Assembly of Human Stem Cell-Derived Cortical Spheroids and Vascular Spheroids to Model 3-D Brain-like Tissues

Human cerebral organoids derived from induced pluripotent stem cells (iPSCs) provide novel tools for recapitulating the cytoarchitecture of human brain and for studying biological mechanisms of neurological disorders. However, the heterotypic interactions of neurovascular units, composed of neurons, pericytes, astrocytes, and brain microvascular endothelial cells, in brain-like tissues are less investigated. The objective of this study is to investigate the impacts of neural spheroids and vascular spheroids interactions on the regional brain-like tissue patterning in cortical spheroids derived from human iPSCs. Hybrid neurovascular spheroids were constructed by fusion of human iPSC-derived cortical neural progenitor cell (iNPC) spheroids, endothelial cell (iEC) spheroids, and the supporting human mesenchymal stem cells (MSCs). Single hybrid spheroids were constructed at different iNPC: iEC: MSC ratios of 4:2:0, 3:2:1 2:2:2, and 1:2:3 in low-attachment 96-well plates. The incorporation of MSCs upregulated the secretion levels of cytokines VEGF-A, PGE2, and TGF-β1 in hybrid spheroid system. In addition, tri-cultured spheroids had high levels of TBR1 (deep cortical layer VI) and Nkx2.1 (ventral cells), and matrix remodeling genes, MMP2 and MMP3, as well as Notch-1, indicating the crucial role of matrix remodeling and cell-cell communications on cortical spheroid and organoid patterning. Moreover, tri-culture system elevated blood-brain barrier gene expression (e.g., GLUT-1), CD31, and tight junction protein ZO1 expression. Treatment with AMD3100, a CXCR4 antagonist, showed the immobilization of MSCs during spheroid fusion, indicating a CXCR4-dependent manner of hMSC migration and homing. This forebrain-like model has potential applications in understanding heterotypic cell-cell interactions and novel drug screening in diseased human brain.

Brain organoids derived from human induced pluripotent stem cells (hiPSCs) emerge as powerful model systems for neurological disease modeling, drug screening, and for studying Zika virus infections [1][2][3][4][5] , which affect over one billion people globally 6 . However, generating brain-region specific organoids with defined structure and function remains a critical challenge because the heterotypic cell-cell interactions to mimic human brain have not yet been fully understood [7][8][9] . Recently, fusion of human forebrain spheroids of different regions (e.g., human dorsal spheroids with ventral spheroids) has been investigated to model interneuron migration and the interactions of different neuronal subtypes [10][11][12] . However, the interactions of neuronal cells with other cell types, such as endothelial cells, have not been fully studied in brain organoids 5 .
Neural-vascular interactions, known as neural-vascular unit, play an important role in brain structure and function 13 . It has been suggested that organ-specific endothelial cells secrete a unique set of growth factors that regulate tissue morphogenesis into desired tissue types 14 . Vascular cells can form spheroids to assemble blood vessels or as building blocks for scaffold-free tissue fabrication 15,16 . In vitro vascularization of organoids has been attempted for cardiac organoids, showing the enhanced cardiac cell function 17 . In vivo vascularization of organoids was realized for the hiPSC-derived organ buds, in which the mixed hiPSC-derived progenitors and endothelial cells efficiently self-organize into functional and vascularized liver or kidney in vivo respectively 18,19 . In particular, blood-brain barrier (BBB) is involved in various neurological diseases development, drug

Results
Aggregate fusion and cell localization of tri-cultured hybrid spheroids. The initial optimization of mixing sequence to achieve maximum spheroid fusion for the tri-cultured hybrid spheroids was summarized in Supplementary Materials (Supplementary Information 1 and Figs S1-S4). Briefly, two methods were evaluated: (A) iNPC-MSC-iEC; and (B) iNPC-iEC-MSC (Fig. 1A) 30 . The squared aspect ratio of contact length between the two aggregates over maximum diameter was calculated to assess the spheroid fusion process. The iNPC-MSC spheroids and iEC spheroids, at different iNPC: iEC: MSC ratios of Bi-(4:2), Tri-(3:2:1), Tri-(2:2:2), and Tri-(1:2:3), fused into the aggregates with squared aspect ratio of 0.5~0.7 after 7 days. For iNPC-iEC-MSC spheroids, hMSCs were added into the well of day 14 iNPC-iEC hybrid spheroids and cultured for another 7 days or MSCs and iEC spheroids were added together to the wells containing iNPC spheroids ( Fig. 1B-D). hMSCs were integrated with the iNPC-iEC hybrid spheroids and migrated toward the spheroid center. The squared aspect ratio was in the range of 0.9-1.0 (Fig. 1E). The low concentration of ROCKi did not impact the integration of MSCs into iNPC-iEC spheroids. Due to the higher squared aspect ratio (indicating good fusion kinetics), iNPC-iEC-MSC spheroids were mainly used in the following experiments.
