Engineering biomimetic tissue barrier models on chips: From design and fabrication to applications in disease modeling and drug screening

Abstract

Replicating the in vitro properties of tissue barriers-such as the blood-brain barrier, gut, skin, lung, kidney, retina, nasal epithelium, and placenta-is crucial for many applications, including drug screening, studying molecular transport, drug delivery, and disease modeling in preclinical studies. Organ-on-a-chip (OoC) platforms are advanced three-dimensional (3D) in vitro models that aim to replicate various aspects of organ functionality within microfluidic systems by providing microenvironments akin to native tissue. When used to model the interface between two different tissue compartments, OoC technology offers a promising platform for more accurately replicating the physiology and pathophysiology of various tissue barriers in the body. This review focuses on the state-of-the-art biomimetic tissue barrier models, ranging from two-channel tissue barrier-on-a-chip systems with a thin porous membrane to hydrogel-based membrane models. Specifically, it explores the engineering of tissue barrier-on-a-chip platforms, highlighting various fabrication techniques for microfluidic chips and membranes, as well as methods for functional characterization of the engineered tissue barriers. Additionally, we discuss the development of organ-specific barrier models and multi-organ-on-a-chip systems for studying inter-organ communication. Finally, we highlight the current challenges in the field and future directions in advancing tissue barrier modeling using OoC technology.

Keywords: Disease modeling; Drug screening; Drug transport; Membrane; Microfabrication; Microfluidics; Organ-on-a-chip; Tissue-tissue interface.

Figure 1.

Schematic overview of the tissue barrier-on-a-chip concept, highlighting key components such as the cellular toolbox, organ-on-a-chip functional units, and different tissue barrier models. These include membrane-based systems with a thin porous membrane separating two channels and hydrogel-based tissue barrier models. The illustration also highlights the ultimate goal of organ-on-a-chip technology in advancing the 3Rs principle (Replace, Reduce, and Refine) to minimize animal use in preclinical research. Created with BioRender.com.

Figure 2.

Fabrication approaches for engineering microfluidic chips and assembly of a membrane in tissue barrier-on-a-chip. A) Photolithography steps, reproduced with permission from Ref. (120), Copyright Royal Society of Chemistry. B) 3D printing procedures for chip fabrication, reproduced with permission from Ref. (121), Copyright Elsevier, Under a Creative Commons license. C) Hot embossing for chip fabrication and assembly, reproduced with permission from Ref. (52), open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license. D) Tape-based chips for PC membrane assembly. Reproduced with permission from Ref. (119), Copyright Royal Society of Chemistry.

Figure 3.

