Human iPSC-Derived Blood-Brain Barrier Chips Enable Disease Modeling and Personalized Medicine Applications

Human iPSC-Derived Blood-Brain Barrier Chips Enable Disease Modeling and Personalized Medicine Applications

Resource Human iPSC-Derived Blood-Brain Barrier Chips Enable Disease Modeling and Personalized Medicine Applications Graphical Abstract Authors Gad ...

7MB Sizes 0 Downloads 16 Views

Resource

Human iPSC-Derived Blood-Brain Barrier Chips Enable Disease Modeling and Personalized Medicine Applications Graphical Abstract

Authors Gad D. Vatine, Riccardo Barrile, Michael J. Workman, ..., Zhaohui Chen, Jennifer Van Eyk, Clive N. Svendsen

Correspondence [email protected] (G.D.V.), [email protected] (C.N.S.)

In Brief The blood-brain barrier (BBB) is a multicellular neurovascular unit that tightly regulates brain homeostasis and is disturbed in several neurological diseases. Combining patient-specific stem cells and microfluidic technologies, Vatine et al. have generated a personalized human BBB-Chip, which recapitulates the human BBB and can predict variability between individuals.

Highlights d

iPSC and microfluidic technologies were combined to generate a human BBB-Chip

d

Flow-induced shear and co-cultures enhance barrier performance

d

The BBB-Chip exhibits physiologically relevant TEER and can predict CNS penetrance

d

Personalized BBB-Chips can detect interindividual variability in BBB performance

Vatine et al., 2019, Cell Stem Cell 24, 995–1005 June 6, 2019 ª 2019 Elsevier Inc. https://doi.org/10.1016/j.stem.2019.05.011

Cell Stem Cell

Resource Human iPSC-Derived Blood-Brain Barrier Chips Enable Disease Modeling and Personalized Medicine Applications Gad D. Vatine,1,2,5,* Riccardo Barrile,1,3,5 Michael J. Workman,1,5 Samuel Sances,1 Bianca K. Barriga,1 Matthew Rahnama,1 Sonalee Barthakur,3 Magdalena Kasendra,3 Carolina Lucchesi,3 Jordan Kerns,3 Norman Wen,3 Weston R. Spivia,4 Zhaohui Chen,4 Jennifer Van Eyk,4 and Clive N. Svendsen1,6,* 1The

Board of Governors Regenerative Medicine Institute, Cedars-Sinai Medical Center, Los Angeles, CA 90048, USA Department of Physiology and Cell Biology and the Regenerative Medicine and Stem Cell (RMSC) Research Center, Ben-Gurion University of the Negev, Beer Sheva 84105, Israel 3Emulate, Inc., 27 Drydock Avenue, Boston, MA 02210, USA 4Advanced Clinical Biosystems Research Institute, Heart Institute, Cedars-Sinai Medical Center, Los Angeles, CA 90048, USA 5These authors contributed equally 6Lead Contact *Correspondence: [email protected] (G.D.V.), [email protected] (C.N.S.) https://doi.org/10.1016/j.stem.2019.05.011 2The

SUMMARY

The blood-brain barrier (BBB) tightly regulates the entry of solutes from blood into the brain and is disrupted in several neurological diseases. Using Organ-Chip technology, we created an entirely human BBB-Chip with induced pluripotent stem cell (iPSC)-derived brain microvascular endotheliallike cells (iBMECs), astrocytes, and neurons. The iBMECs formed a tight monolayer that expressed markers specific to brain vasculature. The BBBChip exhibited physiologically relevant transendothelial electrical resistance and accurately predicted blood-to-brain permeability of pharmacologics. Upon perfusing the vascular lumen with whole blood, the microengineered capillary wall protected neural cells from plasma-induced toxicity. Patient-derived iPSCs from individuals with neurological diseases predicted disease-specific lack of transporters and disruption of barrier integrity. By combining OrganChip technology and human iPSC-derived tissue, we have created a neurovascular unit that recapitulates complex BBB functions, provides a platform for modeling inheritable neurological disorders, and advances drug screening, as well as personalized medicine. INTRODUCTION The blood-brain barrier (BBB) comprises a multicellular neurovascular unit (NVU) in which pericytes, astrocytes, and neurons directly contact brain microvascular endothelial cells (BMECs). In turn, BMECs form a specialized transporter barrier created by tight junctions (TJs) and polarized efflux pumps. This finetuned cellular architecture permits the blood-to-brain passage

of crucial nutrients and metabolic molecules while prohibiting the transport of deleterious factors and most pharmacologics (Pardridge, 2005). Several neurological diseases involve BBB dysfunction (Agrawal et al., 2017; El-Habashy et al., 2014; Spencer et al., 2018), creating the need to understand BBB physiology and transport mechanisms in both health and disease. Marked differences in BBB substrate specificity and transporter expression and activity exist between humans and ani€nen et al., 2009; Vatine et al., 2017). Therefore, a mals (Syva human-specific model of the BBB would enhance the understanding of human diseases and the discovery of central nervous system (CNS)-permeable drugs. Human induced pluripotent stem cells (iPSCs) differentiated into BMEC-like cells (iBMECs) provide a robust source for human BBB models. iBMECs display molecular, structural, and functional BBB properties, including transendothelial electrical resistance (TEER) (Lippmann et al., 2014) closely resembling human brain vasculature. iPSCs can be derived from nondiseased and diseased subjects (Lippmann et al., 2012), and culturing iBMECs from patients with genetically inherited neurological disorders can identify disease mechanisms and abrogated signaling pathways (Lim et al., 2017; Vatine et al., 2017). These models of the BBB, however, used 2-dimensional (2D) Transwell inserts, which have a rigid surface, do not permit direct cell-cell interactions between BMECs and other NVU cells, and lack physiological mechanical forces such as shear stress. The inability of 2D cultures to develop and/or maintain the complex in vivo microenvironment of the BBB ultimately limits their utility and translation to the patient. Organ-Chip technology offers a 3D engineered microscale system that mimics the cellular microenvironment (Bhatia and Ingber, 2014). These systems permit tissue and organ functionality by recreating multicellular architectures, tissue-tissue interfaces, mechanical forces, physiochemical microenvironments, and vascular perfusion. The utility of Organ-Chip technology was demonstrated by modeling different aspects of the BBB, including continuous luminal perfusion, real-time TEER monitoring, live-cell imaging for permeability measurements, and Cell Stem Cell 24, 995–1005, June 6, 2019 ª 2019 Elsevier Inc. 995

metabolism pathways (Booth and Kim, 2014; Cho et al., 2015; Cucullo et al., 2011; Griep et al., 2013; Maoz et al., 2018; Prabhakarpandian et al., 2015; Yeon et al., 2012). Several models demonstrated increased TEER in response to shear stress. However, the highest TEER value reported was still an order of magnitude lower than the physiological state, reducing the ability to assess permeability (Wolff et al., 2015). A recent study using iBMECs on a semi-microfluidic system achieved physiologically relevant TEER values (Wang et al., 2017); however, this system used rat primary astrocytes rather than a complete humanbased model. Here, we report the establishment of a novel platform of the human BBB that combines iPSC and Organ-Chip technologies that has the potential to be used in personalized medicine. This iPSC-based BBB-Chip recapitulates interactions between human neural and endothelial cells. Functionally, the human BBB-Chip demonstrates physiologically relevant TEER values, low paracellular permeability, response to inflammatory cues at the organ level, transport of soluble biomarkers, and active efflux pumps. Importantly, personalized iPSC-based BBB-Chips generated from several individuals demonstrate consistency across healthy persons yet also detect functional differences in iPSC-based BBB-Chips derived from patients with neurological diseases. Furthermore, the iBMECs form a blood-vessel-like structure capable of sustaining whole human blood flow, thus mimicking multiple features of the BBB. RESULTS iBMECs Cultured under Laminar Flow Express TissueSpecific Markers of the Human Brain Microvascular Capillary Wall All experiments were performed using the previously published Organ-Chip (Emulate) (Huh et al., 2013; Jain et al., 2018; Sances et al., 2018), which is composed of a flexible, transparent poly(dimethylsiloxane) (PDMS) elastomer containing two closely opposed, parallel microchannels separated by a porous flexible PDMS membrane. iBMECs were generated from healthy control iPSCs as previously described and seeded on the collagen IV-fibronectin pre-coated bottom channel of the Organ-Chip (Figure 1A). In most cases, to facilitate study, an automated platform from Emulate was used that enables the simultaneous culture of up to 12 Organ-Chips (Figure S1A). Plated cells formed a monolayer that expressed the endothelial marker PECAM-1; the TJ markers zona occludens-1 (ZO-1), OCCLUDIN, and CLAUDIN-5; and the BBB glucose transporter GLUT-1 (Figure 1B). Notably, iBMECs formed a continuous band of ZO-1containing TJs along cell borders lining the entire ‘‘blood side,’’ thereby creating a hollow blood vessel-like structure (Figure 1C). In the living organism, the BBB is exposed to laminar blood flow. We therefore tested whether laminar flow in the OrganChip can promote iBMEC maturation. The iBMECs were seeded and allowed to attach with no flow for one day, after which the Organ-Chips were either maintained in static conditions with daily media changes or exposed to flow using a peristaltic pump at a rate of 30, 1,500, or 6500 mL/h (equivalent to 0.01, 0.5, or 2.4 dyn/cm2) for 48 h. RNA sequencing (RNA-seq) analysis revealed that expression of the TJ CLAUDIN gene family (CLDN; Figure 1D) as well as some endothelial markers, 996 Cell Stem Cell 24, 995–1005, June 6, 2019

