henhaobio
Sign in

Advanced ScienceBioengineeringPeer-reviewed journal article

Cerebral Organoids with Integrated Endothelial Networks Emulate the Neurovascular Unit and Mitigate Core Necrosis

Josep Fumadó NavarroiD, Siobhan CrillyiD, Wai Kit ChaniD, S. BrowneiD, John O. MasoniD, Catalina Vallejo‐GiraldoiD, Abhay S. PanditiD, Mihai LomoraiD

PDFDOIPreprint
In-depth full-text digest available · Jump to digest ↓

Abstract

Cerebral organoids (COs) are multicellular, self-organized, in vitro, 3D brain-like tissues used for developmental biology, disease modelling and drug screening. However, their lack of vascularity renders them less physiologically accurate. Vascularization of COs remains challenging due to the different requirements between COs and vascular cells, limited vascular network penetration within the organoid, and the absence of luminal perfusion. Here, we devised an encapsulation approach in which human brain microvascular endothelial cells (HBMVECs) were delivered to developing COs from progressively degrading extracellular matrix (ECM)-based hydrogel droplets. By tuning this hydrogel concentration and media composition, we observed enhanced vascular-like network formation that expanded within the organoid tissue. Using pathway inhibitors, we showed that a subset of the endothelial cells (ECs) originated from the CO itself, promoting network integration. Endothelial networks displayed blood-brain barrier (BBB) features, including astrocytic end-footlike interactions, pericyte wrapping, and collagen-laminin basal lamina. Vascularized COs exhibited greater media internalization and up to three-fold lower apoptosis than non-vascularized COs. This comprehensive 3D neurovascular model is a promising platform for cerebrovascular research and drug testing applications. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=132 SRC="FIGDIR/small/650161v1_ufig1.gif" ALT="Figure 1"> View larger version (53K): org.highwire.dtl.DTLVardef@175edb7org.highwire.dtl.DTLVardef@1499b91org.highwire.dtl.DTLVardef@184567borg.highwire.dtl.DTLVardef@149c8e4_HPS_FORMAT_FIGEXP M_FIG Graphical abstract C_FIG HighlightsO_LIAddition and angiogenic stimulation of human brain microvascular endothelial cells (HBMVECs) co-cultured with cerebral organoids (COs) generate multicellular vascular-like networks C_LIO_LIVascularization induces changes in organoid morphology but not on tissue stiffness C_LIO_LIEndothelial networks morphological features are correlated with the concentration of the supporting matrix C_LIO_LIA number of the endothelial cells (ECs) in the networks originate from the organoid itself C_LIO_LIEndothelial networks integrate within the organoid tissue interacting with astrocytes and pericyte-like cells, and are surrounded by basement membrane-like depositions C_LIO_LIVascularized COs exhibit higher media diffusion, and a reduced necrotic core compared to non-vascularized COs C_LI

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (26 Apr 2025) · CC BY 4.0

Embedding day-8 cerebral organoids in Geltrex loaded with human brain microvascular endothelial cells, plus VEGF-supplemented mixed media, yields integrated endothelial networks and almost three-fold less apoptosis.

Cerebral organoids lack vasculature, so millimetre-scale tissue develops hypoxic necrotic cores, and earlier routes (ETV2 overexpression, added HUVECs or vessel organoids, bioprinting or chips) yield variable, mostly superficial networks with little intra-organoid perfusion. This work offers a simpler embedding-step strategy with commercially available brain-specific endothelial cells and reagents that produces neurovascular-unit-like structures inside the tissue.

Key results

  1. ShownVascularized organoids had almost three-fold fewer apoptotic cells, and the remaining cell death was spread evenly instead of concentrating in the dense core seen in controls.

    TUNEL on central sections: apoptotic cell percentage reduced almost three-fold in both A- and S-iPSC lines (two-way ANOVA vascularization effect p < 0.0001; N = 7-14, three or four batches per line); controls showed core-enriched death (Tukey p < 0.05), while vascularized COs showed no regional difference (ns). · Vascularized COs are more permeable to nutrients…; TUNEL assay; Figure legends

  2. ShownVascularized organoids contain internal CD31+ networks with neurovascular-unit hallmarks: VE-cadherin junctions, GFAP+ astrocytes in contact with the tubes, NG2+ pericyte-like cells, and basement-membrane-like collagen IV and laminin alpha5 deposits.

    Internal CD31+ area/nuclei area was higher in vascularized than in batch-matched control cryosections (two-way ANOVA p < 0.001; N = 8-14, three or four batches per line). Networks sat mostly in low-density edge regions, not in neuroepithelium, and were fewer in S-COs. Masson's trichrome and human-specific antibodies confirmed the matrix deposits. · Vascular-like networks are present within the…; Figure legends

  3. SuggestedVascularized organoids took up more 70 kDa dextran and contained lumen-bearing networks. This fits a leaky, partly open endothelial network but does not show true perfusion.

    Retained d-RhoB per organoid area was higher in vascularized COs (Welch's t-test p < 0.01; N = 14-17), and the size-fluorescence regressions differed (p = 0.0089). d-RhoB+ section area was higher in vascularized A-COs (Mann-Whitney p < 0.05; N = 20-38 areas) but not in S-COs. H&E showed open lumens. · Vascularized COs are more permeable to nutrients…; 70 kDa dextran penetration tests; Figure legends

  4. ShownThe networks mix the added endothelial cells with endothelial cells made by the organoid itself. How much endogenous endothelium forms depends on the iPSC line, and dual SMAD inhibition in cortical organoids suppresses it.

