Fibroblasts from patients were obtained from the INSERM UMR1052 COPERNIC cell collection (16). This collection was approved by the regional Ethical Committee (CPP Sud-Est, Lyon, France) and cell lines were declared under the numbers DC2008-585 and DC2011-1437 to the Ministry of Research. The database derived from the COPERNIC collection is protected under the reference IDDN.FR.001.510017.000.D.P.2014.000.10300.

All the anonymous patients were informed and signed consent according to the ethics recommendations. This collection is composed of cancer patients presenting with overreactions in normal tissues after radiotherapy. Severity of side-effects was graded for each patient according to the Common Terminology Criteria for Adverse Events scale, version 4.03 (17). Sampling was performed in non-irradiated, non-photo-exposed anatomical region after local anesthesia. Standardized dermatological punch and untransformed fibroblast cell strains were prepared from skin biopsies. In the present study, 16 breast cancer patient cells were studied, comprising eight cell strains from grade 2 patients, here referenced as P1 to P8, and eight cell strains from grade 3 patients (P9 to P16). Cells were studied between passage 7 to 10 in culture (mean population doublings: 35 to 50), before any senescence occurrence.

As control, primary dermal fibroblasts were obtained from eight non-irradiated female healthy donors (C1 to C8). Surgical samples were obtained from the Hospitals Board of Lyon, France (Hospices Civils de Lyon), with the subjects’ informed consent. Cells were subcultured up to seven passages and studied between passage 7 to 10, to have similar age in culture as patient cells (mean population doublings: 35 to 50). As fibroblasts are quiescent cells in the dermis, most studies were performed on confluent cells, in the G0/G1 cell cycle phase, both for patient and control cells.

Cultures were maintained in Dulbecco’s Modified Eagle Medium–Glutamax medium (ThermoFisher Scientific, Illkirch, France) supplemented with 10% fetal bovine serum (ThermoFisher) and 1% penicillin/streptomycin (Sigma-Aldrich, Saint-Quentin-Fallavier, France).

Primary dermal fibroblasts were irradiated after reaching confluency with 2 Gy using an XRAD320 X-ray generator (Precision X-Ray, North Brandford, CT, USA) at a dose rate of 0.8 Gy.min⁻¹ and then further cultured for indicated times depending on the assay.

Dermal fibroblasts were irradiated with 2 Gy X-rays after reaching confluency and seeded at low density (5 to 40 cells/cm²) 24 h after irradiation. Two weeks later, cell cultures were fixed with EtOH 100% for 15 min and stained with hematoxylin/eosin. Only colonies formed by more than 50 fibroblasts were considered for calculating survival fraction at 2 Gy (SF2), expressed as the ratio between colonies formed with and without irradiation.

Irradiated fibroblasts were further cultured for the indicated times (0, 15 min, 2, 6, and 24 h) and fixed with 4% paraformaldehyde for 15 min. Then, cells were permeabilized (0.1% Triton X-100 and 0.1 M Glycine) and incubated in blocking buffer (5% goat serum, 2% BSA, 0.1% Triton X-100, and 0.05% Tween-20) for 15 min prior to immunostaining with anti-γH2AX antibody (05-636, Millipore) or anti-53BP1 antibody (PA1-46147, ThermoFisher). For immunodetection, goat anti-rabbit IgG or goat anti-mouse IgG Alexa Fluor-488 or -546 conjugated secondary antibody (ThermoFisher) was incubated for 1 h and nuclei were counterstained with 4′-6-diamidino-2-phenylindole dihydrochloride (DAPI). The resulting foci were counted in at least 50 nuclei per condition using an Eclipse Ti-E inverted microscope (Nikon).