including OCLN, PECAM1, caveolins, and Von Willebrand factor (VWF) (Figure 1E), were flow dependent. Principal component analysis (PCA) performed on the same samples showed that cells clustered along principal component 1 (PC1) (Figure 1F), representing 33.4% of total variance across samples, in a flowrate-dependent manner that did not differ at the higher flow rates of 0.5 and 2.4 dyn/cm2. Gene ontology analysis of the PC1 genes (Figure S1B) revealed flow-dependent pathways such as cholesterol biosynthesis, smooth muscle contraction, cell migration and proliferation, and angiogenesis that may be involved in BBB function and maturation. Unsupervised hierarchical clustering with Pearson correlation confirmed that cells cluster in a shear-stress-dependent manner (Figure S1C), which was also observed in an unbiased clustering analysis of 459 BBB-related genes, including solute carrier (SLC) and ATP-binding cassette (ABC) transporters (Figure S1D) (Lim et al., 2017). Notably, physiological shear forces did not change the morphology of iBMECs (Figure S1E) as previously demonstrated (Katt et al., 2016). In order to separate the effects of shear stress (2.4 dyn/cm2 versus 0.01 dyn/cm2) from media replenishment (0.01 dyn/cm2 versus static), we analyzed differentially expressed genes between the different shear rates. Interestingly, the majority of the top 10 upregulated genes in 2.4 dyn/cm2 versus 0.01 dyn/cm2 (Table S1) are in agreement with previous studies testing shear stress on endothelial cells (DeStefano et al., 2017), suggesting that these effects are mediated mainly by mechanical force rather than by media replenishment. Collectively, these results suggest that laminar flow promotes the expression of pathways that are related to BBB maturation in iBMECs and that shear stress further enhances specific pathways associated with metabolism and vascular function. In addition to endothelial genes, iBMECs also express many epithelial genes. We therefore performed PCA and hierarchical clustering analysis to compare iBMECs cultured on the Organ-Chip with various endothelial and epithelial cell sources, including acutely purified human BMECs (Zhang et al., 2016), human umbilical vein endothelial cells (HUVECs), human liver sinusoidal endothelial cells (LSECs) (Ginsberg et al., 2012), human choroid plexus epithelium (Kim et al., 2016b), and human lung epithelium (Hackett et al., 2012) (Figures S1F and S1G). Results show that iBMECs most closely correlate with cells of an endothelial lineage (HUVECs and LSECs) as shown by their position along PC1, while the choroid plexus epithelium is closer to the purified human BMECs (Zhang et al., 2016). These results suggest that iBMECs cultured on the Organ-Chip resemble endothelial cells but share some similarities with epithelial cells. Co-cultured iBMECs and Human Neural Cells Show Cellular Interactions and BBB Functionality To recreate a multicellular human NVU, iBMECs were again seeded on the bottom channel to constitute the blood side and primary human astrocytes and pericytes at a 3:1 ratio, respectively, were seeded on the top, larger channel to form the ‘‘brain side’’ (Figure 2A). Transcriptome analyses showed that iBMECs expressed P-glycoprotein1 (PGP1), breast cancer resistance protein (BCRP), and several multidrug-resistance-associated proteins (MRPs) (Figure S2A) similar to HUVECs. However, some of these levels were significantly lower than seen in acutely purified human brain endothelial cells (Zhang et al., 2016). To

Chip – top view

A

B

GLUT-1

PECAM-1

GLUT-1/PECAM-1

CLDN-5

OCLN

ZO-1

‘blood side’ Chip – vertical section human iPSCs

iBMECs ‘blood side’

C

ZO-1 DAPI

2

5 0.

01

2. dy 4 n/ cm

−2−1 0 1 2

0.

ic

Row Z−score

JAM3 HIF1A FLT1 F11R VCAM1 ICAM3 JAM2 VWF TJP3 OCLN CDH5 CAV1 CAV2 TJP1 ICAM2 PECAM1 ICAM4 ICAM1 MCAM TJP2

F PC2 (14.1% of total variance)

E

at

2. dy 4 n/ cm

5 0.

0. 01

St

at

ic

2

CLDND1 CLDN5 CLDN3 CLDN19 CLDN14 CLDN15 CLDN10 CLDN16 CLDN20 CLDN11 CLDN18 CLDND2 CLDN2 CLDN4 CLDN23 CLDN7 CLDN6 CLDN1 CLDN24 CLDN12 CLDN9

St

D

10

0

−10

−20

−20

−10

0

10

PC1 (33.4% of total variance)

Static 0.5 dyn/cm2

0.01 dyn/cm2 2.4 dyn/cm2

Figure 1. Modeling the Capillary Wall of the BBB on an Organ-Chip (A) Schematic of seeding strategy. iPSCs are differentiated into endothelial cells and seeded in the bottom channel of the Organ-Chip to constitute the blood side. Top: schematic of a top view of the microengineered Organ-Chip. Bottom: schematic of a vertical section of the Organ-Chip. (B) iBMECs derived from healthy donor CS83iCTR form a tight monolayer on the blood side of the Organ-Chip and express the BMEC markers glucose transporter 1 (GLUT-1, green), PECAM-1 (red), CLDN-5 (green), OCLN (red), and ZO-1 (green). Scale bar, 200 mm. (C) The iBMEC monolayer (ZO-1, green; DAPI, blue) covers the walls surrounding the blood side and forms a full lumen similar to a blood vessel. Scale bar, 200 mm. (D) Relative expression of tight junction Claudin (CLDN) genes in response to shear stress. (E) Relative expression of junctional-related genes in response to shear stress. (F) PCA of iBMECs cultured under various levels of shear stress. See also Figure S1 and Table S1.

further assess PGP1 activity, we used immunocytochemistry to stain for the protein, which showed some specific weak staining in the iBMECs (Figure S2B). As there were only low levels of this critical BBB protein, we wanted to test the function of these transporters using a traditional assay of intracellular accumulation of rhodamine 123. This was monitored in the presence or absence of the specific PGP1 inhibitor vinblastine. We show that iBMECs exhibited a 2-fold increase in rhodamine 123 accumulation in the presence of vinblastine (Figure S2C), demonstrating efflux pump functionality. Assessing the interaction of co-cultured iBMECs with astrocytes and pericytes by immunocytochemistry showed that iBMECs marked by phalloidin and ZO-1 formed a uniform monolayer on the blood side (Figures 2B–2D), while astrocytes expressing GFAP and pericytes expressing alpha-smooth muscle actin (a-SMA) (Figure 2E), platelet-derived growth factor receptor b (PDGFRb), and NG2 (Figures S2D–S2F) formed an intricate network within the brain compartment. Confocal anal-

ysis revealed that astrocytes, pericytes, and iBMECs remained in their respective channels. However, astrocyte processes were seen to protrude through the pores, and created direct contact with iBMECs on the blood side (Figure S3A). Furthermore, co-culture with human pericytes and astrocytes significantly decreased the blood-to-brain leakage of fluorescein isothiocyanate (FITC)-labeled dextran (3 kDa) compared to iBMECs cultured alone (Figure 2F). Together, these data suggest that primary human astrocytes and pericytes can promote functional BBB maturation on the BBB-Chip. Immunoglobulin G (IgG), albumin, and transferrin are among the most abundant proteins in human blood, with IgG and albumin being mostly confined to the blood side while transferrin is transported across the BBB by receptor-mediated transcytosis (RMT) (Pardridge et al., 1987). To further test the functionality of the human BBB-Chip, we assessed whether these proteins could be selectively filtered. Indeed, IgG and albumin remained confined to the blood side, while transferrin accumulated on Cell Stem Cell 24, 995–1005, June 6, 2019 997

A

B

C

F

G

J

D

H

K

E

I

L

Figure 2. A Fully Human BBB Organ-Chip (A) Schematic of BBB-Chip seeding paradigm, with iBMECs seeded on the blood side and primary human astrocytes and pericytes seeded on the brain side. (B) Tile imaging of immunocytochemistry of the human BBB-Chip 5 days post-seeding. The blood side is populated with CS83iCTR iBMECs that form a monolayer and express phalloidin (magenta). The brain side is seeded with primary human astrocytes that express GFAP (green) and pericytes. Scale bar, 1 mm. (C) High magnification of the human BBB-Chip in (B). Scale bar, 500 mm. (D) iBMEC monolayer also expresses membrane ZO-1 (white). Scale bar, 200 mm. (E) Primary human astrocytes express GFAP (white) and pericytes express a-SMA (green). Scale bar, 100 mm. (F) Blood-to-brain permeability of dextran-FITC (3 kDa) was measured on Organ-Chips seeded with iBMECs alone (black bar) and iBMECs with primary human astrocytes and pericytes (gray bar) cultured under continuous flow for 3 days. Student’s t-test (p < 0.05). n = 7. (G–I) Blood-to-brain filtration of the human proteins immunoglobulin (IgG; 700 mg/mL; G), albumin (300 mg/mL; H), and transferrin (50 mg/mL; I). *p < 0.05; ***p < 0.0001 (Student’s test); n = 6. (J) Tumor necrosis factor a (TNF-a), interleukin-1 beta (IL-1b), or IL-8 treatment of the blood side reduces the relative ZO-1 expression on the blood side (see Figures S3B–S3E). (K) Blood-to-brain permeability to dextran (3 kDa) is disrupted in a concentration-dependent response to TNF-a, IL-1b, or IL-8. (L) Percentage of endothelial surface covered with astrocyte endfeet-like processes is disrupted in a concentration-dependent response to perfusion of the blood side with TNF-a, IL-1b, or IL-8 (see Figures S3F–S3I). *p < 0.05; **p < 0.01; ***p < 0.0001 (one-way ANOVA with Dunnett’s multiple comparisons test); n = 3–6 (J and K). (F–L) Error bars represent SEM. See also Figures S2 and S3.

998 Cell Stem Cell 24, 995–1005, June 6, 2019

the brain side (Figures 2G–2I), indicating that active blood-tobrain transferrin RMT was recapitulated on the BBB-Chip. Interestingly the IgG/albumin ratio within the brain compartment, a clinically relevant parameter for diagnostics of neurodegenerative diseases such as multiple sclerosis (LeVine, 2016), was nearly zero (0.1–0.01), and the transferrin/albumin ratio was 1.5, both consistent with a healthy BBB in vivo (Kere´nyi et al., 1980). To test for organ level functionality, we perfused the blood compartment overnight with cytokines typically involved in vascular inflammation, specifically tumor necrosis factor a (TNF-a), interleukin-1 beta (IL-1b), or IL-8 (Rochfort and Cummins, 2015). The response to inflammatory stimulation altered the expression of the TJ marker ZO-1 (Figures 2J and S3B–S3E) and consequently led to a dose-dependent increase in blood-to-brain leakage of dextran (Figure 2K). Interestingly, changes in the blood side were followed by a retraction of astrocyte protrusions and a reduction in vascular endfeet-like coverage of the vascular surface (Figures 2L and S3F–S3I), suggesting that a functional link between the brain and blood compartments was achieved. Taken together, these results demonstrate that the BBB-Chip recreates organ-level structures and functions that resemble the interface of the NVU. Co-cultured iBMECs and Human iPSC-Derived Neural Cells Show Cellular Maturity To generate a truly personalized and translational BBB model, we created a completely iPSC-based BBB-Chip by replacing the primary pericytes and astrocytes with human iPSC-derived neural cells. Isogenic iPSCs were used for both neural cell and iBMEC differentiations (Figure 3A). EZ-spheres (pre-rosette neural progenitor cells) (Ebert et al., 2013; Shelley et al., 2014) were seeded on the brain side and differentiated into a mixed neural culture containing nestin+ progenitors, GFAP+/S100b+ astrocytes, and bIII-tubulin+/MAP2ab+/neurofilament+ neurons (Figure 3B). bIII-tubulin+ cells also expressed the synaptic marker synaptophysin. Calcium imaging demonstrated that neurons consistently exhibited spontaneous neuronal activity (Figure 3C; Video S1) that was blocked by tetrodotoxin (TTX; 1 mM), suggesting synaptogenesis and network formation on the iPSC-based BBB-Chip. Confocal analysis showed that similar to primary astrocytes, iPSC-derived astrocytes on the brain side sent processes that protruded through the porous membrane to form direct cell-to-cell interactions with the iBMECs on the blood side (Figure S4A; Video S2). To test whether iPSC-derived neural cultures could enhance the iBMEC barrier, the permeability of dextran-FITC was compared in iBMEC cultures alone or with human primary astrocytes and pericytes or iPSC-derived neural cultures. The primary cell-based and iPSC-based BBB-Chips showed similar dextran permeability, both of which were significantly lower than iBMECs cultured alone (Figure 3D), indicating that iPSC-derived neural cultures can support the functional maturation of iBMECs in the absence of pericytes. To further compare the conditioning effect of cells on the brain side on iBMECs, we developed an approach in which fluorescently labeled iBMECs (derived from the CS83iCTR-ACTB::nGFP) were cultured alone or with human primary astrocytes and pericytes or iPSC-derived neural cultures. After 4 days in culture, all cells were collected and

