A plug-and-play multi-organ-on-chip for studying interaction between central nervous system and peripheral monocytes following neuroinjury from radiation and Parkinson's disease

Abstract

Following neuroinjury (e.g., from radiation or Parkinson's disease), peripheral monocytes infiltrate the CNS, promoting neuroinflammation and cognitive decline. However, studying this is challenging due to a lack of suitable in vitro models. Multi-organ-on-a-chips (MOCs) address this gap with interactable in vitro models. We developed a plug-and-play multi-organ-chip (PPMOC) to mimic CNS-monocyte interactions. The PPMOC features two chambers, each with an insert for CNS or monocyte models, interconnected by a channel. A PDMS-based fabrication method using laser cutting of polymethyl methacrylate was developed for its fabrication. Finally, the PPMOC was employed to investigate CNS-monocyte interactions in radiation-induced neuroinjury and Parkinson's disease (PD). Results show that radiation caused neuroinjury, manifesting as decreased viability and morphological damage in nerve cells, and increased permeability of the blood–brain barrier (BBB). Further analysis revealed that radiation-induced neuroinjury inhibits the proliferation of THP-1 cells and promotes their activation and differentiation. Similarly, in PD, increased BBB permeability and activation of THP-1 cells were observed. Analysis of exosomes from the medium revealed upregulation of miR-151a-5p and miR-423-3p. Both models showed upregulated expression of inflammatory proteins and cytokines (e.g., CD14, TLR-2, IL-6, TNF-α, CCL-20), indicating monocyte activation. To sum up, PPMOC, a user-friendly and flexible multi-organ-on-a-chip, will become an important tool for studying CNS-monocyte interactions.

A plug-and-play multi-organ-chip (PPMOC) with interconnected CNS insert and monocyte insert mimics CNS-monocyte interaction, where monocyte activation is induced by γ-ray neuroinjury or Parkinson's disease.

1. Introduction

The interaction between the central nervous system (CNS) and the peripheral immune system is a crucial component of the regulatory network of neuroimmune interactions, which maintains the homeostasis of the CNS. Especially following neuroinjury, interactions with peripheral monocytes often lead to a series of effects, including monocyte infiltration into the CNS and activation, which are pertinent to neuroinflammation and cognitive dysfunction in patients after neuroinjury or disease.¹

Studies have shown that ionizing radiation can induce neuroinjury.² Ionizing radiation directly induces neuronal apoptosis and suppresses neurogenesis and oligodendrogenesis within the hippocampal dentate gyrus and subventricular zone.^3,4 In parallel, radiation damages cerebral microvascular endothelial cells and disrupts the blood–brain barrier, further amplifying the inflammatory and oxidative injury to the brain parenchyma.³ As a central cellular mediator of this neuroinflammatory response, microglia undergo rapid activation characterized by a shift to a pro-inflammatory phenotype, releasing numerous inflammatory mediators that perpetuate chronic neuroinflammation, disrupt neurogenesis and astrocyte development.⁵ These changes, in turn, also induce non-target effects in peripheral monocytes, such as monocyte proliferation in peripheral blood, differentiation into macrophage-like cells with upregulated CD14 expression, and exhibit chemotaxis.⁶ These monocytes affected non-target can subsequently infiltrate the brain parenchyma and, together with microglia, contribute to the ongoing neuroinflammatory response, impacting cognitive functions.^7–10 As ionizing radiation is present in both radiotherapy and space environments, studying these effects is crucial for astronauts and patients who have received cranial radiotherapy.

Similarly, in patients with Parkinson's disease (PD), there is a comparable interaction between the CNS and peripheral monocytes. Parkinson's disease is driven by redox imbalance and oxidative stress, leading to selective loss of dopaminergic neurons in the substantia nigra through a dynamic interplay of multiple cell death pathways, and chronic neuroinflammation mediated by activated microglia and astrocytes, forming a self-amplifying vicious cycle.¹¹ Neuroinjury not only activates microglia within the brain but also induces abnormal activation, phagocytosis, and proliferation of peripheral monocytes along with their infiltration into the nervous system.^12,13 Additionally, higher activation levels in monocytes have been associated with poorer cognitive functions.¹⁴

In summary, cognitive dysfunction linked to neuroinjury from ionizing radiation and PD closely relates to the interaction between the CNS and peripheral monocytes. In conventional studies, researchers often use in vitro and in vivo models to investigate the interactions between the CNS and peripheral monocytes. However, traditional in vitro models usually enable the study of cumulative effects over time but fail to accurately reproduce the three-dimensional structure of tissues, which significantly differs from real-time interactions between cells and tissues in vivo. In contrast, in vivo models involve complex interactions between multiple organs, making it challenging to evaluate specific interactions precisely. Multi-organ-on-a-chip (MOC) technology can integrate multiple organs, constructing interactable in vitro organ models,¹⁵ particularly suitable for studying the interactions between CNS and peripheral monocytes. In its simplest form, a MOC consists of the combination of two organ. To date, several dual-organ MOC have been reported, including the gut–liver,¹⁶ liver–nerve,¹⁷ , and liver–kidney.¹⁸ These MOCs rely on microchannels to interconnect tissues from different organs, with the flow of culture medium between compartments simulating inter-tissue interactions. Chen et al. reported a gut–liver MOC in which human hepatocytes and Kupffer cells were cultured in one chamber, while enterocytes, goblet cells, and dendritic cells were cultured in another, with microchannels connecting the two compartments. A pneumatic pump–valve system drove the recirculation of medium between the gut and liver chambers. Using this platform, the authors investigated gut–liver crosstalk under both normal and inflammatory conditions.¹⁶ MOCs have also been applied to study the interaction between the CNS and the immune system. Zhao et al. designed a neuro-immune MOC that utilized a micropillar array to establish a barrier, enabling co-culture and real-time in situ observation of SH-SY5Y cells (nerve cell) and THP-1 cells (monocyte), as well as quantitative assessment of monocyte migration.¹⁹ Micheli et al. constructed a BBB–monocyte MOC to evaluate the feasibility of monocyte-mediated delivery of oncolytic HSV-1 for the treatment of glioblastoma.²⁰

