EWS::FLI1-DHX9 interaction promotes Ewing sarcoma sensitivity to DNA topoisomerase 1 poisons by altering R-loop metabolism

Abstract

Drug resistance is an ill-defined cause of dismal outcomes in cancer. Ewing sarcoma (EwS), a pediatric cancer characterized by high therapy failure rates, is driven by a single oncogenic event generating EWSR1::ETS gene fusions (primarily EWSR1::FLI1) in a silent genomic background. This provides a straightforward model to study the impact of gene fusions on drug responses. Here, we describe a novel mechanism of sensitivity to DNA topoisomerase 1 poisons in EwS. We discovered that EWS::FLI1 prevents the resolution of R-loops induced by these drugs via sequestering DHX9 helicase, ultimately resulting in R-loop accumulation, replication stress, and genome instability. In turn, excessive DHX9 or reduced EWS::FLI1 levels render EwS cells resistant to the active metabolite of irinotecan (SN-38) independent of proliferation and global transcription rates. This resistance helps explain how elevated DHX9 levels predict worse clinical outcomes. Overall, our research demonstrates the impact of a dominant mutation on cancer drug sensitivity, highlighting its significant clinical implications.

Fig. 1. EwS cell lines and tumors are highly sensitive to TOP1 poisons.

A Comparative analysis of drug activity between 15-17 EwS (carrying the EWSR1::FLI1 fusion oncogene) and more than 600 non-EwS cell lines using the Genomics of Drug Sensitivity in Cancer website. Each dot is a tested drug. B Evaluation of SN-38 sensitivity between EwS (A4573, A-673, TC-71, and EW-7) and non-EwS cell lines (U2OS (osteosarcoma), SK-UT-1 (leiomyosarcoma), HT-1080 (fibrosarcoma), and 93T449 (liposarcoma)). After 72 h of treatment with indicated concentrations, cell viability was analyzed by MTT. Data represent the mean (+SEM) of the percentage of survival (relative to DMSO); n = 2 independent experiments. C Evaluation of tumor growth of PDX models (EwS: IEC-073, HSJD-ES-002, HSJD-ES-006; osteosarcoma, IEC-036; UPS, IEC-056) over 21 days following the beginning of irinotecan treatment. The dotted line indicates the end of treatment. Data represent the mean (+SEM) of tumor volume. D Evaluation of tumor growth inhibition of PDX models at the end of irinotecan treatment. Data represent the mean of the percentage of tumor growth inhibition. E Analysis of SN-38-induced DSBs comparing EwS (A4573, A-673, TC-71 and EW-7) and non-EwS cell lines (U2OS, SK-UT-1, HT-1080 and 93T449) by γH2AX IF. Cells were treated with 5 µM of SN-38 (30 min). Left, representative images. DAPI counterstain. Scale bar, 20 µm. Right, data represent the mean (±SEM) of nuclear γH2AX intensity; n ≥ 2 independent experiments (~75 cells were analyzed per replicate). F Similar to (E) in S-phase population (determined by EdU incorporation). Cells were incubated with 10 µM of EdU (20 min), washed and treated with 5 µM SN-38 (30 min). Data represent the mean (±SEM) of nuclear γH2AX intensity in EdU-positive cells (a total of ~100 cells were analyzed). G Determination of transcriptional rates between EwS and non-EwS cell lines by EU incorporation. Cells were incubated with 0.5 mM of EU for 45 min. Data represent the mean (±SEM) of nuclear EU intensity; n ≥ 2 independent experiments (~100 cells were analyzed per replicate). H Study of HR efficiency by SCEs assay in EwS, non-EwS, and BRCA1-mutated breast cancer cell lines (MDA-MB-436 and HCC-1937). Cells were treated with 2.5 µM etoposide for 30 min. Left, representative image. Scale bar, 5 µm. Asterisks indicate SCEs events. Right, data represent the mean (±SEM) of the frequency of SCEs per metaphase; n = 3 independent experiments (~20 metaphases were analyzed per replicate). P-value was determined by t-test.

Fig. 2. EWS::FLI1 impairs drug-induced R-loop resolution promoting genome instability and cytotoxicity.