Fibroblast DNA repair was measured on ExSy-SPOT assay (LXRepair). Protein extracts from lysed cells were applied directly on the biochip containing plasmids with well-characterized DNA lesions (8-Oxoguanine, Ethenobase, Abasic site, Glycols, Photoproducts and Cisplatin adducts) and incubated with fluorescent nucleotides to allow DNA repair. Effective DNA repair was quantified for each type of lesion by CT measurement of the resulting fluorescence signal. Thus, fluorescence level was proportional to cell ability to repair the specific DNA damage within the prescribed time.

Total RNA was isolated from fibroblasts at confluency with the NucleoSpin RNA plus kit (Macherey–Nagel, Hoerdt, France) or RNeasy Plus Minikit (Qiagen, Courtaboeuf, France) according to the manufacturer’s instructions.

For next-generation RNA sequencing at CNRGH (CEA, Evry, France), RNA sequences were captured using a TruSeq RNA Library Prep Kit v2 (Illumina, Evry, France) with input of 1 µg. Paired-end RNA sequencing was performed on HiSeq4000 with 100 bp paired-end reads. Sequencing data quality control was performed using FastQC 0.11.7 before and after adapter trimming by Cutadapt 1.13 (parameters: -q 15 -a 5’-AGA TCG GAA GAG CAC ACG TCT GAA CTC CAG TCA C-3’ – 5-AAG ATC GGA AGA GCG TCG TGT AGG GAA AGA GTG TAG ATC TCG GTG GTC GCC GTA TCA TT-3’). The reads were mapped to the human genome (GRCh37/hg19) using HISAT2 2.0.5. For a single gene, sequences were aligned versus all known exons of all gene isoforms. The resulting BAM files were sorted by read pairs (using SAMtools 1.3.1) and counted using the HtSeq-count tool of HtSeq 0.6.1.

For transcriptome analysis, filtering was applied to genes with low expression (mean number of reads in the training set <10). Principal component analysis and hierarchical clustering were performed using the R DEseq2 and stats packages. VennDiagram and plots were made using the VennDiagram and ggplot2 packages. Functional annotation of the gene list was performed using the WEB-based GEne SeT AnaLysis Toolkit (WebGestalt) and the KEGG database 2019. Gene Ontology (GO) was performed using the clusterProfiler package in R software (18). GO and KEGG enrichment analyses were based on a false discovery rate (FDR) threshold of <0.05. The Enrichedplot package was used for graphical visualization of the result from enriched analysis. Transcriptome data have been deposited into the GEO database and are available under the accession number GSE154559.

An equal amount of total RNA (500 ng) was used as template for reverse transcription with PrimeScriptTM RT reagent kit (Takara, Shiga, Japan) and analyzed by Real-Time QPCR using SYBR^® Premix ExTaqII (Takara) on an AriaMx Realtime PCR system (Agilent Genomics, Santa Clara, CA, USA). All primers listed below were provided by Eurogentec. NFATC2 expression level was normalized to TBP and RPS17 housekeeping gene expression level.

Total proteins were extracted using RIPA buffer (50 mM Tris-HCl pH = 8, 150 mM NaCl, 1.5 mM KCl, 1% NP-40, 0.1% SDS, 0.5% sodium deoxycholate, 0.1% Triton X-100, 1 mM EDTA) containing protease inhibitor cocktail (cOmplete mini, Roche Diagnostics) and phosphatase inhibitor cocktail (5 mM NaF, 50 mM β-glycerophosphate, 5 mM orthovanadate). Proteins were quantified using the Pierce BCA Protein Assay Kit (ThermoFisher), loaded on an 8% SDS-polyacrylamide gel and transferred to a nitrocellulose membrane (Bio-rad). The membrane was blocked for 1 h at room temperature in TBS-Tween20 0.1–5% BSA and immunoblotted overnight at 4°C for primary antibodies specific to NFATC2 (#4389, Cell Signaling Technology) or VINCULIN (V9131, Sigma-Aldrich). After washing, goat anti-mouse IgG or goat anti-rabbit IgG HRP-conjugated secondary antibodies (Bio-rad) were incubated for 1 h at room temperature. Proteins were revealed using SuperSignal West Pico PLUS Chemiluminescent Substrate (ThermoFisher) and the signal was detected by the Fusion Fx system (Vilber Smart Imaging). Immunoblot quantifications were performed using GelAnalyzer software.