Studying radiation-induced neuroimmune interactions requires constructing in vitro models where only a portion of the organ is irradiated. However, traditional organ-on-a-chip devices are often irreversibly sealed, fixing the tissue within the chip,²¹ which makes it inconvenient to create models for partial tissue irradiation. Similarly, modeling agents such as 6-OHDA (6-hydroxydopamine) can inadvertently affect other cells when developing PD models. Plug-and-play multi-organ-on-chip incorporates multiple detachable modules, allowing for rapid and reversible assembly and disassembly of the chip and its cultures, significantly enhancing operational convenience. More importantly, plug-and-play multi-organ-on-chip enables researchers to remove specific modules from the chip for irradiation or to construct PD models using modeling agents, without worrying about the impact on monocytes. For example, Ronaldson-Bouchard et al. reported a MOC in which four types of matured human tissues (heart, liver, bone, and skin) were interconnected by recirculating vascular flow. This MOC adopted a plug-and-play modular design, with each tissue housed in an independent modular chamber cultured in its own optimized microenvironment and separated from the common vascular flow by a selectively permeable endothelial barrier.²² Therefore, plug-and-play multi-organ-on-chip (PPMOC) are particularly suitable for studying the interaction between the CNS and monocytes.

In our previous work, we developed a microfluidic chip fabrication method based on directly using laser cutting to fabricate individual components and epoxy resin bonding to assemble all the parts and porous membrane.²³ However, that design exhibited certain limitations in terms of plug-and-play convenience. To address this issue, we established a new low-cost, short-cycle, high-aspect-ratio fabrication method building upon our previous work. Specifically, rather than fabricating the main body of PPMOC directly by laser cutting and epoxy bonding, we first applied the previous method to create a mold, and then used the mold together with polydimethylsiloxane (PDMS) to produce the main body of PPMOC. The PPMOC can install inserts, making it easy to investigate non-target effects on peripheral monocytes induced by neuroinjury, and enabling the setup of partial irradiation and localized 6-OHDA treatment. Finally, we used the PPMOC to investigate the effects on peripheral monocytes induced by radiation and CNS damage caused by PD.

2. Materials and methods

2.1. Fabrication of PPMOC

The fabrication of main body of PPMOC involves the following steps (Fig. 1): ① laser cutting of PMMA for male mold fabrication. Laser cutting was performed on PMMA using the following parameters: maximum cutting power set at 100% with a laser scanning speed of 2 mm s⁻¹, to fabricate the male mold. ② Plasma-treated male mold and PS dish were bonded with epoxy resin. The plasma-treated bonding surface of the male mold and the PS dish were bonded using epoxy resin to form the chip mold. ③ Epoxy curing. The bonded mold was cured at room temperature for 24 hours or at 60 °C for 3 hours to achieve complete curing. ④ PDMS pre-mix adding and curing. PDMS pre-mix was prepared at a ratio of 1 : 10, poured into the mold, and cured at room temperature for 24 hours or at 60 °C for 3 hours. ⑤ PDMS chips demolded and bonded. The PDMS was carefully demolded from the mold to obtain the channel layer of the main body. Meanwhile, PDMS pre-mix with a thickness greater than 10 mm was poured into a PS dish without the male mold. After curing, two through-holes for placing inserts and a pair of inlet/outlet holes connecting the channel layer were cut at corresponding positions on the PDMS using die-cutting blades and a PDMS puncher, resulting in the chamber layer of the main body. Finally, plasma treatment was used to bond the channel layer and chamber layer, resulting in the main body of PPMOC.

Fig. 1. Schematic illustration of the method for fabricating a high aspect ratio PDMS chip and its mold.

The fabrication of insert of PPMOC involves the following steps (Fig. S1): ① preparation of PDMS film. Pour 1 mm and 4 mm thick PDMS sequentially onto a PS dish, and after curing, obtain the PDMS film used for fabricating the insert. ② Cutting. Use Φ32 mm and Φ18 mm die-cutting blades to cut two concentric rings from the 1 mm-thick PDMS film (Fig. S1a and b). ③ Bonding PET membrane (Fig. S1c). Attach a porous membrane (Φ25 mm, 0.4 µm) to the center of the ring, and then plasma bond the PDMS to fix the porous membrane. ④ Bonding the insert body and PDMS film for connection to the main body. Cut a Φ32/22 mm ring (insert body) from the 4 mm-thick PDMS film and a Φ45 mm film (PDMS film for connection to the main body) from the 1 mm-thick PDMS film (Fig. S1e). Bond them sequentially using the aforementioned method. Finally, carefully cut an inlet for the 3D cell culture scaffold along the inner edge of the insert body on the PDMS film for connection to the main body to complete the insert (Fig. S1f).

The fabrication of clamps of PPMOC involves the following steps: ① laser cutting. Laser cutting was performed on PMMA (4 mm-thick) using the following parameters: maximum cutting power set at 100% with a laser scanning speed of 0.5 mm s⁻¹, to fabricate the top clamp and bottom clamp. ② Bonding of top clamp. Bond the two layers of top clamp and joint with epoxy resin, then cured at room temperature for 24 hours or at 60 °C for 3 hours. Then, pour epoxy resin into the ring of the top clamp where the joint is placed, and after complete curing, the top clamp is obtained. ③ Embed an M4 flange nut into the bottom clamp to obtain the bottom clamp.

2.2. Cell culture

To establish an in vitro CNS–monocyte interaction model, four cell lines, hCMEC/D3, SH-SY5Y, U-87 MG, and THP-1, were selected for their advantages in reproducibility and accessibility. The human cerebral microvascular endothelial cell line hCMEC/D3, widely used for in vitro blood–brain barrier modeling, was employed to represent the BBB.²⁴ To model the CNS, SH-SY5Y cells (neurons) and U-87 MG cells (glia) were co-cultured directly at a ratio of 1 : 1.65 (neurons: glia). This ratio follows our previously established model^6,23 that approximately recapitulates the neuron-to-glial cell ratio of the human olfactory bulb,²⁵ a brain region sensitive to radiation injury.^26,27 The human monocytic cell line THP-1 was selected as the monocyte model, also based on our previously established model.^6,23 THP-1 cells, derived from the peripheral blood of a patient with acute monocytic leukemia, are a well-characterized human leukemia monocytic cell line widely adopted for monocyte/macrophage modeling. Morphologically and functionally, THP-1 cells closely resemble primary human monocytes/macrophages. Upon appropriate induction, they can be polarized into M1 or M2 macrophage subtypes and effectively recapitulate key monocyte/macrophage functions, including phagocytosis, antigen presentation, and cytokine secretion.²⁸

2D cell culture. Human cell lines hCMEC/D3, SH-SY5Y, U-87 MG and THP-1 were purchased from the Cell Center in the Chinese Academy of Medical Sciences. DMEM and RPMI-1640 were purchased from Gibco Scientific Inc. Fetal bovine serum (Cat. P30-3302) was purchased from PAN BIOTECH. Cells were replaced with digestion and collection performed by trypsin (T1300, Solarbio, China), and terminated with fetal bovine serum.

