Ultrasound-Controlled Blood-Brain Barrier (BBB) Opening in a BBB-Glioblastoma Microfluidic Chip for Drug Delivery Studies

Summary

This protocol describes the construction of a microfluidic BBB‑glioblastoma (BBB‑GBM) chip model. The model is validated for BBB integrity and enables investigation of low‑intensity ultrasound‑induced transient and largely reversible BBB opening, thereby allowing quantitative evaluation of ultrasound‑mediated nanomicelle transport across the BBB and tumor-targeting efficiency.

Abstract

The blood-brain barrier (BBB) is a major obstacle to treating glioblastoma (GBM) by restricting the entry of most therapeutic molecules into the brain. Interspecies differences in BBB structure and function complicate the clinical translation of animal models, highlighting the need for human-relevant in vitro BBB platforms for assessing drug permeability in GBM therapy. Here, a protocol is described for constructing a microfluidic BBB-GBM chip model. This model establishes a tri-culture system comprising human cerebral microvascular endothelial cells (HCMECs), primary astrocytes (ACs), and U87-MG cells within a Matrigel-embedded microfluidic platform. BBB integrity is verified through continuous zonula occludens-1 (ZO-1) immunostaining and FITC-dextran permeability assay. Low-intensity ultrasound (US) (1 MHz, 1 W/cm², 30 s) is then employed to induce transient and largely reversible BBB opening, allowing tumor-targeting nanomicelles (SFN@RB@SPMs) to efficiently traverse the barrier and accumulate in GBM cells. A key feature of this protocol is the integration of real-time BBB permeability and drug delivery measurements within a single ultrasound-responsive chip, providing a powerful platform to dissect the mechanisms of ultrasound-augmented drug delivery in GBM.

Introduction

The treatment of glioblastoma (GBM) is strongly limited by the BBB, which severely limits the entry of nearly all therapeutic agents into the brain^1,2. The BBB is formed by non-fenestrated endothelial cells joined by tight junction proteins (e.g., occludin, ZO-1), with additional support from pericytes and astrocytic end-feet³. Although animal models have provided valuable insights into BBB function, species-specific variations in physiology and metabolism limit their ability to accurately predict human drug responses^4,5. Consequently, physiologically relevant human BBB models are indispensable for evaluating drug permeability in GBM therapy.

Traditional in vitro BBB models, such as the Transwell culture system, fail to recapitulate the complexity of the BBB, including cell-cell interaction, dynamic fluidic change, and flow-induced shear stress⁶. In contrast, microfluidic BBB models have emerged as superior platforms that closely mimic these in vivo-like characteristics⁷. Such BBB-on-a-chip systems have been widely used to study brain tumor metastasis⁸, investigate BBB dysfunction in neurodegenerative diseases such as Alzheimer’s disease⁹, and evaluate drug transport across the BBB^10,11. However, most existing BBB-chip models are mainly used for drug permeability testing and lack the capability to incorporate external physical stimuli that modulate BBB permeability. Currently, focused ultrasound (FUS) has emerged as a promising strategy to transiently open the BBB with substantial recovery, enabling enhanced delivery of therapeutic agents to the brain^12,13. Previous mechanistic studies have demonstrated that ultrasound-induced mechanical stimulation can transiently modulate barrier function by remodeling cell-cell junctions and enhancing endothelial permeability. For example, Silvani et al. showed that ultrasound-induced mechanical forces generated reversible interendothelial gap formation in in vitro vessel-on-a-chip¹⁴. Consistent with these findings, Beekers et al. demonstrated that ultrasound-mediated mechanical bioeffects induced a transient opening of endothelial cell-cell contacts, thereby increasing endothelial permeability¹⁵. More recently, Qiao et al. reported that endothelial barrier permeability is critically regulated by the ultrasound exposure parameters, particularly pulse length, emphasizing the importance of optimizing acoustic parameters to achieve safe and controllable BBB modulation¹⁶. Despite this progress, in vitro platforms that simultaneously allow quantitative analysis of ultrasound-mediated BBB opening and its impact on nanomedicine delivery remain limited.

To address the above gap, a protocol is presented for constructing a microfluidic BBB-GBM chip. The microfluidic device supports a tri-culture of HCMECs, primary ACs, and U87-MG GBM cells embedded in Matrigel. The reconstructed BBB exhibited both structural and functional characteristics of an intact barrier, demonstrated by continuous ZO-1 localization along intercellular junctions and limited diffusion of 40 kDa FITC-dextran across the vascular interface. Importantly, this chip can be coupled with controlled ultrasound stimulation to induce transient BBB opening, thereby enhancing the subsequent transport and tumor-targeting efficiency of nanomicelles. Using this system, the transport and tumor-targeting efficiency of nanomicelles SFN@RB@SPM can be quantitatively assessed. Overall, this platform provides a useful tool for studying the relationship between ultrasound parameters, BBB modulation with reversible characteristics, and enhanced nanomicelles delivery in GBM therapy.