DNA was extracted from fibroblasts using the QIAamp DNA Mini Kit (Qiagen) according to the manufacturer’s instructions. To assess methylation of a specific DNA region, DNA was converted with bisulfite treatment, using the EpiTect Bisulfite Kit (Qiagen) according to the manufacturer’s instructions. Then, HRM PCR was performed using the EpiTect HRM PCR kit (Qiagen) to amplify the specific DNA region and to measure the melting temperature of the amplicon on an AriaMx Realtime PCR system (Agilent Genomics). One hundred percent methylated DNA, 100% unmethylated DNA, and bisulfite unconverted DNA from the EpiTect Control DNA Set kit (Qiagen) were used as controls. Figures with methylation peaks were produced with Agilent Aria 1.5 Software (Agilent Genomics). Primers were designed with Methyl Primer Express Software v1.0 (ThermoFisher Scientific) and were provided by Eurogentec. The following primers were used for methylation study:

Control fibroblasts were transduced with lentiviral vectors from NFATC2 Human shRNA Plasmid Kit (OriGene). Lentiviral vector particles were produced by the vector facility at SFR BioSciences Gerland-Lyon Sud (Lyon, France) as previously described (19). Control cells were infected at 40% confluency with lentiviral particles (MOI at 10) containing a vector with a shRNA targeting NFATC2 (sh-NFATC2) or a plasmid with a non-effective shRNA sequence (sh-SCR) for 12 h. At confluency, cells were trypsinized and seeded in another plate. Then, transduced cells were maintained under puromycin selection for 1 week and then selected cells were amplified for 1 week before analysis.

Statistical significance was calculated by Student’s t-test, one-way analysis of variance (ANOVA), two-way analysis of variance (ANOVA2), or Pearson correlation using Prism software (version 8.0, GraphPad Software). Mean differences were considered statistically significant when P < 0.05. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001.

We analyzed ionizing radiation toxicity in cultured dermal fibroblasts from sixteen breast cancers patients with severe radiotherapy (RT) side-effects (eight grade 2, eight grade 3) and eight control individuals. To estimate radiation-induced toxicity, we used the reference method of colony survival fraction measurement after a standard dose of 2 Gy X-irradiation (SF2). The colony survival assay showed that dermal fibroblasts from the eight control biopsies presented a mean 48% survival fraction, whereas cells from patients with RT over-reaction exhibited significantly lower SF2, with 28 and 27% survival fraction for cells of grade 2 and grade 3 patients respectively ( Figure 1 ). Since no significant difference in cellular radiosensitivity was observed between cells from grade 2 and grade 3 patients, the 16 cell strains were pooled for the following experiments.

To study patient cell DNA repair ability, we first performed immunofluorescence against γH2AX and 53BP1, two early markers of DNA double-strand breaks (DSB), in four control and four radiosensitive cell strains after 2 Gy irradiation. The number of γH2AX and 53BP1 foci 15 min after 2 Gy irradiation was identical in control and patient fibroblasts, with a mean 40 foci per nucleus in both ( Figures 2A, B ). However, more γH2AX foci were detected in patient than control cells 2, 6, and 24 h after irradiation: 2.7, 4.13, and 2.03 additional foci per cell respectively ( Figure 2A ). More 53BP1 foci were also detected 6 and 24 h after 2 Gy irradiation in patient than control fibroblasts: respectively +2.65 and +5.55 foci per cell ( Figure 2B ).

