Study design
This study aimed to develop a microphysiological system that mimics the human GBA to investigate systemic inflammation and neuropathogenesis associated with E. coli K1 meningitis. To achieve this, we created a three-compartment microfluidic platform using photolithography and soft lithography, allowing for the culture of five different types of cells spanning the human GBA, including gut epithelial cells (EP), brain endothelial cells (EC), neurons (Neu.), astrocytes (AC), and microglia (MG). The final platform consisted of three GBA microfluidic devices integrated into a glass slide at the bottom and an acrylic array at the top, serving as a medium reservoir (Supplementary Fig. 1c). This design enables three replicates within a single platform, supports long-term cell survival, and facilitates easy imaging. To assess the complex interplay of E. coli K1 infection in gut-blood-brain-barrier damage, a critical aspect of meningitis, we developed a gut-blood-brain-barrier model by co-culturing E. coli K1, gut EP, and brain EC in separate chambers. Then, soluble IL-6 was introduced to the brain EC chamber 24 hours after infecting the gut EP chamber with E. coli K1. This experimental approach was designed to mimic the peripheral immune responses that occur during the progression of bacterial meningitis. Specifically, the delay in IL-6 addition reflects the time-dependent activation of systemic inflammation, where gut-originated infections trigger the release of IL-6 into circulation. These cytokines subsequently contribute to the development of meningitis. Barrier damage was then identified through immunostaining of tight junction proteins. The fluorescent signal was normalized by the non-stimulated group to get fold changes.
To demonstrate detrimental neuroinflammation during E. coli K1 meningitis, we established an immune surveillance model by co-culturing human immortalized astrocytes and microglia in the CNS chamber with or without E. coli K1 infection. Immunostaining against neuroinflammatory markers, along with living cell assays and analysis of multiple cytokines, were utilized to evaluate inflammatory responses during E. coli K1 meningitis. To investigate the role of astrocytic IFN-γ on microglial-mediated neuroinflammation, soluble IFN-γ was introduced into single cultures of microglia 4 hours prior to stimulation with E. coli K1. This pre-treatment period was selected based on evidence that IFN-γ primes microglial activation, facilitating the upregulation of inflammatory signaling pathways and the release of pro-inflammatory mediators. The signals induced by this treatment were subsequently normalized to non-stimulated microglia to account for baseline activation. Next, to evaluate IFN-γ-impaired microglial phagocytosis of E. coli K1 and neurotoxic proteins, we co-cultured E. coli K1 and microglia with pHrodo-Aβ with/without prior stimulation with IFN-γ. Finally, we mimicked the neurons-glia interaction model by tri-culturing hNPC-derived neurons and astrocytes with immortalized microglia in the CNS chamber to identify the role of neuroinflammation in neurodegeneration during E. coli K1 meningitis. We used immunostaining against neurodegenerative markers and other living cell assays to characterize neuropathogenesis associated with E. coli K1 meningitis.
Preparation of Escherichia coli K1
Escherichia coli K1 (ATCC, Cat. 700973) was proliferated in Tryptic Soy Broth (Sigma-Aldrich, Cat. 22092-500 G). The culture was placed at 37 °C with shaking at 100 × rpm overnight. The bacterial culture was diluted 100-fold until the optical density at 600 nanometers wavelength attained a value of 1 under identical experimental conditions. Finally, the bacterial pellets were separated from the bacterial conditioned medium by centrifuging at 10,000 × g in 10 min.
Microfluidic chip fabrication
We utilized photolithography to fabricate the SU8 master mold and soft lithography methods to reconstruct a microfluidic platform comprising three chambers for reconstructing multi-organ interaction (Supplementary Fig. 1), by pouring a mixture of 10% (v/v) of polydimethylsiloxane base and 1% (v/v) of curing agent (Sylgard 184 A/B, Dow Corning, Midland, MI, USA) into the master mold. Then, the mixture is vacuumed for 20 min to completely remove all remaining bubbles, followed by curing at 60 °C for at least two hours for polymerization. Next, the PDMS sheets are removed from the mold and punched 2 mm diameter holes to make the inlets and outlets while the medium reservoirs are fabricated by laser cutting the 6 mm thick acrylic sheets (Zing 24, Epilog Laser, Golden CO., USA). Then, the PDMS and acrylics are bonded together by using a mixture of uncured PDMS and curing agents. The assembled platforms were placed at 60 °C for at least 4 hours for full binding. Lastly, the glass slides were bonded with the PDMS using oxygen plasma (PX-250, March Plasma System Petersburg FL, USA) with settings of 350 mW power for 2 min.
