Establishment and validation of an immortalized in vitro human blood-brain barrier (BBB) model for drug permeability studies, and application to natural product derived leads

In the human brain, the endothelial cells lining the cerebral microvessels form a uniquely tight cellular layer separating the brain tissue from the bloodstream. This cellular barrier, designated as “blood-brain barrier” (BBB), prevents the entry of xenobiotics and neurotoxic metabolites into the central nervous system (CNS) and thus protects the nerve tissue from chemical damage. More than 98% of small molecule drugs have been estimated to not cross the BBB. For drugs targeting the CNS, however, low BBB permeability may lead to limited brain penetration, culminating in insufficient drug concentrations at the target sites and thus therapeutic failure. On the other hand, low BBB permeation is desirable for non-CNS drugs, as this reduces the risk of CNS-related side effects. Regardless of the therapeutic area, drug lead candidates should therefore be screened for their ability to permeate the BBB already at an early stage of the drug development process, in order to reduce their attrition rate at a later stage.
In the past years, a broad spectrum of cell-based in vitro BBB models has been developed and implemented in academia and industry to bring forward molecules with high potential for CNS exposure. Despite considerable efforts, there is still an urgent need for reliably predictive BBB models, in particular human ones. Primary human cells are difficult to obtain on ethical grounds, are laborious to cultivate, suffer from batch-to-batch variation, and are thus suitable only for low throughput screenings. To overcome these limitations, immortalized human brain microvascular endothelial cell lines have been generated by transfection with tumor genes. Unlike primary cultures, immortalized cells are easy to cultivate, proliferate indefinitely, and maintain their differentiating properties even after repeated passaging. These properties render them highly suitable for standardized screenings amenable to higher throughput. Regrettably, currently available immortalized human brain capillary endothelial cell lines often show deficiencies such as low barrier tightness, relatively high leakage of barrier integrity markers, and insufficient expression of key transporter systems. Consequently, careful optimization and validation of human cell line-based in vitro BBB models have to be carried out prior to their application to permeability screening of drug candidates.
The aim of this thesis was to establish a human in vitro BBB model based on an immortalized human brain capillary endothelial cell line, to validate it with a representative series of drug substances known to cross the BBB to a varying extent, and to apply it to BBB permeability studies of promising lead compounds of natural origin.
To establish an improved in vitro human BBB model, we evaluated in a first step four currently available immortalized human brain capillary endothelial cell lines (hCMEC/D3, hBEMC, TY10, and BB19) regarding their ability to produce endothelial cell monolayers with sufficient barrier tightness in a 24-well Transwell system. Transendothelial electrical resistance (TEER) values were recorded in real-time using an automated CellZscope system to obtain highly standardized data. Culture conditions (growth medium composition, tissue culture insert material, coating material and procedure, cell seeding density) were systematically optimized, the impact of co-cultured immortalized human astrocytes (SVG-A cell line) and pericytes (HBPCT cell line) on barrier integrity of endothelial cell monolayers was investigated, and biochemical and immunocytochemical characterization of cell-type specific cellular junction proteins was performed. Under the conditions examined in our experiments, mono-cultures of hBMEC cell line exhibited highest TEER values (around 40 Ωcm2) and lowest leakage of two fluorescent barrier integrity markers (sodium fluorescein, Na-F; and lucifer yellow, LY) (apparent permeability coefficients (Papp) in the range of 3–5 x 10-6 cm/s). Furthermore, hBMEC cells were shown to express the tight junction proteins ZO-1 and claudin-5, and the endothelial marker protein VE-cadherin, confirming their endothelial lineage. Thus, we concluded that hBMEC cell line was the most suitable cell line in terms of barrier tightness for the establishment of an immortalized in vitro human BBB model.
The hBMEC cell line-based in vitro human BBB model was validated in a next step with a representative series of structurally diverse compounds known to cross the BBB to a different extent. Antipyrine, caffeine, diazepam, and propranolol were selected as positive controls, while atenolol, cimetidine, quinidine, and vinblastine served as negative controls. For each compound, a quantitative ultra-high performance liquid chromatography tandem mass spectrometry (UHPLC-MS/MS) assay in Ringer HEPES buffer (RHB) was developed and validated in terms of selectivity, precision, and reliability according to current international guidelines. During method validation, numerous biological and analytical challenges were encountered, demonstrating that major precautions have to be taken prior to quantification, and underlining the importance of careful method development. All compounds were screened in the in vitro human BBB model with the barrier integrity marker Na-F in parallel, and endothelial permeability coefficients (Pe) across hBMEC monolayers were determined by means of the validated UHPLC-MS/MS methods. The in vitro human BBB model correctly predicted BBB permeability of the selected compounds, with the exception of one negative control (quinidine, a small basic lipophilic P-glycoprotein (P-gp) inhibitor and substrate). A limitation of the model may thus be the lack of discrimination between passively diffusing compounds and substrates of active efflux. Complementary assays to determine efflux pump interaction are therefore recommended. Nevertheless, we conclude that our model represents a promising tool for early BBB permeability assessment of lead candidates in drug discovery, as it is of human origin (thus reducing the risk for data confounded by species differences), easy and fast to set up, and thus amenable to moderate to higher throughput screening.
After validation, we screened the alkaloid piperine from black pepper (Piper nigrum L.) and five selected piperine analogs with positive allosteric γ-aminobutyric acid type A (GABAA) receptor modulatory activity for their ability to permeate the BBB in the immortalized in vitro human BBB model. Since GABAA receptors are expressed in the CNS, lead compounds modulating this target need to cross the BBB to reach their sites of action. For comparative purposes, the compounds were screened in parallel in a human stem cell-derived and in a well-established primary animal (bovine endothelial/rat astrocytes co-culture) in vitro BBB model. For each compound, a quantitative UHPLC-MS/MS assay in the corresponding matrix was developed, and permeability coefficients in each model were determined. In vitro predictions from both human models were in good agreement, while permeability data from the animal model differed to some extent. In all three BBB models, piperine and the semisynthetic analog SCT-64 displayed highest BBB permeability, which was corroborated by in silico prediction data. For the other piperine analogs, BBB permeability was low to moderate in the two human models, and moderate to high in the animal model. Efflux ratios (ER) calculated from bidirectional permeability experiments indicated that the compounds were not substrates of active efflux transporters.
In addition to GABAA receptor modulating compounds, the indolinone derivative (E,Z)-3-(4-hydroxy-3,5-dimethoxybenzylidene)indolin-2-one (indolinone) from woad (Isatis tinctoria L.) was screened in our immortalized in vitro human BBB model. The compound had previously been shown to possess potent histamine release inhibitory and anti-inflammatory properties, and thus represents a promising lead candidate for the development of new anti-allergic drugs. In vitro data from the immortalized in vitro BBB model indicated a high BBB permeation potential for indolinone, which was corroborated by in vitro permeability data obtained from two well-established primary animal models, and by in silico prediction data. Furthermore, P-gp interaction of the compound was assessed with the aid of two specific efflux pump interaction assays. Both assays suggested that no active mediated transport mechanism was involved for the compound.
In conclusion, we have successfully established and validated an easy and fast to set up human in vitro BBB model, and applied it to in-house BBB drug permeability assays of promising lead candidates of natural origin.