Different examples of BBB-on-a-chip. A) The BBB model by Park et al. i. Photograph (left), schematic (center), and immunofluorescence images (right) of a 2-channel microfluidic Organ Chip. iPS-BMVECs line the basal vascular channel, while primary astrocytes and pericytes are on the upper membrane surface in the apical parenchymal channel. Top right: z-stack of pericytes (yellow, F-actin) and astrocytes (white, GFAP) in the top channel. Bottom right: side view of BMVECs (blue, ZO-1) in the vascular channel. ii. Impedance values of the human BBB in the TEER chip over 7 days, measured across 0.1 Hz–100 kHz, with iPS-BMVECs differentiated under hypoxia (closed circles) or normoxia (open triangles). Reproduced with permission from Ref.(158) Copyright Springer Nature, Creative Commons CC BY license. B) The iPSC-based BBB model presented by Vatine et al., i. Schematic of whole human blood perfusion through the iPSC-based BBB-Chip. ii. Blood flows at 3,600 μL/h through the vascular channel, restricted by iBMECs. Blood diffuses into the brain side when neural cells are cultured alone or with TNF-α (10 ng/mL green arrows). iii. Fluorescent dextran-FITC permeability after overnight perfusion shows low diffusion with iBMECs (white bar) but increases with neural cells alone (gray bar) or TNF-α treatment (black bar). iv. Immunocytochemistry of βIII-tubulin+ neurons (red) and GFAP+ astrocytes (green) shows neural protection with iBMECs but reduction without them. Scale bar, 200 μm. v. LDH quantification confirms iBMECs protect against blood-induced cytotoxicity (white bar), while absence of iBMECs (gray bar) or TNF-α treatment (black bar) increases LDH release. Reproduced with permission from Ref. (130) Copyright Elsevier. C) The BBB-on-a-chip model presented by Matthiesen et al. integrated with a sensor for barrier integrity assessment. i. Photograph of the hiBBB-on-chip system showing vascular (magenta) and perivascular (green) compartments perfused with colored solutions above and below the permeable membrane with interdigitated gold electrodes. ii–iii. Barrier permeability data by condition. Time-resolved normalized ECIS R, with SIN-1 (+) conditions present only for transwells. Reproduced with permission from Ref.(35) Copyright John Wiley & Sons, under the Creative Commons CC-BY-NC license. D) 3D hydrogel-based BBB-on-a-chip model developed by Campisi et al., i. (left) Schematic of the BBB and (right) Diagram of the 3D BBB microvascular network (μVN) model replicating brain microvasculature. Confocal image of the self-assembled μVN model with iPSC-ECs (CD31, green), PCs (F-actin, red), ACs (GFAP, magenta), and nuclei (DAPI, blue). ii. Microfluidic device with gel, cells, and media loading, along with a photo of the device. Reproduced with permission from Ref. (164), Copyright Elsevier.

Figure 4.

(A) The vessel-on-a-chip model presented by Tang et al. (176) i. 3D depiction of microcapillary vessel replication in glass/Si microfluidic chip using a SixNy nanoporous membrane, separating endothelial cells and pericytes. (ii) Exploded diagram of the OOC system featuring wet-etched channels in top and bottom glass layers, a silicon frame with a Si_xN_y membrane, and a silicone rubber sealing layer. (iii-v) Microscopic images show MMECs on a gelatin-coated SixNy nanoporous membrane within a closed chip, demonstrating cell growth and detail. (iii) MuMEC cells form a uniform monolayer shortly after self-loading seeding. (iv) One day after pre-cultivation, they maintain even coverage, supported by capillary forces and spacious chip inlets. (v) Close-up views show the Si_xN_y membrane’s nanopore regions and provide additional cell details. Reproduced with permission from Ref.(176) Copyright, Frontiers Media S.A. Under the Creative Commons Attribution License (CC BY). B) vessel-heart tissue barrier model integrated by TEER-MEA chip presented by Maoz et al. (114), i) different components of the chip, the layers include a platinum (Pt) MEA connector layer (gray) for electrical signal transmission, a gold TEER electrode layer for resistance measurement, and a microfluidic chip layer positioned on top of the MEA electrodes, ii) TEER and MEA setup schematics, iii) The comparison of the impedance value for tissue barrier in normal condition vs treated with TNF-α. Reproduced with permission from Ref.(114) , Copyright the Royal Society of Chemistry.

Figure 5.

Gut-on-chip modeling. (A) Different types of gut-on-chip models: i) 2D culture models with upper and lower channels separated by a permeable membrane. ii) Scaffold-based 3D model with villi-like structures. iii) Stretching-based 3D model to stimulate gut cells. (B) The GoC model presented by Kim et al. (196) i) Device with microfluidic channels. ii) Micrograph of intestinal structures. iii) Confocal image of intestinal villi. Reproduced with permission from Ref.(196) Copyright Royal Society of Chemistry. (C) The GoC model presented by Jalili-Firoozinezhad et al. (203) i) Cross-section images of villi-like structures before and after radiation. ii) Villus height changes post-radiation. iii) Villus height distribution. iv) Changes in apparent permeability coefficient (P_app) post-radiation. Reproduced with permission from Ref.(203), Copyright Springer Nature. (D) The presented model by Fadel et al. (204), i) Morphological comparison of healthy and intestines with EED under normal and deficient conditions. ii) Cross-section images of villus-like structures. iii) Villus height differences. iv) P_app permeability coefficient differences. Reproduced with permission from Ref(204), Copyright Springer Nature.

