prenylated proteins were separated using the Triton X-114 phase-separation method according to [25]. Briefly, cells were resuspended in PBS 2% pre-condensed Triton X-114 for 30 min at 4 • C, followed by centrifugation at 13,000 rpm for 10 min at 4 • C. Phase separation was achieved by incubating at 37 • C for 10 min, followed by centrifugation at 13,000 rpm for 10 min at room temperature. The detergent phase containing prenylated protein was collected and washed 2 times using PBS 0.1% pre-condensed Triton X-114. The proteins were extracted with the Methanol/chloroform method (as described by [25]) by adding 1/5 volume of methanol and 4/5 volume of chloroform for 20 min to the detergent phase. After centrifugation at 13,000 rpm for 30 min at 4 • C, the upper phase was removed and 9 volumes of methanol was added to the lower phase, followed by centrifugation at 13,000 rpm for 30 min at 4 • C. The pellet was resuspended in Laemli Buffer containing 1 mM DTT. The detergent phase was resolved by SDS-PAGE and transferred onto a nitrocellulose membrane (Bio-Rad). Blots were incubated with the antibodies rabbit anti-RhoA (1:1000; #2117; cell signaling), mouse anti-Rac1 (1:500; 610650; BD Biosciences), rabbit anti-Ras (1:10,000; ab52939; abcam), and goat anti-HLA-A (1:500; sc-23446; Santa Cruz) and the secondary antibodies HRP conjugated mouse anti-goat IgG (1:10,000; sc-2354; Santa Cruz) and HRP conjugated sheep anti-mouse IgG (1:3000; NA931; Sigma Aldrich).

Xenograft of Ewing PDX Models
Tumor fragments of IC-pPDX-87, IC-pPDX-3, and IC-pPDX164 were grafted into the fat pad of 12 mice per PDX model. At a tumor size of around 32-60 mm 3 , mice were randomly assigned to Atorvastatin or DMSO groups. Mice were treated every day by intraperitoneal injection of DMSO or Atorvastatin (10 mg/kg/day) until one mouse reached the ethic size volume of 2000-2500 mm 3 , measured by the formula V = D × d 2 /2, where D is the largest diameter and d the smallest one. Tumor volume was measured every 2 or 3 days with a caliper.
This study was performed in accordance with the recommendations of the European Community (2010/63/UE) for the care and use of laboratory animals. Experimental procedures were approved by the ethics committee of Institut Curie CEEA-IC #118 (Authorization APAFIS#11206-2017090816044613-v2 given by National Authority) in compliance with the international guidelines.

Statistics
IC50 values were determined by nonlinear fit dose-response curves using prism graphpad 8 software and normalized to cells treated with DMSO. All statistic tests except for Bayesian Student's t-tests were performed using prism graphpad 8 software. Bayesian Student's t-tests were performed using the Limma R package. Boxplots are represented from min to max with median representation. Barplots, growth curves, and last-day tumor volume are represented as mean +/− SEM when the repliquat number between cell lines was different, otherwise +/− SD was used. EwS cells' higher sensibility to statin was analyzed using Bayesian t-tests as used in depmap portal high-throughput screening. The statistical evaluation of Mevalonate pathway expression was performed with two-tailed unpaired Student's t-tests with Welch's correction. The Atorvastatin effect or rescue effect was analyzed using two-tailed unpaired Student's t-tests, whereas a two-tailed paired Student's t-statistic was used to access the in vivo growth inhibitory effect of Atorvastatin, and a comparison of the mean volume of the tumor on the last day treatment was performed using a nonparametric Wilcoxon rank test. The p-value was considered significant at p < 0.05.

