Developing an RNA Signature for Radiation Injury Using a Human Liver-on-a-Chip Model

Radiation exposure in a therapeutic setting or during a mass casualty event requires improved medical triaging, where the time to delivery and quantity of medical countermeasures are critical to survival. Radiation-induced liver injury (RILI) and fibrosis can lead to death, but clinical symptoms manifest late in disease pathogenesis and there is no simple diagnostic test to determine RILI. Because animal models do not completely recapitulate clinical symptoms, we used a human liver-on-a-chip model to identify biomarkers of RILI. The goals of this study were: 1. to establish a microfluidic liver-on-a-chip device as a physiologically relevant model for studying radiation-induced tissue damage; and 2. to determine acute changes in RNA expression and biological pathway regulation that identify potential biomarkers and mechanisms of RILI. To model functional human liver tissue, we used the Emulate organ-on-a-chip system to establish a co-culture of human liver sinusoidal endothelial cells (LSECs) and hepatocytes. The chips were subject to 0 Gy (sham), 1 Gy, 4 Gy, or 10 Gy irradiation and cells were collected at 6 h, 24 h, or 7 days postirradiation for RNA isolation. To identify significant expression changes in messenger RNA (mRNA) and long non-coding RNA (lncRNA), we performed RNA sequencing (RNASeq) to conduct whole transcriptome analysis. We found distinct differences in expression patterns by time, dose, and cell type, with higher doses of radiation resulting in the most pronounced expression changes, as anticipated. Ingenuity Pathway Analysis indicated significant inhibition of the cell viability pathway 24 h after 10 Gy exposure in LSECs but activation of this pathway in hepatocytes, highlighting differences between cell types despite receiving the same radiation dose. Overall, hepatocytes showed fewer gene expression changes in response to radiation, with only 3 statistically significant differentially expressed genes at 7 days: APOBEC3H, PTCHD4, and GDNF. We further highlight lncRNA of interest including DINO and PURPL in hepatocytes and TMPO-AS1 and PRC-AS1 in LSECs, identifying potential biomarkers of RILI. We demonstrated the potential utility of a human liver-on-a-chip model with primary cells to model organ-specific radiation injury, establishing a model for radiation medical countermeasure development and further biomarker validation. Furthermore, we identified biomarkers that differentiate radiation dose and defined cell-specific targets for potential radiation mitigation therapies.