Microengineered human blood-brain barrier platform for understanding nanoparticle transport mechanisms

Abstract

Challenges in drug development of neurological diseases remain mainly ascribed to the blood-brain barrier (BBB). Despite the valuable contribution of animal models to drug discovery, it remains difficult to conduct mechanistic studies on the barrier function and interactions with drugs at molecular and cellular levels. Here we present a microphysiological platform that recapitulates the key structure and function of the human BBB and enables 3D mapping of nanoparticle distributions in the vascular and perivascular regions. We demonstrate on-chip mimicry of the BBB structure and function by cellular interactions, key gene expressions, low permeability, and 3D astrocytic network with reduced reactive gliosis and polarized aquaporin-4 (AQP4) distribution. Moreover, our model precisely captures 3D nanoparticle distributions at cellular levels and demonstrates the distinct cellular uptakes and BBB penetrations through receptor-mediated transcytosis. Our BBB platform may present a complementary in vitro model to animal models for prescreening drug candidates for the treatment of neurological diseases.

Fig. 1. Microengineered human blood–brain barrier (BBB) model.

a Schematic description of the BBB consisting of endothelial cells (ECs) along the blood vessel under continuous blood flow, pericytes covering the endothelial monolayer, and astrocytes with aquaporin-4 (AQP4) expression at their end-feet near the blood vessel. b Schematic description of our microengineered human BBB model. c 3D configuration of the BBB model showing human brain microvascular endothelial cells (HBMECs) (ZO-1, red) and human astrocytes (HAs) (GFAP, white) (scale bar = 100 µm). d Explosion view of the device consisting of upper vascular layer, porous membrane, lower perivascular layer, and glass slide. e A photo of the device after completing fabrication of the device (blue: upper channel and red: lower channels) (scale bar = 500 µm). f Lower layer consisting of three parallel channels separated by series of micropillars (red: center channel) (scale bar = 500 µm). g Cross-section of the device after fabrication (along A-B in Fig. 1f) (scale bar = 200 µm). h Cell metabolic activities assessed by a (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) (MTS) assay (E + G: 1:1:1 mixture of endothelial medium, astrocyte medium, and microglia medium, E + G + P: 1:1:1:1 mixture of endothelial medium, astrocyte medium, microglia medium, and pericyte medium) (Data represent mean ± s.d. of n = 6 for each condition, *p < 0.05 and ****p < 0.001 versus each cell culture medium by student t-test). i Bottom view of the device with endothelial monolayer (ZO-1, red) and astrocytic network (GFAP, white) (scale bar = 50 µm). j Endothelial monolayer (ZO-1, red) supported by a layer of human brain vascular pericytes (HBVPs) (α-SMA, green) (scale bar = 50 µm). k Aquaporin-4 (AQP4, yellow) and α-syntrophin (α-syn, magenta) expressions at astrocytic end-feet (GFAP, white) underneath a porous membrane (indicated as the dotted line) in the lower channel (Blue arrows indicate co-localization of AQP4 with α-syn.) (scale bar = 50 µm). All images are representative ones from at least five biological and three technical replicates.

Fig. 2. Barrier integrity of the endothelial monolayer for the BBB chip.

a Heat map of RT-qPCR results of HBMECs in mono-culture and tri-culture systems (n = 3 for each condition). b, c Gene expression of HBMECs in mono-culture and tri-culture systems including junctional proteins (b) and receptor proteins (c) (n = 3 for each condition, *p < 0.05 by student t-test). d Tight endothelial monolayer (ZO-1, red; DAPI, blue) formed in the upper channel of the device. e Pericytes cultured underneath the porous membrane where an endothelial monolayer is constructed on the other side (α-SMA, green; DAPI, blue). f, g Astrocytes with star-shaped morphology labeled with GFAP (GFAP, white) (f) and S100β (S100β, magenta) (g). h Bi-layer of an endothelial monolayer (ZO-1, red) and HBVPs (α-SMA, green). i Astrocytic end-feet stretching to the endothelium in 3D cellular network (GFAP, white; DAPI, blue). j 3D BBB structure constructed in a device (ZO-1, red; α-SMA, green; GFAP, white; DAPI, blue). The fluorescence intensity profiles indicate the distribution of ZO-1, α-SMA, and GFAP in the image. k Transendothelial electrical resistance (TEER) measured across the membrane between the upper and lower layers with an endothelial monolayer (EC) and an endothelial monolayer with pericytes and astrocytes (BBB) (n = 11 for EC and n = 12 for BBB, **p < 0.01 by student t-test). l TEER measured from BBB models under different levels of shear stress (n = 5 for No shear, n = 4 for 0.4 dyne cm⁻¹, and n = 12 for 4 dyne cm⁻², *p < 0.05 by student t-test). m Permeability coefficients calculated from the diffusion of 4 kDa and 40 kDa FITC-dextran through a membrane (No cell), an endothelial monolayer (EC), an endothelial monolayer co-cultured with pericytes and astrocytes (BBB) (n = 4 for each condition, *p < 0.05 and ****p < 0.001 vs. No cell, ^#p < 0.05 and ^##p < 0.01 vs. EC, all by student t-test). Data are presented as mean ± s.e.m. All scale bars = 50 µm. All images are representative ones from at least five biological and three technical replicates.