We also assessed patient cell ability to repair various DNA lesions, using the ExSy-SPOT chip, a microsystem developed to measure excision-synthesis activity in immobilized plasmid DNA (20). Repair activity was reflected by the incorporation of fluorescent nucleotides at the lesion site. Fluorescence in plasmids containing 8-oxoGuanine and abasic sites was higher in control than patient samples ( Figure 2C ), and in plasmids containing glycol-damaged bases, although the difference did not quite reach significance (P = 0.062) ( Figure 2C ). Thus, ability to repair 8-oxoGuanine, abasic site and glycol-damaged bases, three types of DNA damage induced by oxidative stress and known to be repaired by the BER pathway, seemed to be impaired in dermal fibroblasts from patients with severe radiotherapy side-effects.

To investigate the mechanisms underlying individual radiosensitivity at cellular level, we used next-generation RNA sequencing to profile the whole genome transcriptome of the 16 patient and 8 control fibroblast cultures. Principal component analysis (PCA) of whole gene expression data clearly separated controls from over-reacting patients ( Figure 3A ). However, one patient’s cell strain was classified as being in the control group by hierarchical cluster analysis ( Figure 3B ). Within the patient group, there was no clear separation between grades 2 and 3, whether on PCA or hierarchical cluster analysis ( Figures 3A, B ).

We identified 1,338 genes differentially expressed (adjusted p-value <0.05) between fibroblasts from grade 2 patients and controls ( Figure 3C ), and 804 genes differentially expressed between fibroblasts from grade 3 patients and controls ( Figure 3C ). Five hundred forty of these differentially expressed genes were in common between grade 2 patients vs controls and grade 3 patients vs controls ( Figure 3C and Supplementary Table S1 ). Interestingly, no genes were differentially expressed between grade 2 and grade 3 patients, in agreement with the PCA and hierarchical cluster classification of the transcriptome data.

Four hundred forty-five of the 540 differentially expressed genes were protein-coding and were used for functional enrichment analysis. Gene ontology analysis highlighted 93 significantly enriched GO biological processes with FDR <0.05 ( Figure S1 Supplementary data ). Functions with the smallest FDRs comprised regulation of GTPase activity regulation, of organ development and of cell adhesion and junction ( Figure S1 Supplementary data ). All these functions might be involved in cellular radiation toxicity, but correspond to multiple intracellular pathways. We then searched for enriched pathways involving the 445 differentially expressed protein-coding genes using the WebGestalt and the KEGG database 2019. We found only one significantly enriched pathway ( Supplementary Table S2 ): Arrhythmogenic right ventricular cardiomyopathy (adjusted p-value 0.0065733—10 genes out 72 modulated), a pathology of cardiac muscles with progressive loss of myocytes replaced by adipocytes. Systematic analysis of our gene list revealed that none of these genes was directly involved in any known genetic syndrome leading to increased radiosensitivity.

Among the most differentially expressed genes between control and patient fibroblasts identified by our RNA sequencing approach, we focused on NFATC2, which encodes a transcription factor initially described in T-cell activation and involved in numerous cellular functions such as apoptosis and the cell cycle (21). We therefore carefully analyzed NFATC2 gene and protein expression in patient cells extracted from non-irradiated skin. Dermal fibroblasts from patients with severe radiotherapy side-effects exhibited much lower NFATC2 gene expression than control cells ( Figure 4A ). Similarly, NFATC2 protein expression was lower in patient’s cells ( Figures 4B, C ). NFATC2 gene and protein expressions were assessed after irradiation to determine whether NFATC2 could be a radiation-responding gene, potentially involved in cellular radiation response. NFATC2 gene overexpression was detected in response to 2 Gy irradiation, with a 4-fold peak at 3 h in control fibroblasts and a 15-fold peak in patient fibroblasts, with return to baseline after 24 h ( Figure 4D ). NFATC2 protein was detected in greater quantities (×1.8) after 2 Gy irradiation with a peak between 3 and 6 h in normal fibroblasts, but the difference was not statistically significant due to interindividual variability (P = 0.1192 and P = 0.1553, respectively) ( Figures 4E, F ). Furthermore, NFATC2 protein was barely detected in cells from radiosensitive patients ( Figures 4E, F ). Interestingly, there was a significant correlation between NFATC2 transcriptional expression and irradiated cell survival (Pearson correlation coefficient, R² = 0.4949, P = 0.0001254) ( Figure 4G ), suggesting a possible role for NFATC2 in cellular radiosensitivity. Taken together, these results suggest that NAFTC2 is a radiation-responding gene potentially involved in cellular response to ionizing radiation.