Cell proliferation, metabolic activity, and cytokine secretion. DNA content of hybrid spheroids was measured to evaluate the proliferation potential of the hybrid sspheroids. At day 1 after co-culture, Bi-(4:2) group showed the highest DNA content, while the DNA content in the other three group was comparable. After 7 days of co-culture, all four groups showed cell proliferation with increased DNA content ( Fig. 2A). Increased DNA content was observed for Tri-(2:2:2) and Tri-(1:2:3) spheroids with the treatment of 5% GelTrex (Fig. 2B). The treatment of 0.05 wt% HA showed no significant difference in DNA content except for Bi-(4:2) spheroids, while 0.025 wt% HA treatment showed no significant difference in DNA content for all the groups ( Fig. 2C and Supplementary Fig. S6A). Higher MTT activity was observed for the Bi-(4:2) spheroids compared to the other three groups, and the treatment of ROCKi increased MTT activity (Fig. 2D). 5-Bromo-2′-deoxyuridine (BrdU) assay showed that the cells in S-phase of cell cycle were not homogeneously happening in the spheroids, but tended to be localized at the interface of the spheroids or the spheroid surface ( Supplementary Fig. S6B).
The trophic factors secreted by MSCs (e.g., FGF2 and VEGF-A) can enhance the angiogenesis, neurogenesis, and axonal growth during neural tissue regeneration 36,37 . MSCs also secret anti-inflammatory factors, such as TGF-β1 and PEG2, to regulate the immune response 38,39 . The secretion levels of FGF2, VEGF-A, PGE2, and TGF-β1 from hybrid spheroids was characterized (Fig. 3). The highest secretion levels were observed for the hMSC-only group. For FGF2, higher secretion was observed for Tri-(2:2:2) spheroids compared to iNPC-only spheroids (Fig. 3A). The VEGF-A concentration increased with the relative ratio of hMSCs in the hybrid Figure 1. Fusion kinetics of iNPC spheroids and hMSCs with iEC spheroids to construct hybrid spheroids. (A) (i) Schematic illustration of endothelial cell (iEC) spheroid derivation from hiPSCs. (ii) Schematic illustration of generating hybrid spheroids from iNPCs. hMSCs were added to iNPCs for iNPC-MSC-iEC spheroids (method A), before iEC transfer. Or hMSCs were added to the well containing iNPC and iEC spheroids for iNPC-iEC-MSC spheroids (method B). hMSCs were labeled with CellTracker Red. (B) Phase contrast images of iNPC-iEC-MSC spheroids morphology at day 1, 3, 5, and 7 (total day [15][16][17][18][19][20][21]. iEC spheroids and MSCs were added to the preformed iNPC aggregates. Scale bar: 400 μm. (C) (i) Schematic illustration of calculation of squared aspect ratio of contact length between two aggregates over maximum diameter ( ) Lneck Lmax 2 . The aggregation kinetics were evaluated by (ii) the squared aspect ratio over 7 days and (iii) the inter-sphere angle formed by the two aggregates. (D) Overlay of phase contrast images (iNPCs and iECs) with fluorescent images (hMSCs) of hybrid spheroids (i) with or (ii) without ROCKi Y27632 (10 μM) when MSCs were added to the culture one week after fusion of iNPC spheroids and iEC spheroids. Scale bar: 400 μm. (E) The aggregation kinetics were analyzed in (Ei) and (Eii), respectively. *Indicates p < 0.05 for the different test conditions. www.nature.com/scientificreports www.nature.com/scientificreports/ from MSCs is maintained in hybrid spheroid culture and the amount is dependent on the ratio of incorporated hMSCs.