FACS-sorted based on green fluorescent protein (GFP) expression, resulting in pure populations of co-cultured iBMECs that were collected for RNA-seq analysis. Pearson correlation, unsupervised hierarchical clustering, and PCA revealed that iBMECs co-cultured with iPSC-derived neural cells were distinct from iBMECs cultured alone or co-cultured with primary astrocytes and pericytes (Figures S4B–S4D). Interestingly, gene ontology analysis of the PC1 genes revealed that iPSC-derived neural cells induced pathways that are related to DNA replication and cell proliferation (Figure S4E), which was reflected in the greater number of iBMECs found in these cultures (Figure S4F). These different molecular responses to the co-cultured cells demonstrate the plasticity of endothelial cells. To further evaluate the functionality of the iPSC-based BBBChip, real-time TEER was assessed using a specially engineered version of the BBB-Chip that incorporated gold electrodes on both sides of the porous membrane. Notably, TEER measurements reached values as high as 1,500 U 3 cm2 2 days postseeding and levels were maintained above 1,000 Uxcm2 for 5 days (Figure 3E), indicating that the iPSC-based BBB-Chip obtained physiologically relevant values (Mantle et al., 2016). Interindividual Variability across BBB-Chips Can Detect BBB Alterations in Diseased Patients Demonstrating organ-specific disease modeling is important to validating the iPSC-based BBB-Chip as a relevant system for personalized medicine. We recently reported an iPSC-based model of Huntington’s disease (HD) in which HD-iBMECs showed altered barrier functions (Lim et al., 2017). In order to assess the possibility of using a personalized BBB-Chip as a predictor for patient-specific brain penetrability of candidate molecules, we examined the permeabilities of fluorescently labeled dextrans of varying molecular weights across iPSC-based BBB-Chips derived from three healthy donors (CS617iCTR, CS172iCTR, and CS188iCTR) and an HD patient with 71 CAG repeats within the HUNTINGTIN gene (CS81iHD). Notably, there were no significant variations observed across healthy individuals, yet there was a significant increase in dextran-FITC molecule permeability in the HD BBB-Chip (Figure 3F). Furthermore, the permeability of the dextran molecules across BBB-Chips from healthy controls correlated with previously reported in vivo rodent brain uptake (Yuan et al., 2009) (R2 = 0.96), demonstrating that the iPSC-based BBB-Chip formed and maintained a barrier that could selectively separate molecules based on size (Figure 3G). We have also recently reported an iPSC-based model for monocarboxylate transporter 8 (MCT8) deficiency, a severe form of psychomotor retardation, and showed that thyroid hormone (triiodothyronine [T3]) transport across iBMECs requires functional MCT8 (Vatine et al., 2017). In order to further assess organ-specific disease modeling, we generated an iPSC-based BBB-Chip for MCT8 deficiency, which used the above healthy control lines as well as the following published iPSC lines: (1) control line CS03iCTR; (2) CRISPR/Cas9-mediated MCT8 mutation CS03iCTRmut; (3) patient line CS01iMCT8; and (4) MCT8 mutation line with corrected mutation using CRISPR/Cas9mediated homologous recombination CS01iMCT8cor. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) showed that while T3 transport was consistent across healthy control BBB-Chips, MCT8-deficient BBB-Chips showed Cell Stem Cell 24, 995–1005, June 6, 2019 999

A

B

Chip – top view ‘brain side’ EZ Spheres (neural progenitors)

Nestin

GFAP

hNF

MAP2ab

S100

‘blood side’ Chip – vertical section

human iPSCs

‘brain side’ iBMECs

III-tubulin/SYP

‘blood side’

0.5 0.0

-0.5

0

180

Time (sec)

360

4×10 -7 3×10 -7 2×10 -7 1×10 -7 0

al

s /P r EC 1 A eu iN + + s EC CS M E iB BM i

F 2000

1.5×10 -6

1500 1000 500 0

0 1 2 3 4 5 6 7 8 9 10

Time (days)

M

CS83iCTR

I 4×10 -6

kDa

0

5×10 -7 1×10 -6

In vivo Pe (cm s-1)

0

0

20

40

60

CS0617iCTR

CS0172iCTR

2×10 -6 1×10 -6 0

1.0×10 -5 5.0×10 -6 0

R et ig ab Le in ve e tir ac et am C ol ch ic in e 2N B D D ex G tr an 4 kD a

70

0

Papp (cm s-1)

20

kDa

r

5×10 -7

5×10 -7

1.5×10 -5

3×10 -6

C S0 6 C 1 7i S0 C 17 TR C S0 2iC 18 TR 8 C iCT S C 03 R S0 iC 3i TR C T C S0 R mu t 1i C S0 M 1i CT M 8 C T8 c o

1×10 -6 R2=0.96 4 kDa

1×10 -6

Molecular weight (kDa)

H T3 Papp (cm s-1)

BBB-Chip Papp (cm s-1)

iB

G

CS0617iCTR CS0172iCTR CS0188iCTR CS81iHD

2×10 -6

Papp (cm s-1)

F/F

1.0

E

5×10 -7

x cm2)

D TTX (1μM) μ

1.5

TEER (

C

Dextran (3 kDa) Papp (cm s-1)

Astrocytes Neurons Neural progenitors

Figure 3. Personalized iPSC-Based BBB-Chip Detects Interindividual Variability (A) Schematic of seeding strategy. iPSCs are differentiated into EZ-spheres (early neural progenitors) in suspension. EZ-spheres are then dissociated and seeded on the brain side, where they further differentiate into mixed neural cultures. Isogenic iPSCs are differentiated into iBMECs and seeded on the blood side as described above. (B) Immunocytochemistry on the CS83iCTR iPSC-based BBB-Chip 7 days post-seeding demonstrates that EZ-sphere-derived neural cultures populating the brain compartment express nestin+ neural progenitors (red); GFAP+ (red) and S100b+ (green) astrocytes; and the neuronal markers heavy chain neurofilament (hNF, red), MAP2ab (green), and bIII-tubulin (red), as well as the synaptic marker synaptophysin (SYP, green). Scale bar, 200 mm. (C) At 8 days post-seeding, calcium imaging revealed that neural cultures exhibit spontaneous neuronal activity on the Organ-Chip that is inhibited by tetrodotoxin (TTX). (D) Blood-to-brain permeability of dextran-FITC (3 kDa) was measured on Organ-Chips seeded with iBMECs alone (black bar), iBMECs with primary human astrocytes and pericytes (red bar), or iBMECs with iPSC-derived neural cells (iNeural, blue bar). *p < 0.05 (one-way ANOVA with Dunnett’s multiple comparisons test). (E) TEER measurements reach 1,500 U 3 cm2 2 days post-seeding and remain above 1,000 U 3 cm2 for 5 days. (F) Permeability of dextran molecules of 4, 20, or 70 kDa compared in BBB-Chips derived from healthy control iPSCs (CS617iCTR, CS0172iCTR, and CS188iCTR) or a Huntington’s disease patient iPSCs (CS81iHD). *p < 0.05 (two-way ANOVA with Tukey’s multiple comparison test). (G) The permeability of dextran molecules across healthy control iPSC-derived BBB-Chips correlated (R2 = 0.96) with previously reported in vivo rodent brain uptake (Pe, cm s1). (H) Blood-to-brain T3 permeability compared in BBB-Chips derived from healthy donors (CS617iCTR, CS172iCTR, CS188iCTR, and CS03iCTR), an MCT8deficient patient (CS01iMCT8), an isogenic line where a nonsense mutation was introduced into the CS03iCTR line (CS03iCTRmut), and an isogenic line where a point mutation was corrected in the CS01iMCT8 line (CS01iMCT8cor). *p < 0.05 (nested t-test between MCT8 mutants and control lines). (I) Dextran-FITC and the fluorescent glucose analog 2-NBDG or retigabine, levetiracetam, and colchicine were spiked simultaneously to the ‘‘blood channel’’ (10 mg/mL each), and their permeability and concentrations were evaluated on the brain side by means of fluorescence or liquid chromatography-tandem mass spectrometry (LC-MS/MS). ***p < 0.001 (one-way ANOVA with Tukey’s multiple comparison test); n = 3–4 BBB-Chips per drug per cell line. Data for samples run on LC-MS/MS were averaged from 3 separate injections per sample. (D, F, H, and I) Error bars represent SEM. See also Figure S4.

