Wnt activation of immortalized brain endothelial cells as a tool for generating a standardized model of the blood brain barrier in vitro

Abstract

Reproducing the characteristics and the functional responses of the blood-brain barrier (BBB) in vitro represents an important task for the research community, and would be a critical biotechnological breakthrough. Pharmaceutical and biotechnology industries provide strong demand for inexpensive and easy-to-handle in vitro BBB models to screen novel drug candidates. Recently, it was shown that canonical Wnt signaling is responsible for the induction of the BBB properties in the neonatal brain microvasculature in vivo. In the present study, following on from earlier observations, we have developed a novel model of the BBB in vitro that may be suitable for large scale screening assays. This model is based on immortalized endothelial cell lines derived from murine and human brain, with no need for co-culture with astrocytes. To maintain the BBB endothelial cell properties, the cell lines are cultured in the presence of Wnt3a or drugs that stabilize β-catenin, or they are infected with a transcriptionally active form of β-catenin. Upon these treatments, the cell lines maintain expression of BBB-specific markers, which results in elevated transendothelial electrical resistance and reduced cell permeability. Importantly, these properties are retained for several passages in culture, and they can be reproduced and maintained in different laboratories over time. We conclude that the brain-derived endothelial cell lines that we have investigated gain their specialized characteristics upon activation of the canonical Wnt pathway. This model may be thus suitable to test the BBB permeability to chemicals or large molecular weight proteins, transmigration of inflammatory cells, treatments with cytokines, and genetic manipulation.

Figure 1. Box plots showing the genetic comparisons between immortalized mouse endothelial cell lines and primary brain microvascular endothelium as reference.

A. Global distribution of the BBB-specific genes in the bEnd5, H5V and lung cells in comparison with MBMECs (set to 1; dotted line). A non-parametric test (i.e., Wilcox test, alpha value set at 0.05) was used to determine whether or not the differences between these cell types are significant. *p<0.1; **p<0.05; n.s., not significant. B. Global distribution of the BBB-specific genes in the bEnd5 cell systems, compared with MBMECs (set to 1; dotted line). Various conditions were tested: conditioned medium from Wnt3a-transfected L-cells (Wnt3aCM) undiluted (pure) or diluted 1 to 3 in growing medium; two different commercial recombinant Wnt3a preparations (100 ng/ml; Peprotech and R&D); BIO and 6-BIO (2.5 µM). Undiluted Wnt3aCM treatment was for 3 days (first boxplot from the left) or 24 hours (second boxplot from the left), as all the other cell activations. β-CTA (cells infected with LEFΔN-βCTA) and +Astros (co-culture with astrocytes). No significance differences were detected between these conditions, although a negative trend was seen for R&D Wnt3a and BIO conditions, as median values are lower than 1.

Figure 2. β-catenin transcriptional activity improves impedance of bEnd5 cells A–D.

Transendothelial electrical resistance (TEER, top panels) and the corresponding capacitance (Ccl, bottom panels) of the endothelial monolayers at various times after they reached confluence. A. Representative TEER/Ccl measurement comparing MBMECs, bEnd5 and bEnd5 infected with LEFΔN-βCTA (βCTA-bEnd5), indicating that β-catenin transcriptional activation leads to increased electrical resistance in bEnd5 cells. Vertical line at 125 hours indicates media exchange and boxed insert shows magnification of the Ccl curves after the media exchange, highlighting pronounced lower values for the βCTA-bEnd5 compared to bEnd5 controls. B. Parental bEnd5 cells in comparison to the primary mouse MBMECs cells (n = 3). C. bEnd5 cells infected with lenti-LEFΔN-βCTA in comparison to the lenti-GFP control. D. bEnd5 cells treated with the GSK3α/β inhibitor 6-BIO (2.5 µM) in comparison to the DMSO-treated cells. E. bEnd5 cells co-cultured with astrocytes (+AC) in comparison to bEnd5 cell monocultures (−AC).

Figure 3. β-catenin transcriptional activity reduces dextran permeability of bEnd5 cells.

A,B. Endothelial monolayer permeability to FITC-labeled 38-kDa dextran, was measured as percentage (%) of relative fluorescence units (RFUs). A. bEnd5 cells infected with LEFΔN-βCTA in comparison to GFP as control. B. bEnd5 cells treated with Wnt3a conditioned medium (Wnt3aCM) in comparison to the control medium (controlCM). p values were obtained by a 2-tailed paired t-test (Graphpad Prism 5.0), using values from n = 3 independent experiments, and pairing for time points.

Figure 4. LiCl treatment improves the BBB-specific phenotype of hCMEC/D3 cells.

A. Basal mRNA expression of the BBB endothelial cell-related genes in hCMEC/D3 cells (as indicated; +++: 20<Δ²-Ct<25; ++: 25<Δ²-Ct<30; +: 30<Δ²-Ct<35). (For details see statistical analysis paragraph in Materials and Methods). B. qRT-PCR analysis from hCMEC/D3 cells treated with 10 mM LiCl compared with untreated cells. The RNA level obtained from untreated cells was set to 1 and the ratio LiCl treated versus control is shown for each gene. * p<0.05. Cldn, Claudins; VE-cad, VE-cadherin; Abcb1b, multidrug resistance protein 1; Abcg2, ATP-binding cassette transporter G2; Slc2a1, Solute carrier family 2 (facilitated glucose transporter) 1. C. hCMEC/D3 cell permeability (Pe) to Lucifer Yellow. Cells were untreated (EBM2) or treated with controlCM, 50% Wnt 3aCM or 10 mM LiCl (Pe values normalized to EBM2 Pe = 1.7×10⁻³ cm/min). D. hCMEC/D3 cell permeability (Pe) to Lucifer Yellow. Cells were incubated with DMSO as control or 20 µM XAV939 (XAV), 10 mM LiCl, or 10 mM LiCl plus 20 µM XAV939 (LiCl+XAV; Pe values normalized to DMSO Pe = 1.65×10⁻³ cm/min). All cell treatments were performed for 6 days.