To elucidate the regulatory mechanisms underlying NFATC2 differential expression in non-irradiated fibroblasts from controls and patients, we compared gene methylation level between patients and controls. The methylation pattern of the NFATC2 gene region was extracted from genome-wide methylation profiling using methylation bead chips performed on patient and control fibroblasts (data not shown), identifying 34 differentially methylated sites (CpGs) (33 hyper- and 1 hypo-methylated) ( Figure 5 ). This hypermethylation of the NFATC2 gene in patients’ fibroblasts was consistent with its lower gene expression. Interestingly, the upstream region of the transcription starting site (TSS) exhibited the same methylation profile in patient and control cells ( Figure 5 ). However, among the hypermethylated CpGs, cg00418183, cg00401091, cg11086066, cg10226546, cg11074047, cg18302534, cg16419175, cg21610125, cg15497991, cg26408896, cg22243637, cg09740920, cg00498368, cg08637147, cg09465142, cg03986956, and cg00689890 belong to gene regions identified as promoter-associated regions according to the ENCODE consortium. This hypermethylation of these regulatory regions could, at least in part, explain the weak expression of NFATC2 in patient fibroblasts.

To confirm these results by an independent method, we performed HRM PCR to evaluate the methylation state of the CpG00498368 in four patient and two control cell strains. HRM PCR is a PCR measuring the melting temperature of a specific amplicon. After bisulfite conversion, an amplicon comprising a methylated CpG would exhibit a higher melting temperature than an amplicon with an unmethylated CpG. We detected a higher melting temperature peak, corresponding to a methylated state of this CpG in patient fibroblasts and to an unmethylated state in control fibroblasts ( Figure S2B Supplementary data ), in agreement with the global methylome data ( Figure S2A Supplementary data ).

These results suggest that NFATC2 down-regulation could, at least in part, be due to hypermethylation of the gene in fibroblasts from patients with severe radiotherapy side-effects.

To evaluate the functional impact of NFATC2 in cellular radiosensitivity, NFATC2 expression was silenced in cells from healthy individuals by stable RNA interference mediated by a lentiviral vector. We tested four lentiviral vectors, each containing a different short hairpin RNA (shRNA) targeting NFATC2, and chose the most efficient in terms of silencing for the further experiments ( Figure S3 Supplementary data ). NFATC2 expression was reduced by 48% on average at gene level ( Figure 6A ) and by 70% on average at protein level in transduced fibroblasts expressing the shRNA targeting the NFATC2 transcript ( Figures 6B, C ). Colony survival assays revealed that cells with lower NFATC2 expression (HNF sh-NFATC2) exhibited a significantly lower SF2 compared to their control (HNF sh-SCR) (20% decrease in cell survival in response to irradiation) ( Figure 6D ). To better understand the molecular mechanisms of this cell death elevation in sh-NFTAC2 expressing cells, we investigated their DNA DSB repair ability, and detected more residual γH2AX and 53BP1 foci 24 h after 2 Gy irradiation in HNF sh-NFATC2 cells than in control cells (HNF sh-SCR) (respectively, +1.49 γH2AX and +1.46 53BP1 foci per nucleus on average) ( Figures 6E, F ), corresponding to an excess of residual unrepaired DNA double-strand breaks. These results suggest that the NFACT2 down-regulation observed in patient cells is involved in their cellular radiation sensitivity.