Protocol

All animal procedures adhered to the Guidelines (GB/T35892-2018) for Ethical Review of Experimental Animal Welfare and were approved by the Institutional Animal Care and Use Committee (IACUC) of Ningbo Institute of Life and Health Industry, UCAS (Approval No. GK-2023-XM-0073). The reagents and the equipment used are listed in the Table of Materials.

1. Fabrication and preparation of the BBB-GBM organ-on-a-chip device

1. Chip design and photomask layout
   1. Design the chip using Auto CAD software.
      ​NOTE: The chip consists of three parallel microfluidic channels, including a blood channel, a brain channel, and a tumor channel. The widths of the blood channel and tumor channel are both 1000 µm, while the width of the brain channel is 1300 µm. The height of the three channels is 150 µm, and adjacent channels are separated by trapezoid micropillar arrays with a width of 200 µm, which allow intercompartmental molecular diffusion and cellular communication. The detailed photomask design layout is provided in Supplementary Figure 1 and Supplementary Figure 2. The SU-8 photoresist used in the present study is 3050.
2. Fabricate a silicon master mold using standard photolithography with SU-8 photoresist.
   1. Place a blank silicon wafer on the spin coater vacuum chuck and activate the vacuum to securely immobilize the wafer.
   2. Slowly dispense 10 mL of SU-8 photoresist onto the center of the wafer.
   3. Perform a two-step spin-coating process to obtain a uniform photoresist layer: first at 500 rpm for 10 s, then at 2500 rpm for 90 s.
   4. Transfer the coated wafer to a hot plate and perform a soft bake at 95 °C for 30 min to evaporate the solvent.
   5. Allow the wafer to cool to room temperature and measure the thickness of the SU-8 layer using a thickness gauge.
      NOTE: For optimal BBB formation, a channel height of 110–150 µm is recommended. A single spin-coat (step 1.2.3) typically yields a film thickness of 40–50 µm; therefore, two to three spin-coating processes are required to reach the desired total thickness.
3. Photolithographic patterning
   1. Set the ultraviolet (UV) light intensity of the mask aligner to 14–15 mW/cm².
   2. Align the chromium photomask with the SU-8-coated silicon wafer and perform UV exposure for 10 s.
   3. Immediately transfer the silicon wafer to a hot plate for post-exposure baking under the following conditions: 65 °C for 10 s, followed by 95 °C for 300 s.
   4. After the silicon wafer has cooled to room temperature, immerse it in developer solution (propylene glycol monomethyl ether acetate, PGMEA) and allow it to develop for 4–6 min.
   5. Rinse the wafer with isopropanol to terminate development, followed by washing with deionized water.
   6. Dry the wafer using nitrogen gas and inspect the patterned microchannel structures under an optical microscope.
      ​NOTE: If undeveloped photoresist residues remain, repeat the developing step until the microstructures are clearly defined.
   7. Bake the completed silicon master in an oven at 200 °C for 1 h to improve adhesion between the SU-8 photoresist and the silicon wafer surface.
4. PDMS chip fabrication
   1. Prepare polydimethylsiloxane (PDMS) by thoroughly mixing the base prepolymer and curing agent at a 10:1 weight ratio (w/w) for 30 min to obtain a homogeneous PDMS prepolymer.
   2. Pour the PDMS mixture onto the fabricated silicon master mold, ensuring that the microstructures are fully covered.
   3. Place the master mold in a vacuum desiccator for degassing to remove air bubbles.
   4. Transfer the master mold to an oven and cure it at 65 °C for 1.5–2 h.
   5. After curing, carefully cut the PDMS layer along the chip boundaries using a craft knife and gently peel it from the silicon master mold.
   6. Immerse the PDMS chip in 70%–75% ethanol for 30 min for preliminary sterilization, then rinse with deionized water and allow to air-dry.
   7. Punch inlet and outlet ports using a 2 mm biopsy punch.
5. Plasma bonding and device assembling
   1. Place the PDMS chip and a clean glass slide into the chamber of a plasma cleaner, ensuring that the microchannel side of the PDMS is facing upward.
   2. Evacuate the chamber until the pressure drops below 30 Pa, then introduce oxygen gas.
   3. Maintain the oxygen flow for 90 s while keeping the chamber pressure below 40 Pa.
   4. Set the Radio Frequency (RF) power to 55% and perform plasma treatment for 90 s to activate the PDMS and glass surfaces.
   5. Immediately bring the activated PDMS chip into contact with the glass slide and gently press to achieve irreversible bonding.
   6. Place the assembled device in an oven at 65 °C for 24 h to stabilize the bonding and restore its hydrophobicity.
   7. Prior to cell seeding, sterilize the bonded device by ultraviolet irradiation for 30 min.
      NOTE: Maintain sterile conditions by storing the sterilized chips in sealed Petri dishes inside a biosafety cabinet until use.