A Evaluation of the association between CPT and B SN-38 activity (AUC), and EWSR1::FLI1 expression levels in EwS cell lines (n = 16). FLI1 gene was used as a surrogate marker of EWSR1::FLI1, since FLI1 is not expressed in EwS cells. Lineal correlation was determined by Pearson correlation coefficient. C EWS::FLI1 and LOX protein levels by WB in shA673 cells after DOX incubation. Loading control: GAPDH. Molecular weight in kDa. D SN-38-induced apoptosis upon EWSR1::FLI1 knockdown by PARP1 and CASP3 WB. Cells were pre-incubated with DOX for 24 h before 5 µM SN-38 treatment (3 h). After washout, cells were cultured in drug-DOX-free medium for 24 h. Loading control: GAPDH. Molecular weight in kDa. E Similar to (D) by AnnexinV FACS. Data represent the mean (+SEM) of the percentage of AnnexinV-positive cells, n = 5 independent experiments. F Similar to (D) by MTT assay. Cells were treated with indicated concentrations of SN-38 (3 h). After treatment, cells were cultured in drug-DOX-free medium for 48 h. Data represent the mean (+SEM) of the percentage of survival (relative to DMSO), n = 5 independent experiments. Dotted lines indicate IC50. G SN-38-induced γH2AX upon EWSR1::FLI1 knockdown by IF. Cells were pre-incubated with DOX for 24 h and treated with 5 µM SN-38 (30 min). Left, representative images. DAPI counterstain. Scale bar, 20 µm. Right, data represent the mean (±SEM) of nuclear γH2AX intensity, n = 5 independent experiments (~100 cells were analyzed per replicate). H SN-38-induced chromosomal breaks upon EWSR1::FLI1 downregulation. After DOX incubation, cells were treated with 2.5 µM SN-38 (30 min). Left, representative image. Scale bar, 10 µm. The arrow indicates a chromosomal break. Right, data represent the mean (±SEM) of the frequency of chromosomal breaks per chromosome, n = 3 independent experiments (20 metaphases were analyzed per replicate). I EWS::FLI1 protein levels in HeLa EF cells after 72 h of DOX incubation. Loading control: GAPDH. Molecular weight in kDa. J SN-38-induced γH2AX upon EWSR1::FLI1 overexpression by IF. HeLa EF cells were pre-incubated with DOX for 72 h and treated with 5 µM SN-38 (30 min). Other details as in (G), n = 3 independent experiments (~100 cells were analyzed per replicate). K SN-38-induced R-loops upon EWSR1::FLI1 knockdown by s9.6 IF. Cells were pre-incubated with DOX and treated with 5 µM SN-38 (30 min). Left, representative images. DAPI counterstain. Scale bar, 20 µm. Right, data represent the mean (±SEM) of nuclear s9.6 intensity, n = 3 independent experiments (~75 cells were analyzed per replicate). L Effect of RNH1 overexpression on SN-38-induced DSBs. Cells were transfected with RNH1:GFP or control plasmids 24 h before treatment with 5 µM SN-38 (30 min). Left, representative images. DAPI counterstain. Scale bar, 20 µm. Right, data represent the mean (±SEM) of nuclear γH2AX intensity in GFP-positive cells, n = 3 independent experiments (~50 cells were analyzed per replicate). P-value was determined by t-test.

Fig. 3. Loss of EWS::FLI1-DHX9 interaction alleviates R-loops accumulation promoting drug resistance.

