Assessment of Drug-Induced Toxicity Biomarkers in the Brain Microphysiological System (MPS) Using Targeted and Untargeted Molecular Profiling.

Early assessment of adverse drug effects in humans is critical to avoid long-lasting harm. However, current approaches for early detection of adverse effects still lack predictive and organ-specific biomarkers to evaluate undesired responses in humans. Microphysiological systems (MPSs) are representations of human tissues and provide organ-specific translational insights for physiological processes. In this study, a brain MPS was utilized to assess molecular signatures of neurotoxic and non-neurotoxic compounds using targeted and untargeted molecular approaches. The brain MPS comprising of human embryonic stem (ES) cell-derived neural progenitor cells seeded on three-dimensional (3D), chemically defined, polyethylene glycol hydrogels was treated with the neurotoxic drug, bortezomib and the non-neurotoxic drug, tamoxifen over 14-days. Possible toxic effects were monitored with human N-acetylaspartic acid (NAA) kinetics, which correlates the neuronal function/health and DJ-1/PARK7, an oxidative stress biomarker. Changes in NAA levels were observed as early as 2-days post-bortezomib treatment, while onset detection of oxidative stress (DJ-1) was delayed until 4-days post-treatment. Separately, the untargeted extracellular metabolomics approach revealed distinct fingerprints 2-days post-bortezomib treatment as perturbations in cysteine and glycerophospholipid metabolic pathways. These results suggest accumulation of reactive oxygen species associated with oxidative stress, and disruption of membrane structure and integrity. The NAA response was strongly correlated with changes in a subset of the detected metabolites at the same time point 2-days post-treatment. Moreover, these metabolite changes correlated strongly with DJ-1 levels measured at the later time point (4-days post-treatment). This suggests that early cellular metabolic dysfunction leads to later DJ-1 leakage and cell death, and that early measurement of this subset of metabolites could predict the later occurrence of cell death. While the approach demonstrated here provides an individual case study for proof of concept, we suggest that this approach can be extended for preclinical toxicity screening and biomarker discovery studies.