Ewing sarcoma (ES) is an aggressive bone and soft tissue neoplasia that mainly affects early ages between childhood and young adulthood. ES arises from a distinct molecular background that is characterized by the presence of recurrent reciprocal chromosomal translocations in a percentage as high as 85-90% of cases (1). Specifically, a driver fusion gene is generated after the binding of the Ewing sarcoma breakpoint region 1 (EWSR1) gene to the DNA-binding domain of a transcription factor gene belonging to the erythroblast transformation-specific (ETS) family. Although five different ETS factors that fuse with EWSR1 have been reported in Ewing sarcoma, Friend leukemia virus-induced erythroleukemia-1 (FLI1) is the most common partner, contributing to malignant transformation in approximately 95% of ES cases (2). The oncogenic EWS/FLI1 chimeric protein is a key tumor-initiating transcriptional mediator of Ewing sarcoma pathogenesis and disease progression.

The generation of the oncogenic EWSR1/FLI1 fusion gene, which is recognized as the hallmark of ES, can be attributed to the simultaneous introduction of two DNA double-strand breaks (DSBs) at distinct loci on two nonhomologous chromosomes and the subsequent joining of the newly exposed DNA ends (3). Under normal circumstances, these duplex breaks are repaired through the induction of intricate DNA repair mechanisms; however, a compromised DSB repair function can lead to genomic rearrangements, including the t(11;22)(q24;q12) chromosomal translocation in ES (4). There are generally two major repair pathways that can restore these DSBs: nonhomologous end joining (NHEJ) and homologous recombination (HR). On one hand, NHEJ is active throughout the cell cycle, and it ligates DSBs without requiring extensive sequence homology to join the DNA ends at the repair site, resulting in a flexible but error-prone repair (5). On the other hand, HR is a more accurate process that requires the use of homologous sequences to align DSB ends prior to ligation; this homology-directed pathway is activated during the S/G2 phase, where a homologous donor (the sister chromatid) is physically in proximity (6). Errors in the process that the cell uses to repair a break may cause genomic instability and chromosomal rearrangements that might eventually induce cell death (7).

Recent evidence suggests an excess of deleterious germline variants in patients with translocation-associated sarcomas, including the pathogenic variation of genes involved in DNA repair mechanisms, such as BRCA1 and BRCA2, which are key members of the HR pathway (8,9). In a broader context, homologous recombination seems to be functionally suppressed in ES since expression of the EWS/FLI1 genomic fusion leads to the blockage of BRCA1 repair and the decreased survival of ES cell lines (10). Additional quantitative PCR studies have confirmed the downregulation of other DNA-repair genes beyond BRCA1, such as GEN1, and ATM, while higher levels of PARP1 have been identified repeatedly (11). These findings harbor important clinical implications since tumors with mutations in DSB repair factors might respond to available targeted therapies (e.g., poly ADP-ribose polymerase inhibitors) (12).

The increased sensitivity of ES cell lines to PARP inhibitors was first revealed following a large-scale screen of several hundred cancer cell lines (13). Despite the somewhat limited success of Olaparib as a single agent therapy in clinical trials to date (14), the aforementioned efforts introduced a novel pharmacogenomic approach that highlights the importance of elucidating impaired DNA repair functions to guide clinical decisions in ES. In this context, the aim of this work was to study the expression of DNA repair genes involved in the NHEJ and HR pathways, in a series of ES patients, in an effort to provide a molecular basis for further deciphering ES pathogenesis and tumor biology.

Herein, we have analyzed the expression levels of 6 genes involved in NHEJ (XRCC4, XRCC5, XRCC6, POLλ, POLμ) and 9 genes involved in HR (RAD51, RAD52, RAD54, BRCA1, BRCA2, FANCC, FANCD, DNTM1, BRIT1) using real time PCR, in 36 pediatric ES samples.

Histopathology. Fresh frozen tissue from biopsy or tumor excision was obtained from 35 patients. Hematoxylin and eosin (H&E) staining was accompanied by immunohistochemistry. ES diagnosis was based on both immunopathologic characteristics and further confirmed by the detection of the characteristic gene fusion. Positive immunohistochemistry with CD99 (clone 12E7, DAKO and clone O-13, Invitrogen, 1;100) and Fli-1 (Santa Cruz) were used as diagnostic histopathological markers. Furthermore, ki67 expression (clone M7240, DAKO, Agilent, 1:100) as well as necrosis (as a percentage of necrotic/ischemic non-viable tissue detected) were estimated.

RNA extraction. Total RNA was extracted from fresh or formalin-fixed paraffin embedded tissue sections (FFPE) using the PureLink TotalRNA and PureLink FFPE Total RNA Isolation Kit (Invitrogen, Thermo Scientific, Waltham, MA, USA). Total RNA (10μl) was reverse-transcribed in 20 μl reactions using the SuperScript II First-strand Synthesis Super Mix (Invitrogen, Thermo Scientific) and random hexamers as primers.