2. Cell seeding and establishment of the BBB-GBM model

1. Prepare the following reagents:
   1. ACs culture medium (for the brain channel): DMEM/F12 basal medium containing 20% fetal bovine serum (FBS) and 1% penicillin/streptomycin (P/S).
   2. HCMECs culture medium (for the vascular channel): Endothelial cell medium (ECM) supplemented with 1% P/S.
   3. GBM cell culture medium (for the glioma channel): MEM basal medium containing 10% FBS and 1% P/S.
   4. Cell dissociation reagent: Use 0.25% trypsin-EDTA solution for detaching adherent cells.
      NOTE: Prolonged exposure to trypsin may reduce cell viability. Monitor cell detachment under a microscope and neutralize trypsin promptly with complete medium.
   5. Maintain all cells (ACs, HCMECs, and U87-MG) in 100 mm-culture dishes prior to seeding into the microfluidic device.
2. Primary ACs isolation
   1. Isolate primary ACs from cerebral cortices of neonatal rats (postnatal day 1–3). After carefully removing the meninges and blood vessels on ice, mince the cortical tissue with sterile surgical scissors and subject it to enzymatic digestion.
   2. Wash the dissociated cell suspension with ice-cold PBS, and treat with trypsin to achieve single-cell dissociation, and collect the pellets by centrifugation (400 x g, 5 min, room temperature).
   3. Resuspend the cells and seed them into T25 culture flasks containing ACs culture medium.
   4. Identify the ACs by GFAP through immunofluorescence staining. The image is provided in Supplementary Figure 3.
3. Cell seeding
   ​NOTE: Day 1 - Seeding of ACs in the brain channel.
   1. Introduce 1 mL of 0.25% trypsin-EDTA into the ACs-containing culture dish. Gently swirl to ensure even coverage.
   2. Incubate at 37 °C for 1–2 min until the cells detach from the culture dish.
   3. Add 3 mL ACs culture medium (from step 2.1.1) to neutralize trypsin. Pipette gently to obtain a single-cell suspension.
   4. Collect the cell suspension into a centrifuge tube and centrifuge at 300 × g for 3 min at 4 °C.
   5. Remove the supernatant and resuspend the cell pellet in 200 µL of AC medium.
   6. Determine the cell concentration using an automated cell counter and adjust the density to 2.5 × 10⁶ cells/mL.
   7. Mix the ACs cell suspension with Matrigel at a 3:5 (v/v) ratio (AC cells: Matrigel).
   8. Using a micropipette, dispense 10 µL of the Matrigel-ACs mixture into the brain channel.
      NOTE: Perform the loading procedure with the chip maintained on ice to prevent premature Matrigel polymerization.
   9. Incubate the chip at 37 °C in a 5% CO₂ incubator for 30 min to allow Matrigel polymerization and cell stabilization.
      NOTE: Day 1 - Seeding of HCMECs in the vascular channel.
   10. Apply 1 mL of 0.25% trypsin-EDTA and incubate at 37 °C for approximately 30 s until cells detach.
   11. Add 3 mL of HCMECs medium (step 2.1.2) to stop the enzymatic digestion.
   12. Pipette gently to generate a uniform single-cell suspension.
   13. Collect the suspension and centrifuge at 300 × g for 3 min at 4 °C.
   14. Remove the supernatant and resuspend the cell pellet in 200 µL of HCMECs medium.
   15. Count the cells and adjust the concentration to 2 × 10⁶ cells/mL.
   16. Carefully introduce 10 µL of the HCMECs suspension into the vascular channel of the chip.
   17. Position the chip sideways for 30 min in the cell incubator to promote endothelial cell attachment along the channel wall.
   18. Seed an additional 10 µL of HCMECs suspension into the vascular channel to increase endothelial density in the vascular channel.
   19. Incubate the chip for an additional 30–60 min to allow HCMECs adhesion.
   20. Add 100 µL mixed culture medium to the chip inlet.
       NOTE: The mixed medium consists of AC medium and HCMECs medium at a 1:1 ratio (v/v).
       Day 3 - Seeding of U87-MG cells in the glioma channel.
   21. Add 1 mL of 0.25% trypsin-EDTA to the culture dish and incubate at 37 °C for 1–2 min.
   22. Add 3 mL of U87-MG medium (from step 2.1.3) to terminate digestion.
   23. Pipette gently to obtain a homogeneous single-cell suspension.
   24. Transfer the cell suspension to a centrifuge tube and centrifuge at 300 × g for 3 min.
   25. Discard the supernatant and resuspend the cell pellet in 200 µL of U87-MG medium.
   26. Count the cell concentration and adjust the density to 2 × 10⁶ cells/mL.
   27. Introduce 10 µL of the U87-MG cell suspension into the glioma channel of the chip.
   28. Incubate the chip in a 37 °C with 5% CO₂ for 30–60 min to allow cell attachment.
   29. Add 100 µL of U87-MG culture medium to the chip inlet.
   30. Maintain the microfluidic device in a 37 °C environment with 5% CO₂ for an additional 48 h to allow endothelial tight junction formation.