3D cell culture in CellSCAFLD® 3D cell culture scaffold (TDP032012, JET Bio-Fil Co). Following the 2D cell culture method, cells were cultured to the appropriate density. Then, cells were prepared to 180 µL of a suspension containing SH-SY5Y cells and U-87 MG cells, in a ratio of 1 : 1.65 (e.g., 2.72 × 10⁵ cells per mL of SH-SY5Y cells and 4.50 ×10⁵ cells per mL of U-87 MG cells, for a total of 7.22 × 10⁵ cells per mL). Hold one end of the 3D cell culture scaffold with forceps, and carefully add the suspension to the CellSCAFLD® to ensure the cell suspension evenly permeates the entire scaffold without spilling over. Transfer the CellSCAFLD® into a sterile well plate, and then culture it in a 37 °C incubator with 5% CO₂ for 3–6 hours to allow the adherent cells to adhere to the scaffold.

2.3. Design of plug-and-play multi-organ chips (PPMOC)

The structure of the PPMOC is shown in Fig. 2. The PPMOC consists of two chambers connected by a channel, each chamber capable of setting an insert (Fig. 2a). CNS or peripheral monocyte in vitro models were cultured in the inserts to simulate the interaction between the CNS and peripheral monocytes in the body (Fig. 2b). The PPMOC is equipped with two clamps, the top clamp and bottom clamp (Fig. 2c). The top clamp consists of two layers of PMMA (fabricated by laser cutting) and two joints, bonded with epoxy resin. Using epoxy resin to integrate the clamp and joint ensures robustness and seal integrity. It features six symmetrically arranged holes for M4 screws. The bottom clamp (fabricated from PMMA via laser cutting) includes M4 flange nut positions corresponding to the holes in the top clamp. These two clamps apply pressure to the PPMOC main body and the inserts, ensuring the chip remains sealed. The PPMOC main body is composed of two bonded PDMS layers (Fig. 2d), featuring two chambers for insert placement and a perfusion channel connecting them to enable medium flow (Fig. 2e and f). The insert contains a circular chamber for placing CellSCAFLD® or directly culturing non-adherent cells (Fig. 2g). It also incorporates a nuclear track membrane (0.4 µm pore size, 30 mm diameter) at the bottom, for culturing of BBB or retaining non-adherent cells. The circular chamber features a stepped base design, where the PDMS layer above the membrane has a smaller diameter than the 3D cell culture scaffold. This prevents scaffold-induced damage to the BBB cultured on the porous membrane (Fig. 2i and j). On the top of the insert, a PDMS film is set to connect the insert with PPMOC (Fig. 2g and h). The actual image of PPMOC and its inserts were shown in Fig. 2k. The characteristic of plug-and-play allows researchers to control experiments flexibly and effectively avoid damage to the monocytes when constructing in vitro model of partial radiation or 6-OHDA treatment to nerve cells.

Fig. 2. PPMOC schematic. Explosion view (a) and schematic diagram (b) of the PPMOC. (c) Structure of the top and bottom clamp. Explosion view (d), assemble view (e) and actual image (f) of mainbody of PPMOC. Assemble view (g), explosion view (h), section view (i) and actual image (j) of the insert of PPMOC. (k) The actual image of PPMOC is assembled with a clamp and tube.

Due to the poor light transmission of the nuclear track membrane, observing the BBB cell morphology directly is challenging. We designed an observable method for culturing the BBB using gelatin methacryloyl (GelMA) hydrogel, which elevates the positioning of the blood–brain barrier cells, allowing for direct observation of the BBB through an optical microscope without staining. This suggests that GelMA improves cell observation, simplifies the process, and reduces the need for staining, which can adversely affect cells. The relevant results are shown in Fig. S2.

2.4. Cell viability assay

CCK8 was used to determine cell viability based on our previous work.¹⁹ Briefly, the CCK8 detection reagent was added to the chambers at a 10% ratio, followed by incubation for 2 h in the cell culture incubator. Then, 100 µL of the mixture was transferred to a 96-well plate, and OD450 nm was detected by an ELISA Reader (American Botten Instrument Co., Vermont, USA). All experiments were repeated at least three times, and finally, the cell viability was calculated as follows: Cell viability (%) = (OD_experiment − OD_blank)/(OD_control − OD_blank) × 100%.

2.5. Cell seeding, treatment and co-culture on PPMOC

Before the cell seeding, the main body of the PPMOC and the inserts were sterilized by autoclaving, and the clamps were disinfected by soaking in 75% ethanol for more than 30 minutes. Then, the PPMOC was used in 4 stages (Fig. 3).

Fig. 3. The workflow for PPMOC. The in vitro model of the CNS with radiation neuroinjury or PD was placed in the upstream chamber, and the in vitro model of monocytes was placed in the downstream chamber. After continuous perfusion culture for a period of time, the effect of monocytes in the downstream chamber was evaluated.

Stage I: insert preparations. ① nerve cell inoculation. Nerve cells (U-87 MG and SH-SY5Y, at a ratio of 1 : 1.65) were seeded onto 3D scaffolds (TDP032012, CellSCAFLD®, JET co). After 3–6 h of attachment, culture medium was supplemented, and cells were cultured for 72 h. ② BBB construction. 10% GelMA was added to the surface of the PET membrane (0.4 µm pore size, diameter 30 mm) in the upstream insert. Cure it under UV light for 1 minute, then seed hCMEC/D3 cells on its surface and culture for 48 hours. ③ γ ray or 6-OHDA treatment. To simulate neuroinjury of radiation, the scaffold was irradiated with γ-rays (dose of 15 Gy, dose rate of 1 Gy min⁻¹). To simulate neuroinjury in PD patients, the nerve cells were treated with 100 mM 6-OHDA for 6 hours. ④ Assemble of CNS insert. The treated scaffold was placed into the CNS insert containing the cultured BBB. ④ Monocyte inoculation. Seed THP-1 cells in the downstream insert as an in vitro model of peripheral monocytes.

