A transcriptomic dataset evaluating the effect of radiotherapy injury on cells of skin and soft tissue

Radiotherapy injury to cells of the skin and subcutaneous tissue is an inevitable consequence of external beam radiation for treatment of cancer. This sublethal injury to normal tissues plays a significant role in the development of fibrosis, lymphedema, impaired wound healing, and recurrent infections. To elucidate the transcriptional changes that occur in cells of the skin and soft tissues after radiotherapy injury, we performed genome-wide RNA-sequencing comparing irradiated cells (10Gy) with non-irradiated (0Gy) controls in normal human dermal fibroblasts, normal human keratinocytes, human microvascular endothelial cells, human dermal lymphatic endothelial cells, pericytes and adipose derived stem cell populations. These data are publicly available from the Gene Expression Omnibus database (accession number GSE184119). Further insights can be gained by comparing the mRNA signatures arising from radiation injury derived from these data to publicly available signatures from other studies involving similar or different tissue types. These global targets hold potential for manipulation to mitigate radiotherapy soft tissue injury.


a b s t r a c t
Radiotherapy injury to cells of the skin and subcutaneous tissue is an inevitable consequence of external beam radiation for treatment of cancer. This sublethal injury to normal tissues plays a significant role in the development of fibrosis, lymphedema, impaired wound healing, and recurrent infections. To elucidate the transcriptional changes that occur in cells of the skin and soft tissues after radiotherapy injury, we performed genome-wide RNA-sequencing comparing irradiated cells (10Gy) with non-irradiated (0Gy) controls in normal human dermal fibroblasts, normal human keratinocytes, human microvascular endothelial cells, human dermal lymphatic endothelial cells, pericytes and adipose derived stem cell populations. These data are publicly available from the Gene Expression Omnibus database (accession number GSE184119). Further insights can be gained by comparing the mRNA signatures arising from radiation injury derived from these data to publicly available signatures from other studies involving similar or different tissue types. These global targets hold potential for manipulation to mitigate radiotherapy soft tissue injury.

Value of the Data
• These data are useful as they allow us to measure the effects of radiotherapy (RTX) on gene expression at the level of the individual cell types that constitute skin to define a radiation gene signature across all the different cell types profiled. • These data could be used to help clinicians understand and address the adverse effects of RTX on normal soft tissues. • Further insights can be gained by comparing the radiation signatures derived from these data to publicly available signatures from other studies involving similar or different tissue types. ontology (GO) terms (go_terms_lec.csv) and KEGG pathways (kegg_pathways_lec.csv) for differentially expressed genes derived from the LEC data are available in the "analysis" folder.

RNA-seq sample preparation and sequencing
Standardized numbers of LECs, HMECs, NHDFs, NHEKs, PCs and ADSCs were plated in T75 flasks (CELLSTAR ® ) (as per manufacturer's instructions) and were irradiated once 80-90% confluence was achieved. RNA was extracted from tissue culture flasks at 4 h after irradiation or control treatment was completed using Qiazol ® (QIAGEN, Germany) and purified with DNase and RNEasy ® Plus Universal Kit (QIAGEN) as per manufacturer's instructions. Samples were then tested for purity and quality control using the Nanodrop Spectrophotometer (Thermo Fischer Scientific). Each sample was then transported to the Australian Genome Research Facility (AGRF) in Melbourne and underwent RNA sequencing (100 base pair single end) using Illumina HiSeq.

Quality control and data preprocessing
Sequenced reads were first mapped to the hg19 human reference genome using the R/Bioconductor package Rsubread [6 , 7] (version 1.10.5) with default parameters. Mapped reads were then assigned to individual genes, using the featureCounts [8] function with default settings. Genes were annotated using the org.Hs.eg.db [9] R/Bioconductor package (version 3.8.2). Read counts were processed with the R/Bioconductor packages edgeR [10 , 11] (version 3.8.1) and limma [12] (version 3.32.4). Counts were first transformed using the cpm function from edgeR to generate counts per million (cpm) for each gene to account for different library sizes. Genes were retained for further analysis if they had a baseline expression level of 0.5 cpm in at least three samples. Counts were normalized using the trimmed mean of M-values (TMM) method [13] . Multidimensional scaling (MDS) was performed on the transformed counts during exploratory data analysis.

Differential expression analysis
We modelled heteroscedasticity in the gene counts with voomWithQualityWeights from the limma package using radiotherapy treatment as a main effect ( Fig. 1 D), adjusting for cell type and replicate number. Observations that were more variable compared to others were downweighted in the subsequent linear model analysis.
To investigate the global consequences of RTX across the selected cell subtypes, we used limma to fit a linear model and used empirical Bayes moderation of t -statistics [14] to assess differential expression between the contrasts of interest (10Gy vs. 0Gy, 10Gy vs. 2Gyx5, and 2GYx5 vs 0Gy. Gene set analysis was also performed for the same comparisons made on the LEC data using the goana and kegga functions of the limma package.

Ethics Statements
This manuscript adheres to the Elsevier Ethics in publishing standards.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.