3. Evaluation of the BBB integrity

1. Permeability assay
   1. Prepare fluorescein isothiocynate-dextran (FITC-dextran) with molecular weights of 40 kDa in mixed culture medium (ACs medium: HCMEC medium = 1:1, v/v) at a final concentration of 10 µg/mL.
   2. Introduce 10 µL of the FITC-Dextran solution into the vascular channel.
   3. Place the microfluidic chip on the stage of an inverted fluorescence microscope to acquire fluorescence images of both the vascular channel and the adjacent brain channel at 20 min intervals to monitor dextran diffusion across the endothelial barrier.
   4. Quantify the permeability coefficient (P_app, cm/s) using the following formula:
      P_app (cm/s)= 
      Where P_app is the apparent permeability coefficient (cm/s), F_S is the fluorescence intensity at the initial time in the blood channel, F_C is the fluorescence intensity in the brain channel at the end of the diffusion interval, Δt is the diffusion time (s), VT is the diffusion volume of FITC-dextran in the brain channel (cm³), and A is the endothelial surface area (cm²).
      NOTE: A lower permeability coefficient indicates higher BBB integrity and tighter endothelial barrier formation.
   5. Perform data analysis and parameter definitions.
      NOTE: Fluorescence intensity values from both blood and brain channels were background-subtracted using measurements obtained from cell-free regions of the chip. ROIs were defined on both sides of the blood-brain interface based on the fluorescence diffusion length. The endothelial surface area (A) was calculated from the interface area between the vascular and brain channels (using a gap between two micropillars as a reference): 200 µm × 150 µm = 30,000 µm² = 3 × 10^-6 cm². The FITC-dextran diffusion volume (VT) was derived from the diffusion penetration distance in the brain channel multiplied by A. Fluorescence images were acquired every 20 min (1200 s) over a 60-min period; the permeability coefficient was calculated using the 20-min interval data. Permeability values were calculated from three independent chips for each experimental group. All unit conversions were performed before expressing permeability as cm/s.
2. Immunofluorescence analysis of ZO-1
   1. Wash the microfluidic channels three times with phosphate-buffered saline (PBS) to remove remaining culture medium.
   2. Fix the cells by introducing 4% paraformaldehyde (PFA) into the channels and incubating for 10 min at room temperature.
   3. Permeabilize the cells with 0.1% Triton X-100 in PBS for 15 min at room temperature.
   4. Block nonspecific antibody binding by incubating the channels with 2% bovine serum albumin (BSA) in PBS for 45 min at room temperature.
   5. Introduce the anti-ZO-1 primary antibody (1:200 dilution) into the channels and incubate at 4 °C overnight.
   6. Wash the channel thoroughly with PBS to remove unbound primary antibody.
   7. Incubate the chips with an Alexa Fluor 555-conjugated anti-rabbit secondary antibody (1:1000 dilution) for 45 min at room temperature in the dark.
   8. Counterstain the nuclei with 1 µg/mL 4’6-diamidino-2-phenylindole (DAPI) for 1 min at room temperature in the dark.
      NOTE: DAPI is light sensitive and should be protected from light during staining.
   9. Wash the channels three times with PBS.
   10. Image the microfluidic device using an inverted fluorescence microscope and a confocal laser scanning microscope to visualize ZO-1 localization along the endothelial cell junction.

