Tissue-engineered blood-brain barrier models via directed differentiation of human induced pluripotent stem cells

Abstract

Three-dimensional (3D) tissue-engineered models of the blood-brain barrier (BBB) recapitulate in vivo shear stress, cylindrical geometry, and cell-ECM interactions. Here we address four issues associated with BBB models: cell source, barrier function, cryopreservation, and matrix stiffness. We reproduce a directed differentiation of brain microvascular endothelial cells (dhBMECs) from two fluorescently labeled human induced pluripotent stem cell lines (hiPSCs) and demonstrate physiological permeability of Lucifer yellow over six days. Microvessels formed from cryopreserved dhBMECs show expression of BBB markers and maintain physiological barrier function comparable to non-cryopreserved cells. Microvessels displaying physiological barrier function are formed in collagen I hydrogels with stiffness matching that of human brain. The dilation response of microvessels was linear with increasing transmural pressure and was dependent on matrix stiffness. Together these results advance capabilities for tissue-engineered BBB models.

Figure 1

Directed differentiation of human brain microvascular endothelial cells (dhBMECs) from two induced pluripotent stem cell lines (BC1-GFP and C12-RFP). (a) Schematic illustration of differentiation over eight days. (b,c) Phase and fluorescence images of a representative BC1-GFP differentiation. (d,e) Phase and fluorescence images of a representative C12-RFP differentiation. Cytoplasmic fluorescence is maintained across both differentiations.

Figure 2

Transendothelial electrical resistance (TEER) and immunocytochemistry of dhBMECs derived from the BC1-GFP and C12-RFP iPSC lines. (a) Schematic illustration of timeline for cell seeding following differentiation. (b) Maximum TEER for each cell line. (c) TEER values for each cell line over two weeks. TEER data represent mean ± SEM for 3–4 technical replicates across 3 independent differentiations for BC1-GFPs and 4 independent differentiations for C12-RFPs. (d,e) Fluorescence images of representative dhBMEC monolayers stained for CD31, ZO-1, claudin-5, occludin, GLUT-1, p-glycoprotein (Pgp), and VE-cadherin (violet or green). Nuclei (blue).

Figure 3

Formation and live-cell imaging of 3D blood-brain barrier microvessels. (a) Schematic illustration of microvessel fabrication. (b) Formation of a confluent monolayer occurs within two days post-seeding. (c) Confocal reconstructions demonstrate lumen formation on day 2. (d) Immunocytochemistry of the tight junction protein claudin-5 at the bottom microvessel plane on day 2. Microvessels were formed in 7 mg mL⁻¹ collagen I matrix cross-linked with genipin and perfused at 1 dyne cm⁻².

Figure 4

Permeability of 3D blood-brain barrier microvessels. (a) Microvessels were maintained under continual perfusion (~1 dyn cm⁻²) and fluorescent dyes introduced into the flow loop to quantify permeability. (b) Schematic illustration showing timeline for permeability experiments. Representative images are taken prior to introducing the solute, when the solute appears in the microvessel lumen, and after 60 minutes. (c,d) Representative phase / fluorescence images of Lucifer yellow and 10 kDa dextran permeability. (e,f) Quantification of Lucifer yellow and 10 kDa dextran permeability. Microvessels were formed in 7 mg mL⁻¹ collagen I matrix cross-linked with genipin. Data represent mean ± SEM for 3 microvessels from 3 independent differentiations. D. L. detection limit.

Figure 5

Utilization of cryopreserved dhBMECs in 2D and 3D BBB models. (a,b) Maximum TEER and time course over two weeks. Data represent mean ± SEM for 3 technical replicates over 3 independent differentiations. (c) Immunocytochemistry of cryopreserved BC1-GFP dhBMECs. (d) Representative phase / fluorescence overlays of Lucifer yellow permeability. (e,f) Permeability of Lucifer yellow and 10 kDa dextran conducted on cryopreserved BC1-GFP dhBMECs on day 2 following seeding. Data represent mean ± SEM for 3 microvessels from 3 independent differentiations. D. L. detection limit.

Figure 6

Role of ECM stiffness on structure and function of blood-brain barrier microvessels. (a) Comparison of Young’s modulus for bulk hydrogels: (1) 7 mg mL⁻¹ collagen cross-linked with genipin, (2) 7 mg mL⁻¹ collagen I, and (3) 5 mg mL⁻¹ collagen I, and mouse brain. Data represent mean ± SEM for at least 8 bulk gels across 3 independent preparations, and 5 independent mouse samples. (b) Representative time course images of microvessels for each matrix condition. Monolayers on the softest gel condition, 3 mg mL⁻¹ type I collagen, failed to reach confluence by day 2; such vessels were subsequently failed by collapse of the lumen, as indicated by the red outline on the corresponding time course image. (c) Survival (%) of microvessels over time based on characterization of confluent endothelium and continuous perfusion. Data represents longevity for at least five microvessels across at least 3 independent differentiations. (d) Representative phase/fluorescence overlays of Lucifer yellow permeability of microvessels in 7 mg mL⁻¹ and 5 mg mL⁻¹ collagen I on day 2 following seeding. (e,f) The influence of matrix stiffness on permeability of Lucifer yellow and 10 kDa dextran permeability on day 2. Data represent mean ± SEM for 3 microvessels from at least 2 independent differentiations. (g,h) Dilation of microvessels in response to changes in transmural pressure in 7 mg mL⁻¹ microvessels cross-linked with genipin and 5 mg mL⁻¹ microvessels. Under baseline conditions, the head was 5 cm water corresponding to a transmural pressure of approximately 0.05 Pa. Data represent mean ± SEM for 3 microvessels from 2 independent differentiations.