Characterization of human fetal brain endothelial cells reveals barrier properties suitable for in vitro modeling of the BBB with syngenic co-cultures

Abstract

Endothelial cells (ECs) form the basis of the blood-brain barrier (BBB), a physical barrier that selectively restricts transport into the brain. In vitro models can provide significant insight into BBB physiology, mechanisms of human disease pathology, toxicology, and drug delivery. Given the limited availability of primary human adult brain microvascular ECs ( aBMVECs), human fetal tissue offers a plausible alternative source for multiple donors and the opportunity to build syngenic tri-cultures from the same host. Previous efforts to culture fetal brain microvascular ECs ( fBMVECs) have not been successful in establishing mature barrier properties. Using optimal gestational age for isolation and flow cytometry cell sorting, we show for the first time that fBMVECs demonstrate mature barrier properties. fBMVECs exhibited similar functional phenotypes when compared to aBMVECs for barrier integrity, endothelial activation, and gene/protein expression of tight junction proteins and transporters. Importantly, we show that tissue used to culture fBMVECs can also be used to generate a syngenic co-culture, creating a microfluidic BBB on a chip. The findings presented provide a means to overcome previous challenges that limited successful barrier formation by fBMVECs. Furthermore, the source is advantageous for autologous reconstitution of the neurovascular unit for next generation in vitro BBB modeling.

Keywords: BBB model; Blood–brain barrier; human fetal brain endothelial cells; syngenic.

Figure 1.

Immunofluorescence staining of protein localization in primary human fetal brain microvessels. Human fetal brain microvessels aged 7 weeks, 12 weeks, and 19 weeks were stained for tight junction proteins. Vessels were imaged using confocal microscopy. (a) Immunofluorescence staining of primary human fetal brain microvessels (7 and 12 weeks) showing continuous expression pattern of ZO-1 (green) and tricellulin (yellow), while occludin (Cyan) and claudin-5 (white) show an intracellular diffuse expression pattern. In contrast, vessels isolated post-barrier development (19 weeks) demonstrate continuous expression of tight junction proteins at the cell borders for occludin (Cyan), ZO-1 (green), claudin-5 (white) and Tricellulin (yellow). White arrows indicate continuous TJP expression while white ellipses highlight diffuse intracellular staining. Scale bar at 20 µm. (b) Bar graph quantification of line intensity profiles obtained using MeasurementPro (Imaris, Bitplane) for vessels of similar caliber (<15 µm diameter) for each of the gestational ages (Avg ± SEM).

Figure 2.

mRNA expression of tight junction proteins, adhesion molecules, and transporters in fetal and adult brain microvascular endothelial cells (BMVECs). mRNA from five fBMVEC and five aBMVEC donors was analyzed for gene expression of tight junction proteins (occludin, ZO-1, claudin-5), transporters (GLUT-1, BCRP-1, MRP-1, MRP-5, PGP-1) and a ubiquitous endothelial adhesion molecule (PECAM-1). Box and whisker plots are shown ofΔCT values for occludin and ZO-1 (a), BCRP-1, MRP-1, MRP-5, and PGP-1 (b), which were all non-significant between fBMVEC and aBMVEC donors. Fetal donors expressed lower levels of Claudin-5 (a) and Glut-1 (b) and higher levels of PECAM-1 (c) when compared to adults. Student’s t-test *p < 0.05.

Figure 3.

Protein expression analysis and immunofluorescence staining of fBMVECs. (a) Cell lysates for fetal and adult donors separated by SDS-page and probed for zona occludens-1 (ZO-1), multidrug resistance-associated protein-5 (MRP-5), platelet endothelial cell adhesion molecule-1 (PECAM-1), transferrin receptor (TfR), occludin, glucose transporter-1 (Glut-1), claudin-5, and β-Actin. Representative images for two fetal and two adult donors are shown. (b) Bar graph densitometry quantification showing non-significant differences in expression levels between fetal donors and adult donors. (c) Immunofluorescence staining of TJP occludin expression (white) in fBMVECs (top) and aBMVECs (bottom) showing occludin localization at cell boundaries (Dapi, Cyan). 10 × images. Scale bars 50 µm.