4. Ultrasound-enhanced nanomicelles delivery across the BBB to GBM cells

1. Ultrasound stimulation
   1. Carefully take the chip out of the culture dish and place it inside a biosafety cabinet.
   2. Apply ultrasound stimulation using a therapeutic ultrasound system equipped with a 1 MHz transducer (effective radiating area: 5 cm²).
   3. Apply a 3 mm thick medical-grade acoustic coupling pad between the ultrasound probe and the surface of the microfluidic chip to ensure efficient acoustic transmission.
   4. Position the ultrasound probe directly above the vascular channel region of the microfluidic device.
      ​NOTE: Air bubbles between the transducer and the chip significantly reduce acoustic energy transmission and should be avoided.
   5. Set the ultrasound parameters as follows: duty cycle: 60%; frequency: 1 MHz; intensity: 1 W/cm²; exposure duration: 30 s; pulse repetition frequency (PRF): 100 Hz. The detailed ultrasound parameters are provided in Supplementary Table 1.
   6. For the untreated control group, place the ultrasound transducer in the same position but ensure to keep inactive to serve as a sham control, ensuring identical handling procedures without acoustic exposure.
   7. After ultrasound exposure, remove the coupling pad and immediately return the chip to a sterile culture dish.
   8. Add fresh culture medium to the inlet ports of the chip.
   9. Incubate the device at 37 °C with 5% CO₂ until further analysis.
2. Evaluation of ultrasound-induced BBB modulation
   1. Perform ZO-1 immunofluorescence staining immediately after ultrasound treatment (0 h) and at 24 h post-treatment to assess changes in tight junction integrity.
   2. Follow the immunofluorescence staining procedure described in step 3.2.
   3. Compare the ZO-1 staining distribution along the z-axis to evaluate the effect of ultrasound stimulation on BBB tight junction integrity.
3. Preparation and characterization of SFN@RB@SPM nanomicelles
   1. Prepare the SFN@RB@SPM nanomicelles according to the previously established protocol¹⁷. Briefly, stock solutions of the amphiphilic peptide (designated as SPM, 6 mg/mL), RB (0.6 mg/mL), and SFN (6 mM) were mixed at a volume ratio of 1:1:1.
   2. Place the mixture in a sonicator and sonicate at 28 kHz, 25 °C for 30 min.
   3. Dialyze the mixture (molecular weight cut-off: 1000 Da) for 48 h to remove free RB and SFN.
   4. Measure the size and zeta potential of the nanomicelles by dynamic light scattering (DLS). The characterization data are provided in Supplementary Figure 4.
4. Evaluation of ultrasound-stimulated nanomicelles delivery and glioma cell accumulation
   1. Immediately following ultrasound exposure, introduce the nanomicelles-containing culture medium into the vascular channel of the microfluidic device.
   2. To evaluate nanomicelles transport across the BBB, place the microfluidic chip on the stage of an inverted fluorescence microscope and acquire fluorescence images of both the vascular channel and the adjacent brain channel at 20 min intervals over 60 min.
   3. Quantify the fluorescence images in the brain channel to evaluate ultrasound-enhanced nanomicelles permeability.
   4. To assess nanomicelles accumulation in glioma cells, incubate the device under standard culture conditions (37 °C, 5% CO₂) for 3 h after ultrasound exposure, then image the glioma channel using a fluorescence microscope to evaluate intracellular nanomicelles uptake.
5. Statistical analysis
   1. Express all quantitative data as mean±standard error of the mean (SEM).
   2. Perform each experiment with three independent biological replicates (n = 3 chips per experimental condition). For each biological replicate, at least three technical replicates (ROI per channel) were measured to ensure data reliability.
   3. Calculate the permeability coefficient (P_app, cm/s) from each independent chip served as the primary unit of analysis for all statistical comparisons.
      NOTE: Comparisons between two groups were performed using an unpaired two-tailed Student’s t-test. Comparisons among three or more groups were analyzed using one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test for multiple comparisons. For experiments involving two independent variables, two-way ANOVA followed by Tukey’s multiple comparisons test was performed. Statistical analyses were conducted using statistical and graphing software. A p-value of <0.05 was considered statistically significant.

Results

To recapitulate the in vivo structure of the BBB, a tri-culture microfluidic chip was developed by incorporating HCMECs, primary ACs, and U87-MG cells (termed the “chip model”). As shown in Figure 1A,B, three parallel channels are arranged within the device: the vascular (left, blue), the Matrigel-filled brain (central, pink), and the tumor (right, green) channels, which are seeded with HCMECs, ACs, and U87-MG cells, respectively. The channels are separated by micropillar arrays that maintain matrigel confinement in the central compartment while permitting inter-channel molecular and cellular exchange. After device sterilization, ACs (2.5 × 10⁶ cells/mL) suspended in Matrigel at a 3:5 (v/v) ratio were delivered into the brain channel. Upon Matrigel polymerization, HCMECs (2 × 10⁶ cells/mL) were seeded into the left vascular channel. Following a 2-day co-culture of ACs and HCMECs, U87-MG cells (2 × 10⁶ cells/mL) were introduced into the right tumor channel to establish the GBM compartment. The chip model was fully established by day 5. Barrier integrity was then assessed via ZO-1 immunostaining and 40 kDa FITC-dextran permeability measurements. As illustrated in Figure 1C, ZO-1 displayed a continuous, linear distribution along the HCMEC-ACs interface, confirming a tight junction formed at the BBB interface. Confocal 3-D reconstruction further revealed prominent ZO-1 signals at the vascular-brain channel interface (z-y plane, Figure 1D), confirming the formation of an intact endothelial barrier within the chip. To further determine the function of the barrier, the permeability for 40 kDa FITC-dextran was evaluated. Permeability was quantified at 20-min intervals over 1 h. As shown in Figure 1E,F, the permeability coefficient is (3.46 ± 0.50) × 10^-6 cm/s at 60 min, which is comparable to in vivo rat microcirculation¹⁸ and advanced induced pluripotent stem cell (iPSC)-based models¹⁹, and it represents a significant improvement over conventional endothelial-only models¹³. Together, these structural and functional data validated the successful establishment of a physiologically relevant BBB-GBM platform.

As ultrasound has been reported to induce transient and largely reversible BBB opening, with barrier function typically restored within 24 h²⁰. The direct impact of US stimulation on the BBB function was next investigated on the chip model at 0 h or 24 h post-US exposure. The chip model was exposed to US (1 W/cm², 1MHz, 30 s). Immunofluorescence analysis of ZO-1 showed comparable intensity at 0 h and 24 h after US treatment (Figure 2A), suggesting maintenance of tight junction integrity. Quantitative ZO-1 fluorescence analysis confirmed comparable signal intensity at the vascular-brain interface between the control and US-treated group (Figure 2B). To further assess whether US exposure enhanced BBB permeability, 40 kDa FITC-dextran was used as a permeability tracer. As shown in Figure 2C,D, US exposure promptly accelerated tracer diffusion across the BBB, with the permeability coefficient increased from 4.20 x 10^-6cm/s in the untreated group to 1.33 x 10^-5cm/s. At 24 h post-US exposure, the permeability coefficient decreased to 8.04 x 10^-6cm/s, representing substantial recovery compared to the immediate post-US group, although it remained slightly higher than the untreated control level. These results demonstrate that low-intensity US induces transient BBB opening followed by substantial recovery of barrier function while preserving the structural integrity in the chip model.