Stage II: PPMOC assemble. ⑤ Assemble the inserts into PPMOC. Then, place PDMS gaskets of appropriate thickness under the connector of the top clamp to prevent liquid leakage, tighten the screws, and connect the tubing, peristaltic pump, and reservoirs. Finally, seal the top of the upstream and downstream chambers with microplate sealers to avoid contamination.

Stage III: continuous perfusion. ⑥ Continuous perfusion was performed for 48 hours to simulate fluid flow and cell interactions in vivo.

Stage IV: analysis. ⑦ Post-culture, critical metrics assessment.

For the control group, the CNS insert was processed identically to the experimental group, except it was not irradiated or treated with 6-OHDA.

2.6. Flow simulation

COMSOL 6.0 was used for flow simulations. When PPMOC was perfused at the minimum speed of 16 mL h⁻¹ using a peristaltic pump, the velocity at the inlet of the chip channel was approximately 0.00111 m s⁻¹. Based on the actual perfusion conditions, the parameters of the flow simulation were set below: the temperature was set to 311.15 K; the fluid was set as water with an inlet velocity of 0.00111 m s⁻¹, and the outlet pressure was set to one atmosphere. Additionally, due to the use of a PET porous membrane in the chip insert, the laminar flow physics was coupled with the porous medium during the simulation, with the porosity of the medium set to 0.5.

2.7. RNA extraction, cDNA synthesis, and real-time PCR analysis

Total RNA of THP-1 cells was extracted using Trizol reagent (Invitrogen, Carlsbad, CA, USA). The microRNAs in exosomes were extracted using Urine Exosome Purification Mini Kit (Norgen Biotek,Cat no.57700, Canada) and Exosomal RNA Isolation Kit (Norgen Biotek, Cat no.58000, Canada).cDNA was synthesized using a GoScript™ Reverse Transcription system (Promega, Cat. No. A5001, USA) or Mir-X miRNA First-Strand Synthesis Kit (Takala, Cat No. 638315, Japan) in accordance with the manufacturer's instructions. Primer sequences are detailed in Table 1. Quantitative PCR was conducted in a total volume of 20 µL with 40 cycles of 10 s at 95 °C and 10 s at 60 °C (TB Green™ Premix Ex Taq™, Takala, Cat No. RR420A, Japan) Table 2.

Table 1. Primer sequence of investigated genes.

Gene	Forward sequence (5′–3′)	Reverse sequence (5′–3′)
GAPDH	AAGGTGAAGGTCGGAGTCA	GGAAGATGGTGATGGGATTT
TNF-α	TCTCGAACCCCGAGTGACAA	TGAAGAGGACCTGGGAGTAG
IL-6	ACTCACCTCTTCAGAACGAATTG	CCATCTTTGGAAGGTTCAGGTTG
TLR-2	AGTTGATGACTCTACCAGATG	GTCAATGATCCACTTGCCAG
CD14	CGAGGACCTAAAGATAACCGGC	GTTGCAGCTGAGATCGAGCAC
CXCL-1	GCGCCCAAACCGAAGTCATA	ATGGGGGATGCAGGATTGAG

Table 2. Primer sequence of investigated miRNA.

miRNA	Sequence (5′–3′)
miR-151a-5p	TCGAGGAGCTCACAGTCTAGT
miR-423-3p	AGCTCGGTCTGAGGCCCCTCAGT

2.8. Transepithelial electrical resistance (TEER) assay

TEER was measured using a RE1600 Epithelial Volt-Ohm Meter (KingTech, Beijing, China). Briefly, one electrode was placed in the upstream insert (above the BBB). The other electrode was placed in the downstream chamber connected to the bottom channel of the chip (below the BBB). After the reading stabilized, the resistance value (R) with the BBB and the blank resistance value (R₀) without cells were recorded. The TEER value was then calculated using the following formula: TEER (Ω cm²) = [ R (Ω) − R₀ (Ω) ] × S (cm²). S represents the area of the blood–brain barrier in contact with the culture medium.

2.9. Giemsa staining and image analysis

Giemsa staining was performed based on the previous literature.⁶ Briefly, the nuclear track membrane in the downstream insert was washed with PBS and treated with 4% paraformaldehyde before staining. Then the membranes were stained for 8 min with Giemsa working solution in a mixture of Giemsa stock solution and Giemsa buffer = 1 : 9 (Solarbio, G1010), then washed for 30 s. Finally, the membrane was observed under an Olympus-IX71 microscope. Microscopic images were analyzed after binarization processing by Adobe Photoshop 2020. The microscopic images were analyzed using MATLAB by counting the white pixels. Finally, the percentage of the area of adherent cells were calculated using the following formula:Percentage of area of adherent cells = Number of white pixels/Number of total pixels × 100%.

2.10. Statistical analysis

Data mapping based on mean and standard error. Data statistics based on t-test and ANOVA. p < 0.05 marking *, p < 0.01 marking**; p < 0.001 marking ***.

3. Results

3.1. Establishment of a method for fabricating a high aspect ratio PDMS chip and its mold

To fabricate a PPMOC that is easy to assemble and disassemble and suitable for constructing partially irradiated in vitro models, it is necessary to use a cell culture scaffold that facilitates assembly. JET Co.'s CellSCAFLD® is a rigid 3D cell culture scaffold made of polystyrene, with surface properties similar to those of commonly used cell culture dishes. Also, it can be directly disassembled and reassembled using tweezers. Since adopting CellSCAFLD®, the cell density in PPMOC has increased, requiring an increase in the volume flow rate of the culture medium to fulfil the needs for nutritional and waste exchange for the cells. However, increasing the volume flow rate inevitably raises the fluid linear velocity during perfusion within the PPMOC channel, thereby reducing chip stability and shortening the medium's residence time in the channels. This, in turn, affects exchange of nutrient and waste for cells within the 3D scaffold. Increasing the channel aspect ratio is a viable solution, but it demands a fabrication method that can achieve high aspect ratio channels.

Soft lithography using PDMS, currently the most widely utilized method for organ-on-a-chip fabrication, can produce high-aspect-ratio chambers through methods like punching, yet struggles to create chip channels with a high aspect ratio. Typically, the maximum height of the chip channel reaches only 200 µm.²⁹ Despite this limitation, PDMS is extensively used to fabricate organ-on-a-chip devices due to its high gas permeability, excellent biocompatibility, and inertness. Since PPMOC employs 3D cell culture scaffolds for high-density cell culture, insufficient dissolved oxygen in the chip can hinder normal cell growth. Therefore, it is preferable to use gas-permeable materials for PPMOC fabrication, such as PDMS.

