X32 - Mimicking the Neurovascular Niche: 3D Co-Culture of Astrocytes and Endothelial Cells in Hydrogels with Tunable Stiffness and Localized Nanoparticle-Drug Delivery

Introduction: : Neurological disorders, including stroke and neurodegeneration, are among the most devastating health burdens globally. Despite decades of research, effective therapies remain elusive. Traditional 2D culture systems and rodent models do not replicate key features of human brain physiology, leading to poor clinical translatability of preclinical findings. In vitro human brain models also do not typically replicate the cellular and structural complexity of the human brain, as they lack mimicry of the neurovascular unit and its tightly regulated blood-brain barrier (BBB).
New biomaterials and 3D bioprinting technologies have opened new frontiers in engineering brain-mimetic platforms. However, existing bioinks frequently fall short in balancing mechanical tunability with cellular compatibility – a balance that is especially important when co-culturing mechanosensitive glial and endothelial cells in 3D. To address this gap, we developed a brain extracellular matrix-inspired hydrogel composed of methacrylated hyaluronic acid (HAMA), gelatin methacroyl (GelMA), and poly(ethylene glycol) diacrylate (PEGDA), which was further designed for extrusion-based 3D printing. This formulation enables fine control over matrix stiffness and spatial configuration, while supporting the viability and phenotype of both astrocytes and endothelial cells.
Here, using spinning disc confocal microscopy, we characterized astrocyte and endothelial cell morphology, polarity, and spatial distribution in brain-mimetic 3D hydrogels. We discovered an emergence of distinct, organized cell layers within the hydrogel—mimicking the astrocyte-endothelial cell colocalization observed in vivo within the BBB. Our hydrogel platform allows for BBB modelling, neuroinflammation studies, and CNS-targeted therapeutic development.

Materials and

Methods: : We engineered a hydrogel composed of HAMA (0 – 2 wt%), GelMA (0-10 wt%), and PEGDA (0-1.95 wt%). HAMA was synthesized in-house with a 50-60% extent of methacrylation by Fischer esterification with methacrylic anhydride. GelMA and PEGDA were procured commercially. Poly(lactic-co-glycolic acid) Resomer 502H (PLGA) (initially dissolved in acetone) was fabricated into nanoparticles via nanoprecipitation and solvent evaporation. PLGA nanoparticles were incorporated into select formulations at up to 3 wt% to assess potential for drug-loading applications. Photo-crosslinking was achieved using 0.25% (w/v) lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) and UV light exposure (365 nm).
We examined the impact of matrix stiffness (based on tuning the bioink composition) on printability and cell response. Rheological behavior was assessed via amplitude sweep testing (0.1–50% strain at 1 Hz) and compressive modulus measurements to characterize the linear viscoelastic region (LVER) and structural robustness of each formulation. Primary human astrocytes were purchased from ScienCell and encapsulated within the hydrogel bulk and stained for aquaporin-4 (AQP4), while human brain endothelial cells (hCMEC/D3) were seeded on the surface and stained for ZO-1 to assess expression of membrane biomarkers correlated with tight junction integrity. Cell membranes were labeled with wheat germ agglutinin (WGA), and nuclei with Hoechst to enable detailed morphological analysis. Immunostaining protocols were optimized to preserve 3D architecture and antigen localization. Imaging was performed using spinning disc confocal microscopy, allowing for volumetric analysis of cell morphology, polarization, and spatial organization. Astrocyte and endothelial cell viability was quantified using Live/Dead staining and confocal microscopy at 24- and 48-hours post-encapsulation.

Results, Conclusions, and Discussions:: Compositional tuning of the hydrogel system enabled precise control over its mechanical behavior. Increasing GelMA concentration (2–10%) progressively elevated the compressive modulus and broadened the linear viscoelastic region (LVER), (Fig. B), indicating improved mechanical integrity and deformation tolerance. Amplitude sweep testing (0.1–50% strain, 1 Hz) showed that the LVER expanded from ~12% strain at 2% GelMA to ~30% at 10% GelMA, with storage modulus (G′) remaining within ±5% of its initial value throughout each range—demonstrating stable elastic behavior essential for maintaining construct fidelity during extrusion and culture. The H2G5 formulation (2% HAMA, 5% GelMA) emerged as an optimal balance between printability and compliance.
Primary human astrocytes were encapsulated within the hydrogel during crosslinking, while hCMEC/d3 endothelial cells were seeded on the surface post-gelation to mimic neurovascular organization, as shown in Fig. (A). Astrocyte viability remained high, with 97.85% ± 0.51% and 98.32% ± 0.63% at 24 and 48 hours, respectively. Morphology was matrix dependent: at 5% GelMA, astrocytes exhibited stellate morphology with extended processes and polarized AQP4 expression. Softer (2%) and stiffer (10%) matrices induced more compact forms—suggesting matrix-guided regulation of process extension and phenotype.
Spinning disc confocal microscopy revealed spatially compartmentalized co-culture: astrocytes (AQP4, green) populated the hydrogel bulk, while endothelial cells (ZO-1, red) formed surface monolayers (Fig. C). Continuous ZO-1 staining confirmed tight junction formation, and astrocytes retained polarized AQP4 expression—resembling native neurovascular organization. In H2G5, WGA and Hoechst staining showed cobblestone-like endothelial morphology and strong cell–cell contacts. In contrast, formulations with higher or lower GelMA concentrations showed reduced confluency and irregular shapes, suggesting that H2G5 supports a local optimum in cellular organization without attributing changes solely to matrix stiffness.
Together, these results validate the H2G5 hydrogel as a mechanically tunable, biologically responsive platform for engineering human 3D neurovascular models. The integration of optimized immunostaining and spinning disc confocal microscopy allowed insight into the spatial architecture and cell–matrix interactions within the construct. By capturing both form and function, this system supports the development of physiologically relevant in vitro models for blood-brain barrier research, CNS drug screening, and regenerative neuroengineering—bridging a critical gap between engineered biomaterials and translational neuroscience.

Acknowledgements and/or References (Optional):: Funding is gratefully acknowledged from the Institute of General Medical Sciences (NIGMS) of the National Institutes of Health under award number 1R35GM150970 and an Institutional Development Award (IDeA) from the National Institute of General Medical Sciences, the National Institutes of Health, under grant number P20 GM135009.