The permeability of the nanomicelles SFN@RB@SPM across the BBB under US stimulation was next evaluated. In our previous study, these self-assembled nanomicelles were shown to effectively cross the BBB and accumulate at the tumor site in a xenografted glioma mouse model¹⁷. To examine whether US modulation enhances SFN@RB@SPM transport in this chip model, we applied the same sonication parameters as in Figure 2C. US exposure markedly accelerated SFN@RB@SPM across the BBB, elevating the permeability coefficient from (2.42 ± 0.15) × 10^-5 cm/s to (4.58 ± 0.07) × 10^-5 cm/s at 60 min post-treatment, representing a 1.9-fold enhancement over the control group (Figure 3A,B). To verify active tumor targeting, the red fluorescence signal of RB in the glioma channel was measured following nanomicelles injection into the vascular channel. As shown in Figure 3C, a marked increase in red fluorescence was observed in the glioma channel 3 h post US exposure, indicating enhanced nanomicelles transport across the BBB. In the absence of the US, only negligible fluorescence was detected, confirming that BBB disruption is essential for SFN@RB@SPM to reach and enrich within the tumor site.

Figure 1: BBB-GBM on-a-chip fabrication and characterization. (A) Schematic of the three-channel microfluidic device: vascular channel (blue), brain channel (pink), and tumor channel (green). (B) Photograph of the assembled PDMS-glass chip. (C) ZO-1 immunostaining (red) at endothelial junctions, with DAPI nuclear counterstain (blue). Scale bar: 50 µm. (D) Orthogonal confocal views (x-z and y-z) showing ZO-1 enrichment (red) at the vascular-brain interface. (E) Representative time-lapse images of 40 kDa FITC-dextran diffusion from the vascular to the brain channel over 60 min (20-min intervals). Scale bar: 200 µm. (F) Quantified permeability coefficients, confirming size-selective barrier function. Data represent mean ± SEM from three independent chips (n = 3). Statistical significance was determined using one-way analysis of variance (ANOVA), followed by Tukey’s post hoc test for multiple comparisons. The exact p-value is shown in the figure. Please click here to view a larger version of this figure.

Figure 2: BBB opening and subsequent recovery of barrier function induced by ultrasound. (A) ZO-1 immunostaining at the vascular-brain interface immediately (0 h) and 24 h post-sonication. 3-D confocal reconstructions are shown. (B) Normalized ZO-1 fluorescence intensity along the z-axis (depth) at the vascular-brain interface in the control and US-treated group. (C) Diffusion kinetics of 40 kDa FITC-dextran from the vascular to the brain compartments at 0 h and 24 h, with or without US exposure (image acquired at 20-min intervals over 60 min). Scale bar: 200 µm. (D) Quantified permeability coefficient for 40 kDa tracer at 0 h or 24 h post-US, with or without sonication. Data represent mean ± SEM from three independent chips (n = 3). Statistical significance was determined using two-way analysis of variance (ANOVA), followed by Tukey’s post hoc test for multiple comparisons. Statistical significance: p < 0.0001. Please click here to view a larger version of this figure.

Figure 3: US-enhanced SFN@RB@SPM permeability and targeting to U87-MG cells. (A) Permeability measurement of SFN@RB@SPM across the vascular-brain interface, with or without US exposure. Scale bar: 200 µm. (B) Quantified permeability coefficients of SFN@RB@SPM under control and US-treated conditions. (C) Accumulation of SFN@RB@SPM in the tumor channel following US-mediated BBB opening, red fluorescence was recorded at 3 h post-injection. Scale bar: 50 µm. Data represent mean ± SEM from three independent chips (n = 3). Statistical significance was determined using an unpaired two-tailed Student’s t-test. Statistical significance: p < 0.0001. Please click here to view a larger version of this figure.

Supplementary Figure 1: CAD design layout of BBB-GBM microfluidic chip.Please click here to download this file.

Supplementary Figure 2: The detailed design of the BBB-GBM chip. The design consists of three parallel microfluidic channels; the width of both the side channels is 1000 µm, while the width of the central channel is 1300 µm.Please click here to download this file.

Supplementary Figure 3: Characterization of the primary astrocytes by GFAP (green), and the staining of the nuclei with DAPI (blue). Scale bar: 50 µm.Please click here to download this file.

