Purdue University Graduate School
Kelsey Lubin Dissertation FINAL VERSION 7-25-2019 (2).pdf (2.89 MB)

Design of Experiment Based Optimization of a Direct Contact Blood Brain Barrier In Vitro Model for Neuroactivity Screening

Download (2.89 MB)
posted on 2020-12-16, 18:41 authored by Kelsey E LubinKelsey E Lubin
Neurotherapeutics are an essential drug class that is often forgotten or neglected due to the difficulties associated with pharmaceutical development and approval. These compounds face high rates of attrition in clinical trials and late stage development predominantly due to the restrictiveness of the blood brain barrier (BBB). The inherent role of the BBB is to protect and maintain the homeostatic environment around the neuronal cells in the brain parenchyma. This is accomplished by the BBB posing not only as a physical barrier through its restrictive tight junctions that prevent paracellular permeation, but also through the high expression levels of efflux transporters and drug metabolizing enzymes that prevent transcellular permeation of potential drug compounds. In attempting to deliver compounds to the brain the intended outcome is often over-shot to the point of causing neurotoxic implications. One way to mitigate the difficulties associated with drug delivery to the brain and early evaluation of potential toxic compounds is to develop in vitro cell-based models that mimic the in vivo BBB and neurovascular unit (NVU). The mainstays of the BBB phenotype are presented in the brain microvessel endothelial cells (BMECs) and are regulated and influenced by the close contacts of supporting cells of the NVU such as astrocytes, pericytes, and neurons. An in vitro model that can mimic the close contacts between these four cell types and is capable of being implemented in pharmaceutical development for BBB permeability and neuroactivity screening could lead to better selection of hit and lead candidates, and ultimately reduce the attrition rates of neurotherapeutics.
Direct contact coculture and triculture models have been developed in our laboratory that mimic the in vivo cell-cell contacts between the different cell types of the NVU and provide increased barrier properties in comparison to other models utilizing indirect contact between cell types. Early development and optimization of these models was accomplished using the human cerebral endothelial cell line hCMEC/D3. Although this cell line proved useful in early validation stages, it was decided that a different endothelial cell source would be sought out. Work was done using iPSC derived endothelial cells (iCell® endothelial cells) and an alternative immortalized human brain endothelial cell line (HBEC-5i). Both cell lines proved to be amenable to the direct contact coculture and triculture models, with the iCell® models showing greater barrier properties in comparison to those using the HBEC-5i cell line. However, drawbacks of the iCell® model were observed in extending culturing of the cells causing the cells to “roll” to the middle of the filter and proving to be cost prohibitive for extensive optimization. Ultimately, the HBEC-5i cell line was chosen for continued development and optimization due to its immortalized origin and potential for replacing the hCMEC/D3 cell line in the direct contact models.
Optimization of the direct contact triculture using the HBEC-5i cell line was required as all of the previous development was performed using the hCMEC/D3 cells. Typically, optimization of in vitro systems is performed in a one factor at a time manner or not at all. Given the large number of factors that can influence the outcome of this model, a design of experiments (DOE) based optimization approach was taken. DOEs are traditionally used in process optimization of non-biologically based systems; however, the production of the direct contact triculture is a process that could greatly benefit from extensive optimization. The seeding densities of all three cell types used in the triculture (astrocytes, pericytes, and HBEC-5i), the extracellular matrix used, and the length of culture time post endothelial cell plating were the factors chosen for the optimization process given the observations made during early development of the model. The conditions were optimized for barrier tightness by measuring the permeability of a 4 kD dextran as a paracellular marker because the model would have limited utility without adequate tight junction formation. Based on the results of this work, optimized conditions were determined in a significantly reduced amount of time as compared to traditionally used cell model optimization methods and an in vitro BBB screening tool that mimics the physiology of the NVU was developed. Given the outcomes of these studies it can be seen that a DOE optimization approach should be considered for development of biologically based systems to understand interactions between key system factors and to reduce the time to develop these necessary systems.
BBB permeability is not the only factor that slows development of neurotherapeutics. The intent of many of these compounds is to elicit an effect on the neuronal environment; however, permeability and neuroactivity are often evaluated separately even though they are inherently linked in vivo. Further enhancement of the optimized direct contact triculture was done to develop a screening tool that could assess neuroactivity of a compound as it is related to its brain permeability. The in vitro NVU model was developed by adding human neurons to the basolateral chamber of the direct contact triculture so permeating compounds would accumulate in the receiver chamber and their neuronal effects could be measured. During development of this model it was seen that the addition of neurons both increases tightness of the apical BBB model, but also increases viability of the neurons themselves. This is likely due to the facilitation of cross-talk between the four cell types of the NVU due to the proximity of the cells in the model system. The BBB permeability linked neuroactivity of marker compounds was measured by neuronal viability and neurite outgrowth in response to compound accumulation over the neurons during the course of BBB permeation. The results of this assay showed that the model is capable of being used to assess both BBB permeability and the subsequent neuroactivity of a given compound, and that the inclusion of additional cell types from the NVU further increases the physiological relevancy of the model. This work shows that the NVU model is an enhancement of the direct contact triculture model and can be easily implemented in the early development stages of neurotherapeutic compounds. Ultimately, this model has the potential to increase the number of brain targeting compounds by facilitating early, predictive assessment and rank ordering of large compound libraries for continued development.


Degree Type

  • Doctor of Philosophy


  • Industrial and Physical Pharmacy

Campus location

  • West Lafayette

Advisor/Supervisor/Committee Chair

Gregory T. Knipp

Additional Committee Member 2

Stephen Byrn

Additional Committee Member 3

Rodolfo Pinal

Additional Committee Member 4

Jean-Christophe Rochet

Additional Committee Member 5

Yoon Yeo

Usage metrics



    Ref. manager