1000 Cell Stem Cell 24, 995–1005, June 6, 2019

significantly lower permeability (Figure 3H), confirming the necessity of MCT8 for the blood-to-brain transport of T3 across the BBB. Altogether, these results suggest that patient-specific iPSC-based BBB-Chips may be used to predict inter-patient variability in BBB functions. We next tested whether the iPSC-based BBB-Chip could be used to predict the relative permeability of additional molecules, including the fluorescent glucose analog 2NDBG, the marketed drug colchicine (gout treatment, moderate permeability), and the anti-epileptic drugs levetiracetam and retigabine, which can efficiently penetrate the BBB. These molecules were spiked simultaneously into the blood side and their concentration on the brain side was evaluated using fluorescence or LC-MS/MS (Figure 3I). The diffusion of these molecules across the iPSC-based BBB-Chip demonstrated the expected differences in permeability, suggesting that this model can predict human CNS drug penetrability. Whole Human Blood Perfused through the BBB-Chip Is Restricted to the Blood Side, which Protects against Blood-Induced Toxicity and Enhances Disease Modeling In most culture systems, neural cells are maintained in medium containing nutrients and growth factors, which does not account for vascular blood flow. However, exposing neural cells to unfiltered blood may lead to cytotoxicity, as observed in brain hemorrhage (Kim et al., 2016a). Thus, a functional BBB is required to filter the blood in order to safely provide supportive nutrients and growth factors to neural cells. Whole human blood treated with the anticoagulant sodium citrate was perfused at a physiologically relevant shear rate (3,600 mL/hr, equivalent to 5 dyn/cm2) (Cucullo et al., 2011) through the blood side, while the brain compartment with iPSC-derived neural cells was perfused with neural media (Figure 4A). Measuring dextran-FITC (3 kDa) permeability during blood perfusion showed that iBMECs were able to maintain a functional barrier that confined blood to the blood side (Figures 4B and 4C). In contrast, BBB-Chips with no iBMECs or with iBMECs treated with TNF-a did not have a functional barrier, resulting in blood leakage into the brain side (Figures 4B and 4C). Immunocytochemistry analysis and a lactate dehydrogenase (LDH) viability assay showed that neural cells with the iPSC-based BBB-Chip did not exhibit bloodinduced toxicity, while neural cultures without iBMECs or treated with TNF-a exhibited significant toxicity (Figures 4D and 4E). These results demonstrate that iBMECs in the iPSC-based BBB-Chip protect neural cells on the brain side from bloodinduced cytotoxicity, thereby recapitulating one of the key in vivo functions of the BBB. The incorporation of whole human blood into the iPSC-based BBB-Chip introduces another important in vivo BBB interface to this platform. In the blood, over 90% of T3 is bound to proteins that act as carriers and only a small portion remains unbound and active (Refetoff et al., 1970). Thus, to test the MCT8-deficiency model under these physiological conditions, whole human blood spiked with T3 was perfused through the blood side, and neural media without T3 was perfused through the brain side (Figure 4F). T3 permeability across MCT8-deficient iPSC-based BBB -Chips was significantly lower compared with iPSC-based BBB-Chips from healthy control lines (Figure 4G). These results demonstrate that blood-carried T3 transport across the BBB is MCT8

dependent. Notably, the overall permeability of T3 across healthy control iPSC-based BBB-Chips was 5-fold lower with wholeblood perfusate compared to media (Figure 3H), suggesting that measuring transport in the context of whole blood recapitulated some aspects of the physiological BBB. These experiments demonstrate that whole blood can be used in conjunction with the BBB-Chip platform to further establish the role of the BBB in genetic-based neurological disease. DISCUSSION The BBB tightly regulates traffic of molecules between the blood and CNS, and its disruption is associated with several neurological diseases. Lack of a robust, scalable, and physiologically relevant model of the human BBB has hampered the study of BBB in health and disease. The source of human BMECs is limited, as primary BMECs have a low yield (<0.1% of CNS cells), a tendency to de-differentiate in culture, and donor variability (Cecchelli et al., 2007). Immortalized cell lines overcome these limitations but display poor barrier properties that limit their use in permeability assays (Kamiichi et al., 2012). A further challenge is obtaining a multicellular NVU and flow. Here, we combined iPSC and microengineered Organ-Chip technologies to produce a personalized human BBB system that could successfully model disease and screen drugs. Human iPSCs are readily available and can generate BMEC-like monolayers, which express endothelial-, TJ-, and BBB-related markers. iBMECs share molecular similarities with endothelial cell sources; however, they also express epithelialassociated genes, suggesting that further modifications to the differentiation protocol may produce iBMECs that are closer transcriptomically to endogenous BMECs. However, iBMECs produced in this study and others display functional properties that we think can be used to model the BBB (Lim et al., 2017; Lippmann et al., 2012, 2014; Vatine et al., 2017). Culturing iBMECs in a Transwell system has shown the utility of iBMECs but did not assess the BBB at the multicellular organ level. The ability to culture iPSC-derived neural cells that promote the functional maturation of iBMECs originating from the same patient has been demonstrated (Canfield et al., 2017). It is now possible to shift from an incomplete BMEC-centric to a multicellular BBB model that can fully interrogate molecular and cellular mechanisms involved in BBB health and disease. The Transwell system is simple and inexpensive, but it poorly supports the complexity of multicellular organ systems (Cucullo et al., 2011; Hawkins and Davis, 2005; Santaguida et al., 2006; Sweeney et al., 2016). In contrast, the Organ-Chip technology offers cell-cell interactions, physiochemical microenvironments, mechanical forces, and vascular perfusion to provide a superior level of tissue and organ functionality. The complicated application of microengineered Organ-Chips can lead to low sample size and has hindered intra-laboratory reproducibility and widespread use. However, the current report employs optimized Organ-Chips combined with a commercially available instrument, facilitating simple and large-scale simultaneous culture of multiple Organ-Chips with controlled laminar flow and shear stress forces. The Organ-Chip microfluidic channels, and the porous membrane that separates them, permit the reconstitution of multicellular architectures and tissue-tissue interfaces. Recently, Cell Stem Cell 24, 995–1005, June 6, 2019 1001

A

B ‘brain side’

‘blood side’

TNFα

‘blood side’

Whole human blood

+ Triiodothyronine (T3)



+

+ –

+ +

6×10 -7 4×10 -7 2×10 -7 0 co r

‘brain side’

iNeural TNFα

G F

+ + –

C T8

– iBMECs

iM

+ iBMECs

C T8

+ +

0

iBMECs

iM

+ –

1

C S0 1

+

2

ut

TNFα



3

m

iNeural

+ + –

+ +

iC TR

iBMECs

Absorbance (A.U.)

5×10 -6

+

+ –

4

iC TR

1×10 -5



E

Whole human blood perfusion

C S0 3

1.5×10 -5

βIII-tubulin GFAP

D

Dextran (3 kDa) Papp (cm s-1)

C

0

iNeural

+ + –

C S0 3

+ Dextran-FITC +/- TNFα

iBMECs

C S0 1

Dextran-FITC permeability LDH viability assay

T 3 Permeability (Papp, cm s-1)

Whole human blood

Cell line

Figure 4. Whole Human Blood Filtration with the iPSC-Based BBB-Chip Can Assess Neurocytotoxicity and Enhance Disease Modeling (A) Schematic of perfusion of whole human blood through the iPSC-based BBB-Chip. (B) Whole human blood was perfused through the blood side at 3,600 mL/hr (equivalent to 5 dyn/cm2). iBMECs restrict whole blood to the blood channel. When neural cells are cultured alone or when iBMECs are treated with TNF-a (10 ng/mL), whole human blood can diffuse to the brain side (highlighted by the green arrows). (C) Measuring blood-to-brain permeability of fluorescent dextran-FITC (3 kDa) following overnight perfusion of whole human blood showed low diffusion levels when neural cells were cultured with iBMECs (white bar) and a significant increase in permeability when iPSC-derived neural cells were cultured alone (gray bar) or following TNF-a treatment (10 ng/mL, black bar). *p < 0.05; ***p < 0.0001 (one-way ANOVA). (D) Immunocytochemistry of bIII-tubulin+ neurons (red) and GFAP+ astrocytes (green) showed that neural cells are protected by iBMECs but are reduced when cultured alone. Scale bar, 200 mm. (E) Assessing neural toxicity by quantifying LDH confirmed that iBMECs provide a functional barrier that can protect the brain side from blood-induced cytotoxicity (white bar). In contrast, Organ-Chips without iBMECs (gray bar) or with TNF-a treatment (black bar) led to vascular leakage and a significant increase in LDH release. *p < 0.05; **p < 0.01 (one-way ANOVA with Dunnett’s multiple comparisons test). (F) Experimental design for MCT8-deficiency disease model. (G) Measuring blood-to-brain T3 (100 nM) permeability across iPSC-based BBB-Chip containing control lines (filled bars) and MCT8-deficienct lines (empty bars) showed that iPSC-based BBB-Chips with the MCT8 mutation have significantly reduced transport compared to BBB-Chips with no mutation. *p < 0.05 (one-way ANOVA with Tukey’s multiple comparisons test). Color coding represents CRISPR/Cas9-edited isogenic iPSC lines. (C, E, and G) Error bars represent SEM.

iBMECs and rodent-derived astrocytes co-cultured on a semimicrofluidic platform demonstrated physiological TEER, without the application of shear forces (Wang et al., 2017). However, here we obtained further biological relevance with human primary or iPSC-derived astrocytes, which partially covered the iBMEC monolayer, thereby resembling parts of the BBB in which astrocyte endfeet partially cover the vasculature (El-Khoury et al., 1002 Cell Stem Cell 24, 995–1005, June 6, 2019

2006). The iPSC-based BBB-Chip reached physiologically relevant TEER values that were maintained for several days, suggesting that the iPSC-based BBB-Chip maintains BBB function under high shear stress. These results mirror a previous study showing that unlike human and animal-derived endothelial cells from other tissues, iBMECs resist applied physiological shear stress when cultured under flow (DeStefano et al., 2017).

Moreover, iBMECs cultured under different laminar flow rates allowed the investigation of the effects of both constant media replenishment permitted by low flow rates and shear stress with higher flow rates, thereby distinguishing biomolecular availability and mechanical force sensing. Shear-stress-activated pathways such as cholesterol metabolism, proliferation, and angiogenesis, which reflect previously reported shear-stressinduced activation of cholesterol biosynthesis (Liu et al., 2002), stimulation of endothelial cell proliferation (White et al., 2001), and angiogenesis (Conklin et al., 2002). TJ-related gene expression was also affected by laminar flow. Interestingly, TJP1/ZO-1 expression was increased in all laminar flow conditions, including low shear stress (0.01 dyn/cm2), whereas OCLN, and the endothelial adhesion and junctional-related genes PECAM-1, and CDH5/VE-Cadherin showed greater increases in expression specifically in higher shear-stress conditions (0.5 and 2.4 dyn/cm2). Despite recirculating media in each flow condition, it is difficult to separate the effects of shear stress from media replenishment. Nonetheless, these results suggest that the dynamic microenvironment provided by the OrganChip system can be beneficial for modeling the complex vascular environment. The endothelial-glial interface that was mimicked on the Organ-Chip demonstrated organ-like functional responses to multiple inflammatory stimulations applied on the blood side, creating a response on both the endothelial barrier and the retraction of astrocyte endfeet-like processes. Furthermore, iPSC-derived neural cultures led to the functional maturation of iBMECs in the absence of pericytes, which corroborates our finding that neural-derived conditioned media without pericytes was sufficient to derive functional BMECs (Canfield et al., 2017). In future studies, isogenic iPSC-derived brain pericytes (Stebbins et al., 2019) could be added to this system in order to enhance this in vitro model and more closely recapitulate the physiological NVU. Finally, the neural and endothelial cell interaction promoted neuronal function demonstrated by spontaneous activity observed on the brain side of the iPSC-based BBB-Chip after only 8 days in culture. This effect of endothelial cells is supported by our recent demonstration that iBMECs promote functional maturation of iPSC-derived spinal motor neurons when co-cultured on an Organ-Chip (Sances et al., 2018). The possibilities of seeding iBMECs at any stage after neural cell seeding (data not shown) and the co-culture with any neural cell type will allow future examination of mature versus developing BBB as well as the heterogeneity in specific brain regions of the BBB. Various nutrients, ions, hormones, and growth factors in the blood are selectively filtered through the BBB to reach the CNS. The endothelial compartment of the iPSC-based BBBChip formed a blood vessel-like structure that sustained whole human blood perfusion, further recapitulating physiology. The iPSC-based BBB-Chip not only permitted perfusion of whole blood but also protected neural cells from blood-induced cytotoxicity. Without iBMECs or upon pro-inflammatory treatment, the ability to protect the neural cells from the blood was lost and plasma leakage into the brain side led to neuronal toxicity, resembling harmful events that follow vasogenic edema (Kim et al., 2016a). In the future, neural cells could be cultured within a BBB environment that filters blood-to-brain molecule passage