Therefore, we conceived a chip-fabrication technique that combines laser cutting of polymethyl methacrylate (PMMA) with polydimethylsiloxane (PDMS) molding. Based on our preliminary experience in fabricating microfluidic chips using PMMA with laser cutting technology,²³ PMMA, epoxy resin, and polystyrene culture dishes exhibit poor adhesion to PDMS. This allows for easy demolding of PDMS from these three materials. After plasma treatment, these materials can be securely bonded using epoxy resin. Additionally, due to the ease of processing high aspect ratio microfluidic chips with PMMA through laser cutting, they can be conveniently fabricated into high aspect ratio molds using PMMA, epoxy resin, and polystyrene culture dishes. Therefore, we designed and validated a process for fabricating high aspect ratio chip molds, as illustrated in Fig. 1. The process involves using laser-cut PMMA as a positive mold, which is then plasma-treated and bonded to a polystyrene dish with epoxy resin. After curing, the chip mold is obtained, and PDMS pre-mix is poured into it. Upon further curing, a PDMS chip is formed. Finally, the chip is bonded to the PDMS substrate after trimming and finishing.

To evaluate this method, we used laser cutting to fabricate various PMMA positive molds (including letters and symbols). Following the above method, we fabricate the PDMS chip, as shown in Fig. 4. The results show that patterns are clear without leakage. The results of the biocompatibility tests for the chips are shown in Fig. 4a–d. The results indicate that the cells were in good condition under the microscope. Both groups of cells could attach and grow normally (Fig. 4a). There was no significant difference in cell adhesion area (Fig. 4b) and in cell viability (Fig. 4c). These results demonstrate that the high aspect ratio chips fabricated using this method are suitable for constructing in vitro models. To investigate the range of aspect ratios achievable with this method, we tested the fabrication of chips using PMMA positive molds with thicknesses from 1 mm to 5 mm (Fig. 4d). The results showed that our method can produce PDMS chips with an aspect ratio greater than 5 : 1. Compared to soft lithography, this method significantly improves the aspect ratio of the chips, meeting our requirement.

Fig. 4. Method of fabricating a high aspect ratio PDMS chip and its mold. (a) Cell morphology of the fabricated PDMS chip. (b) Comparison of adherent cell area between the control and demolded groups; ns represents p > 0.05, indicating no significant difference. (c) Cell viability was measured using the CCK8 of the fabricated PDMS chip; ns represents p > 0.05, indicating no significant difference. (d) PDMS chips with chamber depths ranging from 1 to 5 mm (aspect ratio at the narrowest part of the chamber is 1 : 1 to 5 : 1).

3.2. Flow simulation of the PPMOC

To verify the capability of the nuclear track membrane to reduce shear stress, COMSOL Multiphysics 6.0 was used to perform fluid dynamics simulations of the shear stress distribution on both sides of the nuclear track membrane. The results are shown in Fig. 5. The results indicate that during perfusion, the shear rate of the fluid in the channel (Fig. 5a), particularly at the center of the channel, is relatively high, with a maximum shear rate of approximately 129 s⁻¹, which corresponds to a shear stress of about 8.92 × 10⁻² Pa (estimated using the viscosity of water at 37 °C as 691.52 µPa s, the same approximation is used below). On the lower surface in the center of the nuclear track membrane (Fig. 5b, the fluidic contact surface), the shear rate of the fluid is approximately 9.8 s⁻¹ (corresponding to a shear stress of about 6.78 × 10⁻³ Pa). On the upper side at the corresponding position (the cells contact surface), the shear rate of the fluid is approximately 0.01 s⁻¹ (corresponding to a shear stress of about 6.92 × 10⁻⁶ Pa). The shear rate distribution across the section of the upstream chamber and downstream chamber similarly confirms this conclusion (Fig. 5c and d). In summary, the nuclear track membrane in the insert reduces the fluid shear stress within the insert chamber.

Fig. 5. Flow simulation to PPMOC. (a) Shear rate distribution of the fluid within the channel. (b) Shear rate distribution of the fluid within the insert. (c) Shear rate distribution across the section of the downstream chamber of the chip. (d) Shear rate distribution across the section of the upstream chamber of the chip.

3.3. Radiation neuroinjury induced peripheral monocyte activation

The experimental procedures based on the PPMOC are shown in Fig. 3. After the experiment, radiation-induced neuroinjury and its effects on peripheral monocytes were investigated. For the control group, all treatments are identical except irradiation. TEER monitoring confirmed that the BBB was properly established in the PPMOC, remaining stable from day 2 to day 6 and thus suitable for subsequent experiments (Fig. S3).

As shown in Fig. 6a and b, the irradiated group exhibited a significantly lower cell number than the control group (Fig. 6a), which was paralleled by a decrease in cell viability (Fig. 6b), indicating radiation-induced neuroinjury. CNS injury usually induces an increase in the permeability of the blood–brain barrier.³⁰ Therefore, BBB damage induced by radiation-induced neuroinjury was investigated. The results show that compared to the control group, holes were forming in the cells on the BBB in the radiation group (Fig. 6c). Furthermore, a decrease in TEER values (Fig. 6d) and an increase in the permeability coefficient of sodium fluorescein (Fig. 6e) also indicated the reduced permeability of the BBB after radiation-induced neuroinjury.

Fig. 6. Radiation-injured CNS induced BBB damages and peripheral monocyte activation. The ratio of SH-SY5Y to U-87 MG cells was 1 : 1.65, with 2.6 × 10⁵ cells seeded per scaffold and cultured for 72 hours before irradiation (15 Gy, 1 Gy min⁻¹). After perfusion for 48 hours, analyses were performed on the CNS, BBB, and monocytes. (a) Morphology of nerve cells in the scaffold before and after irradiation. (b) Cell viability of nerve cells in the scaffold was measured by CCK8 assay 48 hours post-irradiation (**p < 0.001). (c and d) Radiation-injured CNS induced BBB damage (post-perfusion 48 hours). (c) Morphology of BBB cells. (d) TEER changes in the BBB induced by radiation-injured CNS. (e) Changes in the sodium fluorescein permeability coefficient of the BBB. (f–h) Radiation-injured CNS induced adhesion and differentiation of THP-1 monocytes (post-perfusion 48 hours). (f) Microscopic images of THP-1 monocytes adhered to the PET membrane of the downstream insert, stained with Giemsa stain. (g) Percentage of area of THP-1 cells. (h) Cell viability of THP-1 cells within the insert (****p < 0.0001). (i–n) Analysis of RNA relative expression levels in THP-1 monocytes, including relative expression levels of IL-6 (i), TNF-α (j), CD14 (k), TLR-2 (l), CXCL-1 (m), and CCL-20 (n) (ns – no significant difference, **p < 0.01, ***p < 0.001, ****p < 0.0001).

