An isogenic neurovascular unit model comprised of human induced pluripotent stem cell-derived brain microvascular endothelial cells, pericytes, astrocytes, and neurons

IMR90C4 (WiCell, Madison, WI) iPSCs were utilized for differentiations of all cell types. iPSCs were cultured between passages 30–50 on Matrigel (BD Biosciences, San Jose, CA) with daily mTESR1 (WiCell) medium changes. BMECs were differentiated as previously described [24]. In brief, iPSCs were singularized with Accutase (Life Technologies, Carlsbad, CA) and expanded to 30,000 cells/cm² prior to the initiation of the differentiation. Unconditioned medium (UM; 100 mL Knock-out serum replacement, 5 mL non-essential amino acids (NEAA), 2.5 mL of gluta-max, 392.5 mL of Dulbecco’s modified Eagle’s medium (DMEM)/F12 (1:1) (Life Technologies) and 3.5 μL of β-mercapto-ethanol (Sigma, St Louis, MO) was exchanged daily for 6 days. UM was exchanged with EC +/+ medium (200 mL human Endothelial Serum-Free medium (hESFM; Life Technologies), 20 ng/mL basic fibroblast growth factor (bFGF; WiCell) and 1% platelet-derived bovine serum (PDS; Biomedical Technologies, Tewksbury, MA)) for 48 h supplemented with (treated) or without (non-treated) 10 μM retinoic acid (RA; Sigma). Neurons and astrocytes were derived from intermediary EZ-sphere and astro-sphere populations (respectively) as previously described [25]. EZ- and astro-spheres were singularized with Accutase and seeded at 25,000 cells/cm² onto Matrigel coated plates. Neurons were differentiated and cultured in neuron medium [DMEM/F12 (70:30) supplemented with 1% penicillin–streptomycin, 2% B27 minus Vitamin A (Life Technologies) and 2 μg/mL heparin (Sigma)] with every-other day medium exchanges for 2 weeks. Astrocytes were differentiated and cultured in astrocyte medium (DMEM/F12 (1:1) supplemented with 1% NEAA, 1% N2 supplement; Life Technologies, and 2 μg/mL heparin) with every-other day medium exchanges for 2 weeks. Pericytes were differentiated from iPSCs as previously described [36]. Pericytes were derived in a two-step process. Initially iPSCs were differentiated in E6 medium [DMEM/F12, l-ascorbic acid-2-phosphate magnesium (Sigma), sodium selenium (Sigma), insulin (Sigma), NaHCO₃ (Sigma), transferrin (Sigma)] supplemented with dorsomorphin (Tocris, Bristol, United Kingdom), SB431542 (Tocris) CHIR99021 (Tocris), bFGF (Waisman Biomanufacturing, Madison, WI), and heparin (Sigma)) to a neural-crest stem cell population. CD271 positive cells were enriched with neural crest stem cell microbeads (Miltenyi, Bergisch Gladbach, Germany) and seeded at 10,000 cells/cm² onto non-coated plates (Corning). Following 6 days of expansion in E6 medium supplemented with 10% FBS (Thermofisher), pericytes were seeded at 10,000 cells/cm² for co-culture studies with BMECs.

3T3 mouse fibroblasts (ATCC, Manassas, VA) were cultured on Matrigel coated-plates with daily medium changes of DMEM (Life Technologies) supplemented with 10% FBS (Life Technologies). Mouse 3T3 fibroblasts served as a non-neural co-culture control.

BMECs were differentiated from iPSCs as previously discussed. Following differentiation, BMECs were singularized with Accutase and seeded onto Transwells (Corning Transwell polyester filters, 0.4 μm pore diameter; 1.12 cm², Corning, NY) coated with collagen IV/fibronectin (Sigma) at a density of 1 × 10⁶ cells/cm². Immediately following seeding, BMEC-coated Transwells were then placed into plates with either no cell types (monoculture), neuron: astrocyte (1:3) mixture, 3T3 fibroblasts, pericytes, or pericytes for the initial 24 h followed by a mixture of neurons: astrocytes (1:3) for the duration of the experiment. All experimental groups were in EC +/+ medium (hESFM supplemented with 1% PDS and 20 ng/mL bFGF) with or without retinoic acid for the initial 24 h and switched to EC ± medium (hESFM supplemented with 1% PDS) for the remainder of the experiment. All experiments occurred immediately following 48 h in co-culture unless otherwise stated.