for better recapitulation of the in vivo neural milieu. In addition, studying transporter-based pharmokinetics may be optimal with blood, which would contain protein-binding molecules possibly absent in culture medium. Evaluating several proteins and drugs suggests that this new BBB model predicted CNS penetrability. Importantly, the blood-to-brain ratio of IgG and albumin was comparable to the physiological range in healthy patients (Kere´nyi et al., 1980). Increased transferrin permeability supports a recent demonstration of RMT in iBMECs (Ribecco-Lutkiewicz et al., 2018) and suggests that the iPSC-based BBB-Chip could predict the penetrability of RMT-based neuro-pharmaceuticals. The barrier function is stably maintained for several days, which permits multiple molecules to be tested in each Organ-Chip, increasing throughput of the system. Moreover, the ability to combine this methodology with MS detection allows simultaneous quantification of permeability for multiple molecules, as demonstrated here for three drugs and T3. Testing additional molecules and biologics with various physiochemical properties and mechanisms is underway to further qualify the system. However, the consistency of molecule permeability among BBB-Chips derived from iPSCs of several healthy donors demonstrates the reliability of the system. Furthermore, calcium imaging of spontaneous neuronal activity in patient-specific iPSC-based BBB-Chips permits assessment of function on the brain side. These capabilities enable future applications to quantify both BBB penetrability and neuronal effects, permitting better evaluation of drug safety and efficacy. Variable human genetics may cause differences in patient response to drugs, which has implications for both efficacy and safety. The ability to generate individual BBB-Chips may facilitate predictive personalized medicine applications by allowing interindividual variability to be addressed. Indeed, the two disease models provided here support this exciting possibility. Notably, MCT8-deficient T3 transport was also observed when whole blood was perfused through the iPSC-based BBB-Chip. As such, we have provided a patient-specific BBB-Chip for neurological disease modeling. This human BBB-Chip permits a better understanding of the BBB’s role in health and disease, thereby supporting the development of new drug treatments. STAR+METHODS Detailed methods are provided in the online version of this paper and include the following: d d d d

KEY RESOURCES TABLE CONTACT FOR REAGENT AND RESOURCE SHARING EXPERIMENTAL MODEL AND SUBJECT DETAILS METHOD DETAILS B BBB-Chip Microfabrication and Culture B Generation of EZ-spheres from iPSCs B Differentiation of EZ-spheres into Neural Cultures B Differentiation of iPSCs into iBMECs B Primary Cell Culture and Organ-Chip Preparation B Immunocytochemistry B Assessment of Vascular Integrity and Blood-to-Brain Transport of Soluble Biomarkers B Image Analysis of Astrocyte Endfeet Coverage and Cell Junction Integrity Cell Stem Cell 24, 995–1005, June 6, 2019 1003

B

Permeability Assays Blood Perfusion and Shear Stress B Efflux Pump Activity B Viability Assay B Fluorescence-Activated Cell Sorting (FACS) of iBMEC Co-cultures B Transcriptional Analysis of iBMECs B Gene Ontology Analysis B Calcium Imaging B Measurements of Iodothyronines by LC/MS/MS QUANTIFICATION AND STATISTICAL ANALYSIS DATA AND SOFTWARE AVAILABILITY B

d d

SUPPLEMENTAL INFORMATION

Bhatia, S.N., and Ingber, D.E. (2014). Microfluidic organs-on-chips. Nat. Biotechnol. 32, 760–772. Booth, R., and Kim, H. (2014). Permeability analysis of neuroactive drugs through a dynamic microfluidic in vitro blood-brain barrier model. Ann. Biomed. Eng. 42, 2379–2391. Canfield, S.G., Stebbins, M.J., Morales, B.S., Asai, S.W., Vatine, G.D., Svendsen, C.N., Palecek, S.P., and Shusta, E.V. (2017). An isogenic bloodbrain barrier model comprising brain endothelial cells, astrocytes, and neurons derived from human induced pluripotent stem cells. J. Neurochem. 140, 874–888. Cecchelli, R., Berezowski, V., Lundquist, S., Culot, M., Renftel, M., Dehouck, M.P., and Fenart, L. (2007). Modelling of the blood-brain barrier in drug discovery and development. Nat. Rev. Drug Discov. 6, 650–661. Cho, H., Seo, J.H., Wong, K.H., Terasaki, Y., Park, J., Bong, K., Arai, K., Lo, E.H., and Irimia, D. (2015). Three-dimensional blood-brain barrier model for in vitro studies of neurovascular pathology. Sci. Rep. 5, 15222.

Supplemental Information can be found online at https://doi.org/10.1016/j. stem.2019.05.011.

Conklin, B.S., Zhong, D.S., Zhao, W., Lin, P.H., and Chen, C. (2002). Shear stress regulates occludin and VEGF expression in porcine arterial endothelial cells. J. Surg. Res. 102, 13–21.

ACKNOWLEDGMENTS

Cucullo, L., Hossain, M., Puvenna, V., Marchi, N., and Janigro, D. (2011). The role of shear stress in blood-brain barrier endothelial physiology. BMC Neurosci. 12, 40.

We would like to thank Dr. Soshana Svendsen and Dr. Sifis Pediaditakis for critical writing and editing. This work was supported by California Institute for Regenerative Medicine grant ID DISC1-08800, the Sherman Family Foundation, Israel Science Foundation grant 1621/18, NIH-NINDS grant 1UG3NS105703, and The ALS Association grant 18-SI-389. AUTHOR CONTRIBUTIONS G.D.V., R.B., M.J.W., and C.N.S. designed the experiments and wrote and edited the manuscript. G.D.V., R.B., and M.J.W. performed in vitro experiments with assistance from S.S., B.K.B., M.R., S.B., and M.K.. C.L., Z.C., W.R.S., and J.V.E. performed the mass spectrometry analysis. R.B., S.B., and M.K. designed and performed experiments with shear stress and inflammatory cytokines. M.J.W. processed and analyzed gene sequencing data. Organ-Chip engineering assistance was provided by N.W. and J.K. DECLARATION OF INTERESTS Cedars-Sinai owns a minority stock interest in Emulate, the company that produces the study’s microfluidic Organ-Chips. An officer of Cedars-Sinai also serves on Emulate’s Board of Directors. Emulate provided no financial support for this research. R.B., S.B., M.K., C.L., N.W., and J.K. are employees and shareholders of Emulate. Cedars-Sinai and Emulate have patents filed related to this work. Received: May 10, 2018 Revised: February 24, 2019 Accepted: May 13, 2019 Published: June 6, 2019 REFERENCES Agrawal, M., Ajazuddin, Tripathi, D.K., Saraf, S., Saraf, S., Antimisiaris, S.G., Mourtas, S., Hammarlund-Udenaes, M., and Alexander, A. (2017). Recent advancements in liposomes targeting strategies to cross blood-brain barrier (BBB) for the treatment of Alzheimer’s disease. J. Control. Release 260, 61–77.

DeStefano, J.G., Xu, Z.S., Williams, A.J., Yimam, N., and Searson, P.C. (2017). Effect of shear stress on iPSC-derived human brain microvascular endothelial cells (dhBMECs). Fluids Barriers CNS 14, 20. Ebert, A.D., Shelley, B.C., Hurley, A.M., Onorati, M., Castiglioni, V., Patitucci, T.N., Svendsen, S.P., Mattis, V.B., McGivern, J.V., Schwab, A.J., et al. (2013). EZ spheres: a stable and expandable culture system for the generation of pre-rosette multipotent stem cells from human ESCs and iPSCs. Stem Cell Res. (Amst.) 10, 417–427. El-Habashy, S.E., Nazief, A.M., Adkins, C.E., Wen, M.M., El-Kamel, A.H., Hamdan, A.M., Hanafy, A.S., Terrell, T.O., Mohammad, A.S., Lockman, P.R., and Nounou, M.I. (2014). Novel treatment strategies for brain tumors and metastases. Pharm. Pat. Anal. 3, 279–296. El-Khoury, N., Braun, A., Hu, F., Pandey, M., Nedergaard, M., Lagamma, E.F., and Ballabh, P. (2006). Astrocyte end-feet in germinal matrix, cerebral cortex, and white matter in developing infants. Pediatr. Res. 59, 673–679. Ginsberg, M., James, D., Ding, B.S., Nolan, D., Geng, F., Butler, J.M., Schachterle, W., Pulijaal, V.R., Mathew, S., Chasen, S.T., et al. (2012). Efficient direct reprogramming of mature amniotic cells into endothelial cells by ETS factors and TGFb suppression. Cell 151, 559–575. Griep, L.M., Wolbers, F., de Wagenaar, B., ter Braak, P.M., Weksler, B.B., Romero, I.A., Couraud, P.O., Vermes, I., van der Meer, A.D., and van den Berg, A. (2013). BBB on chip: microfluidic platform to mechanically and biochemically modulate blood-brain barrier function. Biomed. Microdevices 15, 145–150. Hackett, N.R., Butler, M.W., Shaykhiev, R., Salit, J., Omberg, L., RodriguezFlores, J.L., Mezey, J.G., Strulovici-Barel, Y., Wang, G., Didon, L., and Crystal, R.G. (2012). RNA-seq quantification of the human small airway epithelium transcriptome. BMC Genomics 13, 82. Hawkins, B.T., and Davis, T.P. (2005). The blood-brain barrier/neurovascular unit in health and disease. Pharmacol. Rev. 57, 173–185. Huh, D., Kim, H.J., Fraser, J.P., Shea, D.E., Khan, M., Bahinski, A., Hamilton, G.A., and Ingber, D.E. (2013). Microfabrication of human organs-on-chips. Nat. Protoc. 8, 2135–2157.

Barrile, R., van der Meer, A.D., Park, H., Fraser, J.P., Simic, D., Teng, F., Conegliano, D., Nguyen, J., Jain, A., Zhou, M., et al. (2018). Organ-on-chip recapitulates thrombosis induced by an anti-CD154 monoclonal antibody: translational potential of advanced microengineered systems. Clin. Pharmacol. Ther. 104, 1240–1248.

Jain, A., Barrile, R., van der Meer, A.D., Mammoto, A., Mammoto, T., De Ceunynck, K., Aisiku, O., Otieno, M.A., Louden, C.S., Hamilton, G.A., et al. (2018). Primary human lung alveolus-on-a-chip model of intravascular thrombosis for assessment of therapeutics. Clin. Pharmacol. Ther. 103, 332–340.