EWSR1-FLI1 and EGR2 Regulate Lipid/Cholesterol/Mevalonate Pathway
We previously showed that EWSR1-FLI1 directly regulates the expression of the EGR2 susceptibility gene in Ewing cells by binding to a GGAA repeat motif [12,13]. As EGR2 encodes a transcription factor that is essential to Ewing oncogenesis [13], we undertook a characterization of its mechanisms of action by investigating downstream targets. Upon downregulation of EGR2 with two different siRNAs (siEGR2 #1 and #2) in the A673 and SKNMC Ewing cell lines, we found a common set of 462 modulated genes (log2FC > 0.5) upon EGR2 knock-down (EGR2-KD) in both cell lines ( Figure 1A). Genes down-regulated upon EGR2-KD (280 genes) demonstrated a very significant enrichment in genes involved in the cholesterol/steroid synthesis pathway ( Figures 1B and S1A). This result is in full agreement with the previously identified central function of EGR2 in myelinization programs [15,16] and with its direct role in the regulation of the lipid/cholesterol biosynthetic pathway [17]. Of particular interest was the presence of a set of seven genes highlighting the role of EGR2 in the regulation of the mevalonate (MVA) pathway. This pathway is key for the biosynthesis of cholesterol and sterol biosynthesis but also for the production of isoprenoid lipids (Fpp, Farnesyl pyrophosphate, and GGpp, Geranyl-Geranyl pyrophosphate). Fpp and GGpp are essential for post-translational protein prenylation and for the synthesis of ubiquinone, which is implicated in the mitochondrial respiratory chain and the production of ROS ( Figure 1C). In agreement with EGR2 being downstream of EWSR1-FLI1, we observed a highly significant overlap between EWSR1-FLI1-and EGR2-regulated genes ( Figure S1B). The pathway enrichment analysis of down-regulated genes confirmed that EWSR1-FLI1 and EGR2 regulate MVA pathway genes in EwS ( Figure S1C). This finding was further extended by the RNAseq analysis of the expression of MVA pathway enzymes ( Figure 1D, in blue) and prenyltransferases ( Figure 1D, in green) upon EWSR1-FLI1 knockdown (EWSR1-FLI1-KD) in six Ewing cell lines (A673, SKNMC, TC71, CHLA10, MHHES1, and EWS1), as well as in the A673/TR/shEF1 DOX-inducible system (shEF1) [18]. This analysis showed that 7 out 16 enzymes were consistently down-regulated upon EWSR1-FLI1-KD ( Figure 1D). Using a CRISPR/Cas9 approach, an EWSR1-FLI1 translocation was recently generated in primary Mesenchymal Stem Cells (MSCs) isolated from an Ewing patient [19]. As shown in Figure 1D, MVA pathway genes were up-regulated in EWIma1 cells as compared with the parental MSCs and were down-regulated in the EWlma1 clone upon EWSR1-FLI1-KD. ChIPseq experiments further documented a direct binding of EGR2 to the promoter regions of HMGCS1 and HMGCR, two essential genes of the MVA pathway, as well as to promoters of MVK and MVD ( Figure 1E, Table S1).
These findings were confirmed at the protein level, showing that EWSR1-FLI1 and EGR2-KD led to a dramatic decrease in the expression of enzymes of the MVA pathway in four different EwS cell lines ( Figure 2). Altogether, these data indicate that EWSR1-FLI1 regulates the expression of key mevalonate pathway enzymes via EGR2.

Ewing Sarcoma Cells Are Dependent on the Mevalonate Pathway
The observation that EWSR1-FLI1 and EGR2 regulate enzymes of the MVA pathway prompted us to investigate whether Ewing cells' growth may be dependent upon this pathway. We interrogated the DepMap portal [26], which reports the genome-wide screening of dependencies by CRISPR/Cas9 invalidation in 1054 cell lines corresponding to 27 different cancer types, including 25 different Ewing cell lines [27][28][29][30]. As shown in Figure 3A, we noticed that 9 out of 16 genes of the MVA pathway (including the 5 prenyltransferases) demonstrated median dependency scores lower than the threshold value of −0.5, suggesting that EwS cells are strongly dependent upon the MVA pathway. Moreover, in agreement with the MVA pathway being regulated by the EWSR1-FLI1/EGR2 axis, the global expression of 16 MVA pathway enzymes, and specifically the 2 rate-limiting genes HMGCR and HMGCS1, was significantly higher in Ewing cell lines as compared with other cancer cell lines ( Figures 3B and S2A).  