Cortical neural differentiation of tri-cultured hybrid spheroids. Hybrid spheroids were replated to investigate their cellular composition ( Fig. 4 and Supplementary Fig. S8). The expression of vascular markers, CD31 and VE-cadherin, was observed for all the groups (Fig. 4A). The fused spheroids also expressed ZO1, the tight junction protein of brain microvascular cells. Co-staining of CD31 and Nestin showed that CD31 + cells interacted with the Nestin + neural cells (Fig. 4B). The expression of deep cortical layer VI marker TBR1 (and a little cortical layer II-IV BRN2 expression) indicated cortical identity of the hybrid spheroids, although hindbrain marker HOXB4 was also expressed at this stage. Extensive MAP2 expression also showed neuron population, while the expression of GFAP indicated the existence of glial progenitors (Fig. 4C). Additional markers for astrocyte lineage, including S100B, vimentin, and Aldolase C, were also detected ( Supplementary Fig. S9). Cells from tri-cultured spheroids had more E-cadherin expression (heterogeneous signal intensity), while Bi-(4:2) group had homogenous expression.
Histological sections were evaluated to assess the in situ distribution and localization of iECs and neural cells within the hybrid spheroids (Fig. 5A). Numerous CD31 + vascular cells interacted with β-tubulin III + neurons were observed throughout the spheroids (Supplementary Figs S10 and S11). In addition, the distribution of CD31 was more homogenous for Tri-(2:2:2) and Tri-(1:2:3) groups. More ZO1 expression was observed for tri-cultured spheroids. Confocal images of intact spheroids showed the FOXG1 + layers and the lumens of CD31 + cells ( Fig. 5B and Supplementary Fig. S12). The expression of TBR1 inside the fused spheroids was observed, but the expression of BRN2 was minimal at this early stage ( Supplementary Fig. S13A). In addition, one side of spheroids expressed HOXB4.
The electrophysiological properties of the outgrowth cells of the derived spheroids/organoids were examined via patch clamping. As cells within the dense core of the spheroid cannot be visualized by phase contrast microscopy while in the recording chamber, outgrowth cells toward the boundary of the spheroid were chosen for these experiments. Recorded cells displayed fast inward currents and long-lasting outward currents during voltage-clamp recording, suggesting the presence of functional voltage-gated Na + and K + channels, respectively (Supplementary Fig. S17). In addition, a subpopulation of the cells fired rebound action potentials in response to hyperpolarizing current injection during current clamp recording. Spontaneous postsynaptic currents were observed in the absence of stimulation during continuous voltage clamp recording. Cellular morphology was stereotypically neuron-like, with small cell bodies and extensive long and thin projections. Together, these results suggest that the hybrid spheroids have the functional and morphological properties of neurons including synaptic activity.
