This study investigated the effects of drug formulation and shear stress on cellular uptake, viability, and real-time reactive oxygen species (ROS) sensing using a microfluidic bilayer liver-on-chip (BLC) system. To improve the biomimetics of the liver chip model and form vascular structures, we designed a bilayer model in which human umbilical vein endothelial cells and epithelial hepatocarcinoma cells (HepG2) were co-cultured through a porous membrane. Human serum albumin (HSA)-oleic acid nanoparticles (AONs) were synthesized using the desolvation method, and doxorubicin hydrochloride (DOX-HCl) was efficiently loaded into the nanoparticles (NPs) via an incubation process involving electrostatic interactions. Because of continuous blood flow in the human body, differences in cellular uptake and ROS generation of NPs occur between dynamic and static environments, affecting delivery efficiency. Therefore, cellular internalization and cell viability based on the drug formulation were evaluated under both static and dynamic conditions. Under dynamic conditions, we observed a notable enhancement in the cellular internalization of DOX-AONs compared to that under static conditions, leading to improved efficiency in killing cancerous cells. Simultaneously, real-time ROS production was detected using a sensor, and it was confirmed that with an increase in the drug concentration, cell viability decreased, and ROS production increased. It was concluded that there was a correlation between ROS production and cell survival rate. Additionally, to verify in vivo correlation of the liver chip data, IC50 values were used to predict LD50 values according to the regression formula in ICCVAM (log LD50 (mg/kg) = 0.7092 log IC50 (μg/mL) + 0.4005). This value was similar to LD50 of DOX in a mouse model, indicating that the model can be used as an alternative to animal testing. Keywords: bilayer liver-on-chip, HepG2 cell, real-time ROS sensing, nanoformulation, shear stress, doxorubicin, cell viability, IC50 value, predictive