Gene portal performed on the 280 genes down-regulated upon EGR2-KD with a log2FC < −0.5 comparing siEGR2 vs. siCTL. (C) Schematic representation of the mevalonate pathway. The blue boxes indicate the 2 rate-limiting enzymes. (D) Heat map representation of the differential expression of mevalonate pathway enzymes (in blue) and prenyltransferases (in green) in 6 Ewing cell lines after invalidation with siRNA targeting EWSR1-FLI1 (siEF1) for 48 h (A673 and SKNMC), 72 h (A673, TC71, CHLA10, and EW1), and 96 h (SKNMC, TC71, and MHHES1) and in the A673/TR/shEF1 inducible system (shEF1) treated with Dox for 7 days [24] and parental MSC cells (MSC_Pat_R1) as well as in EWIma1 cells invalidated or not for 72 h by siEF1. EWIma1 was generated by engineering an EWSR1-FLI1 translocation in a Mesenchymal Stem Cell (MSC_Pat_R1) [19]. The red box shows the 7 enzymes commonly up-regulated by EGR2 and EWSR1-FLI1. (E) CHIP-seq profiles for EWSR1-FLI1, EGR2, H3K27Ac, and H3K4me3 binding to HMGCS1, HMGCR, MVK, and MVD promotors in the POE shEGR2_4#22 clone [13] expressing a tagged EGR2 (EGR2HA). The endogenous EGR2 was invalidated by shRNA induced upon Dox treatment (+EGR2HA) or not (−EGR2HA). mean of two siEGR2 (#1 and #2) (absolute log2FC > 0.5) and Western blot analysis of EGR2 expression cells for 48 h after siEGR2 or siCTL transfection. The volcano plot shows the differentially expressed genes in A673 and SKNMC Ewing cells upon siEGR2 vs. siCTL transfection. The dotted line corresponds to the absolute value of log2FC = 0.5 and the red dots point out mevalonate pathway genes down-regulated by EGR2-KD. (B) Gene Ontology and pathway analysis using the Topp-Fun Gene portal performed on the 280 genes down-regulated upon EGR2-KD with a log2FC < −0.5 comparing siEGR2 vs. siCTL. (C) Schematic representation of the mevalonate pathway. The blue boxes indicate the 2 rate-limiting enzymes. (D) Heat map representation of the differential expression of mevalonate pathway enzymes (in blue) and prenyltransferases (in green) in 6 Ewing cell lines after invalidation with siRNA targeting EWSR1-FLI1 (siEF1) for 48 h (A673 and SKNMC), 72 h (A673, TC71, CHLA10, and EW1), and 96 h (SKNMC, TC71, and MHHES1) and in the A673/TR/shEF1 inducible system (shEF1) treated with Dox for 7 days [24] and parental MSC cells (MSC_Pat_R1) as well as in EWIma1 cells invalidated or not for 72 h by siEF1. EWIma1 was generated by engineering an EWSR1-FLI1 translocation in a Mesenchymal Stem Cell (MSC_Pat_R1) [19]. The red box shows the 7 enzymes commonly up-regulated by EGR2 and EWSR1-FLI1. (E) CHIP-seq profiles for EWSR1-FLI1, EGR2, H3K27Ac, and H3K4me3 binding to HMGCS1, HMGCR, MVK, and MVD promotors in the POE shEGR2_4#22 clone [13] expressing a tagged EGR2 (EGR2HA). The endogenous EGR2 was invalidated by shRNA induced upon Dox treatment (+EGR2HA) or not (−EGR2HA). In parallel to the investigation of EGR2 downstream signaling, we also performed a medium-throughput drug-repositioning screening using a set of more than 1200 off-patent drugs in a microscopy image-based assay. We used one dose of treatment (10 µM) and two time points (24 h and 72 h) to investigate the sensitivity of the A673 Ewing cell line to these compounds (Table S2). The selection criteria of cytotoxic drugs in A673 cells are described in detail in Figure S3. In total, 187 compounds were identified as positive hits from the screening as inducing a significant decrease in Ewing cell growth at 24 and/or 72 h of treatment. Among these, 110 compounds could be assigned to 10 functional and structural families ( Figure 3C, Table S2). As expected, given the sensitivity of Ewing cells to chemotherapy, we observed an important enrichment of drugs classified as "Oncology and Antineoplastic" therapeutic drugs (22%), including the inhibitors of topoisomerases class II used in the treatment of Ewing sarcoma, such as Doxorubicin and Etoposide, as well as compounds regulating DNA synthesis via the folate cycle (Amethopterin and Methotrexate) or purine synthesis (Table S2). The Enolid family, which includes regulators of the Na+/K+ pump, also showed up in this screening (Table S2). Interestingly, the overexpression of the ATP1A1 subunit (ATPase Na+/K+ transporting subunit alpha 1) of the Na+/K+ pump was recently reported as a downstream target of EWSR1-FLI1 [31]. Of particular interest was the observation that five out of the six tested statins, which are competitive inhibitors of HMGCR, a rate-limiting enzyme of the MVA pathway, demonstrated cytotoxicity in the A673 Ewing cell line at 72 h ( Figure 3C). The only statin that did not show up in this screening was the hydrophilic pravastatin, which is known to require an organic anion transporter (SLCO1B1) specifically expressed in hepatocytes but not in Ewing cells. Interestingly, the results from this screening are in full agreement with the DepMap database of pharmacological dependencies [32], whereby Ewing sarcoma cell lines demonstrated the highest sensitivity to Atorvastatin and Simvastatin ( Figure 3D) as compared with cell lines from other malignancies. This prompted us to further define the IC50 of three statins (Atorvastatin, Simvastatin, and Lovastatin) for four