It has been reported that cell migration in cerebral organoids depends on CXCR4 (a cell homing receptor) activity 12 . To understand the mechanisms of spheroid fusion and the self-sorting behavior of different cell types, the effects of CXCR4 antagonist AMD3100 was investigated (Fig. 8). The MTT activity indicated that AMD3100 treatment had little influence on cell proliferation ( Supplementary Fig. S21). iEC spheroids (with CellTracker Red) gradually fused with the iNPC spheroids. Small hMSC (with CellTracker Green) areas sparsely spread over the fused spheroids for Tri-(3:2:1) group. For Tri-(2:2:2) and Tri-(1:2:3) groups, one large area of hMSCs occupied the interface of iEC and iNPC spheroids (Fig. 8Ai,Bi). For AMD3100 treatment, the area occupied by hMSCs was smaller than the control groups (Fig. 8Aii,Bii). Analysis of the relative ratio of area occupied by hMSCs to the total area of fused spheroids showed that the aspect ratio of hMSCs decreased from day 4 to day 8 with AMD3100 treatment (Fig. 8Bii). These results indicate that hMSC migration and invasion into the iEC and iNPC spheroids may be mediated by a CXCR4-dependent manner.

Discussion
Current hiPSC-derived brain organoids show promising results in modeling different neurological diseases such as microcephaly 3 , lissencephaly 43 , and ZIKV infections 2 . However, the lack of interactions with other cell types such as endothelial cells in current brain organoids model limits their applications 5,9 . As neurological diseases such as BBB breakdown and dysfunction in Alzheimer's and stroke involve multiple cell types, in vitro models such as brain organoids must include relevant cell types to better reconstruct cellular microenvironment 13 . As www.nature.com/scientificreports www.nature.com/scientificreports/ vascular system is an essential component of brain tissue, incorporating neural-vascular interactions in forebrain organoids is an important step in developing brain organoids in vitro.
This study applied a neural-vascular co-patterning method through the fusion of independently derived cortical spheroids and isogenic endothelial spheroids from hiPSCs to introduce vascular cells into the cortical www.nature.com/scientificreports www.nature.com/scientificreports/ spheroids. To our knowledge, it is the first study to introduce vasculature into the cortical spheroids/organoids through spheroid fusion. Most existing study directly mixed the endothelial cells with the other cell types for 3-D co-culture systems 17,19,44 . The advantage of spheroid fusion method over direct mixing method is that (1) it avoids cell dissociation and re-association process, which could lose many cells; (2) the hybrid spheroid structure can be pre-controlled with special compartment arrangements.
Spheroid fusion method has been used for assembly of dorsal and ventral forebrain spheroids to study interneuron migration and the assembly of iPSC-derived endothelial progenitor spheroids and smooth muscle progenitor spheroids 10,16 . Fusion of spheroids of different cell types is most likely driven by minimization of interfacial free energy and cellular thermodynamics, differential cellular adhesions (e.g., E-cadherin expression), or cortical tension redistribution 28,45,46 . The fusion kinetics was found to be affected by ROCK inhibitor, ECMs in the medium, and mixing sequence based on our study. The addition of high concentration of ROCKi Y-27632 during the initial aggregation delayed fusion process, indicating that cortical tension regulates spheroid fusion process, and actomyosin may play a key role in spheroid fusion 47,48 . The presence of Geltrex and HA (an ECM component in the brain) at an appropriate concentration promotes spheroid fusion. Since MSCs condense into the center of the hybrid spheroids, adding MSCs to iNPC spheroids before iEC spheroid fusion (iNPC-MSC-iEC) would constrict MSCs as the iNPC spheroid core. So the better mixing sequence is iNPC and iEC spheroid fusion in the presence of MSCs (iNPC-iEC-MSC).

Neural-vascular interactions through spheroid fusion.