To investigate the effect of radiation neuroinjury induced on peripheral monocytes, a series of assays were conducted on the THP-1 cells in the downstream insert of the PPMOC. These assays included cell viability, the number of adherent cells on the nuclear track membrane, and relative transcriptional expression of some key proteins or cytokines of monocyte activation. The results show that inhibition of monocyte proliferation was induced by radiation neuroinjury (Fig. 6h), alongside an upregulation of monocyte differentiation levels (Fig. 6f and g). Furthermore, the transcriptional levels of several monocytic differentiation markers, inflammatory factors, and cytokines in THP-1 cells were examined (Fig. 6i–n), including CD14, CXCL-1, CCL-20, TNF-α, IL-6 and TLR-2. CD14, a surface marker of monocytes that plays a crucial role in their differentiation and activation, when upregulated, indicates that THP-1 cells are differentiating towards macrophages.^28,31 TLR-2, a critical immune receptor involved in recognizing and responding to bacteria, viruses, and other pathogens,³² can also serve as a marker for differentiating THP-1 cells into macrophages when upregulated.²⁸ Our results showed that the relative expression levels of CD14 (Fig. 6k) and TLR2 (Fig. 6l) in THP-1 cells were upregulated, indicating a tendency for THP-1 cells to differentiate towards macrophages. These results are consistent with our previous in vitro experiments and in vivo studies in rats.⁶

IL-6 and TNF-α are critical pro-inflammatory cytokines that play key roles in various inflammatory responses. In this study, we observed increased expression levels of the pro-inflammatory cytokines, IL-6 and TNF-α, in THP-1 cells (Fig. 6i and j), particularly a substantial upregulation of IL-6 RNA expression. This finding indicates that when the CNS is subjected to radiation damage, monocytes become activated and begin secreting pro-inflammatory cytokines, thereby promoting the onset and progression of inflammation.

As part of the CC chemokine family, CCL20 (MIP-3 alpha) demonstrates a strong chemotactic effect on lymphocytes and a weak chemotactic effect on neutrophils.³³ THP-1 cells did not constitutively express CCL20 mRNA,³⁴ but can be strongly induced by many inflammatory factors in macrophages.³⁵ Our results showed that expression levels of CCL20 were increased after CNS injured by radiation (Fig. 6n). Similarly, CXCL-1 is also a chemokine that is expressed by macrophages, neutrophils, and epithelial cells, and it possesses neutrophil chemoattractant activity. Additionally, CXCL-1 can serve as a marker for the polarization of macrophages towards the M1 phenotype.²⁸ These results indicated that peripheral monocytes have been activated by radiation-induced neuroinjury. This may also suggest that the infiltration and activation of monocytes could further attract other lymphocytes, thereby participating in the immune response.

3.4. Peripheral monocyte activation in Parkinson's disease

Similarly, the experimental procedures for investigating the effects of peripheral monocytes in Parkinson's disease, based on the PPMOC, are illustrated in Fig. 3. For the control group, all treatments were identical to those of the experimental group, except the 6-OHDA treatment.

Firstly, to verify the effect of 6-OHDA in modeling Parkinson's disease (PD) in vitro, we investigated the cell viability of 6-OHDA on the nerve cells. The results indicate that the cell viability of nerve cells treated with 100 mM 6-OHDA was significantly lower than that of the control group (Fig. 7a), suggesting that the PD model was successfully established. Furthermore, focal and partial loss of monolayer confluence of BBB cells (Fig. 7b) together with a decrease in TEER values (Fig. 7e) indicated an increased BBB permeability following 6-OHDA-induced neuronal damage. Finally, we examined the changes in the number of adherent cells on nuclear track membrane cells in the downstream plug of the PPMOC. As shown in Fig. 7c and d, the number of adherent THP-1 cells on the membrane in the treatment group was significantly higher compared to the control group. These results are similar to those observed in 3.3, where radiation-induced neuroinjury led to peripheral monocytes activation and differentiation. This is also consistent with previously reported findings in neurobiological studies.¹

Fig. 7. Peripheral monocyte activation in Parkinson's disease. The ratio of SH-SY5Y to U-87 MG cells was 1 : 1.65, with 2.6 × 10⁵ cells seeded per scaffold and cultured for 72 hours before 6-OHDA treatment (100 mM, 6 h). After perfusion for 48 hours, analyses were performed on the CNS, BBB, and monocytes. (a). The viability of nerve cells in the scaffold was measured using a CCK8 assay post-treatment. (b) Morphology of BBB cells. (c) Microscopic images of THP-1 monocytes adhered to the PET membrane of the downstream insert, stained with Giemsa stain. (d) Adhesion area of THP-1 cells. (e). TEER changes in the BBB induced by CNS treated with 6-OHDA. (f and g) Analysis of RNA relative expression levels in exosomes extracted from culture medium, including relative expression levels of miR-423-3p (f) and miR-151a-5p (g). (h–m) Analysis of RNA relative expression levels in THP-1 monocytes, including relative expression levels of IL-6 (h), TNF-α (i), CD14 (j), TLR-2 (k), CCL-20 (l) and CXCL-1 (m) (ns – no significant difference, **p < 0.01, ***p < 0.001, ****p < 0.0001).

