The blood-brain barrier (BBB) is a unique barrier of the central nervous system (CNS) that has a highly selective barrier function that prevents most drugs from entering the brain, leading to a high failure rate in the development of therapeutics for the CNS diseases. Currently there is a large unmet need for the development of new therapeutics for the CNS diseases with an increasing death rate of patients with CNS diseases like Alzheimer’s disease (AD). One innovative approach to address this challenge is to develop a microengineered model of the human BBB that can mimic the pathophysiological conditions of the human brain. Yet, there is no physiologically relevant in vitro human BBB models that can incorporate shear stress, direct cell interactions, and 3D glial physiology of the human brain. Here we present a novel microengineered human BBB model designed to create a 3D co-culture of human brain endothelial cells (HBMECs), human brain vascular pericytes (HBVPs), human astrocytes (HAs) with the physiological morphology and interaction. Furthermore, we construct a neuroinflammation model with 3D incorporation of human microglia (HM) to understand the role of micrglia-mediated neuroinflammation in AD pathogenesis.
Our microfluidic BBB model consists of two layers separated by a porous membrane to mimic the luminal and abluminal regions of the BBB. After growing HBVPs on the abluminal side of the porous membrane, HAs were cultured in the same channel. As HAs cultured in 3D Matrigel showed more physiologically relevant morphologies with lower expression of reactive markers compared to that in 2D, the abluminal layer of our model was filled with HAs that are embedded in 3D Matrigel. This Matrigel was confined by surface tension from the two side channels that are designed for culturing HMs. HBMECs were cultured on the luminal side of the porous membrane and exposed to physiological shear stress that mimics blood flow in the brain microvasculature. As a result, our microengineered human BBB model showed a physiological network of the BBB cells with a polarized expression of aquaporin-4 to the luminal channel in astrocytic end-feet, highly specialized phenotypes of the brain endothelial cells with increased expressions of junctional proteins and solute-carrier genes, and a significantly decreased permeability of the endothelial monolayer, as compared to the monoculture of endothelial cells. Neuroinflammation model was constructed by adding HMs that were exposed to IFN-γ into the side channels of the stabilized human BBB model. HMs cultured in the side channels migrated into the Matrigel and showed a dynamic interaction with HAs, resulting in neurotoxic reactivities of HAs. Our microengineered BBB model can be utilized as a tool for the study of the pathological mechanisms of neuroinflammation and exploration of future therapeutic or preventive strategies for AD.