The results of this study indicate that assembly of vascular spheroids and cortical spheroids enhanced the glucose transporter, GLUT-1 (specifically expressed in endothelial cells in brain 13 ), and a polarized efflux transporter BCBP. Structurally, the tight junction protein ZO1 was promoted in the tri-culture, indicating that neurovascular co-patterning promotes the specification of iECs toward brain microvascular cells. Human iPSC-based 2-D co-culture systems of multiple cell types (i.e., neurons, astrocytes, pericytes and brain microvascular endothelial cells) have been recently reported to mimic BBB function with higher trans-endothelial electrical resistance (TEER) properties and study drug permeability in vitro [21][22][23][24] . www.nature.com/scientificreports www.nature.com/scientificreports/ Neural-vascular interactions result in the special structure and function of BBB. However, the 3-D BBB models have not been well established due to the complex BBB feature and the difficulty to form micro-vessels structure in 3-D. Some studies use hollow fiber system with perfusion culture or artificially creating microchannels for 3-D vascularization 20 . Although our system is not yet an accurate and perfusable 3-D BBB model, the system recapitulates the anatomical features of the BBB using human stem cells. The inclusion and characterization of additional cell types (e.g., astrocytes), complex 3-D capillary network development (need novel biomaterials design), and perfusion flow study with bioreactors or microfluidics may be explored in future 49 .
Neural-vascular interactions also impact brain tissue patterning. Our results showed the elevated β-tubulin III and CD31 expression, as well as higher TBR1 and Nkx2.1 gene expression in tri-culture, in particular Tri-(1:2:3) group. In addition, the "inside-out" development of cortical superficial layer and deep layer in forebrain is faster for Tri-(1:2:3) group compared to other groups. It has been reported that neural differentiation of hPSCs requires direct association with vascular cells 50 as well as interactions with hMSCs, possibly through mitogen-activated protein kinase (MAPK) and PI3K-Akt signaling (involved FGF2) 51 . In particular, our tri-culture system promotes the expression of Notch-1, the key protein in Notch signaling which is responsible for cell-cell contact interactions involved ECs 52,53 . In addition, activation of Notch signaling can promote the neural stem cell self-renewal, glial cell differentiation, and neuron regeneration 54,55 . The results in this study indicate that direct contact among iNPCs, iECs, and hMSCs at a given ratio accelerates the development of 3-D cortical tissue structure containing vascular cells to model human brain development.

the role of MsC in neural-vascular interactions-insoluble eCMs. The interactions of hMSCs and
iNPC have been discussed in our previous study 30 . In this study, the presence of hMSCs in addition to iECs enriches ECM localization and affects matrix remodeling. As brain ECMs have limited fibril ECMs such as collagens 20 , the expression of collagen IV in hybrid spheroids was mainly attributed to the incorporation of hMSCs. The elevated MMP-2 and MMP-3 expression indicates the active matrix remodeling, which is required for maintaining the function of neural stem cells 42 . The formation of hMSC aggregates upregulates several types of MMPs (MMP-2, -9, and -1/13) 56 , which were reported to enhance neuronal differentiation through NF-кB signaling 57 . ECM remolding may also be an important contributing factor involved in the migration and invasion of the hMSCs. In this study, the immobilized hMSCs after CXCR4 (receptor for CXCL12/stromal-derived factor-1 chemokine) inhibition reveals that the migration of hMSCs contributes to the fusion of the hybrid spheroids. www.nature.com/scientificreports www.nature.com/scientificreports/ the role of MsC in neural-vascular interactions-soluble cytokines. Secretion of cytokines and neurotrophin is a critical function of hMSCs, which can enhance neurogenesis of hiPSCs 58 . The influence of TGF-β1 and PGE2 secreted by hMSCs on hiPSC-neural differentiation was discussed in our previous study 30 . The upregulated TGF-β1 and PGE2 secretion by hMSCs promotes Nestin and β-tubulin III expression. In this study, two additional growth factors FGF2 (mitogen) and VEGF-A were measured. Our results showed that the main source of VEGF-A is hMSCs, although vascular cells also contribute to VEGF-A secretion. Previous studies have suggested that brain vascular ECs promote neural cell functionality, such as synaptic activities, via the modulation of VEGF signaling and the VEGF receptors were activated by neural cell-secreted nitric oxide 59,60 . In our study, iECs may be a minor source of VEGF-Aas shown in Bi-(4:2) group, as they are not mature enough compared to the bone marrow-derived hMSCs. The elevated VEGF-A in tri-culture (in particular Tri-(1:2:3) group) regulated by Notch signaling is a result of close cell-cell contacts of neural-vascular-mesenchymal cells through autocrine, paracrine, and juxtacrine interactions 61,62 . It was noted that the amount of cytokine produced did not correlate with the DNA content increase in different hybrid spheroids (Fig. 3A), since the DNA content increase was mainly attributed to iNPCs and iECs, but not hMSCs.