Fig. 3. 3D culture of astrocytes with AQP4 polarization in the BBB chip.

a, b Representative morphology of human astrocyte (HA) cultured on a Matrigel-coated 2D surface (a) and in 3D Matrigel (b) (GFAP, white; DAPI, blue) (scale bars = 50 µm). c Cell body size of HAs cultured in 2D and 3D (n = 69 for 2D and 39 for 3D, ***p < 0.005 by student t-test). d Process length of HAs cultured in 2D and 3D (n = 1352 for 2D and 1302 for 3D, ****p < 0.001 by student t-test). e Gene expression of reactive gliosis markers in HAs cultured in 2D and 3D (n = 4 for each condition, ***p < 0.005 and ****p < 0.001 by student t-test). f, g Gene expression of lipocalin-2 (LCN2) in HAs cultured on a Matrigel-coated 2D surface (f) and within 3D Matrigel (g) with interleukin-1β (IL-1β) treatment demonstrating the ability to model reactive astrocytes more effectively in 3D (n = 4 for each condition, **p < 0.01, ***p < 0.005, and ****p < 0.001 by student t-test). h Quantitative analysis of AQP4 polarization by measuring AQP4 distribution in vascular and parenchymal side in the perivascular channel. i Co-localization of AQP4 (AQP4, yellow) and α-syn (α-syn, magenta) at astrocytic end-feet (scale bars = 20 µm). j Distribution of AQP4 (AQP4, yellow) along the cell bodies of HAs (GFAP, white; DAPI, blue) in the channel (scale bars = 50 µm). k Polarized expression of AQP4 to the vascular side in the perivascular channel (n = 4 for each condition, *p < 0.05 by student t-test). Data are presented as mean ± s.e.m. All images are representative ones from at least five biological and three technical replicates.

Fig. 4. On-chip BBB transport analysis of HDL-mimetic nanoparticles.

a Discoidal engineered HDL-mimetic nanoparticle with apolipoprotein A1 (eHNP-A1) consisting of lipid, apolipoprotein A1 and fluorescent marker. b Transmission electron microscopy (TEM) image of the synthesized eHNP-A1 (scale bar = 20 nm). c Size distribution of the synthesized eHNP-A1. d Composition of the eHNP-A1. e Biodistribution of the eHNP-A1. f Quantification of the relative fluorescence intensity in each organ (Data represent mean ± s.e.m from n = 4 for each condition). g eHNP-A1 accumulated in the mouse brain (scale bar = 50 µm). h Schematic description of eHNP-A1 distribution in the BBB model showing (1) eHNP-A1 remaining in the vascular channel, (2) eHNP-A1 interact with endothelial cells (HBMECs), (3) eHNP-A1 translocated to the perivascular channel, and (3-1) eHNP-A1 interact with astrocytes (HAs). i, j Confocal images showing eHNP-A1 within the HBMEC monolayer (i) and HAs (j) in a BBB chip (scale bars = 50 µm). k, l Relative fluorescence intensity of sampled culture medium containing eHNP-A1 from the upper channel (k) and the lower channel (l) after 2 h of eHNP-A1 incubation in the vascular channel (k: n = 12; l: n = 5, **p < 0.01 by student t-test). Data show mean ± s.e.m. m, n Distribution of eHNP-A1 in control (m) and the block lipid transport-1 (BLT-1) treated microengineered BBB model (n). o, p Representative fluorescence-activated cell sorting (FACS) plot for the numbers of eHNP-A1 positive HBMECs and HAs in control (o) and BLT-1 treated BBB models (p). q Cellular uptake of eHNP-A1 in the BBB chip quantified from FACS analysis (n = 3). Data are presented as mean ± s.e.m. All images are representative ones from at least three biological and three technical replicates.