Dextran (Alexa-Fluor^® 488, 10 kDa, 10 μM; Sigma) was used to measure the amount of transcytotic activity in the BMECs. Transwells were removed from co-culture and replaced in EC ± medium in the basolateral chamber. Dextran, diluted in EC ± was added to the apical side of the transwells on a rotating platform at 37 °C or 4 °C for 2 h. To determine the rate of transcytosis, media was collected from the basolateral chamber and read on a fluorescent plate reader. BMECs were then rinsed twice in phosphate buffer solution (PBS; Sigma) and lysed with radioimmunoprecipitation assay (RIPA; Sigma). After trituration the lysate was collected and quantified using a fluorescent plate reader, this value is the Dextran accumulated at this time point. Both transcytosis and accumulation values were normalized to protein content as determined by a bicinchoninic acid assay (BCA; Sigma).

Barrier tightness was measured by the voltage difference from the movement of ions across the Transwell membrane. Epithelial Volt/Ohm Meter (EVOM) with STX2 electrodes (World Precision Instruments) was used to measure the TEER value in ohms starting at day 9 of differentiation and continuing up to a week after purification to create an extended barrier tightness profile. The values are normalized by subtracting the background (TEER from a blank well) and then multiplied by the surface area (1.12 cm²) of the transwell filter and reported as ohms × cm².

Cells were fixed with cold methanol (100%; Sigma) or 2% paraformaldehyde (PFA; Sigma) in PBS for 15 min at room temperature. Following three washes in PBS the cells were blocked in 10% goat serum (Sigma) for 30 min. Cells were incubated with primary antibodies at 4 °C overnight on a rotating platform (Table 1). Following three PBS washes, cells were incubated with secondary antibodies for 1 h on a rotating platform at room temperature. Cells were incubated with a nuclear counterstain 4′,6-Diamidino-2-phenylindole dihydrochloride (DAPI; Sigma) for 15 min then washed in PBS and viewed on the Olympus epifluorescence microscope (Center Valley, PA). Images were analyzed using Image J, tight junction proteins (claudin-5, occludin and zo-1) were analyzed using threshold analysis and perimeter counting tools to determine the area of each image that exhibited occludin, claudin-5, or zo-1 immunoreactivity to determine area fraction index as previously discussed [25].

BMECs were removed from co-culture groups and placed in EC ± . Cyclosporin A (CsA, 10 μM; Sigma) was used as an inhibitor to P-glycoprotein efflux transporter (PGP), whereas the small molecule Rhodamine123 (Rh123, 10 μM; Sigma) is a PGP substrate. Half the wells of each experimental group were treated with CsA for 30 min in both apical and basolateral chambers of the transwell. A solution of diluted Rh123 in EC ± with and without CsA, was added to the apical side of all transwells for 1 h at 37 °C on a rotating platform. Media collected from the basolateral chamber is quantified on the fluorescent plate reader. To ensure that barrier integrity was maintained for the duration of the experiment, TEER values were monitored throughout the experimental timeline. NVU co-culture did not affect baseline Rh123 transport therefore the data is presented as a percentage change from no CsA treatment with each respective co-culture condition.

Data for Figs. 2, 3, 4 are presented as mean ± SD of three independent differentiations. Due to variable TEER baseline recordings, TEER values presented in Fig. 1 are mean ± SD of three replicates from a single differentiation. Two additional independent differentiations were repeated for verification of reported statistical trends. SigmaStat 3.0 software (Systat Software, San Jose, CA) was used for statistical analysis. Student’s t-test and one-way analysis of variance (ANOVA) with Holm-Sidak correction for all comparisons were utilized when appropriate and described within the figure legends for each particular data set.