Ewing cell lines and four primary cultures from Ewing PDX, as well as four non-Ewing cell lines ( Figure 3E and Figure S2B, Table 1). All Ewing cell lines as well as primary cultures from PDX tumors exhibited an IC50 in the micromolar range. On the contrary, apart from the malignant rhabdoid tumor cell line KD, other control cell lines demonstrated much higher IC50 values, in the 10-100 micromolar range (Table 1). Lovastatin, which did not show up in the DepMap screening, also demonstrated growth inhibitory properties in our analysis of Ewing cells. Altogether, these results indicate that two completely independent strategies, one based on the functional analysis of EGR2, a critical target of EWSR1-FLI1, and the other based on an agnostic screening of Ewing cells' pharmacological vulnerabilities, point out the MVA pathway as a major dependency in Ewing sarcoma.

Atorvastatin Impacts Cell Proliferation of Ewing Cell Lines through Altered Anchorage on the Membrane of Small G-Proteins and Increased Intra-Cellular ROS
As all three tested statins give very similar results on Ewing cell lines, we selected Atorvastatin to assess its impact on cell-cycle progression and on apoptosis in three Ewing cell lines. No significant effects on the cell cycle were observed at the IC50. We therefore increased concentrations to 2× IC50, 1 µM for A673 cells, 2.6 µM for TC71, and 2.4 µM for POE. Again, only modest effects were observed: Atorvastatin treatment for 72 h had no significant effects on the different cell-cycle phases of A673 and TC71 cells and significantly reduced the proportion of S-phase cells for POE with a parallel increase in G0/G1 cells ( Figure 4A). Cancers 2022, 14, 2327 12 of 24   However, in all three cell lines, we observed that Atorvastatin induced a decreased EdU incorporation, suggesting that this compound may induce a reduced incorporation of nucleotides during the S-phase ( Figure 4B). To further explore this hypothesis, we performed EdU pulse-chase experiments and monitored S-phase progression at 0, 2, 4, 6, 8, and 10 h. We thus observed delayed S-phase entry and exit in Atorvastatin-treated cells ( Figure 4C,D and Figure S4A,B). To further confirm these results, we synchronized cells at the G1/S boundary with double thymidine treatment ( Figure S4C). After release of thymidine, we observed a delay of around 2 h in the progression into the S-phase of cells treated with Atorvastatin ( Figure S4C). Atorvastatin also induced a strong increase in p21 (CDKN1A) expression ( Figure 4E). These data strongly suggest that the inhibition of growth by Atorvastatin induces an inhibition of cell proliferation by acting on the DNA replication velocity. We also observed that Atorvastatin induced an increase in Annexin-V-positive cells, as well as increased levels of cleaved Caspase 3 and cleaved PARP forms, as compared with DMSO-treated cells ( Figure 4F). As the expected effect of statins is to inhibit HMGCR by competitive inhibition of HMG-CoA ( Figure 1C), we reasoned that statin inhibition of HMGCR should be rescued by supplementing the growth medium with downstream products of the MVA pathway. We hence evaluated the ability of mevalonate acid (MEV), geranylgeranyl pyrophosphate (GGpp), or farnesyl pyrophosphate (Fpp) to rescue the Atorvastatin-induced growth inhibition. Highly significant rescues of cell growth ( Figure 4G) and of p21 induction ( Figure 4E) were observed in three different Ewing cell lines with the exception of A673, the cell growth of which was not rescued by Fpp. The slowdown of S-phase progression by Atorvastatin was also rescued by MEV ( Figure 4H). Altogether, these data show that the inhibition of HMGCR mostly accounted for the statin-induced cell growth decrease. The products of the MVA pathway play important roles in protein prenylation, a post-translational modification essential for the biological activities of numerous proteins. Indeed, prenyltransferases catalyze the addition of isoprenoid, farnesyl, or geranylgeranyl to the C-terminus of proteins ( Figure 1C), thus creating a hydrophobic domain and enabling protein anchorage on the membrane. Small G-proteins, including the RAS, RHO, and RAC families, are dependent on prenylation for their localization to cell membranes and for their activity in cellular processes, including cell proliferation and migration (for a review, see [34,35]). To assess the impact of Atorvastatin on protein prenylation, we studied the expression of the prenylated forms of RHOA, RAS, and RAC1 by Western blot after extraction with the triton X-114 fractioning method. We observed that Atorvastatin induced a decrease in the membrane-bound forms of these proteins that can be rescued by MEV ( Figure 5A).