Figure 6.

Skin-on-a-chip modeling. (A) The presented model by Barros et al. (216), schematic illustrating the creation of (i) a 3D-printed HUVEC layer, 3D hDFs in hydrogels, and multi-layered human keratinocytes (hKCs). (ii) A representative optical image showing the location imaged in 3D using a confocal microscope, along with a schematic of the skin-on-a-chip fabrication process. (iii) Confocal imaging at the same location displaying different 3D skin equivalent layers from (ii), with HUVECs in blue, hDFs in red, and hKCs in green. (iv) A representative 3D cross-sectional confocal image illustrating the structure of the fabricated skin layers. Reproduced with permission from Ref. (216), Copyright IPO Science. (B) The presented model by Lee et al. (220), i. Diagram of the device featuring three parallel cell culture chambers and TEER measuring electrodes. ii. The image of the fabricated chip is shown in the inset. iii. TEER values dropped to 82% of their original value after 30 minutes of UV irradiation. iv. Tight junction staining appeared as dotted filaments, in contrast to the more continuous and solid filaments observed before UVB irradiation. v) (* P < 0.05). Reproduced with permission from Ref. (220), Copyright Royal Society of Chemistry.

Figure 7.

(A) i-iii Exploded view of a lung-on-a-chip proposed by Sellgren et al. (238) using PTFE and PET membranes (SEM images). Reproduced with permission from Ref. (238), Copyright The Royal Society of Chemistry. (B) A lung-on-a-chip model proposed by Huh et al. (229) fabricated using a PDMS membrane and soft lithography. Three PDMS layers are aligned and permanently bonded to create two sets of three parallel microchannels, separated by a 10 μm-thick PDMS membrane with an array of through-holes, each with an effective diameter of 10 mm. A vacuum is applied to these chambers to induce mechanical stretching. Images of the actual lung-on-a-chip microfluidic device are shown. Reproduced with permission from Ref. (229), Copyright the American Association for the Advancement of Science (AAAS). (C) The proposed lung-on-a-chip model by Si et al. (239), (i) Schematic of a cross-section of the Airway Chip. (ii) Immunofluorescence staining of the human Airway Chip showed a pseudostratified epithelial layer consisting of CK5+ basal cells and β-tubulin IV+ ciliated cells. DAPI was used to stain the nuclei. (iii) Fold change in TMPRSS2 mRNA levels in well-differentiated primary human airway epithelium on the chip compared to undifferentiated cells. (iv) Immunofluorescence images showing ACE2 receptor expression in well-differentiated primary human airway epithelium on the chip, compared to the same cells before differentiation. Blue indicates DAPI-stained nuclei, and green indicates ACE2 expression. Scale bars: 50 μm. Reproduced with permission from Ref. (239), Copyright Springer Nature.

Figure 8.

A) i. This figure illustrates the replication of the glomerulus in a Tissue Glomerulus Chip (TGC) presented by Pajoumshariati et al. (249) with schematics depicting the glomerulus’s key elements and cell interactions. It also includes images of an acellular mesangium, the TGC with cell markers, and cross-sections of acellular and cell-seeded chips. ii) The images show top views of cell-seeded chips under light and fluorescent microscopy, with a marked lumen. Scale bars measure 200 μm. Reproduced with permission from Ref. (249), Copyright John Wiley & Sons, B) i. Physiology of the proximal tubule and the schematic of proximal tubule-on-a-chip model presented Asif et al. (253), ii) Tissue chip components, Reproduced with permission from Ref. (253) Copyright Springer Nature, C. i. Illustration of glomerular barrier and Glomerulus-on-a-chip microdevice proposed by Wang et al. (257). ii) Time-lapse images show glomerular filtration barrier (GFB) permeability using fluorophore-labeled IgG tracers. Red TRITC-IgG was added to the capillary channel with the GFB, and green FITC-IgG to the capsule channel without a cell layer, both perfused at 2.5 μl/min. iii. GFB permeability coefficients were compared using the ratio of fluorescent dye diffusion distance to time. Reproduced with permission from Ref. (257), Copyright the Royal Society of Chemistry.

