Development of Co-cultured Microfluidic Platform for Mimicking the Blood-Brain Barrier

A blood-brain barrier (BBB) is a biochemical, physical barrier found in the brain vasculature that only allows selective transport of molecules in order to protect the brain from potential damages. Due to its complex structure of cellular arrangement, there is a lack of a physiologically relevant BBB model that could be used to test the efficacy of drugs that treat brain cancers, such as glioblastoma multiforme. Therefore, this study aims to introduce an in vitro BBB model on a microfluidic platform that captures the dynamic nature of BBB, mainly permeability of the BBB. The proposed BBB model, or BBB on a chip (BBBoC), incorporates human primary cells that comprise a human BBB, including human brain microvascular endothelial cells (HBMECs), human brain vascular pericytes (HBVPs), and normal human astrocytes (NHAs). To validate the proposed BBB model, this study tests for permeability of different combinations of the primary brain cells to observe the effect of cellular composition on permeability. Finally, an effective in vitro BBB model is introduced. Background Glioblastoma multiforme (GBM) is one of the most frequent, fatal forms of primary brain tumor. 1,10 GBM is characterized and induced by overexpression, activation, and dysregulation of various membrane receptors, signaling pathways, and other factors. 2 At the present time, the two-year survival of the patients diagnosed with GBM is approximately 28% with median survival of 16.5 months. 9 The current standard procedure for treating GBM is invasive and includes resection followed by chemotherapy, which aims to eliminate the tumor. 2 However, due to the infiltrative nature of GBM, a gross-total resection (GTR) often does not have a significant influence on a microscopic level. 11 It is also important to note that GTR will not always be a primary option because GTR is often limited in size as brain is a vital organ. As a result, chemotherapy is used as a follow-up or an alternative procedure. Although many combinations of drugs have been introduced, there is a lack of ‘personalized’ therapy, and the level of cytotoxicity of the new strategies has yet to be determined. 2 The drug screening for GBM is often difficult to evaluate due to the presence of blood-brain barrier (BBB). BBB is a complex structure in the brain that consists of endothelial cells, pericytes, astrocytes, neurons and microglia, which comprise the neurovascular unit (NVU). 3 The interactions between the different cell types contribute to the specialization of the endothelial cells that regulate passage of molecules in and out of the brain. 4 Thus, investigating the transport pathway through the BBB under GBM condition is crucial in order to not only enhance the efficacy of the drugs but also potentially find better candidates to treat GBM. In pursuit of gaining insight into the complexity that surrounds the BBB, there have been various efforts to model BBB in vitro. At an earlier stage, the role of endothelial cells was viewed more significant to other types of cell that constitutes the BBB. Monoculture of brain endothelial cells was viewed as a sufficient model to study the transport phenomenon through the BBB. 4 The endothelial monolayer was cultured on top of a semi-porous membrane and under static condition. 4 Although these models may enlighten us on the response of isolated endothelial cells to a particular drug, the results are questionable due to the lack of physiological relevance in their design, such absence of dynamic flow and NVU. In the earlier example, the model does not incorporate fluid flow. Wolff et al. and many authors argue that this creates significant problem as the endothelial cells are cultured under less favorable environment, as shown by lower transendothelial electrical resistance (TEER) values compared to those of the cells cultured under shear stress. 4 Santaguida et al. even go further and claim that the static model is flawed as the absence of shear stress reduces the tightness, shown by TEER values, and the permeability to polar molecules through the monolayer compared to the values in vivo. 6 Although agreeing on these drawbacks in using static model, Wilhelm et al. raises another view on the static model by pointing out that the static model is inexpensive and is easy to handle, and therefore, for the sake of high-throughput screening, the simple endothelial monolayer BBB model is quite adequate. 5 It seems agreeable that the simple model would drastically remove grossly inadequate candidates and thus save time and money. However, because the integrity of the monolayer is suboptimal in the absence of shear stress, the simple model may further complicate, rather than simplifying, the candidate selection. In addition, for the sake of researching the BBB in the context of GBM, the monolayer model would not be suitable as the cancer would further induce leaky vasculature and disrupt the integrity of the monolayer, and therefore the model would not produce any meaningful results. In an effort to increase physiological relevance, many researchers adopted microfluidic ‘organ-on-a-chip’ approach to introduce fluid flow. Santaguida et al. describes one microfluidic model produced by Prabhakarpandian et al., which has two compartments segregated by a porous membrane and is able to induce microcirculation within the compartments. 