The MVA pathway is a major cellular redox regulator that acts on the biosynthesis of ubiquinone (CoQ10) implicated in the regulation of intra-mitochondrial reactive oxygen species (ROS) via the respiratory chain (for a review, see [36]) and on the maturation of glutathione peroxidase 4 (GPX4) by modulating lipid peroxidation and the ferroptosis process (for a review, see [37]) ( Figure 5B). Excessive ROS generation can result in DNA damage and programmed cell death, including apoptosis. In Ewing sarcoma, it has been demonstrated that the balance of intracellular ROS is tightly regulated [38][39][40]. To elucidate whether ROS generation may be involved in the cytotoxic effect of Atorvastatin on Ewing cell lines, we first quantified the intracellular ROS levels by flow cytometry and observed that Atorvastatin treatment induced a dose-dependent increase in intracellular ROS in three Ewing cell lines ( Figure 5C). We then showed that this increase can be rescued by MEV and GGpp but not by Fpp in A673 cells ( Figure 5D). As Fpp cannot rescue Atorvastatin-induced growth inhibition either ( Figure 4G), this suggests a correlation between increased ROS production and cell growth inhibition. To explore this correlation, we performed rescue experiments with antioxidant N-Acetyl Cysteine (NAC) ( Figure 5E). In agreement with increased ROS production being responsible for altered cell growth, NAC treatment both abolished the increased intra-cellular ROS ( Figure 5E, bottom) and significantly rescued the growth inhibition induced by Atorvastatin treatment (Figure 5E, top).  To investigate whether increased ROS levels may lead to DNA-damage-induced ROS, we analyzed the expression of γH2AX in A673 cells after 72 h of Atorvastatin treatment in the absence or presence of MEV, Fpp, and GGpp. Atorvastatin led to a dramatic increased expression of γH2AX that could be rescued by MEV and GGpp but not Fpp ( Figure 5F). Moreover, NAC treatment also allowed a partial reversion of γH2AX induction by Atorvastatin ( Figure 5G). The MVA pathway end product Isopentyl-pp is also essential for the maturation of GPX4, a key enzyme to limit the peroxidation of lipids and ferroptosis ( Figure 5B). We assessed this function with the bodipy fluorescent probe, which enables the detection of free-radical-mediated lipid peroxidation. We observed that Atorvastatin treatment induced a dose-dependent increase in lipid peroxidation in three Ewing cell lines ( Figure 5H) that was rescued by MEV and GGpp but, again, not by Fpp in A673 cells ( Figure 5I).
Altogether, these data show that statins induce pleiotropic effects on Ewing cells as a consequence of the inhibition of the MVA pathway, including altered protein prenylation, as well as increased intracellular ROS and lipid peroxidation.