The cellular ratio indicates that Tri-(1:2:3) group better promotes neural-vascular interactions than the other groups, indicated by higher cytokine secretion, neural patterning marker expression, BBB-related gene expression, and the cortical layer separation. Consistently, the brain composition is reported to have neuron-to-astrocyte ratio at 1:3 22 , indicating the importance of accessory cells on neural functions. Long-term culture of hMSCs (DNA degradation was observed) is limited in this study using neural differentiation medium in the tri-culture for brain tissue development. Similarly, endothelial cell maturation medium (e.g., EGM-2 medium) was not able to be used to promote mature vascular structure. These results indicate the needs to optimize medium formulations that support all cell types in the co-culture system. In addition, the apoptosis may exist due to upregulated caspase3/7 expression and altered mitochondria bioenergetics on 3-D MSC aggregation due to compaction 63 . An alternative is the use of iPSC-derived MSC (iMSCs) in the tri-culture system as reported by Gao et al. 64 .
Vascularization is crucial in the development of brain organoids in vitro but remains a significant challenge. To date, the actual vascularization was only achieved in vivo 18,19,65 . In vitro vascularization needs the accurate design of ECM amounts, the ECM structure (insoluble), and the soluble secreted factors. Recently, colleagues at UC Davis embedded whole brain organoids derived from hiPSCs in Matrigel with isogenic ECs or coated day 34 spheroids with ECs to achieve in vitro and in vivo vascularization 66 . Another study from Mansour et al. (2018) vascularized brain organoids in vivo showing the integration of microglia and the functional neuronal networks and blood vessels 65 . All these studies indicate that neural-vascular interactions are indispensable for modeling neurological diseases and screening drugs that require 3-D brain tissue structure 44 . This is in particularly true for the use of diseased hiPSC lines such as Alzheimer's disease 31 . Development of Alzheimer's-patient derived cortical organoids containing vascular cells would be important to recapitulate neurodegenerative microenvironment and investigate the cellular response to drug treatments.

Conclusions
This study assembled hiPSC-derived cortical spheroids and isogenic vascular spheroids in the presence of hMSCs to study neurovascular interactions. The presence of hMSCs promotes cortical neural differentiation, layer separation, cytokine secretion, and cell-cell communication. The presence of iECs provides the BBB-related properties inside the cortical spheroids/organoids. Our results indicate that the elevated Notch signaling, matrix remodeling proteins, and the secretion of VEGF-A through the assembly of cortical spheroids, vascular spheroids, and mesenchymal cells contribute to the accelerated cortical tissue development. This study provides insights to heterotypic cell-cell interactions and potential strategy for the fabrication of next generation of forebrain organoids.

Methods
Undifferentiated hiPSC culture. Human iPSK3 cells were derived from human foreskin fibroblasts transfected with plasmid DNA encoding reprogramming factors OCT4, NANOG, SOX2 and LIN28 (kindly provided by Dr. Stephen Duncan, Medical College of Wisconsin, and Dr. David Gilbert, Department of Biological Sciences of Florida State University) 67,68 . Human iPSK3 cells were maintained in mTeSR serum-free medium (StemCell Technologies, Inc., Vancouver, Canada) on 6-well plates coated with growth factor reduced Geltrex (Life Technologies). The cells were passaged by Accutase dissociation every 5-6 days and seeded at 1 × 10 6 cells per well of 6-well plate in the presence of 10 μM Y27632 (Sigma) for the first 24 hours 33,69,70 . Human MsC (hMsC) culture. Standardized frozen hMSCs from multiple donors were obtained from the Tulane Center for Gene Therapy and cultured as previously described 71,72 . The hMSCs were isolated from the bone marrow of healthy donors ranging in age from 19 to 49 years based on plastic adherence, negative for CD34, CD45, CD117 (all less than 2%) and positive for CD29, CD44, CD49c, CD90, CD105, and CD147 markers (all greater than 95%), and possess tri-lineage differentiation potential upon induction in vitro 73,74 . Briefly, hMSCs were expanded at a density of 1.7 × 10 3 cells/cm 2 using αMEM (Invitrogen) medium supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. At approximately 80% confluence, adherent cells were harvested with 0.25% trypsin-EDTA (Sigma-Aldrich) and further propagated.