Gut epithelial cell preparation
The Caco-2 cell line (ATCC, Manassas, VA, USA, Cat. HTB-37), which is derived from human intestinal epithelial cells, was proliferated in Dulbecco’s Modified Eagle’s Medium-high glucose (Sigma-Aldrich, Cat. D5796-500ML). The cells were proliferated using the medium containing 10% (v/v) of fetal bovine serum from Sigma-Aldrich (Cat. F2442) and 1% (v/v) of Penicillin-Streptomycin (Lonza, Cat. 17-745E). Then, the Caco-2 cells were placed at 37 °C and 5% CO2 until reaching approximately 80% of confluency. The Caco-2 culture medium was exchanged every two days with fresh medium.
Brain endothelial cell preparation
Human immortalized brain endothelial cells (EC) were acquired from Cedarlane Laboratories (Ontario, Canada) and cultured in a T25 flask coated with 1% (v/v) of collagen. The cells were proliferated using the medium containing endothelial cell growth basal medium-2 (Lonza, Cat. 190860), 1% (v/v) of penicillin-streptomycin (Sigma-Aldrich, Cat. P4333), 1.4 μM hydrocortisone (Sigma-Aldrich, Cat. H0888-1G), 10 mg/mL L-Ascorbic acid (MedChem, Cat. HY-B0166G), 1% (v/v) of chemically defined lipid concentrate, 10 mM HEPES (Gibco-BRL, Cat. 15630-106), 20 ng/mL bFGF (Stemgent, Cat. 03-2002), and supplement with 5% v/v of FBS (Sigma-Aldrich, Cat. F2442). The cells were placed at 37 °C and 5% CO2 until reaching 80% of confluency. The brain endothelial cell culture medium was exchanged every two days with fresh media.
Proliferation of human microglial cells
The immortalized human microglia (SV40 cell line) were purchased from Applied Biological Material Inc. (ABM, Cat. T0251) and were grown in a T25 flask (SPL Life Sciences, Cat. 70012) containing 5 mL of microglial proliferation medium that contained 90% (v/v) of Pigrow III (ABM, Cat. TM003) and 10% v/v of FBS (Sigma-Aldrich, Cat. F2442), and 1% (v/v) of penicillin-streptomycin (Sigma-Aldrich, Cat. P4333). Then, the cells are maintained at 37 °C with 5% CO2. and the cell culture medium is exchanged every 2 days until it reaches 90% of confluency.
Proliferation of human astrocytes
The immortalized human astrocytes (SV40 cell line) were purchased from Applied Biological Material Inc. (ABM, Cat. T0280) and were grown in a 1% (v/v) of collagen-coated T25 flask (SPL Life Sciences, Cat# 70012) containing 5 mL of microglial proliferation medium that contained 90% (v/v) of Pigrow IV (ABM, Cat. TM004) and 10% (v/v) of FBS (Sigma-Aldrich, Cat. F2442), and 1% (v/v) of penicillin-streptomycin (Sigma-Aldrich, Cat. P4333). Then, the cells are maintained at 37 °C with 5% CO2. and the cell culture medium is exchanged every 2 days until it reaches 90% of confluency.
Proliferation and differentiation of human neural progenitor cells
Human neural progenitor cells (ReN) were purchased from EMD Millipore (Billerica, MA, US, Cat. SCC008) and cultured in a 1% (v/v) of Matrigel-coated T25 flasks (SPL Life Sciences, Pocheon, Korea, Cat# 70012). The 1% (v/v) of Matrigel coating solution is prepared by adding 10 μL of Matrigel (Corning, Cat# 356235) to 0.99 mL of DPBS (Sigma-Aldrich, Cat. D8537). The ReN cells are cultured in a proliferation medium containing DMEM/F12 (Gibco, Cat. 11320033), 0.1% (v/v) of Heparin (2 mg/mL, StemCell Technology, Cat. 7980), 2% (v/v) of B27 (Gibco, Cat. 17504044), and 1% (v/v) of Penicillin/Streptomycin (Sigma-Aldrich, Cat. P4333), bFGF (20 ng/ml, Stemgent, Cat. 03-0002) and hEGF (20 ng/ml, Sigma-Aldrich, Cat. SRP6253)88. The cells were maintained at 37 °C with 5% CO2 and the cell culture medium was exchanged every 2 days until it reached 90% confluency.