Atorvastatin Induces a Reduction in Ewing Tumor Growth In Vivo
Finally, we assessed the impact of Atorvastatin on EwS growth using the 3D spheroid assay and Ewing PDX models. As shown below, using in vitro 2D assays ( Figures 3E and S2B), a dosedependent decreased growth of Ewing 3D spheroids was observed with cells from four Ewing PDX tumors upon Atorvastatin treatment ( Figure 6A). Moreover, when treating three Ewing sarcoma PDX models (IC-pPDX-87, IC-pPDX-3, and IC-pPDX-164) with Atorvastatin at doses (10 mg/kg) calculated to correspond to those used in humans to treat hypercholesterolemia [41], we observed a significant growth reduction of these tumors as compared with those of the control group ( Figure 6B

Discussion
In this manuscript, we show that the MVA pathway is a critical vulnerability of Ewing sarcoma as shown by two independent approaches. Firstly, we demonstrate that, via the regulation of Ewing-susceptibility gene EGR2, EWSR1-FLI1 up-regulates various genes encoding enzymes of the pathway. Secondly, we show that statins, which inhibit HMGCR, a key rate-limiting enzyme of the pathway, decrease cell proliferation and induce apoptosis of Ewing cells at much lower IC50 values than in other tumor types tested here or described in the literature. Indeed, extensive analyses of these publications report that IC50 values surpass 10 micromolar in most cancers, such as breast cancer [42], glioma [43], and pancreas cancer [44]. The sensitivity of Ewing cells to statins is in full agreement with data from large CRSPR/Cas9-based and pharmacological dependency screenings [26][27][28]32]. Some large retrospective studies or meta-analyses of cancer patients have reported a significant decrease in cancer incidence or an increased survival in statin-treated patients, whereas others have failed to detect the significant impact of statins on cancer evolution (reviewed in [45,46]). These studies suggest that different factors may account for the inconsistency in the published results, including variability of follow-up periods, heterogeneity of the investigated populations in terms of types of cancer and treatments, diversity of the statin compounds, dosage, and treatment duration, as well as many inherent biases of such retrospective or meta-analyses. While a number of biomarkers has been proposed to account for statin sensitivity, so far, none of these markers have been shown to be robust enough for patient stratification. There is a general agreement that the MVA pathway is an important tumor vulnerability and that statins constitute the most promising method to block this pathway and should be seriously considered within the anti-cancer drug arsenal. To evaluate this, 58 clinical trials are currently open to more precisely evaluate statins in cancer prevention or treatment (https://clinicaltrials.gov; accessed on 6 May 2022).
The present study therefore constitutes a biologically driven indication that the EWSR1-ETS fusion, which is pathognomonic of Ewing sarcoma, constitutes a strong predictive biomarker of statin sensitivity. Interestingly, some recent reports investigated bisphosphonates, another class of MVA pathway inhibitors in the treatment of bone cancer in children.
Bisphosphonates are widely used medications to treat bone resorption associated with osteoporosis or with primary or secondary bone tumors. They act by inhibiting the farnesyl diphosphate synthase (FDPS), an enzyme of the MVA pathway that is necessary for osteoclast ruffled-border formation. Interestingly, bisphosphonate and, particularly, Zoledronic acid have been shown to demonstrate anti-proliferative and anti-invasive properties in Ewing cells in vitro and in vivo [47][48][49][50]. Though it was not shown in these reports that zoledronic acid acted by inhibiting the MVA pathway, these observations converge with ours to point out the MVA pathway as an important dependency in Ewing sarcoma.
Our observation that the statin effect can be rescued by end products of the MVA pathway clearly points out that the inhibition of this pathway is the key effect of statin. It is interesting to mention that GGpp is generally more efficient than Fpp in rescuing the statin effect. This result is consistent with previous reports showing that the cell death induced by Lovastatin is rescued only by GGpp, suggesting that the genarylgeranylation of proteins is more essential than farnesylation for cell viability [51][52][53][54]. One hypothesis to account for the better rescue effect of GGpp as compared with Fpp is that isopentenyl-pp, an end product of the MVA pathway, is required for the conversion of Fpp to GGpp. Specific GGpp prenylations may thus not be achieved in the presence of Fpp alone. Reciprocally, it has been shown that GGpp can overcome the effect of farnesyl-transferase inhibitors for the prenylation of RAS [55,56]. These data therefore suggest that GGpp is more efficient in complementing Fpp deficiency than the reverse.
We further observed that the inhibition of the MVA pathway had pleiotropic effects on Ewing cells. We observed decreased cell proliferation characterized by a slowdown of the S-phase. Different mechanisms can contribute to this effect on the cell cycle. It may constitute a downstream effect of altered signal transduction. Indeed, GGpp and Fpp, two major downstream products of the MVA pathway, are critical for the isoprenylation of a variety of signal transduction molecules that require anchoring to the plasma membrane. We demonstrated such an altered anchorage for RAC1, RHOA, and RAS, but the anchorage of many other membrane receptors, including GPCRs and growth factor receptors that play a role in the transduction of proliferation signals, is also expected to be impaired. In this regard, it has been shown that RHOA is the direct geranylgeranylated effector signaling to the YAP/TAZ pathway [57], which plays important roles in the tumorigenesis of Ewing sarcoma (for a review, see [58]). The RHO GTPase-MVA pathway-YAP/TAZ axis has also been shown to be required for the proliferation and/or self-renewal of breast cancer and leukemia cells [59,60].