Supplementary Figure 4: Characterization of the self-assembled nanomicelles SFN@RB@SPM. (A) Dynamic light scattering (DLS) analysis showing the size distribution of SFN@RB@SPM, with a mean particle diameter of approximately 64 nm. (B) Zeta potential measurement indicates a surface charge of approximately 150.67 ± 2.85 mV. All measurements were performed using DLS instruments.Please click here to download this file.

Supplementary Table 1: Summary table of Ultrasound parameters.Please click here to download this file.

Discussion

This protocol describes the establishment of a microfluidic BBB-GBM chip for evaluating ultrasound-mediated BBB modulation and nanomicelle delivery. Compared with conventional in vitro BBB models (e.g., Transwell), this microfluidic platform allows spatially defined co-culture of endothelial cells, ACs, and glioma cells, thereby enabling investigation of drug transport across a physiologically relevant BBB. Several critical steps in the protocol require careful optimization to ensure reliable barrier formation. One key parameter is the volume ratio between Matrigel and the AC cell suspension medium used to prepare the mixture loaded into the central brain channel. The ratio is crucial for maintaining appropriate gel viscosity so that the AC-Matrigel mixture remains within the middle channel during loading. If the mixture contains an excessive portion of AC cell suspension medium, the reduced viscosity may cause the Matrigel to leak into adjacent microchannels, allowing AC cells to migrate into neighboring channels. Conversely, an overly concentrated Matrigel may impair cell viability and hinder nutrient diffusion. The optimized 3:5 (v/v) ratio of AC cell suspension and Matrigel (step 2.2.7) provides sufficient mechanical stability to retain the gel within the central channel while supporting AC growth within Matrigel. Similar considerations have been highlighted in other microfluidic BBB models, where proper hydrogel viscosity (e.g., fibrin or Matrigel) is critical for preventing cross-channel leakage, maintaining spatial separation of cell compartments, and mimicking extracellular matrix properties^7,21,22.

Second, immediately after seeding HCMECs into the vascular channel, the microfluidic device must be positioned sideways for approximately 30 min. This step promotes the attachment of HCMECs along the sidewall between the vascular and brain channels, which is necessary to form a continuous endothelial barrier at the cell-cell interface. Without this step, HCMECs primarily settle on the bottom surface of the channel, resulting in incomplete endothelial coverage along the sidewall and reduced barrier integrity. The subsequent additional HCMECs seeding step further ensures sufficient cell density within the vascular channel. Maintaining an adequate cell density is important because low HCMECs density can lead to incomplete monolayer formation and increased apoptosis. Previous BBB-on-a-chip studies have similarly emphasized that achieving a confluent endothelial monolayer is critical for barrier integrity and tight junction protein expression in microfluidic systems^23,24.

Troubleshooting of common issues can substantially improve reproducibility. If Matrigel leakage into adjacent channels occurs during brain channel loading (step 2.2.8): (1) ensure that the Matrigel-cell mixture is kept on ice to prevent Matrigel polymerization; (2) Inject the cell-Matrigel mixture slowly and steadily to avoid overflow into the neighboring channels; (3) adjust the Matrigel-cell ratio to 5:4 (v/v), which helps maintain appropriate viscosity and confinement within the central channel. In addition, the successful formation of the endothelial barrier depends strongly on adequate endothelial cell attachment and density. If the endothelial cells fail to form a continuous monolayer along the vascular channel interface, consider increasing the endothelial cell seeding density or extending the adhesion time to improve endothelial coverage and tight junction formation. If air bubbles appear in the channel after medium addition, gently tap the chip or remove them using a yellow pipette tip with gentle suction, as bubbles trapped near the endothelial barrier may locally disrupt tight junction integrity.

Although the BBB-GBM on-a-chip described here provides a more physiologically relevant model, several aspects can be further optimized in future studies to improve physiological relevance and mechanistic understanding. The first one is the current model using rat-derived primary ACs co-cultured with HCMECs, resulting in a species mismatch. While this approach has been widely used for the construction of a BBB-on-a-chip platform^13,23, the species differences may affect barrier properties and drug permeability responses. Future studies will incorporate human iPSC-derived ACs or primary human ACs to establish a fully humanized model, which would eliminate this limitation and further enhance the translation relevance of the platform. Second, the current model lacks neurovascular unit components, particularly pericytes and immune cells, which are important regulators of BBB function and tumor progression^25,26. Incorporating these components would better recapitulate the complexity of the tumor microenvironment. Third, the device operates under static conditions without continuous medium perfusion. Physiological shear stress generated by blood flow has been shown to enhance BBB properties by upregulating tight junction proteins such as claudin-5 and occludin expression. Therefore, the absence of flow may cause a higher permeability coefficient observed in static chips compared with dynamic systems²⁷. Fourth, BBB integrity was primarily evaluated by combining ZO-1 immunofluorescence staining with FITC-dextran permeability analysis. However, these complementary structural and functional assays are widely used for BBB validation in both conventional transwell BBB models and microfluidic BBB-on-a-chip models^6,28,29. Additional characterization of key regulators governing paracellular and transcellular transport, including claudin-530, MFSD2A, and caveolin-1, would provide a more comprehensive assessment of BBB function^30,31,32. These markers will be incorporated in future studies to further strengthen the physiological characterization of the platform. Finally, the present protocol uses low-intensity therapeutic ultrasound without exogenous microbubbles or direct measurement of acoustic pressure. Unlike conventional focused ultrasound systems, which typically employ microbubble-assisted BBB opening in vivo, the current study was designed to establish a simplified, reproducible in vitro platform to evaluate ultrasound-induced modulation of BBB permeability and nanomedicine transport. This approach minimizes additional variables associated with microbubble concentration, distribution, and cavitation behavior in a microscale microfluidic environment. Nevertheless, future studies incorporating microbubble-assisted ultrasound, temperature monitoring, and detailed characterization of the acoustic field will further enhance the physiological relevance and quantitative understanding of ultrasound-mediated BBB modulation.