6 Although praising the design incorporates shear stress through fluid flow, Santaguida et al. also points out that the immediate effectiveness of microfluidic model is hard to be determined as measurements of TEER or other markers are challenging on microfluidic platform. 6 In the same regards, Wilhelm et al. expresses that the microfluidic model is not well established. 5 The other pressing issue about the simple monolayer is the fact that it is lacking physiological relevance to the human BBB. The BBB is comprised of many cell types, such as astrocytes and pericytes. 3 The endothelial monolayer itself may exhibit the properties of the BBB due to tight junctions; but, in the case like GBM, a deeper insight into the interactions inside the NVU is critical. While some tri-culture models have been proposed, most of the cases of triculture are done in a static environment. There are some cases where co-culture was done on a dynamic environment, such as microfluidic platform, but is typically limited to endothelial cells and astrocytes. 7,8 Besides the previous efforts to produce BBB models, there have been efforts to produce GBM models. For example, Fan et al. utilized hydrogel-based microwells where GBM cancer cells were seeded to produce GBM spheroids. 12 Although this model allowed three-dimensional culture of GBM cancer cells, this model also exhibits the same issues of static environment and lack of physiological relevance. Other studies include in vivo models, such as murine GBM model. Katz et al. injected RCAS-PDGF avian retrovirus into healthy mice subjects, thereby performing a somatic gene transfer of an oncogene PDGF to cause genetic aberrations, such as loss of tumor suppressor on glioma formation, and consequently inducing GBM. 22,25 Despite the fact that both authors directly used cancer cells, their relevance to the human GBM physiology is questionable. Introduction Glioblastoma multiforme (GBM) is the most malignant astrocytic tumor 1,11 and approximately 10,000 new cases of GBM are diagnosed per year. 10 Currently, the standard procedure for GBM includes resection followed by chemotherapy and chemoradiotherapy, which aims to disrupt receptor-ligand interactions or signaling pathways in the cancer development. 2,10 Yet, the presence of blood-brain barrier (BBB) poses a challenge to drug delivery in chemotherapy due to the complex structure and the high selectivity. 3,5 Various in vitro studies have focused on drug delivery through the BBB in order to increase the efficacy of systemically administered drugs. However, the proposed BBB models in the previous studies are lacking in physiological relevance due to their static nature 5,14 and the absence of co-culture of the various human brain cells that compose the BBB. 7,8 There is no adequate BBB model that is suitable to evaluate the drugs used to treat GBM despite the increasing demand for chemotherapy. Therefore, the goal of this study is to develop a physiologically relevant BBB model. In order to address the issues of co-culture and static environment, this study aims to utilize microfluidic technology to reproduce the in vivo conditions of the human BBB. 22 The BBB model proposed in this study is a double-layer microfluidic device that is composed of two different compartments segregated by a semi-permeable membrane. In this model, the top compartment was seeded with human brain microvascular endothelial cells (HBMECs) to mimic human brain capillaries. 15,16,17 In addition, to construct the in vivo structure of the BBB, human brain vascular pericytes (HBVPs) will occupy the bottom side of the membrane and the bottom compartment will be filled with normal human astrocytes (NHAs) in a hydrogel. 16,17,18,19 Because the membrane is separating the two compartments, the HBVPs seeded on the bottom side of the membrane are in contact with the NHAs seeded in the bottom compartment. In this setup, HBMECs are also able to interact with both HBVPs and NHAs. Such proximity would allow the interactions between each cell type observed in vivo. 17,19,20,21 Moreover, the double-layer device has microfluidic channels which allows to model the physiological dynamic conditions in vivo. Then, the proposed BBB model was tested for validation. To validate the model, permeability test was conducted with FITC-Dextran. It was reported that tri-culture of porcine primary pericytes, astrocytes and endothelial cells increases the integrity of endothelial monolayer compared to the monolayer under monoculture. 28 As BBB is a conserved physiology between pigs and human, tri-culture of HBMECs, HBVPs a