    Tissue-cleared day 40 COs contained both Zombie Green iPSC-derived ECs and endogenous CD31+ cells. Brachyury was absent at days 5 and 8. Without HBMVECs, A-COs formed networks but S-COs and A-CortiCOs mostly did not. Superficial CD31+ area was around 5% in BECs CTRL and 10-15% with VEGF. · HBMVECs form multicellular networks on the…; Endothelial networks in COs are a mix of HBMVECs…; Figure legends

  5. ShownHydrogel concentration tunes network morphology. Vascularization does not change organoid stiffness, and supernatant VEGF in HBMVEC-containing cultures stays above the dose added to the medium, which the authors attribute to endothelial release.

    Vessel diameter was wider at 95% than at 80% and 60% Geltrex (Dunn's p < 0.0001; N = 9-22, A-iPSC line). In 10-day gels, 40% Geltrex gave the most network area and 80% barely connected. Young's modulus was 0.5-3 kPa with no CTRL vs BECs MM+VEGF difference (N = 10-13). Supernatant VEGF stayed over 100 ng/mL vs 50 ng/mL dosed. · HBMVECs network formation is modulated by the…; Integration of HBMVECs into cerebral organoids…; HBMVECs form multicellular networks on the…; Figure legends

Methods, data and reproducibility

Lancaster-based STEMdiff cerebral organoids from two iPSC lines were embedded on day 8 in 80% Geltrex carrying 50,000 HBMVECs, then grown to day 40 in 1:7 endothelial:organoid mixed media with 50 ng/mL VEGF. Outcomes were assessed by whole-mount and section imaging, nanoindentation, ELISA, dextran uptake and TUNEL.

  • A-iPSCs: Gibco Human Episomal iPSC Line (hPSCreg TMOi001-A, Cat# A18945, female), now discontinued but kept in the authors' biobank · human
  • S-iPSCs: STEMCELL Healthy Control Human iPSC Line SCTi003-A (Cat# 200-0511, female) · human
  • Zombie Green-expressing iPSC line differentiated into CD31-sorted fluorescent endothelial cells (Z-ECs) for lineage tracing
  • Primary human brain microvascular endothelial cells pooled from multiple donors (Innoprot P10361), used at passage 8 · human
  • Cerebral organoids from A- and S-iPSCs under six conditions (with or without HBMVECs; CTRL, +VEGF, MM+VEGF) cultured to day 40 · human · n = N = 16-25, five batches, two iPSC lines (growth)
  • Cortical organoids (A-CortiCOs) made with dual SMAD inhibition, compared with A-COs and S-COs under identical vascularization conditions · human · n = N = 6-12, two or three rounds per organoid type
  • In vitro HBMVEC network-formation assay: 100,000 cells per well in 40-80% Geltrex for 10 days, varying media ratio and VEGF dose · human · n = N = 6 wells, two repeats
Methods & tools
STEMdiff Cerebral Organoid Kit (Lancaster et al. protocol), Geltrex basement membrane hydrogel, AngioTool64 v0.6a, ImageJ v1.54f, PAVONE Nanoindenter with Optics11 DataViewer V2.5.0 (Hertzian model), Human MMP-9 and Quantikine VEGF ELISA kits, Click-iT Plus TUNEL Assay Kit, FDISCO tissue clearing, 70 kDa dextran-Rhodamine B uptake assay, Masson's trichrome and H&E histology, GraphPad Prism v8.0.2, FlowJo v10 with BD Accuri C6 Plus flow cytometry
Reagents & environment
ImageJ v1.54f; AngioTool64 v0.6a; Optics11 DataViewer V2.5.0; GraphPad Prism v8.0.2; QuPath v0.4.3; ImageScope v12.4.6.5003; FlowJo v10; Fluoview FV31S
Reproducibility gaps
Image-analysis macros and reanalysis details only on request from the lead contact; no raw data repository given; A-iPSC line discontinued commercially but held in the authors' biobank; Key Resources Table empty in the extracted text.
Limitations (6)
  • Cultures were stopped at day 40, so effects of vascularization on longer-term neural maturation and on resemblance to human neurodevelopment were not assessed. · stated by the authors
  • Perfusion and barrier function were not shown: this needs a dynamic flow system and assays such as TEER, and the dextran experiments could not tell whether dye was held inside open lumens. · stated by the authors
  • The two iPSC lines differed markedly in endogenous endothelial differentiation, so wider adoption will need validation studies to keep results reproducible across lines. · stated by the authors
  • Geltrex could not be diluted below 60% without destabilizing the embedding droplet in dynamic culture, and basement-membrane-derived matrices carry batch-to-batch variability. Defined hydrogels are proposed as an alternative. · stated by the authors
  • Apoptosis and dextran comparisons pit BECs MM+VEGF against CTRL organoids only, without MM+VEGF-only or HBMVEC-only arms. The endothelial contribution therefore cannot be separated from that of the added VEGF and ECG media. · noted in this summary
  • Necrotic-core reduction is inferred from TUNEL counts, with no hypoxia or necrosis markers. Cell identities rest on a few immunostains with no transcriptomic profiling of endothelial brain specification or neural off-target effects. · noted in this summary

Read this if you build brain organoid or blood-brain barrier models and want a matrix-embedding route to integrated endothelial networks using commercially available cells and reagents.

Citations
17
FWCI
6.11
Field percentile
top 3%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISBhalla '26Sun '22Fedele '22Shi '20Salmon '21Sato '23
76% semantic matchDevelopmental Biology
Shubhang Bhalla, Belda Gülsuyu, Damian Sanchez+22 more·bioRxiv·Jan 10, 2026· 1 citations