Hybrid iNPC, iEC and hMSC spheroid formation and neural differentiation. Two different methods with different sequences of adding iEC spheroids or hMSCs to the wells containing iNPC spheroids were evaluated.

Effects of Geltrex, hyaluronic acid (HA) hydrogels, and ROCKi Y27632 on spheroid fusion.
To form HA hydrogels, 1% (w/v) HA (Sigma) solution was reacted with 5-fold molar excess amount of methacrylic anhydride (sigma) for 15 h in the dark at 4 °C. The final product was collected by precipitating the solution in 5-fold volume of ethanol twice and purified by dialysis using a membrane (3.5 kDa Mw cut-off, Thermofisher) to remove unreacted reagents. Purified MA-HA was filtered, lyophilized, and stored at −20 °C until further use 80 .
Aggregate fusion analysis. The images of spheroid fusion were captured over time by a phase contrast microscope. The captured images were converted to binary images using ImageJ software (http://rsb.info.nih. gov/ij) and analyzed with the "particle analysis tool". Through particle analysis, the squared aspect ratio of contact length between the two aggregates over maximum diameter (Fig. 1C(i)) was calculated to indicate the aggregate fusion process. The inter-sphere angle was measured as the intersecting angle between the tangent lines of two contacted spheroids to the touch point. At least three images were analyzed for each data point. For some experiments, the number of branching points and the total tube length in the images were evaluated. Biochemical assays. MTT assay. The spheroids of different conditions were incubated with 5 mg/mL 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT, Sigma) solution at day 7 after tri-culture unless otherwise noted. The absorbance of the samples was measured at 500 nm using a microplate reader (Biorad, Richmond, CA).
DNA assay. The DNA content of the hybrid spheroids was determined at day 7 after three cell types were co-cultured unless otherwise noted. DNA standard was prepared by dissolving salmon testes DNA in TEX (10 mM Tris, 1 mM EDTA, 0.1% Triton X-100 at pH 8) and a standard curve was constructed for each assay. The aggregates were lysed with 0.1 mg/mL proteinase K (Fisher Scientific, Pittsburgh, PA) at 50 °C overnight. The lysates (100 μL) were mixed with 100 μL of Picogreen (Molecular Probes) in a 96-well plate. The plate was incubated for 5 min in the dark and then read on a fluorescent plate reader (FLX800, Bioinstrument Inc., Winooski, VT).
Enzyme-linked immunosorbent assay (ELISA) assay. To quantify the growth factors secreted by different spheroids, culture supernatants were collected at day 7 after three cell types were co-cultured. Concentrations of FGF2, PGE2, VEGF, and TGF-β1 were measured by ELISA according to the manufacturers' instructions (R&D Systems, Minneapolis, MN for PGE2 and FGF2; Life Technologies for TGF-β1 and VEGF). www.nature.com/scientificreports www.nature.com/scientificreports/ Immunocytochemistry. Briefly, the samples were fixed with 4% paraformaldehyde (PFA) and permeabilized with 0.2-0.5% Triton X-100. The samples were then blocked for 30 min and incubated with various mouse or rabbit primary antibodies (Supplementary Table S1) for four hours. For surface markers, no permeabilization was performed. After washing, the cells were incubated with the corresponding secondary antibody: Alexa Fluor ® 488 goat anti-Mouse IgG 1 , Alexa Fluor ® 488 or 594 goat anti-Rabbit IgG, or 594 donkey anti-goat IgG (Life Technologies) for one hour. The samples were counterstained with Hoechst 33342 and visualized using a fluorescent microscope (Olympus IX70, Melville, NY) or a confocal microscope (Zeiss LSM 880).