Figure 9.

A) Retina tissue barrier presented by Chen et al. (267), (i-iii) The external Blood-Retina Barrier-on-a-chip consists of ARPE-19 and HUVEC monolayers, which are divided by an artificial membrane (Reproduced with permission from Ref. (267), Copyright Springer Nature. B) i. Exploded view of retinal explant RoC presented by Dodson et al. (269), ii) (a-c) GFP+ microglia in the retina of CX3CR1-GFP mice, exposed to culture media, are imaged over 30 minutes. Microglia in the through-hole (circled) shows no directional migration. ONH denotes the optic nerve head. (d) GFP+ microglia surrounding an access point (Region 1) before LPS delivery. (e) Microglia activation and migration to the access point at 1 minute, and (f) 30 minutes post-LPS delivery. Solid lines mark the microfluidic channel edge; dotted lines show the four sampling region boundaries. Reproduced with permission from Ref. (269), Copyright Springer Nature. C. The proposed model by Chuchuy et al. (60), (i) Photograph of the chip and its components, including the DAPI channel (scale bar: 50 μm). (ii) Schematic of all chip layers. (iii) Electrospinning process to create a flat membrane above the substrate. (iv-vii) Live/Dead assay of human-derived mMVECs and hiPSC-derived RPEs seeded in PLA-only membranes and PLA-GM₂ membranes 2 days post-implant. Reproduced with permission from Ref. (60), Copyright American Chemical Society.

Figure 10.

Nasal Epithelium-on-a-Chip: (A) The proposed model by Gholizadeh et al. (274), (i) Device structure with donor and acceptor channel inlets/outlets. (ii) Cross-section schematic showing RPMI 2650 cells at the air-liquid interface on a permeable membrane. TEER values for cells exposed to IBU-CS-β-GP aerosol at (iii) 1.7 L/min and (iv) 0.5 L/min, and IBU-HBSS aerosol at (v) 0.5 L/min. TEER measurements: before aerosol deposition (TEER1), after deposition and 4-hour pulsatile fluid flow incubation (TEER2), and after overnight growth media incubation (TEER3), with significance indicated (*, p < 0.05; **, p ≤ 0.01). (vi) Microscopic images of cells exposed to 0.5 or 1.7 L/min aerosol flow versus control, showing nuclei (blue), live (green), and dead cells (red). Scale bars: 200 μm. Reproduced with permission from Ref. (274), Copyright Elsevier. (B) The proposed model by Na et al. (276), (i) Nasal mucosa schematic with epithelial and endothelial cells, basement membrane, and gland. Air–liquid interface on the nasal epithelium. (ii) In vitro culture: nasal cells on ECM, the air-liquid interface on the nasal channel. (iii) Microfluidic chip: illustration and dimensions. (iv) In vitro nasal mucosa under air-interface culture: hNECs on PDMS walls, stained and imaged (scale bar: 50 μm). (v) The TEER values graph shows lower permeability at the air-liquid interface. Phase-contrast images of co-cultured cells under different conditions (scale bar: 200 μm). Error bars represent standard deviation (*p < 0.05, **p < 0.01, ***p < 0.001; n = 4 for TEER measurements). Reproduced with permission from Ref. (276), Copyright The Royal Society of Chemistry.

Figure 11.

A) The proposed multi-tissue barrier on a chip by Herland et al. for using the fluidically coupled vascularized OoC system allows for the quantitative prediction of how humans will respond to drugs in terms of pharmacokinetics reproduced with permission from Ref. (11), Copyright Springer Nature. B) The MoC platform proposed by Maoz et al. BBB-Brain-BBB on chips models, reproduced with permission from Ref. (287), Copyright Springer Nature.