Next, the human neural progenitor cells were differentiated in a differentiation medium containing DMEM/F12 (Gibco, Cat. 11320033), 0.1% (v/v) of Heparin (2 mg/mL, StemCell Technology, Cat. 7980), 2% (v/v) of B27 (Gibco, Cat. 17504044), and 1% (v/v) of Penicillin/Streptomycin (Sigma-Aldrich, Cat. P4333) for 14 days to obtain neurons/astrocytes. The differentiated medium was exchanged every two days with fresh media.
Preparation of human GBA models
We reconstructed the gut-brain platform to study the pathogenesis of E. coli K1 meningitis-inducing systemic inflammation and neuropathogenesis. We coated microfluidic chambers with 10 μL of Poly D-lysine (Sigma-Aldrich, Cat. A-003-M), then placed at RT for 20 min. The CNS chamber was incubated with 1% (v/v) Matrigel solution at 37 °C for 30 min. Then, we loaded 10 μL of ReN cells (108 cells/mL) to the CNS chamber and placed the platform at 37 °C and 5% CO2 for 30 min for attachment. Next, we added 100 μL of fresh medium into each reservoir and exchanged every other day for 14 days to get neurons/astrocytes. Prior to gut EPs and brain ECs loading, we coated the gut chamber and BBB chamber with 10 μL of 2 mg/mL collagen Type I at 37 °C for 30 min. Then, we seeded 10 μL of gut EPs (108 cells/mL) and brain ECs (107 cells/mL) to each chambers and placed the platform in a 5% CO2 incubator at 37 °C for at least 5 days to get the monolayer of gut EP and brain EC. Two days prior to infection, we added immortalized microglia (105 cells/mL) to the CNS to recreate neuron-glia interaction. After that, we infected E. coli K1 to the gut EP chamber and placed it in a 5% CO2 incubator at 37 °C for an additional 2 days (Supplementary Fig. 1d).
The microfluidic devices were maintained under semi-dynamic conditions, enabling passive media to flow through a hydrostatic pressure gradient without requiring an external pump (Supplementary Fig. 1e). This gradient was generated by adding 200 μL of fresh medium to the inlet reservoir and 100 μL to the outlet reservoir of each device. Consequently, media flowed sequentially from the inlet through the main chambers of the gut EP, brain EC, and CNS compartments, as well as the interconnecting microchannels, effectively mimicking the physiological communication between these systems. This setup mimics the physiological communication between these compartments. This passive flow ensured continuous nutrient delivery and waste removal for the culture cells.
Preparation of GBBB models
To create GBBB model, we coated microfluidic chambers with 10 μL of Poly D-lysine (Sigma-Aldrich, Cat. A-003-M), incubated at RT for 20 min, and washed with PBS 1X (HanLab, Cat. K19274065). We seeded 10 μL of gut EPs (108 cells/mL) and brain ECs (107 cells/mL) to each chambers and placed the platform in a 5% CO2 incubator at 37 °C for at least 5 days to get the monolayer of gut EP and brain EC, then we infected E. coli K1 to the gut EP chamber.
Preparation of central immune surveillance models
To create immune surveillance-on-chip, we coated microfluidic chambers with 1% v/v of Matrigel and incubated at 37 °C for 30 min. Then, we loaded 10 μL of immortalized microglia/astrocytes (107 cells/mL) and incubated the devices in a 5% CO2 incubator at 37 °C for 1 hour for cell attachment. Next, we added 100 μL of fresh medium to each reservoir. Prior to infection, we exchanged medium without antibiotics and added E. coli K1 directly to the CNS chamber and incubated in a 5% CO2 cell culture incubator at 37 °C.
Preparation of neuron-glial interaction models
To create neuron-glia interaction-on-chip, we coated microfluidic chambers with we coated microfluidic chambers with 1% v/v of Matrigel and incubated at 37 °C for 30 min. Afterward, we loaded 10 μL of ReN cells (108 cells/mL) into the CNS chamber and incubated the platforms at 37 °C and 5% CO2 for 1 hour. Finally, we added 100 μL of fresh medium and exchanged every two days for 14 days. Two days prior to infection, we added immortalized microglia to the CNS to recreate neuron-glia interaction. After that, we infected E. coli K1 directly to the CNS chamber and placed at 37 °C and 5% CO2.