Flow cytometry.
To quantify the levels of various markers, the cells were harvested by trypsinization (0.05% trypsin for 10-15 min followed by pipetting with micro-tips) and analyzed by flow cytometry 81 . Briefly, 1 × 10 6 cells per sample were fixed with 4% PFA and washed with staining buffer (2% FBS in PBS). The cells were permeabilized with 100% cold methanol, blocked, and then incubated with primary antibodies against β-tubulin III, KDR, CD31, and VE-cadherin, followed by the corresponding secondary antibody Alexa Fluor 488 goat anti-Mouse IgG 1 (for β-tubulin III, KDR), or Alexa Fluor 594 donkey anti-goat IgG (for CD31, VE-Cadherin 82 ).
For surface markers, no permeabilization was performed. The cells were acquired with BD FACSCanto ™ II flow cytometer (Becton Dickinson) and analyzed against isotype controls using FlowJo software.
Reverse transcription polymerase chain reaction (Rt-pCR) analysis. Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Valencia, CA) according to the manufacturer's protocol followed by the treatment of DNA-Free RNA Kit (Zymo, Irvine, CA) 83 . Reverse transcription was carried out using 2 μg of total RNA, anchored oligo-dT primers (Operon, Huntsville, AL), and Superscript III (Invitrogen, Carlsbad, CA) (according to the protocol of the manufacturer). Primers specific for target genes (Supplementary Table S2)  Effect of AMD3100. The Day 14 hiPSC-EC spheroids were labeled with CellTracker Red. The hMSCs were labeled with CellTracker Green. The hybrid (iNPC-iEC-MSC) spheroids were cultured in neural differentiation media (control) or media containing the CXCR4 inhibitor AMD3100 (100 nM, Sigma) for additional 10 days 84 . The fusion kinetics and cell localization were captured over time. The cell viability of day 10 hybrid spheroids was determined by MTT activity assay.
Whole-patch clamping for electrophysiology. Whole-cell patch clamp was used to record from iPSK3-derived spheroids cultured on glass covered slips. Cover slips were washed three times with extracellular recording solution containing (in mM) 136 NaCl, 4 KCl, 2 MgCl, 10 HEPES, and 1 EGTA (312 mOsm, pH 7.39) and were incubated in this solution at room temperature during recording. Glass electrodes (resistance 1-5 MΩ) were filled with intracellular solution containing 130 mM KCl, 10 mM HEPES, and 5 mM EGTA (292 mOsm, pH 7.20). Cells were visualized under phase contrast with a Nikon Eclipse Ti-U inverted microscope and attached DS-Qi1 monochrome digital camera. Recordings were made with an Axopatch 200B amplifier (Molecular Devices) and digitized with a Digidata 1440 A system (Molecular Devices). Ionic currents were recorded under a voltage clamp protocol (−60 mV to 135 mV in 15 mV steps, 250 ms in duration). Action potentials were recorded under a current clamp protocol (−100 pA to 200 pA in 20 pA steps, 800 ms in duration). Spontaneous post-synaptic currents were recorded under continuous voltage clamp at −80 mV for 2 min. Signals were filtered at 1 kHz and sampled at 10 kHz. Data was collected and analyzed using pCLAMP 10 software (Molecular Devices). statistical analysis. Each experiment was carried out at least three times (using different batches of cells) with triplicate samples (in some cases spheroids were pooled from more than 12 wells) in each experiment. The representative experiments were presented and the results were expressed as [mean ± standard deviation]. To assess the statistical significance, one-way ANOVA followed by Fisher's LSD post hoc tests were performed. A p-value < 0.05 was considered statistically significant.