Overall, the protocol presented here provides a controlled BBB-GBM microfluidic platform capable of integrating ultrasound stimulation and quantitative analysis of nanomicelle delivery. This system enables direct investigation of ultrasound-mediated BBB opening and the subsequent transport of nanomedicines across the endothelial barrier. Beyond evaluating nanomicelle delivery, the platform may also be used to screen drugs for BBB-penetrating ability, study mechanisms of BBB disruption, and provide a powerful platform for preclinical evaluation of central nervous system (CNS)-directed therapeutics.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Alexa Fluor 555-conjugated anti-rabbit secondary antibody	Cell Signaling Technology, Inc.	4413
anti-ZO-1 antibody	Thermo Fisher Scientific Co., Ltd.	40-2200
Biopsy punch	Zhejiang YoungChip Technology Co., Ltd.	10305130004
Bovine serum albumin	Absin Bioscience Inc.	9048-46-8
Cell counter	Thermo Fisher Scientific Co., Ltd.	Countess TM3
Cell culture dish (100 mm)	Corning Inc.	430167
Centrifuge	Beckman Coulter, Inc.	Avanti J-15R
Chrome mask	Suzhou Chip Scientific Instrument Co.,Ltd.	ZX-GBYM
Confocal Laser Scanning Microscope	Carl Zeiss AG	Zeiss LSM900
DAPI	Beyotime Biotech Inc.	C1002
Developer	Suzhou Chip Scientific Instrument Co., Ltd.	RCD-400
DMEM/F12	ScienCell Research Laboratories	C3130-0500
ECM	ScienCell Research Laboratories	1001
Ethanol	Ningbo Zhenhai Yongfeng Chemical Plant	6923193900015
Fetal bovine serum	ScienCell Research Laboratories	C04001-500
Fiji (Fiji Is Just ImageJ)	Open-source project	N/A
FITC-Dextran (40 kDa)	Thermo Fisher Scientific Co., Ltd.	GC19938
Glass slides	Citotest Scientific Co., Ltd.	188105W
Graphpad prism 9.5	GraphPad Software, Inc.	N/A
Human Cerebral microvascular endothelial cells (HCMECs)	BeNa Culture Collection	BNCC337717
Incubator	Esco Lifesciences Co., Ltd	CLM-170B-8-CN
Inverted fluorescence microscope	Nikon Instruments Inc.	TS2-FL
Isopropanol	Sinopharm Chemical Reagent Co., Ltd.	80109218
Matrigel	Corning Inc.	354234
MEM	Pricella Life Science & Technology Co., Ltd.	PM150410
Nitrogen	Ningbo Baifang Gas CO., Ltd.	N/A
Oxygen	Ningbo Baifang Gas CO., Ltd.	N/A
Paraformaldehyde (PFA, 4%)	Wuhan Servicebio Technology Co., Ltd.	G1101-500mL
PBS	ScienCell Research Laboratories	C3580-0500
PDMS	Corning Inc.	220777
Penicillin/streptomycin	Pricella Life Science & Technology Co., Ltd.	PB180120
Photolithography	Institute of Optics and Electronics,Chinese Academy of Sciences	URE-2000\AL
Photolithography Hot Plate	Suzhou Wenhao Microfluidic Technology Co., Ltd.	WH-HP-02
Plasma Asher	TIANKECHUANGDA Co., Ltd.	PT-05-LF
Primary astrocytes	N/A	N/A	Isolated from neonatal rat (Day1-3) brain
Silicon wafer	Zhejiang YoungChip Technology Co., Ltd.	10204040001
Spin Coater	Suzhou Wenhao Microfluidic Technology Co., Ltd.	WH-SC-01
SU-8	Kayaku Advanced Materials Inc.	1030403002
Triton X-100	Beijing Solarbio Science & Technology Co., Ltd.	9002-93-1
Trypsin	ScienCell Research Laboratories	C3530-0100
U87-MG cells	Pricella Life Science & Technology Co., Ltd.	CL-0238
Ultrasound device	Shenzhen Dongdixin Technology Co., Ltd.	Sonic-Stimu Pro UT1041
Zen3.4(blue edition)	Carl Zeiss AG	N/A