To further elucidate the molecular mechanisms underlying the effects of peripheral monocytes induced by the CNS in PD patients, we conducted additional studies on the culture medium and THP-1 cells. Exosomes, a class of extracellular vesicles generated through the endosomal pathway, participate in cellular communication, immune responses, and cell migration within the body and are widely distributed in various bodily fluids.³⁶ Mounting evidence suggests that exosomes play a crucial role in PD, potentially facilitating the spread of pathological α-synuclein or activating immune cells during the progression of PD.³⁷ Researches have shown that two miRNAs, miR-151a-5p³⁸ and miR-423-3p (data not shown), which are upregulated in exosomes from the plasma of Parkinson's disease patients, have the potential to serve as biomarkers for the diagnosis of Parkinson's disease. The culture medium in the PPMOC functions similarly to peripheral blood in vivo, serving a circulatory fluid role. Therefore, we used RT-qPCR to analyse the changes in transcription levels of the aforementioned two miRNAs in the exosomes extracted from the medium. Our results show that the transcription levels of miR-151a-5p (Fig. 7f) and miR-423-3p (Fig. 7g) were upregulated. These findings are consistent with reported levels of in exosomes from the plasma of Parkinson's disease patients.³⁸

Similar to the study of radiation-induced neuroinjury, we also detected the transcriptional levels of several monocytic differentiation markers, inflammatory factors, and cytokines in THP-1 cells. The results show that, except for CXCL-1 (Fig. 7m), the transcriptional levels of the remaining five proteins, IL-6 (Fig. 7h), TNF-α (Fig. 7i), CD14 (Fig. 7j), TLR-2 (Fig. 7k) and CCL-20 (Fig. 7l) were consistent with those observed in radiation-induced neuroinjury. As previously mentioned, CCL-20 primarily has a chemotactic effect on lymphocytes. At the same time, CXCL-1 mainly affects neutrophils, which may indicate that during PD, monocytes primarily induce lymphocyte chemotaxis with less effect on neutrophils.

4. Discussion

In this research, we developed a method for fabricating mold for microfluidics using laser cutting and successfully fabricated a PPMOC with this mold. Using the PPMOC, we established an in vitro model to mimic the interaction between the CNS and peripheral monocytes.

In brief, the PPMOC we designed offers several advantages. Most notably, its rigid 3D scaffold and plug-and-play design facilitate the construction of radiation and PD models and simplify handling for researchers. This makes the PPMOC particularly suitable for investigating non-target effects, that is, how pathological changes in one organ (e.g., CNS injury) can exert biological effects on a distal organ (e.g., peripheral monocytes). Additionally, its high aspect ratio, combined with 3D cell scaffolds, increases cell culture density and obtained more cells for analysis. Finally, it exhibits versatility and high repeatability, making it applicable to other intercellular interaction studies and drug research.

Ronaldson-Bouchard et al. developed a vascularized MOC based on a plug-and-play concept.²² In this MOC platform, matured human heart, liver, bone and skin tissue niches are linked by recirculating vascular flow containing CD14⁺ monocytes to model circulating immune cells, allowing for the recapitulation of interdependent organ functions. Each tissue is cultured in its own optimized environment and is separated from the common vascular flow by a selectively permeable endothelial barrier. Human umbilical venous endothelial cells (hUVECs) were used to construct the endothelial barrier in this MOC. However, this setup does not recapitulate the real barrier between the CNS and the periphery, the BBB. Extensive evidence has demonstrated that the BBB plays a critical role in maintaining CNS homeostasis in brain radiotherapy.^39–41 In contrast, PPMOC used the hCMEC/D3 cells to establish a BBB model, thereby offering a distinct advantage for studying CNS–monocyte interactions. Furthermore, in comparison with other plug-and-play MOC platforms, the structural design and experimental strategy of PPMOC, particularly its integration with partial irradiation or 6-OHDA treatment, endow it with notable methodological innovation.

The PPMOC was developed based on the concept of MOC, while organoids also represent a current focus of in vitro modeling. Organoids and MOCs systems represent two complementary in vitro research strategies. Organoids offer the advantage of high physiological fidelity, as they self-organize into three-dimensional structures comprising multiple cell types.⁴² However, it remains challenging to construct inter-organ interaction models using organoids alone, because distinct organoids are typically cultured independently in separate systems, lacking fluidic control. This limitation prevents effective modeling of bystander effects or signalling from one organ to another. The core advantage of MOCs lies precisely in addressing this gap. By connecting different organ compartments through microchannels, MOCs maintain fluidic control between spatially separated organ. This design enables intervention in a single compartment (e.g., radiation or drug treatment) while simultaneously monitoring responses in a distal organ, making it an ideal platform for studying inter-organ interactions.

The method for fabricating high aspect ratio microfluidic devices we established also has many advantages and potential applications for the future. Firstly, the microfluidic fabricated by this method maintain a high aspect ratio while also ensuring excellent gas permeability, making them particularly suitable for applications requiring gas exchange, such as cell culture and organ-on-a-chip devices. Secondly, the high aspect ratio of the microfluidics fabricated by this method allows for increased volumetric flow rates without causing excessive pressure in the chambers during perfusion. Meanwhile, the high-aspect-ratio channel design effectively minimizes bubble interference. Finally, this method requires minimal equipment and environmental conditions for fabrication, only a laser cutter, a plasma cleaner, and a heating oven, making it especially suitable for fabricating microfluidics in a basic laboratory setting.

We believe that the microfluidic processing method we have established will play a significant role in future aerospace medicine research. PDMS is currently the most widely used polymer material in fabricating microfluidic chips and organ-on-a-chip devices due to its high biocompatibility, good gas permeability, and strong moldability. Soft lithography is commonly applied to fabricate PDMS devices. However, this method is costly and time-consuming. Furthermore, it is difficult to fabricate high-depth chambers suitable for 3D cell culture scaffolds used in PPMOC with soft lithography. Previously, we developed a method for fabricating PMMA microfluidic chips with high aspect ratio using laser cutting,²³ which can fabricate a chamber suitable for 3D cell culture scaffolds. However, during PMMA processing, thermal bonding can cause chip bending, affecting the airtightness of the assembled chip inserts. To address these issues, we combined PMMA fabrication using laser cutting with PDMS molding to establish a more convenient and economical method for fabricating high aspect ratio microfluidics, which we then applied to producing the PPMOC. On the other hand, in aerospace medicine research, the application of organ-on-a-chip technology in studies in simulated microgravity (SMG) remains limited. There have been only a few reports on microfluidic cell culture chips simulating microgravity, such as those described by Yew et al.⁴³ and Przystupski et al.⁴⁴ The primary reason is that microgravity rotators for SMG environments, such as the random positioning machine (RPM) and the Rotary Cell Culture System (RCCS™), require random rotation along single or dual axes. Existing organ-on-a-chip devices, however, need to be connected to external peripherals such as pumps, valves, and controllers, or they adopt a semi-open design. This makes it difficult to install them directly into microgravity rotators. Furthermore, connecting external devices or setting up semi-open reservoirs on the chip is challenging, so culture medium perfusion cannot be easily applied in microgravity rotators. Therefore, organ-on-a-chip devices used in SMG environments should ideally have fully enclosed chambers with high aspect ratios. This design allows for the storage of more culture media, thereby meeting the metabolic demands of the cells. Because gas permeability is critical for cell culture, closed organ-on-a-chip systems often use PDMS as the primary material due to its excellent gas permeability.⁴⁵ However, soft lithography processing of PDMS makes it difficult to achieve high aspect ratio structures, typically reaching a maximum height of only 200 µm.29 On the other hand, materials commonly used for fabricating high aspect ratio chips, such as PMMA, lack sufficient gas permeability,⁴⁶ significantly increasing the difficulty of cell survival. In summary, an organ-on-a-chip device suitable for SMG environments should feature full enclosure, high aspect ratios, and high gas permeability. In earlier work, we developed a high aspect ratio microfluidic chip using die-cutting blades to fabricate PDMS, and then we used this chip for cell culture under SMG.⁴⁷ Based on this chip and aggregation-induced emission (AIE) probe, we developed a SMG-oriented integrated chip platform capable of 3D cell culture and in situ visual detection of superoxide anion radical (O2˙−). However, this process was also time-consuming since all chip components required manual cutting using a die-cutting blade. The method for fabricating PDMS chips we established in this work meets all the requirements for organ-on-a-chip devices under SMG and offers shorter processing times and higher precision than our previous work. Therefore, we believe this method has significant potential for application under SMG environments.