Benam, K.H., Villenave, R., Lucchesi, C., Varone, A., Hubeau, C., Lee, H.H., Alves, S.E., Salmon, M., Ferrante, T.C., Weaver, J.C., et al. (2016). Small airway-on-a-chip enables analysis of human lung inflammation and drug responses in vitro. Nat. Methods 13, 151–157.

Kamiichi, A., Furihata, T., Kishida, S., Ohta, Y., Saito, K., Kawamatsu, S., and Chiba, K. (2012). Establishment of a new conditionally immortalized cell line from human brain microvascular endothelial cells: a promising tool for human blood-brain barrier studies. Brain Res. 1488, 113–122.

1004 Cell Stem Cell 24, 995–1005, June 6, 2019

Katt, M.E., Xu, Z.S., Gerecht, S., and Searson, P.C. (2016). Human brain microvascular endothelial cells derived from the BC1 iPS cell line exhibit a bloodbrain barrier phenotype. PLoS ONE 11, e0152105. Kere´nyi, L., Koltai, M., and Szirmai, I. (1980). CSF transferrins characterized by the transferrin/albumin index. Clin. Chim. Acta 105, 195–199. Kim, H., Edwards, N.J., Choi, H.A., Chang, T.R., Jo, K.W., and Lee, K. (2016a). Treatment strategies to attenuate perihematomal edema in patients with intracerebral hemorrhage. World Neurosurg. 94, 32–41. Kim, S., Hwang, Y., Lee, D., and Webster, M.J. (2016b). Transcriptome sequencing of the choroid plexus in schizophrenia. Transl. Psychiatry 6, e964. LeVine, S.M. (2016). Albumin and multiple sclerosis. BMC Neurol. 16, 47. Lim, R.G., Quan, C., Reyes-Ortiz, A.M., Lutz, S.E., Kedaigle, A.J., Gipson, T.A., Wu, J., Vatine, G.D., Stocksdale, J., Casale, M.S., et al. (2017). Huntington’s disease iPSC-derived brain microvascular endothelial cells reveal WNT-mediated angiogenic and blood-brain barrier deficits. Cell Rep. 19, 1365–1377. Lippmann, E.S., Azarin, S.M., Kay, J.E., Nessler, R.A., Wilson, H.K., Al-Ahmad, A., Palecek, S.P., and Shusta, E.V. (2012). Derivation of blood-brain barrier endothelial cells from human pluripotent stem cells. Nat. Biotechnol. 30, 783–791. Lippmann, E.S., Al-Ahmad, A., Azarin, S.M., Palecek, S.P., and Shusta, E.V. (2014). A retinoic acid-enhanced, multicellular human blood-brain barrier model derived from stem cell sources. Sci. Rep. 4, 4160. Liu, Y., Chen, B.P.C., Lu, M., Zhu, Y., Stemerman, M.B., Chien, S., and Shyy, J.Y.J. (2002). Shear stress activation of SREBP1 in endothelial cells is mediated by integrins. Arterioscler. Thromb. Vasc. Biol. 22, 76–81. Mantle, J.L., Min, L., and Lee, K.H. (2016). Minimum transendothelial electrical resistance thresholds for the study of small and large molecule drug transport in a human in vitro blood-brain barrier model. Mol. Pharm. 13, 4191–4198. Maoz, B.M., Herland, A., Henry, O.Y.F., Leineweber, W.D., Yadid, M., Doyle, J., Mannix, R., Kujala, V.J., FitzGerald, E.A., Parker, K.K., and Ingber, D.E. (2017). Organs-on-Chips with combined multi-electrode array and transepithelial electrical resistance measurement capabilities. Lab Chip 17, 2294–2302. Maoz, B.M., Herland, A., FitzGerald, E.A., Grevesse, T., Vidoudez, C., Pacheco, A.R., Sheehy, S.P., Park, T.E., Dauth, S., Mannix, R., et al. (2018). A linked organ-on-chip model of the human neurovascular unit reveals the metabolic coupling of endothelial and neuronal cells. Nat. Biotechnol. 36, 865–874. Pardridge, W.M. (2005). The blood-brain barrier: bottleneck in brain drug development. NeuroRx 2, 3–14.

Applicability to study antibody-triggered receptor-mediated transcytosis. Sci. Rep. 8, 1873. Rochfort, K.D., and Cummins, P.M. (2015). The blood–brain barrier endothelium: a target for pro-inflammatory cytokines. Biochem. Soc. Trans. 43, 702–706. Sances, S., Ho, R., Vatine, G., West, D., Laperle, A., Meyer, A., Godoy, M., Kay, P.S., Mandefro, B., Hatata, S., et al. (2018). Human iPSC-derived endothelial cells and microengineered organ-chip enhance neuronal development. Stem Cell Reports 10, 1222–1236. Santaguida, S., Janigro, D., Hossain, M., Oby, E., Rapp, E., and Cucullo, L. (2006). Side by side comparison between dynamic versus static models of blood-brain barrier in vitro: a permeability study. Brain Res. 1109, 1–13. Shelley, B.C., Gowing, G., and Svendsen, C.N. (2014). A cGMP-applicable expansion method for aggregates of human neural stem and progenitor cells derived from pluripotent stem cells or fetal brain tissue. J. Vis. Exp. (88), 51219. Spencer, J.I., Bell, J.S., and DeLuca, G.C. (2018). Vascular pathology in multiple sclerosis: reframing pathogenesis around the blood-brain barrier. J. Neurol. Neurosurg. Psychiatry 89, 42–52. Stebbins, M.J., Gastfriend, B.D., Canfield, S.G., Lee, M.-S., Richards, D., Faubion, M.G., Li, W.-J., Daneman, R., Palecek, S.P., and Shusta, E.V. (2019). Human pluripotent stem cell-derived brain pericyte-like cells induce blood-brain barrier properties. Sci. Adv. 5, eaau7375. Sweeney, M.D., Ayyadurai, S., and Zlokovic, B.V. (2016). Pericytes of the neurovascular unit: key functions and signaling pathways. Nat. Neurosci. 19, 771–783. €nen, S., Lindhe, O., Palner, M., Kornum, B.R., Rahman, O., La˚ngstro¨m, B., Syva Knudsen, G.M., and Hammarlund-Udenaes, M. (2009). Species differences in blood-brain barrier transport of three positron emission tomography radioligands with emphasis on P-glycoprotein transport. Drug Metab. Dispos. 37, 635–643. Vatine, G.D., Al-Ahmad, A., Barriga, B.K., Svendsen, S., Salim, A., Garcia, L., Garcia, V.J., Ho, R., Yucer, N., Qian, T., et al. (2017). Modeling psychomotor retardation using iPSCs from MCT8-deficient patients indicates a prominent role for the blood-brain barrier. Cell Stem Cell. 20, 831–843.e5. Wang, Y.I., Abaci, H.E., and Shuler, M.L. (2017). Microfluidic blood-brain barrier model provides in vivo-like barrier properties for drug permeability screening. Biotechnol. Bioeng. 114, 184–194. White, C.R., Haidekker, M., Bao, X., and Frangos, J.A. (2001). Temporal gradients in shear, but not spatial gradients, stimulate endothelial cell proliferation. Circulation 103, 2508–2513.

Pardridge, W.M., Eisenberg, J., and Yang, J. (1987). Human blood-brain barrier transferrin receptor. Metabolism 36, 892–895.

Wolff, A., Antfolk, M., Brodin, B., and Tenje, M. (2015). In vitro blood-brain barrier models—an overview of established models and new microfluidic approaches. J. Pharm. Sci. 104, 2727–2746.

Prabhakarpandian, B., Shen, M.C., Nichols, J.B., Garson, C.J., Mills, I.R., Matar, M.M., Fewell, J.G., and Pant, K. (2015). Synthetic tumor networks for screening drug delivery systems. J. Control. Release 201, 49–55.

Yeon, J.H., Na, D., Choi, K., Ryu, S.W., Choi, C., and Park, J.K. (2012). Reliable permeability assay system in a microfluidic device mimicking cerebral vasculatures. Biomed. Microdevices 14, 1141–1148.

Refetoff, S., Robin, N.I., and Fang, V.S. (1970). Parameters of thyroid function in serum of 16 selected vertebrate species: a study of PBI, serum T4, free T4, and the pattern of T4 and T3 binding to serum proteins. Endocrinology 86, 793–805.

Yuan, W., Lv, Y., Zeng, M., and Fu, B.M. (2009). Non-invasive measurement of solute permeability in cerebral microvessels of the rat. Microvasc. Res. 77, 166–173.

Ribecco-Lutkiewicz, M., Sodja, C., Haukenfrers, J., Haqqani, A.S., Ly, D., Zachar, P., Baumann, E., Ball, M., Huang, J., Rukhlova, M., et al. (2018). A novel human induced pluripotent stem cell blood-brain barrier model:

Zhang, Y., Sloan, S.A., Clarke, L.E., Caneda, C., Plaza, C.A., Blumenthal, P.D., Vogel, H., Steinberg, G.K., Edwards, M.S.B., Li, G., et al. (2016). Purification and characterization of progenitor and mature human astrocytes reveals transcriptional and functional differences with mouse. Neuron 89, 37–53.

Cell Stem Cell 24, 995–1005, June 6, 2019 1005

STAR+METHODS KEY RESOURCES TABLE

REAGENT or RESOURCE

SOURCE

IDENTIFIER

Nestin

Millipore

Cat# MAB353; RRID: AB_94911

Tuj1a (bIII Tubulin)

Sigma-Aldrich

Cat# T8660; RRID: AB_477590

GFAP

Dako

Cat# Z0334; RRID: AB_10013382

GLUT-1

Thermo Fisher

Cat# MA5-11315; RRID: AB_10979643

Claudin-5

Novex

Cat# AB_2533200

MAP2ab

Sigma-Aldrich

Cat# M1406; RRID: AB_477171

Neurofilament, Heavy

R&D systems

Cat# AF3108; RRID: AB_2149640

Occludin

Invitrogen

Cat# 33-1500; RRID: AB_2533101

PECAM-1 (CD31)

Thermo

Cat# 10333; RRID: AB_2533101

Phallodin

Invitrogen

Cat# T7471; RRID: AB_2620155

S100b

Abcam

Cat# ab66028; RRID: AB_1142710

Synaptophysin

R&D systems

Cat# AF5555; RRID: AB_2198864

ZO-1

Invitrogen

Cat# 33-9100; RRID: AB_87181

PDGFRb

Abcam

Cat# ab32570; RRID: AB_777165

P-glycoprotein-1

Thermo Fisher

Cat# 187246; RRID: AB_86866

a-SMA

ABCAM

Cat# ab5694; RRID: AB_2223021

Matrigel

BD Bioscienses

356234

mTeSR1

StemCell Technologies, Inc.