EWS/FLI1 detection. For the detection of EWSR1-FLI1 fusion gene the cDNA was subjected to TaqMan Real-time RT-PCR amplification. All PCR amplifications were performed in a 20 μl reaction mixture, containing 1× Platinum Quantitative PCR Super Mix (Invitrogen, Thermo Scientific), 10 pmol of each primer and 5 pmol of TaqMan probes. The presence of EWSR1-ERG and other EWSR1-FLI1 variants was assessed by conventional RT-PRCR using Platinum Taq DNA polymerase (Thermo Scientific). Primers and PCR conditions are described in Supplementary material.

Real time PCR. The relative quantification of the XRCC4, XRCC5, XRCC6, POLλ, POLμ, RAD51, RAD52, RAD54, BRCA1, BRCA2, FANCC, FANCD, DNTM1 and BRIT1 genes was assessed by real time PCR using the beta2-microglobulin as reference in 25 available tissue samples. All PCR amplifications were performed in 20 μl reaction mixture, containing 1× Platinum SYBR Green qPCR SuperMix-UDG (Invitrogen, Thermo Scientific), and 10 pmol of each primer. Primers and PCR conditions are described in Supplementary material.

Statistical analysis. Age, sex, location of primary tumor, tumor size, ki67, mitotic count, invasion of adjacent tissues and treatment were analyzed. Overall survival (OS) was calculated from the time of diagnosis until death, loss of follow up or data cut-off. Patient characteristics are presented in Table I.

For statistical analysis, continuous log-scaled data was used to identify significant associations between markers and treatment response in terms of hazard ratio (HR) in Cox models. Benjamini-Hochberg false discovery adjustment method was performed, (adjusted p-Value=0.05) considering the number of comparisons performed. All hypothesis testing was performed at a two-sided significance level of a=0.05. Statistical analysis of clinicopathological and genetic data was performed using SPSS2 (IBM, Armonk, NY, USA).

Survival analysis was performed using the R packages survival and survminer (15). To create the Kaplan-Meier plot representing the association of XRCC4 expression with survival, samples were categorized as either “high” or “low” by using the optimal cutpoint analysis R package survimer. Reported p-values in the plot are associations of the model (log-rank test). The software used for visualization of the data and statistical purposes was R studio Version 1.4.1717.

ES is characterized by the pathognomonic fusion of the EWSR1 gene with several other partner genes, most commonly with Fli1 (16). This cytogenetic translocation demands the formation of a DSB and raises the question whether DSB repair pathways are malfunctioning. We analyzed the expression profile of NHEJ and HR, which are the main DSB repair pathways in ES samples from 35 patients, using real time PCR. Our results are indicative of damage in both NHEJ and HR repair pathways.

Statistically significant differences in expression of DSB genes were found in 4 of the 6 genes involved in NHEJ. XRCC55 was upregulated (p=0.02), polλ was downregulated (p=0.001), POLμ was downregulated (p=0.001) and LIG4 was found to be upregulated (p=0.001). Our results clearly demonstrate a damaged profile regarding the expression of the 6 genes involved in NHEJ.

We have studied 9 genes of the HR repair pathway, and we detected statistically significant discrepancies in the expression of 5 of them. RAD51 and RAD52 were found to be upregulated (p<0.001 for both). RAD54 was found to be downregulated (p=0.03), while BRCA2 was deregulated (p=0.02), being either overexpressed or downregulated in the majority of cases. FANCD was found to be clearly overexpressed (p=0.003). On the other hand, BRCA1 and FANCC were found to show a non-statistically significant trend (p=0.06 and 0.052, respectively) regarding their dysregulation. Expression patterns of the significantly differentially expressed genes tested are shown on Figure 1.

When we tried to study the correlation of gene expression discrepancies with clinicopathological data, there was no statistically significant result. We did not detect any statistically significant associations between treatment modalities and gene expression data. On the other hand, Xrcc4 low expression was related with statistical significance (p=032) with longer OS probability (Figure 2).

Ewing sarcoma is characterized by the fusion of the EWSR1 gene with several partners (17). The formation of these fusion genes requires a DSB which is not repaired. Recently whole genome or whole exome analysis of 175 ES samples revealed with a significant enrichment of mutations or variants in DNA damage repair genes (18). Another sarcoma genetic study which analyzed 430 ES samples detected a strong association between translocation associated sarcomas and mutations in Fanconi anemia genes, especially Fancc (19). In ovarian cancer samples studied with whole exome sequencing several variants of DNA damage response genes were detected, supporting that DNA repair deregulation is a crucial mechanism of carcinogenesis (20). Our expression study, even with the caveats of a small cohort analyzed with real time PCR, highlights that both NHEJ and HR DNA repair pathways are deregulated in ES.

EWS protein has been shown to regulate DNA damage alternative splicing (21). Furthermore, molecular cytogenetic studies with aCGH and expression analysis have revealed genes involved in DNA repair pathway with copy number alterations to be associated with poor prognosis (22). Especially for EWS-FLI1, Gorthi et al. have shown that this fusion gene blocks BRCA1 repair (23). BRCA1 expression in our analysis showed a non-significant decrease (p=006) (23).