Several limitations of the present study should be acknowledged. First, the in vitro model established on the PPMOC employed established cell lines. While this choice facilitates experimental accessibility and reproducibility, cell lines entail well-recognized trade-offs in physiological fidelity when compared to alternative in vitro models, such as primary cell cultures, induced pluripotent stem cells (iPSCs), or organoid. Second, although the PPMOC was designed with a reusable chip body and disposable plugs, offering the potential for repeated use, no systematic performance validation was conducted on reused chips. This represents an important limitation that constrains the future applicability of the PPMOC system. Finally, this study examined radiation-induced CNS–monocyte interactions at a single dose of 15 Gy rather than across multiple dose levels. This decision was guided by two considerations: this dose falls within the clinically relevant range for brain radiotherapy^48,49 and at 15 Gy we observed sufficient CNS injury (Fig. 3a–e) to drive monocyte recruitment and interaction (Fig. 3f). Future studies will compare CNS–monocyte interactions across different doses using the PPMOC model.

5. Conclusions

In this study, to construct a PPMOC suitable for studies of non-targeted effects, building upon our previous method of fabricating components via laser cutting and epoxy resin bonding, we developed a method for processing chip molds and successfully fabricated the PPMOC by casting PDMS in this mold. Based on this PPMOC, we established an in vitro model to mimic the interaction between the CNS and peripheral monocytes. Using this model, we studied the interactions between radiation-induced neuroinjury and the CNS in PD patients and peripheral monocytes. Research outcomes indicate that radiation-induced neuroinjury decreased the viability and altered the morphology of nerve cells, alongside enhanced permeability of the BBB. Furthermore, radiation-induced neuroinjury suppressed the proliferation of THP-1 cells and stimulated their activation. The expression levels of key proteins and cytokines, including CD14, TLR-2, IL-6, TNF-α, CXCL-1 and CCL-20, were increased, suggesting peripheral monocyte activation and inflammatory response. Similarly, in PD model, increased permeability of the BBB and THP-1 cells activation were also observed. Further analysis of exosomes from culture medium revealed upregulated levels of miR-151a-5p and miR-423-3p, consistent with PD patient data. Transcriptional levels of CD14, CCL-20, TNF-α, IL-6, and TLR-2 in THP-1 cells were increased, similar to radiation-induced effects. Results in PD research indicate that 6-OHDA treatment in the PPMOC decreased nerve cell viability and altered cell morphology, establishing a PD model. Similar to radiation-induced neuroinjury, decreased BBB permeability and increased adhesion of THP-1 cells were observed. Analysis of the exosome from culture medium revealed upregulated miR-151a-5p and miR-423-3p levels, consistent with PD patient data. Transcriptional levels of CD14, CCL-20, TNF-α, IL-6, and TLR-2 in THP-1 cells were elevated, similar to effects seen in radiation-induced neuroinjury, indicating that during PD. The PPMOC we designed possesses many advantages, including high aspect ratio, increased cell culture density, user-friendly operation, and versatility. Additionally, the high aspect ratio microfluidics fabrication method provides excellent gas permeability, reduces bubble impact, requires minimal equipment, and offers cost and time advantages over other methods, making it suitable for basic laboratory settings and versatile for various research applications. The PPMOC and its fabricating method, which we developed, are expected to have broader applications in the future.

Author contributions

Yu Chen: investigation, conceptualization, formal analysis, data curation, writing – original draft. Zhirong Wan: methodology, data curation. Beiqin Liu: validation, investigation. Hong Ma: methodology, writing – review and editing. Shuyue Wang: validation, investigation. Yaoyuan Cui: methodology. Yulin Deng: resources, methodology, supervision, writing. Jichen Du: methodology, supervision, writing – review and editing. All authors read and contributed to the manuscript.

Conflicts of interest

There are no conflicts to declare.

Abbreviations

PPMOC

Plug-and-play multi-organ-chip

PD

Parkinson's disease

CNS

Central nervous system

BBB

Blood brain barrier

IL-6

Interleukin 6

TNF-α

Tumor necrosis factor-α

CCL-20

C-C motif chemokine ligand 20

CXCL-1

C-X-C motif chemokine ligand 1

PDMS

Polydimethylsiloxane

PMMA

Polymethyl methacrylate

6-OHDA

6-Hydroxydopamine hydrochloride

GelMA

Gelatin methacryloyl

TEER

Transendothelial electrical resistance

RT-qPCR

Quantitative reverse transcription polymerase chain reaction

SMG

Simulated microgravity

RPM

Random positioning machine

RCCS

Rotary cell culture system

PET

Polyethylene terephthalate

Data availability

Sufficient data are provided in the tables and figures, with additional data available upon request.

Supplementary information (SI): fabrication of the PPMOC insert, the cell images of GelMa⁺ and GelMa⁻ before and after staining, TEER measurement of the BBB in the PPMOC over time. See DOI: https://doi.org/10.1039/d6ra01858k.