85851

Rocki Y-27632

Tocris Biosciences

1254

Accutase

EMD Millipore

SCR005

bFGF

EMD Millipore

01-106

EGF

EMD Millipore

GF144

Antibodies

Chemicals, Peptides, and Recombinant Proteins

Heparin

Sigma-Aldrich

84020

Triiodothyronine, T3

Sigma-Aldrich

T2877

Triiodothyronine-[13C6], T3 internal standard

Isosciences

8165

Human endothelial serum-free medium

Life Technologies

11111-044

All-trans-Retinoic acid

Sigma-Aldrich

R2625

Platelet-poor Plasma Derived Serum, Bovine

Biomedical Technologies, Inc.

BT-214

Human albumin

ABCAM

ab205808

IgG

ABCAM

ab205806

Transferrin

Sigma-Aldrich

T0665

Fluo-4

Invitrogen

F10489

B27 –vitamin A

Thermo Fisher

12587010

N2 Supplement

Thermo Fisher

17502048

BDNF

PeproTech

450-02

Cascade Blue-Dextran, 3,000 MW

Thermo Fisher

D7132

FITC-Dextran, 4,000 MW

Sigma-Aldrich

46944

FITC-Dextran, 20,0000 MW

Sigma-Aldrich

FD20S

FITC-Dextran, 70,0000 MW

Sigma-Aldrich

46945

TRITC-Dextran, 70,000 MW

Thermo Fisher

D1819

2-NBDG, glucose analog

Sigma-Aldrich

72987

Retigabine

Sigma-Aldrich

R-018

Retigabine-D4

Sigma-Aldrich

R-019 (Continued on next page)

e1 Cell Stem Cell 24, 995–1005.e1–e6, June 6, 2019

Continued REAGENT or RESOURCE

SOURCE

IDENTIFIER

Levetiracetam

Sigma-Aldrich

L8668

Levetiracetam-D6

Sigma-Aldrich

L-023

Colchicine

Sigma-Aldrich

C9754

Colchicine-D6

Clearsynth

CS-O-06507

Critical Commercial Assays Oasis HLB 96-well Plate 30 mg

Waters

WAT058951

Kinetex C18-100A (2.6 m, 30 3 3 mm)

Phenomenex

00A-4462-Y0

LiChrospher RP-18

Phenomenex

RP-18

SUPELCOSIL LC-18-DB HPLC Column

Sigma

58978C30 SUPELCO

Human Albumin ELISA Kit

Abcam

ab108788

Human IgG ELISA Kit

Abcam

ab195215

Human Transferrin ELISA Kit

Abcam

ab187391

Pierce LDH cytotoxicity assay kit

Thermo Fisher Scientific

88953

RNeasy Mini Kit

QIAGEN

74104

RNeasy Micro Kit

QIAGEN

74004

PureLink RNA Mini Kit

Invitrogen

12183020

SMART-Seq v4 Ultra Low Input RNA Kit

Takara

091817

This paper

GEO: GSE129290

CS03iCTR

iPSC-core, Cedars-Sinai Medical Center

39 years old healthy male

CS01iMCT8

iPSC-core, Cedars-Sinai Medical Center

2.5 years old MCT8-deficient male

CS01iMCT8cor

iPSC-core, Cedars-Sinai Medical Center

CS01iMCT8

CS03iCTRmut

iPSC-core, Cedars-Sinai Medical Center

CS03iCTR

CS83iCTR

iPSC-core, Cedars-Sinai Medical Center

21 years old healthy female

CS83iCTR-ACTB::nGFP-#46

iPSC-core, Cedars-Sinai Medical Center

Nuclear GFP in AAVS1 safe harbor locus, driven by beta-actin promoter

CS0617iCTR

iPSC-core, Cedars-Sinai Medical Center

79 years old healthy male

CS0172iCTR

iPSC-core, Cedars-Sinai Medical Center

79 years old healthy male

CS0188iCTR

iPSC-core, Cedars-Sinai Medical Center

80 years old healthy male

CS81iHD

iPSC-core, Cedars-Sinai Medical Center

20 years old Huntington’s Disease female

Human primary astrocytes

Sciencell

1800

Human primary pericytes

Sciencell

1200

Deposited Data RNA-seq data reported in this paper Experimental Models: Cell Lines

Software and Algorithms Illumina bcl2fastq

Ilumina

Salmon (version 0.11.3)

https://combine-lab.github.io/salmon/

R version 3.5.2

https://www.r-project.org

DESeq2

Bioconductor

biomaRt

Bioconductor

ComplexHeatmap

Bioconductor

pheatmap

https://cran.r-project.org

tidyverse

https://cran.r-project.org

reshape2

https://cran.r-project.org

DAVID Bioinformatics Resources 6.8

DAVID bioinformatic database

HCImageLive

HCImage

Cell Stem Cell 24, 995–1005.e1–e6, June 6, 2019 e2

CONTACT FOR REAGENT AND RESOURCE SHARING Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Clive N. Svendsen ([email protected]). EXPERIMENTAL MODEL AND SUBJECT DETAILS Primary cultures were obtained under consent and privacy guidelines. All procedures were performed in accordance with the IRB guidelines at the Cedars-Sinai Medical Center under IRB-SCRO Protocols Pro00021505, Pro00032834, Pro00049203, and Pro00049203. All iPSC lines were generated at the iPSC Core at Cedars-Sinai Medical Center. Details regarding origins of cells are provided in the Key Resources Table. METHOD DETAILS BBB-Chip Microfabrication and Culture The basic design of the Organ-Chip used for the BBB-Chip model and its fabrication is based on previously described protocols (Huh et al., 2013). The Chip is composed of a flexible polydimethylsiloxane (PDMS) elastomer that contains two closely opposed and parallel micro-channels (1 3 1 mm and 1 3 0.2 mm, brain and blood channel, respectively) (Jain et al., 2018) separated by a laminin-coated, porous flexible PDMS membrane (50 mm thick, with 7-mm diameter pores with 40 mm spacing, resulting in 2% porosity over a surface area of 0.171cm2 separating the two channels). Chips were coated with laminin (50 mg/ml; Sigma) for the ‘brain side’ and a mixture of collagen IV (320 mg/ml; Sigma) and fibronectin (80 mg/ml; Sigma) in H2O mixture for the ‘blood side’. Coated Chips were then incubated overnight at 37 C and 5% CO2. Typically, Chips were seeded with iBMECs at a concentration of 14,000 cells/ml and allowed to attach to the membrane. After 3-24 hours, EZ-spheres were thawed, dissociated into single cells and seeded at a density ranging between 1000-6000 cells/ml and incubated overnight before flushing with fresh media. Media was then replaced daily with approximately 100 mL in each channel. Generation of EZ-spheres from iPSCs EZ-spheres were generated from iPSC colonies as previously described. Briefly, iPSC colonies grown on Matrigel-coated plates (0.5 mg/plate) in mTESR1 (StemCell Technologies) media were removed by gentle scraping and placed directly into DMEM:F12 7:3 supplemented with 2% B27 (without Vitamin A), basic fibroblast growth factor (bFGF, 100 ng/ml; Chemicon), epidermal growth factor (100 ng/ml; Chemicon), and heparin (5 mg/ml; Sigma) in ultra-low attachment flasks and were passaged weekly using a chopping technique. EZ-spheres were frozen three days post-seeding in Sigma freezing media and stored in liquid N2. Differentiation of EZ-spheres into Neural Cultures As previously defined (Ebert et al., 2013), to induce neural differentiation, EZ-spheres were dissociated with Accutase (EMD Millipore) and seeded onto the laminin (Sigma)-coated ‘brain side’ of the Chip. Neural differentiation was induced by neural media; DMEM:F12 with 2% B27 –vitamin A (Thermo Fisher), 1% N2 Supplement (Thermo Fisher), human brain-derived neurotrophic factor (20 ng/ml; Peprotech 450-02). Differentiation of iPSCs into iBMECs iPSCs were passaged onto Matrigel in mTeSR1 medium for 2–3 days of expansion until reaching a density of 2-3x105 cells/well and then switched to unconditioned medium lacking bFGF for 6 days. Then, human endothelial serum-free medium (hESFM; Life Technologies) supplemented with 20 ng/ml bFGF, all-trans retinoic acid (RA, 10 mM; Sigma) and 1% platelet-poor plasma derived bovine serum (Biomedical Technologies) was then added for 2 days. Cells were then incubated with Accutase for 35 min, gently dissociated and seeded onto the ‘blood side’ of the chip. iBMECs were then grown in endothelial cell medium (ECM) without bFGF and RA. Primary Cell Culture and Organ-Chip Preparation Primary human brain astrocytes and human primary brain vascular pericytes were both purchased from Sciencell, expanded in a T75 flask for one passage and cryopreserved according to manufacturer’s instructions. In order to allow cell recovery and minimize experimental variability, at each experiment one cryovial of astrocytes and pericytes was thawed and expanded in a T75 flask for two days before seeding onto the Organ-Chip. Cells were gently detached with 0.05% trypsin (BD Biosciences) for 2-4 minutes at room temperature. 9 3 105 astrocyte and 3 3 105 pericyte cells/ml in DMEM supplemented with 10% fetal bovine serum (FBS), were introduced into the top compartment. Following attachment, the brain compartment was gently rinsed with fresh cell culture medium and incubated overnight at 37 C. The next day, BBB-Chips were connected to the Human Emulation System and the vascular compartment was perfused at 30 ml/hr to continuously provide fresh media supply. Immunocytochemistry Immunocytochemistry was conducted as previously described (Lippmann et al., 2012; Vatine et al., 2017) with minor modifications. Cells were blocked on the Organ-Chip in phosphate buffered saline (PBS) containing 5%–10% donkey serum (Sigma) at 4 C e3 Cell Stem Cell 24, 995–1005.e1–e6, June 6, 2019