Established protocols were used to differentiate BMECs [24], astrocytes (from EZ spheres) [25], neurons (from EZ-spheres) [25, 37], and brain-like pericytes (from neural crest progenitors) [36] from the IMR90C4 iPSC line. Following differentiation, iPSC-derived BMECs were seeded onto Transwells with different combinations of fully differentiated astrocytes, pericytes and neurons. The co-cultured cells were differentiated and maintained in their respective media prior to co-culture with BMECs. Immediately following seeding onto Transwells, BMECs were co-cultured with either iPSC-derived pericytes alone (P), a neuron and astrocyte mixture (1:3 NA) or a sequential pericyte co-culture followed by a neuron and astrocyte co-culture (PNA) (Fig. 1a). Monocultured BMECs (MC) and 3T3 fibroblast co-cultures (F) were used as comparative controls (Fig. 1a). All co-culture conditions were conducted initially in EC +/+ medium (hESFM supplemented with bFGF and PDS) for 24 h and then switched to an EC ± medium (hESFM supplemented with PDS) for the duration of the time course. To determine the effects that co-cultured cell types had on BMEC barrier properties, transendothelial electrical resistance (TEER) was monitored from 24 h after initiation of co-culture to 96 h (Fig. 1b). TEER values reached a maximum value at 48–72 h in all experimental conditions. Monocultured BMECs exhibited a maximum TEER value of 310 ± 19 Ω × cm², whereas BMECs co-cultured with pericytes reached a maximum TEER value of 564 ± 21 Ω × cm² (p < 0.05) and remained significantly elevated above monoculture and fibroblast co-culture for the duration of the experiment (p < 0.05). As previously demonstrated, EZ-sphere derived neurons and astrocytes (1:3) significantly elevated barrier tightening in BMECs with a maximum TEER of 720 ± 38 Ω × cm² (p < 0.05), and remained significantly elevated (p < 0.05). Sequential BMEC co-culture with pericytes followed by a mixture of neurons and astrocytes resulted in the greatest induction of TEER above all other experimental groups at 48 h (1155 ± 64 Ω × cm²; p < 0.05) and remained significantly elevated above all co-culture groups (p < 0.05). All NVU co-culture groups elevated BMEC barrier tightness above monoculture and 3T3 fibroblast co-culture at all the time points of the experiment (p < 0.05). Finally, mouse 3T3 fibroblasts were utilized as a non-inductive co-culture cell type and did not elevate TEER above monoculture TEER levels at any time point (Fig. 1b, p > 0.05).

We have previously demonstrated that retinoic acid (RA) can significantly elevate the barrier properties of iPSC-derived BMECs [23]. To determine the potential synergistic effects of co-culture on a BMEC population that was exposed to RA during differentiation, we monitored TEER of RA-treated BMECs in all of the previously mentioned NVU co-culture groups (Fig. 1c). Similar to the untreated BMECs, we observed peak TEER values 48 h following seeding onto Transwells and the initiation of co-culture. Monocultured, RA-treated BMECs (M) had a maximum TEER value of 2525 ± 149 Ω × cm² and were undistinguishable from BMECs in co-culture with 3T3 fibroblasts (F) 2321 ± 92 Ω × cm² (p > 0.05). Sequential pericyte co-culture followed by neuron and astrocyte co-culture (PNA) significantly elevated TEER (5486 ± 396 Ω × cm²; p < 0.05) compared with pericyte co-culture (P) (4015 ± 173 Ω × cm²; p < 0.05) or neuron and astrocyte co-culture (NA) (4726 ± 55 Ω × cm²; p < 0.05). Similar to untreated BMECs, NVU co-culture elevated BMEC barrier tightness at all time points throughout the experiment (p < 0.05). To confirm that BMECs, neurons, astrocytes, and pericytes still expressed key markers after co-culture, we verified that RA-treated BMECs expressed the endothelial cell markers PECAM-1 and VE-cadherin, the BBB glucose transporter Glut-1, and tight junction proteins: claudin-5 and ZO-1 (Fig. 1d). Neurons expressed β-tubulin III, astrocytes expressed glial fibrillary acidic protein (GFAP), and pericytes expressed platelet derived growth factor receptor-β (PDGFR-β) and neuron glial antigen-2 (NG2) (Fig. 1e). Taken together, the PNA configuration provided the most significant induction of barrier properties and increased the TEER in RA-treated and untreated BMECs.