A Kaplan–Meier showing overall survival of 166 EwS patient samples stratified by DHX9 mRNA levels. Statistical significance was determined by the Mantel–Cox test. B Determination of R-loops-DHX9 interactions by PLA in shA673 cells upon DOX incubation, and C DOX incubation and 5 µM SN-38 treatment (30 min). Left, representative images (red, PLA foci; blue, DAPI counterstain). Scale bar, 20 µm. Right, data represent the mean (±SEM) of PLA foci, n = 3 independent experiments (40 (B) and 20 (C) cells were analyzed per replicate). D Evaluation of the effect of DHX9 downregulation on SN-38-induced DSBs upon EWS::FLI1 depletion by γH2AX IF. After 48 h of transfection with indicated siRNAs and/or 24 h of incubation with DOX, shA673 cells were treated with 5 µM SN-38 (30 min). Data represent the mean (±SEM) of nuclear γH2AX signal; n = 3 independent experiments (150 cells were analyzed per replicate). E Evaluation of the effect of DHX9 overexpression on SN-38-induced DSBs by γH2AX IF. A-673 and TC-71 cells overexpressing DHX9:GFP or control plasmids were treated with 5 µM SN-38 (30 min). Left, representative images. DAPI counterstain. Scale bar, 20 µm. Right, data represent the mean (±SEM) of nuclear γH2AX signal of GFP-positive cells, n ≥ 3 independent experiments (30 cells were analyzed per replicate). F Effect of DHX9 overexpression on SN-38-induced apoptosis by AnnexinV FACS. Cells overexpressing DHX9:GFP or control plasmids were treated with 5 µM SN-38 (3 h). After washout, cells were maintained in drug-free medium for 24 h. Data represent the mean (+SEM) of the percentage of AnnexinV-positive between GFP-positive cells (relative to SN-38-), n ≥ 3 independent experiments. G Evaluation of the association between SN-38 activity (AUC) and DHX9 expression levels in EwS cell lines (n = 16). Lineal correlation was determined by the Pearson correlation coefficient. H Analysis of the effect of YK-4-279 on SN-38-induced DSBs by γH2AX IF. A-673 cells were pre-incubated with 75 µM YK-4-279 (1.5 h) and treated with 5 µM SN-38 (30 min). Left, representative images. DAPI counterstain. Scale bar, 50 µm. Right, data represent the mean (±SEM) of nuclear γH2AX intensity, n = 3 independent experiments (30 cells were analyzed per replicate). P-value was determined by t-test.

Fig. 4. EWS::FLI1-DHX9 associated R-loop accumulation is a source of replication stress.

A Analysis of the effect of RNH1 overexpression in SN-38-induced replication stress by pCHK1 (Ser345) IF. Cells were transfected with RNH1:GFP or control plasmids 24 h before 5 µM SN-38 treatment (30 min). Left, representative images. DAPI counterstain. Scale bar, 20 µm. Right, data represent the mean (±SEM) of pCHK1 intensity in GFP-positive cells, n ≥ 2 independent experiments (30 cells were analyzed per replicate). B Effect of EWSR1::FLI1 knockdown in SN-38-induced replication stress by pCHK1 (Ser345) WB. shA673 cells were pre-incubated with DOX and treated with 5 µM SN-38 (30 min). Loading control: GAPDH. Molecular weight in kDa. Asterisk indicates non-specific bands. Bottom, quantification of CHK1 phosphorylation (pCHK1/CHK1 band signal). Data represent the mean of 2 independent experiments (relative to SN-38). C Similar to (B), by DNA fiber assay. Cells were incubated with CidU (30 min) and IdU + 5 µM SN-38 (30 min). D Quantification of (C). Data represent the mean (±SEM) of IdU/CidU fiber length ratio, n = 2 independent experiments (more than 250 fibers were analyzed per condition). E Effect of DHX9 overexpression on SN-38-induced replication stress by pCHK1 (Ser345) WB. A-673 and TC-71 cells overexpressing DHX9:GFP or control plasmids were treated with 5 µM SN-38 (30 min). Bottom, quantification of CHK1 phosphorylation (pCHK1/CHK1 band signal). Data are the mean of 2 independent experiments (relative to SN-38 GFP). Other details as in (B). F Effect of YK-4-279 treatment on SN-38-induced replication stress by pCHK1 (Ser345) WB. Cells were pre-incubated with 75 µM YK-4-279 (1.5 h) and treated with 5 µM SN-38 (30 min). Other details as in (B). G Evaluation of the effect of ATR inhibition in SN-38-induced pCHK1 and γH2AX by WB. A-673 and TC-71 cells were pre-incubated with 10 µM AZD6738 for 24 h and treated with 5 µM SN-38 (30 min). Other details as in (B). H Analysis of the effect of ATR inhibition in SN-38-induced chromosomal breaks. Cells were pre-incubated with 10 µM AZD6738 for 24 h and treated with 2.5 µM SN-38 (A-673) or 1.25 µM SN-38 (TC-71) for 30 min. Data represent the mean (±SEM) of the frequency of chromosomal breaks per chromosome, n ≥ 2 independent experiments (20 metaphases were analyzed per replicate). Statistical significance was determined by t-test. I Left, synergy plot representing synergistic effect of AZD6738 and SN-38 combination in A-673 cells. Right, effect of ATR inhibition in cell survival upon SN-38 exposure by MTT assay. Cells were incubated with AZD6738 (250 nM, 24 h) previous to the treatment with indicated concentrations of SN-38 (72 h). Data represent the mean (+SEM) of the percentage of survival, n = 3 independent experiments. Dotted lines indicate IC50. J Similar to (I) in TC-71 cell line. Statistical significance was determined by t-test.