overnight. Triton X-100 0.1% was used to permeabilize membrane when required. Primary antibodies were Nestin 1:1000 (Millipore, MAB353), bIII-Tubulin 1:1000 (Tuj1a, Sigma-Aldrich, T8660), GFAP 1:1000 (Dako, Z0334), Map2AB 1:1000 (Sigma, M1406), hNF 1:100 (R&D systems, AF3108), Synaptophysin 1:50 (R&D systems, AF5555), GLUT-1 1:100 (Thermo Fisher, MA5-11315), PECAM-1 1:100 (Thermo, 10333), OCCLUDIN 1:100 (Invitrogen, 33-1500), CLAUDIN-5 1:100 (Novex, 35-2500). Appropriate fluorescent secondary antibodies (1:000; Invitrogen) were incubated for 1 hr at room temperature. Cells were then counterstained with nuclear dye Dapi. Images were captured with a Leica AF3500 microscope. Assessment of Vascular Integrity and Blood-to-Brain Transport of Soluble Biomarkers Transmission light microscopy was used to assess vascular integrity before and during experiments. In some cases, TEER was measured using a specific chip design incorporating gold electrodes as recently described (Maoz et al., 2017). Additionally, the blood-to-brain dextran-FITC permeability was monitored as described below. Image Analysis of Astrocyte Endfeet Coverage and Cell Junction Integrity In order to quantify the percentage of vascular surface covered with astrocyte protrusions (endfeet like structures), the vascular surface of the BBB-Chips were washed in DMEM (without FBS) and stained with the fluorescently labeled wheat germ agglutinin (WGA-647 Invitrogen), a leptin able to bind to the vascular glycocalyx. After 15 minutes at room temperature, BBB-Chips were rinsed twice with fresh DMEM, fixed in 2% paraformaldehyde for 20 minutes at room temperature and stained for the astrocyte marker GFAP (1:1000, Abcam). Following incubation with the anti-mouse secondary antibody (Alexa Fluor-488, 1:000; Invitrogen) and Hoescht (Abcam), Chips were washed in PBS and imaged via confocal microscopy (Zeiss LSM850). The area underneath the porous membrane was localized using the WGA-647 signal (that stained the whole endothelial surface) and the nuclear fluorescence signal (Hoescht) of endothelial cells. Then, five random areas of each Chip were imaged using the GFAP fluorescent (Alexa 488) signal using a 10X objective (Zeiss). ImageJ software was used to subtract background and then convert the digital signal detected in the GFAP channel into a binary image. Finally, percentage of signal coverage was computed from the binary image as the ratio of bright pixels to the total number of pixels in the image. Estimation of endothelial tight-junctions integrity was performed as followed. Image analysis was performed using Fiji/ImageJ and the ridge detection plugin by analyzing 4 field of view captured with 20X objective on n = 3 chips per condition stained for the tight junction marked ZO-1. Briefly, each image was background subtracted then, in order to estimate the amount of intact cell-edges, the ridge detection was conducted on each captured image using standard settings. The number of ridges detected in each field of view was then divided by the number of cell-nuclei in the same image. The graph reports the percentage of intact edges normalized to the control condition. Image processing and quantification were performed with an automatic macro compiled in ImageJ in order to ensure unbiased signal measurements. Numerical values were collected and statistically analyzed using GraphPad Prism V7. Permeability Assays All transport and permeability assays were conducted by perfusion of both the top and bottom channels with medium flow at 30 ml/hr. The bottom channel was perfused with neural media, human plasma, or whole human blood treated with 3.2% sodium citrate (Research Blood Components) and the top channel was perfused with neural medium. Medium/blood collected from both inputs and effluents from both top and bottom channels were read by fluorescence, luminescence, or MS. Fluorescence (485 nm excitation and 530 nm emission) or luminescence were detected on a plate reader. The values measured were used to calculate Papp values as follow:     mg mg ml Flow Rate Top Output --Top Input sec ml  mg ml  Papp = 2Þ Membrane areaðcm Bottom Input ml In order to assess barrier functions, dextran-FITC (3, 4, 20 or 70 kDa) was spiked in the ‘blood side’ and the concentrations were evaluated in the effluents of both the top and bottom channels. In testing the permeability of proteins, human albumin (ABCAM, ab205808), IgG (ABCAM, ab205806) and transferrin (Sigma, T0665) were used in place of dextran and the ‘brain side’ was maintained static (without flow) in order to maximize accumulation of the biomarkers in this compartment of the Organ-Chip. Detection and quantitation of the proteins was performed using the following ELISA kits from ABCAM: Albumin (ab108788), IgG (ab195215), Transferrin (ab187391) after perfusion overnight. Blood Perfusion and Shear Stress Blood samples anticoagulated in sodium citrate (purchased from Research Blood Components) were reconstituted with calcium as previously described and perfused through the vascular compartment of the BBB-Chip for 15 minutes at 3,600 ml/hr (equivalent to 5 dyn/cm2). For experiments conducted with human plasma, samples were also purchased from Research Blood Components and stored at 80. One day before the experiment, human plasma was thawed at 4 C then loaded with fluorescent-dextran (100 mg/ml) €-CM1) of the and perfused through the blood compartment of the BBB-Chip at 3,600 ml/hr. Because the current Culture Module (Zoe Human Emulation System (Emulate) does not support whole blood perfusion at physiologically relevant shear rate, experiments performed with whole human blood were carried on standard syringe pumps, as previously described (Barrile et al., 2018). Similarly, in Cell Stem Cell 24, 995–1005.e1–e6, June 6, 2019 e4

order to characterize the cellular response of iBMECs to shear stress independently from blood, we perfused the vascular chamber of the Organ-Chips using cell culture media on a standard IPC-N-series peristaltic pump (Ismatec, Switzerlpump), as previously described (Benam et al., 2016). In both cases, pumps were set-up to generate shear stress using the formula here below to convert ‘‘flow-rate’’ (ml/h) in ‘‘shear stress’’ (dyn/cm2):     dyn viscosity x flowrate 109 10 = 6 cm2 height2 x width Height and width of the vascular channels are respectively 200um and 1mm. Viscosity of the cell culture medium was estimated to be 0.00089 Pa.s while the viscosity of the blood was estimated to be 0.00332 Pa.s. Efflux Pump Activity The transporter activity of PGP1 was assessed using the MDR1 Efflux Assay Kit (ECM910, Millipore), per manufacturer’s instructions. Briefly, BBB-Chips were perfused with the cell-permeable, fluorescent PGP1 substrate rhodamine 123 and intracellular accumulation was detected via fluorescent imaging, and measured in the presence or absence of the PGP1 inhibitor vinblastine (0.022 mM, from the Millipore Kit). Each experiment was performed on three BBB-Chips per condition, imaged at three fields of view (20x) and digitally processed using ImageJ software. Viability Assay Measuring lactic dehydrogenase (LDH), which is based on absorbance (A.U.) levels, was performed to assess plasma-induced neural-toxicity following whole blood perfusion. Effluents from the ‘brain side’ of chips were processed within 24 hours using the LDH detection kit (Pierce LDH cytotoxicity assay kit #88953) following manufacturer’s instruction. Increases in LDH release relate to damage to the cell membrane and provides an indicator of cell viability. Fluorescence-Activated Cell Sorting (FACS) of iBMEC Co-cultures CS83iCTR-ACTB::nGFP-#46 iPSC cell line expressing a nuclear GFP integrated into the adeno-associated virus integration site 1 (AAVS1) safe harbor locus was used to differentiate iBMECs. GFP-expressing iBMECs were co-cultured in the BBB-Chip with unlabeled primary astrocytes and pericytes or iPSC-derived neural progenitors. After 4 days in culture, cells were dissociated with Trypsin-EDTA, washed, filtered, and sorted directly into lysis buffer based on GFP expression. RNA isolation and sequencing were performed as described below. Transcriptional Analysis of iBMECs Total RNA was isolated from iBMECs using a RNeasy Mini Kit (QIAGEN) or PureLink RNA Mini Kit (Invitrogen) with on-column DNase digestion. RNA integrity was checked with an Agilent 2100 Bioanalyzer and only samples with RIN >9 were selected for cDNA library construction. Libraries were prepared with SMART-Seq v4 Ultra Low Input RNA Kit (Takara) and sequenced on a NextSeq 500 (Illumina) using single-end 75 bp reads at the Cedars-Sinai Genomics Core. Demultiplexing and conversion of raw sequencing data to FASTQ was performed with Illumina bcl2fastq software. FastQC was used to assess sequence quality and reads were aligned to the GRCh38 genome assembly (GENCODE 28) using Salmon version 0.11.3. Gene-level read counts were quantified with Salmon, and differential gene expression and normalization was subsequently performed using DESeq2. Gene Ontology Analysis Gene ontology analysis was performed using DAVID Bioinformatics Resources 6.8 on the top 500 genes in Principal Component 1, ranked by eigenvalue. Gene expression values were pre-filtered using a row count >10 and normalized using DESeq2 variance stabilizing transformation. Calcium Imaging Fluo-4 (Invitrogen), a cell permeant calcium dye, was resuspended to 5 mg/ml in DMSO with 10% (w/v) Pluronic F127 (Invitrogen). Fluo-4 was then diluted to 10 ng/ml in extracellular solution (ECS; 135mM NaCl, 5mM KCl, 1.2mM MgCl2, 1.25mM CaCl2, 5mM HEPES, 10mM Glucose, pH 7.4) and incubated in the BBB-Chip for 30 min at room temperature in the dark. Organ-Chips were flushed with fresh ECS without Fluo-4, and incubated for an additional 30 min. Organ-Chips were imaged on an Olympus BX51WI upright microscope at 10 frames per second under continuous ECS flow. Tetrodotoxin (TTX, 1 mM) was added to ECS to block sodium channels. Videos were captured with HCImageLive software and processed with Adobe Premiere Pro. Measurements of Iodothyronines by LC/MS/MS T3 was measured by liquid chromatography (dual LC Shimadzu Prominence system, Shimadzu, Columbia, MD) followed by tandem mass spectrometry (Q-Trap 6500, Sciex) with a TurboV ion source. After addition of the internal standard 13C6 – T3 (Iso Sciences) to 100 mL sample of effluent collected from the top or bottom channel, T3 was extracted with 1:4 (v/v) of EtOH:NH4OH (98:2). The combined supernatants were evaporated using a speed vacuum (Thermo Scientific). The residue was then reconstituted in 100 mL of 0.1% formic acid in water and 40 mL of reconstituted extract was injected into Kinetex C18-100A (2.6 m, 30 3 3 mm, Phenomenex, CA) column, protected by a STEM 2190 Phenomenex C18-RP guard cartridge in 40 C column oven. Iodothyronines e5 Cell Stem Cell 24, 995–1005.e1–e6, June 6, 2019

were chromatographed with 0.1% formic acid in deionized water (aqueous mobile phase A) and 0.1% formic acid in methanol (organic mobile phase B). The gradient was 5 to 90% B in 2 min with flow rate of 0.4 ml/min. The positive ion multiple reaction monitoring (MRM) mode was used for detection. The MRM transition monitored was: m/z 651.5 > 605.5 for T3; m/z 651.5 > 605.5 for 13C6 – T3. All the MRM data were processed with Multiquant (Sciex). QUANTIFICATION AND STATISTICAL ANALYSIS All experiments were performed in triplicate or more with a minimum of three independent neural or iBMEC differentiations, with a single clone for each cell line. When comparing two groups, statistical analyses were performed using an unpaired Student’s t test. For multiple group comparisons, data were analyzed by one-way ANOVA, followed by Tukey test to adjust for multiple comparisons. For all testing the level of significance was set at a two-sided p < 0.05 and error bars represent standard error of the mean. DATA AND SOFTWARE AVAILABILITY The accession number for the RNA-seq data reported in this paper is GEO: GSE129290.

Cell Stem Cell 24, 995–1005.e1–e6, June 6, 2019 e6