Figure 4.

Evaluation of barrier trans-endothelial electrical resistance (TEER) for fBMVECs and aBMVECs. (a) fBMVECs, aBMVECs, human coronary artery endothelial cells (HCAECs) and human embryonic kidney cells 293 (HEK293) were grown on electrodes, and resistance (Ohms) was measured over time. After three to five days, the media was exchanged and representative traces for the following 24 h (normalized to HEK absolute resistance values) are shown. Non-barrier forming ECs are ∼10-fold tighter than HEK cells, and barrier-forming ECs (fetal and adult) are ∼50-fold tighter than HEK cells. (b) to (d). fBMVECs (b), aBMVECs (c), and mixed cultures (d) were challenged with barrier tightening (dexamethasone) and disrupting (IL-1β) agents. Pure barrier ECs (b & c) demonstrate changes in resistance in response to barrier tightening and disrupting agents in contrast to impure EC monolayers which fail to form tight barriers. TEER values are represented as the average (line) normalized TEER ± SEM (b) to (d).

Figure 5.

TNF-α induces upregulation of adhesion molecule expression in fBMVECs. Fetal and adult BMVECs were stimulated with TNF-α (100 ng/mL), harvested at 0, 6, 18, 24, 48 h and analyzed by flow cytometry. Bar graph and histograms show the upregulation of adhesion molecules ICAM-1 (a,c) and VCAM-1 (b,d) in response to TNF-α which was not statistically significant between fBMVECs and aBMVECs.

Figure 6.

Functional assays of immune–endothelial interactions using fBMVECs. (a) Adhesion assay: fBMVECs were treated with TNF-α (20 ng/mL) for 18 h. Treatments were removed prior to addition of monocytes. Data are represented as fold difference (mean ± SEM) of adherent cells. Results show a 2.5-fold increased in adherent monocytes on activated as compared to unactivated ECs. (b) Trans-endothelial migration assay: Monocytes were added to the upper chamber of Transwell® membranes and allowed to migrate through fBMVECs towards MCP-1 in the lower chamber. Data are represented as fold difference (mean ± SEM) of migrated cells. Results show a ∼ 2-fold increase in migrated monocytes with the addition of MCP-1. (c) Representative images of the underside of the Transwell® showing migrated Calcein-AM labeled monocytes with or without MCP-1 in the lower chamber. Scare bars 100 µm.

Figure 7.

Proof-of-concept syngenic BBB model in a microfluidic chip using fBMVECs, astrocytes and pericytes derived from the same host. (a) and (c). Bright field images of fBMVECs grown in an outer endothelial compartment and an astrocyte/pericyte mixed culture in a separate inner “CNS compartment”. (b) and (d). Astrocyte/pericyte cultures were labeled with Calcein-AM and then seeded into the “CNS compartment”. Yellow arrowheads indicate areas where astrocytes span the column-matrix to connect with the endothelial compartment (d, insert). (e) and (f). fBMVECs align in the direction of fluid flow in microfluidic chambers. Representative images of fBMVECs before (e) and after (f) exposure to shear stress (10 dyn/cm²). Yellow arrow indicates the direction of the fluid flow. Scale bars at 100 µm. (g) Bright field image (4×) of the microfluidic chip showing a mixed neuronal-glial-pericyte culture in the “CNS compartment” and fBMVECs in the outer endothelial compartment. (h) Bright field close-up of (g). (10×). (i) Permeability of fBMVECs during baseline and LPS insult. Fluorescently labeled tracers (sodium fluorescein, NaFluo, 3 kDa cascade blue dextran, CB-Dex and 40 kDa tetramethyrhodamine dextran, TMR-Dex) were perfused in the endothelial compartment to measure apparent permeability. The “CNS compartment” for the permeability studies include the neuronal–glial–pericyte mixture shown in (g) and (h). (i) Representative images for brightfield, NaFluo, CB-Dex, and TMR-Dex showing baseline and LPS stimulation (200 ng/mL) at time points 0, 15 min, and 30 min. (j) Bar graph quantification of the apparent permeability (P_app) for baseline and LPS insult (Avg ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001).