Four out of six NHEJ genes tested in our population showed significantly altered expression. Xrcc5 expression has been associated with resistance to Doxorubicin in chondrosarcoma cells (24). However, in our analysis detection of Xrcc5 overexpression, cannot be associated with resistance to any of the regimens received by our patients; either Doxorubicin alone or the combination of vincristine, cyclophosphamide, doxorubicin/ifosfamide, and etoposide treatment. LIG4 loss of heterozygosity (LOH) has been shown to promote murine tumorigenesis and especially the development of sarcomas. The overexpression of lig4 found in our population cannot be associated with LOH, though it is important evidence of NHEJ deregulation. Deletion of polμ has been shown to be associated with sarcoma development in p53-/- mice (25). Our data highlight the importance of POLμ downregulation in ES tumor samples. POLλ is a key element of the repair mechanism in DSBs, and our data indicate that its low expression may result in DNA repair deficiency in ES.

Interestingly, XRCC4 expression was significantly associated (p=0.044) with OS probability (Table II). Patients with high expression of XRCC4 had significantly worse survival compared to those with low expression (Figure 1). We are the first to report a connection between XRCC4 expression levels and survival in ES. This finding cannot be associated with staging, since our population included 24 patients with localized disease and only 1 patient with de novo metastatic ES. XRCC4 has been reported to be a downstream target of SYT/SSX fusion in synovial sarcoma (26). In hypopharyngeal cancer, patients with low expression of XRCC4 and KU70 had a better locoregional control, when treated with chemoradiotherapy (27). Furthermore, in patients with cervical cancer who received preoperative chemoradiotherapy and underwent hysterectomy, low XRCC4 expression levels were related to pathologic complete responses (28). Similar data were reported in esophageal cancer, where patient with low expression of XRCC4 had a significantly better OS compared to those with high expression after chemoradiotherapy (29). XRCC4 high expression was reported to be a negative prognostic marker for oral squamous cell carcinoma (30). In breast cancer high XRCC4 expression has been associated with ipsilateral breast tumor recurrence (31). In vitro data from triple negative breast cancer cells show that low expression of XRCC4 sensitizes cells to radiotherapy and improves survival (32). XRCC4 variants are associated with colorectal cancer in a Taiwanese study analyzing 370 healthy donors paired with 370 colorectal cancer patients (33).

The interplay between DNA replication and cancer development has been thoroughly discussed (34). In our analysis, the HR repair mechanism was found to be deregulated with five out of the nine genes studied showing significant altered expression. Furthermore, two more genes (BRCA1 and FANCC) showed a statistical trend regarding their expression patterns. RAD51 and RAD52 overexpression has been published from studies in the sarcoma HT-1080 cell line (35). PARP and ATR inhibition have blocked RAD51 to form HR repair in ES cells (36). Our data regarding the overexpression of these two genes highlight the important role they play in HR of ES samples. RAD54 downregulation has not been studied in sarcomas and our finding indicates the crucial role of this gene in HR. BRCA2 low expression described in our population is clearly in agreement with several publications showing that BRCA2 deletion is associated with HR deficiency, but also with response to PARP inhibition even in case reports of several sarcomas treated with PARP inhibition (37). BRCA1 in our study was found downregulated highlighting a statistical trend (p=0.006) regarding the expression profile of this gene in ES patients. This finding supports functional studies that showed EWS/FLI1 to block BRCA1 repair in ES cells (23). Fanconi anemia genes have not been clearly associated with sarcoma tumorigenesis and our data show the synergistic effect of the HR repair pathway.

DNA repair deficiency offers an interesting therapeutic target with PARP inhibitors being the most studied regimen (38-40). However, ATR and DNA-PK have also been tested in ES (36). Both in cell lines and xenograft models, olaparib enhanced the activity of trabectedin (41). Additionally, PARP inhibition with temozolomide might be a useful combination for ES cells (42). The patients included have not been treated with PARP inhibitors. Furthermore, the population tested has not been treated with any platinum regimens, so there is no indirect indication regarding the potential sensitivity to PARP inhibition due to DNA repair deficiency.

Our study indicates that both NEHJ and HR DSB repair pathways are deregulated in ES. XRCC4 low expression is related to worse survival in ES patients. Expression of BRCA1 gene was low showing a statistical trend regarding this gene’s expression profile in our small series. The significance of these findings needs to be verified in larger studies. The retrospective character and the small number of patients included in this analysis are important caveats of this study. Prospective clinical trials with agents targeting DNA repair pathways, like with PARP inhibition, are required to validate these initial observations.

Available at: https://figshare.com/articles/journal_contribution/Supplementary_material_Table1_and_Table_2/25330393

All Authors have no conflicts of interest for the entire content of this manuscript.

AK was the writer of the article. GS and MM were the two independent investigators, who performed statistical analysis. Expression analysis experiments were performed by VK, LM, ED. Clinical data acquisition was done by AK, EZ, NT, AP, KS, VT, MMP, SP and ANK. Histopathology analysis was performed by VT and KS. ANK and AK contributed to conception and design of the study. Manuscript editing and the revision were performed by ANK and AK. All Authors have read and approved the final manuscript.