Modelling glioblastoma microenvironment for in vitro assessment of cell-based drug delivery

Introduction: Glioblastoma (GBM) is the most hard-to-treat brain tumor due to the heterogeneity of its microenvironment (TME) and to the presence of the Blood Brain Barrier (BBB). Hence, more efficient drug delivery strategies are needed to improve the outcome of GBM patients. For instance, cell-mediated transport of nanoparticles (NPs) has the potential to improve GBM treatment, by exploiting the tumor homing properties of living cells (e.g., microglia). This study aims to design a reliable in vitro GBM model as an alternative tool for the investigation of cell-mediated drug delivery. Methods: A three-dimensional GBM model was realized by encapsulating multicellular spheroids (MS) in different commercial gels. MSs comprised GBM cells and GBM-associated Stem Cells to replicate tumor histology, while the gels reproduced the mechanical properties of GBM extracellular matrix (ECM). MS vascularization was induced through a commercial microfluidic platform, using brain endothelial cells. Network formation was confirmed by immunostaining and perfusion assay. Microglia (μG) were incubated with different concentration of Bortezomib-loaded polyurethane NPs (0.5, 1, 2 mg/mL) for increasing time (2, 24 h). Loading of NPs in μG was optimized by determining cell viability (CellTiter MTS) and internalization by flow cytometry. Treatment efficacy after NPs and μG (incubated for 2 h with 1 mg/mL NPs) administration was assess through viability assay (CellTiter-Glo) on MS. The infiltration in ECM- like gels and extravasation across vessels were tested on the model. Results: MS were effectively encapsulated in gels, which preserved integrity and facilitated invasion. The microfluidic platform successfully induced vascularization, as confirmed by immunostaining for tight junction proteins (ZO-1). NPs internalization in μG increased with concentration and over time, with NPs concentrations above 1 mg/mL resulting in more than 90% internalizatio without significant cytotoxicity (90% viability). BTZ-NPs-μG were able to infiltrate in different gels, reaching the embedded MS, and successfully reducing MS proliferation and invasion. The microvasculature was able to replicate the barrier effect against NPs, while microglia cells penetrated across the vessel towards the MS. Conclusions: This model represents a prototype for a reliable replica of the heterogeneity of GBM microenvironment, which combines cells, biomaterials, and microfluidics. The model provided valuable confirmation of the potential of microglia-mediated drug delivery as a promising strategy for GBM homing. Future studies should introduce further BBB elements into the model and compare the system with in vivo observations.