We also show that the cell-cycle effect is, at least in part, a consequence of increased ROS production. The MVA pathway is essential to the synthesis of coenzyme Q (CoQ10 or Ubiquinone), an essential electron carrier in the electron transport chain that contains a benzoquinone ring linked to a 10-isoprenoid-unit chain. It has recently been shown that statin treatment compromises oxidative phosphorylation and causes severe oxidative stress in mouse KPC cells and in human prostate cancer cells [61]. It has also been suggested, though not yet formally proven, that increased ROS levels may be a cause of statin-induced myopathy. We show that the statin-induced increased ROS level is a direct consequence of the altered MVA pathway, as it can be fully rescued by mevalonate acid or by GGpp. Moreover, ROS levels, increased H2AX, and proliferation defects can, at least in part, be rescued by the NAC antioxidant. Our results showing that statins induce a dramatic slowdown of the S-phase are therefore fully consistent with data demonstrating that oxidative stress alters the dynamics of DNA replication forks during the S-phase [62].
Decreased proliferation is not the sole effect of statin treatment on Ewing cells. Indeed, we observed a significant increase in apoptosis, assessed by increased Annexin V staining, cleaved Caspase 3, and PARP, upon treatment of Ewing cells with statins, in agreement with the many studies that described statins specifically triggering the apoptosis of tumors cells via the inhibition of the MVA pathway (for a review, see [45]).
Finally, and very importantly, our data also demonstrate that, as a single agent, statin treatment significantly inhibited tumor growth in three different PDX Ewing models at doses similar to those used in humans to treat hypercholesterolemia.
Altogether, our data show that the activation of the mevalonate pathway in Ewing sarcoma is a direct consequence of the activation of EGR2 by EWSR1-FLI1 and that this pathway can be efficiently targeted by statins. These FDA-approved, well-tolerated, and inexpensive compounds can hence be added to the therapeutic arsenal used in the treatment of EwS, a deadly disease for which new therapeutic approaches are urgently needed. Phase I dose-escalation studies have shown that statins are well tolerated at doses much higher than typically prescribed for cholesterol management, at least for defined periods of time, suggesting that dosage optimization could be searched for in the treatment of EwS. Even more importantly, new combination therapies can now be considered. The observation that statins induce ROS and DNA damage, as shown by increased H2AX, together with recent reports pointing out a high level of replication stress in Ewing cells [63], may stimulate research on combinations with other DNA-damaging agents or DNA-repair-inhibiting drugs [64]. In this respect, it is interesting to mention that EWSR1-FLI1 was found as a strong predictive biomarker of in vitro sensitivity to PARP inhibitors, suggesting that the combo of statins with these inhibitors may constitute an attractive therapeutic strategy.

Conclusions
We show that the mevalonate pathway is essential in Ewing sarcoma and could be a potential new therapeutic target. In fact, the EWSR1-FLI1 oncogenic protein regulates the expression level of MVA pathway proteins by its direct effect on the EGR2 protein. The statins, which inhibit HMGCR, a key rate-limiting enzyme of the MVA pathway, decrease cell proliferation and induce the apoptosis of Ewing cells in vitro and the tumor growth of PDX Ewing models.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/cancers14092327/s1, Figure S1: Functional analysis of genes regulated by EWSR1-FLI1 and EGR2, Figure S2: Expression of HMGCS1 and HMGCR, and sensitivity of cell lines to Simvastatin and Lovastatin, FigureS3: Workflow of the drug screen, Figure S4: Atorvastatin induces the slowdown of S-phase of cell cycle, Table S1: Prediction of EGR2 binding sites, Table S2: List of compounds significantly impacting Ewing cell growth. File S1: Original Western Blot figures.

Institutional Review Board Statement:
This study was performed in accordance with the recommendations of the European Community (2010/63/UE) for the care and use of laboratory animals. Experimental procedures were approved by the ethics committee of Institut Curie CEEA-IC #118 (Authorization APAFIS#11206-2017090816044613-v2 given by National Authority) in compliance with the international guidelines.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
All data associated with this study are present in the paper or in the Supplementary Materials.