Assessment of microglial phagocytosis assay
The microglia SV40 cells were proliferated in a T25 flask until they reached 90% confluency, as described previously. Then, the cells were detached by incubating with Trypsin EDTA (Sigma, Cat. T3924) at 37 °C and 5% O2 for 2–3 min. Then, the cells were harvested by centrifuging at 1300 × rpm in 3 min and placed in a mixture of 1 mL of Diluent C and 2 μL of the green-fluorescent dye (Sigma-Aldrich, Cat. PKH67GL). The mixture was then placed at room temperature for 5 min and centrifuged at 1300 × rpm for 3 min. The microglia cells were captured in the FITC channel (Alexa488). The Escherichia coli K1 were grown in Tryptic Soy Broth (Sigma-Aldrich, Cat. 22092-500 G) overnight at 37 °C. Then, the cells were harvested by centrifugation at 4500 × rpm for 10 min and were promptly resuspended into a dye solution containing 1% (v/v) of BactoView™ Live red dye (Biotium, Cat. 40101-T). The mixture was placed at 37 °C for 30 min in dark, then the cells were harvested by centrifugation and resuspended into growth medium according to experimental purpose. The bacterial cells were captured in the TRITC channel (Alexa 594).
ROS assessment
ROS was assessed by dichlorofluorescein diacetate (H2DCFDA) (ThermoFisher, Cat. D399) according to the provided guidance from the manufacture. Briefly, the cell samples were incubated with fresh medium containing the 5 μM of ROS indicator for 30 min in the dark at 37 °C and 5% CO2. Then, fresh medium was added into the cells to incubate for an additional 30 min at 37 °C and 5% CO2. Finally, ROS signals were captured using a fluorescence microscope (Nikon TiE microscope, Nikon with FITC channel (Alexa488).
NO assessment
We used DAF-FMTM diacetate (ThermoFisher, Cat. D-23844) to measure NO according to the manufacturer’s protocol. Particularly, the cells were placed with 10 nM DAF-FMTM diacetate in fresh medium at 37 °C and 5% CO2. Then, the NO indicator solution was removed and replaced by fresh medium for an additional incubation at 37 °C with 5% CO2 for 30 min. Lastly, the NO signals were imaged by fluorescence microscope (Nikon TiE microscope, Nikon) with FITC channel (Alexa488).
Multiple cytokines assessment
A multiplex cytokine array kit (R&D Systems, Cat. ARY005) was used to assess the expression of various cytokines and chemokines. Briefly, the cell culture-conditioned medium was harvested and centrifuged to remove the remaining cell pellets. Then, a mixture of conditioned medium and biotinylated detection antibodies was incubated for at least 2 hours while the nitrocellulose membranes were blocked with a blocking buffer. Then, the prepared mixture was added to the blocked nitrocellulose membranes, allowing binding of the target proteins and immobilized capture antibodies. The arrays were placed at 4 °C on a shaker for at least 6 hours and gently washed at least three times at 10-min intervals to remove any unbound antibodies. Chemiluminescent detection solution and streptavidin-horseradish peroxidase were used to assess the intensity signal that is proportional to the level of proteins in the conditioned medium. Then, we used ImageJ software (Wayna Rasband, NIH) to quantify the results.
Calcium imaging
The cell samples were incubated with 5 μM of the calcium-sensitive dye Rhod2-AM (Abcam, Cat. 142780) at 37 °C in the dark for 30 min, following the manufacturer’s guidance. Then, the cells were gently washed twice with calcium-free HBSS and imaged using fluorescence microscope in FITC channel. The fluorescence intensity was normalized to the baseline value F0 and reported as the mitochondrial calcium concentration.
Immunostaining
Cells were fixed with 4% paraformaldehyde (Biosesang, Cat. PC2031-100-00) for 30 min, then washed with phosphate-buffered saline containing 0.1% (v/v) of Tween®20 to remove the excessing paraformaldehyde. Next, the fixed cells were treated with 0.1% (v/v) of Triton-X 100 in PBST for 30 min at and blocked with 3% (v/v) of BSA for 1 hour. Primary antibodies were diluted to an appropriate concentration and treated to the samples overnight at 4 °C (as detailed in Supplementary Table 1). Subsequently, the samples were treated with secondary antibodies diluted 1:200 and DAPI for 2 hours. Finally, the samples were washed five times with PBST before imaging using Nikon fluorescence microscope. The intensity of immune reactivity was quantified with ImageJ software (Wayna Rasband, NIH).
Statistical analysis
All the data were relatively normalized by the control group and reported as mean ± standard deviation. The number of replicates was provided in the description of figure captions. The statistical differences between were analyzed by unpaired t-test and One-way ANOVA followed by Tukey’s post-hoc test using SPSS software (IBM, NY, US). P value less than 0.05 was considered statistically significant. The symbols NS, *, **, and *** denoted no significance, p value < 0.05, p value < 0.01, p value < 0.001, and p value < 0.0001 respectively.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