Immunocytochemistry was utilized to investigate changes in tight junction protein localization during BMEC barrier tightening induced by NVU cell co-culture. Monoculture, RA-treated BMECs expressed junctionally localized occludin, claudin-5 and ZO-1 (Fig. 2a). Compared with monocultured RA-treated BMECs, co-culture with pericytes, a neuron: astrocyte mixture or pericytes followed by a neuron: astrocyte mixture, qualitatively increased occludin localization to the tight junctions (Fig. 2a). Quantification of area fraction index revealed that after 48 h of pericyte co-culture (P), BMECs had a slight, but statistically insignificant elevation in junctional occludin immuno-reactivity (19 ± 11%; p > 0.05) (Fig. 2b). However, BMECs in co-culture with pericytes for 24 h followed by a neuron: astrocyte (1:3) mixture for 24 h (PNA), exhibited significantly elevated junctional occludin levels (54 ± 17%; p < 0.05); however, the most significant boost in junctional occludin occurred after BMECs in co-culture for 48 h with a neuron: astrocyte (1:3) mixture (69 ± 17%; p < 0.05). However, the area fraction index of claudin-5 and ZO-1 were unchanged in BMECs following co-culture with pericytes, neurons, and astrocytes (Fig. 2b). These results suggest that enhanced barrier tightness following co-culture, at least in part, were the result of improved occludin localization and maintenance.

To determine the effects of co-culture on P-glycoprotein (PGP) efflux transporter activity in RA-treated BMECs, the transport of rhodamine 123 across the BMEC monolayer was measured. Rhodamine 123 transport was compared between BMECs in monoculture and BMEC in co-culture with either pericytes (P), a neuron: astrocyte (1:3) mixture (NA), or pericytes followed by a neuron: astrocyte mixture (PNA). Immunocytochemistry revealed that, qualitatively, BMECs either in monoculture or in co-culture expressed similar levels of PGP (Fig. 3a). Correlating with the immunocytochemistry images, PGP function was also unchanged by any co-culture configuration. BMECs in monoculture exhibited an increase in rhodamine 123 following CsA inhibition (45 ± 7%; p < 0.05), benchmarking the baseline PGP activity in RA-treated BMECs (Fig. 3b). A similar increase in rhodamine 123 following CsA inhibition was observed in BMECs following co-culture (P: 45 ± 19%; PNA: 42 ± 7%; NA: 41 ± 5%; p < 0.05); however, these data were indistinguishable between BMECs in monoculture and co-culture (p > 0.05 n.s.). These data reveal that PGP is present and active in RA-treated BMECs as previously demonstrated [23] and co-culture with pericytes, neurons and astrocytes alone or in combination does not affect this activity.

In previous studies, pericytes have been demonstrated to reduce the non-specific transcytosis of tracers and large molecules through BMECs [6, 38, 39]. To determine if co-culture with brain-like pericytes could also elicit similar effects in iPSC-derived RA-treated BMECs, the effects of co-culture on uptake and transcytosis of 10 kDa fluorescently tagged dextran was employed. BMECs in monoculture or 48 h following co-culture with pericytes (P), a neuron: astrocyte (1:3) mixture (NA), or pericytes followed by a neuron: astrocyte mixture (PNA) were incubated with fluorescently-tagged dextran in a Transwell setting. After addition of dextran to the top chamber, transport was measured both as cellular accumulation in BMECs and by accumulation in the lower chamber (transcytosis) then normalized to accumulation and transcytosis using monocultured BMECs (100%) (Fig. 4a, b). Following pericyte co-culture, BMECs exhibited significantly lower accumulation and transcytosis of dextran (65 ± 22% and 45 ± 5%; respectively; p < 0.05). While pericyte co-culture followed by neuron: astrocyte co-culture (PNA) did not yield a statistically significant decrease in dextran accumulation (77 ± 15%; p > 0.05), the amount of dextran that transcytosed across BMECs was significantly reduced (45 ± 8%; p < 0.05). Neuron: astrocyte co-culture (NA) lacking pericyte influences did not significantly affect dextran accumulation or transcytosis across BMECs (103 ± 14%, 83 ± 5%; respectively; p > 0.05). Non-specific transcytosis is minimal in BMECs compared to peripheral endothelial cells, similarly less than 0.05% of the dextran added to the apical chamber accumulated within the cell and less than 0.04% crossed the BMEC monolayer. To rule out the changes in transcytosis being a result of the increased passive barrier properties upon co-culture, dextran accumulation and transcytosis was also conducted at 4 °C conditions where vesicular transport processes are significantly inhibited (< 0.003% of the total dextran), but passive diffusion is comparatively unaffected (Fig. 4a, b). Under these conditions, the dextran transport was unaffected following pericyte co-culture compared to monoculture conditions, indicating that the increases in TEER do not explain the change in dextran transport. Taken together, these data indicate that brain-like pericyte co-culture alone or in conjunction with neurons and astrocytes can decrease the uptake and non-specific transcytosis of large molecules across RA-treated BMECs.

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.