Fig. 5. EWS::FLI1 kidnaps DHX9 into the RNAPII transcriptional complex promoting R-loops formation.

A Analysis of the interaction between DHX9 and RNAPII-Ser2p by RNAPII-Ser2p pull-down. Molecular weight in kDa. Asterisk indicates non-specific bands. B Evaluation of the effect of EWSR1::FLI1 knockdown in DHX9-RNAPII-Ser2p interaction. shA673 cells overexpressing DHX9:GFP or control plasmids were incubated with DOX for 24 h previous to GFP pull-down. Representative immunoblots. Molecular weight in kDa. Data represent the mean (±SD) of the RNAPII-Ser2p band signal, normalized to DHX9:GFP pull-down (relative to DOX-); n = 3 independent experiments. C Evaluation of chromatin distribution of DHX9 by ChIP upon EWSR1::FLI1 knockdown. shA673 cells were treated with DOX for 24 h. Upper, diagram of ACTB and ACTG1 genes; and primers used for qPCR (red lines). Bottom, data represent the mean (+SEM) of DHX9 ChIP signal (normalized to DOX- signal at 3′); n = 3 independent experiments. D Evaluation of the effect of EWSR1::FLI1 overexpression on R-loops levels by s9.6 IF. HeLa EF cells were incubated with DOX for 72 h. Left, representative images. DAPI counterstain. Scale bar, 20 µm. Right, data represent the mean (±SEM) of nuclear R-loops; n = 5 independent experiments (50 cells were analyzed per replicate). E Evaluation of the effect of EWSR1::FLI1 downregulation on R-loops levels by slot blot assay. shA673 cells were incubated with DOX for 24 h. Left, representative immunoblots. Loading control: methylene blue staining. Right, data represent the mean (+SEM) of s9.6 signal (normalized to methylene blue and relative to DOX-), n = 4 independent experiments. F Similar to (D) in A-673 cells after treatment with 25 µM YK-4-279 for 4 h; n = 3 independent experiments. G Similar to (E) in A-673 cells after treatment with 75 µM YK-4-279 for 1.5 h; n = 3 independent experiments. H Similar to (D) in A-673 cells 48 h after transfection with indicated siRNAs; n = 4 independent experiments. I Similar to (E) in A-673 cells 48 h after transfection with indicated siRNAs; n = 3 independent experiments. Statistical significance was determined by t-test. J Analysis of the correlation between R-loops and DHX9 protein levels in a cohort of 59 EwS tumors. Scale bar, 50 µm. Correlation was determined by the Pearson correlation coefficient.

Fig. 6. MODEL.

Under basal conditions, EWS::FLI1 promotes the interaction between DHX9 and the elongating RNAPII leading to the formation of R-loops. In addition, TOP1 poisoning provokes the accumulation of R-loops through the blockage of RNAPII transcription. In non-EwS cells, DHX9 resolves drug-induced R-loops, preventing genome instability. In EwS cells, EWS::FLI1 “hijacks” DHX9 into the RNAPII transcriptional complex, reducing DHX9 recruitment to the R-loops and R-loops resolution. Replication machinery collapse with unresolved R-loops promoting replication stress, genome instability, and cell death.