A Versatile Intestine-on-Chip System for Deciphering the Immunopathogenesis of Inflammatory Bowel Disease

Abstract

The multifactorial nature of inflammatory bowel disease (IBD) necessitates reliable and practical experimental models to elucidate its etiology and pathogenesis. To model the intestinal microenvironment at the onset of IBD in vitro, it is important to incorporate relevant cellular and noncellular components before inducing stepwise pathogenic developments. A novel intestine-on-chip system for investigating multiple aspects of IBD's immunopathogenesis is presented. The system includes an array of tight and polarized barrier models formed from intestinal epithelial cells on an in-vivo-like subepithelial matrix within one week. The dynamic remodeling of the subepithelial matrix by cells or their secretome demonstrates the physiological relevance of the on-chip barrier models. The system design enables introduction of various immune cell types and inflammatory stimuli at specific locations in the same barrier model, which facilitates investigations of the distinct roles of each cell type in intestinal inflammation development. It is showed that inflammatory behavior manifests in an upregulated expression of inflammatory markers and cytokines (TNF-α). The neutralizing effect of the anti-inflammatory antibody Infliximab on levels of TNF-α and its inducible cytokines could be explicitly shown. Overall, an innovative approach to systematically developing a microphysiological system to comprehend immune-system-mediated disorders of IBD and to identify new therapeutic strategies is presented.

Keywords: immunopathogenesis; inflammatory bowel disease; innate immunity; intestine-on-chip systems.

Figure 1

Schematic representation of key structural and cellular components of a) the in vivo IEB and b) an in vitro IEB model formed within the MultiU‐Int microfluidic chip.

Figure 2

a) Schematic representation of the individual layers of the MultiU‐Int microfluidic chip, their arrangement, and materials. The top layer was a commercially available, multi‐well slide made from polystyrene plastic material (ibidi sticky‐slide 18 well). The middle layer of the chip was fabricated by sandwiching CCC strips between two pressure‐sensitive adhesive (PSA) foils. The upper foil (①) featured oval areas where fibroblasts and, later, immune cells were allowed to interact with the IEB model (fibroblast seeding areas). The lower foil (②) featured the same pattern of channel structures that were hot‐embossed into the bottom layer of the chip. The bottom layer was fabricated by hot embossing the elastomer Flexdym. b) Schematic cross‐sections of the MultiU‐Int microfluidic chip showing details of the layer alignment: i) bottom view, and ii) side view. c‐i) Top view, and c‐ii) side view photographs of the MultiU‐Int microfluidic chip. For visualization, apical channels and their reservoirs were filled with red fluid, and the basal compartments were filled with blue fluid. The cell seeding areas on the basal side are marked with white dashed lines. Scale bars: 10 mm.

Figure 3

a) Experimental timeline for the formation of the IEB model on chip. b) Characterization of on‐chip IEBs formed under (i) static and (ii) dynamic culture conditions for 5 d. The IEB model with well‐established E‐cadherin and ZO‐1 networks was obtained only under dynamic culture conditions, as shown by z‐projected confocal images of 40–50 µm‐thick z‐stacks. Scale bars: 100 µm. c‐i) A z‐cross section of a polarized IEB model – formed under dynamic culture conditions for 5 d – shows cells with increased height and polarized localization of E‐cadherin and ZO‐1 proteins. Scale bar: 50 µm. ii) Densely packed and well‐defined microvilli, shown by F‐actin staining, were observed on the apical surface of the on‐chip IEB model. Scale bars: 10 µm. iii) Fibroblasts were observed near the IEC layer (marked with white dashed lines). Scale bars: 50 µm. d) IF staining showing the expression of i) enterocyte (ALPi) and (ii) goblet cell markers (membrane‐bound mucus, MUC2). Scale bars: 100 µm. e) Permeability of the IEB model to two fluorescence tracers on day 5, represented as Papp values (n  =  12 basal compartments of 3 individual IEB models). Given values include means ± standard deviation (SD). Statistical analysis by two‐way ANOVA with Tukey‐Kramer post‐hoc test (ns: not significant, ****: p < 0.0001).

Figure 4

a) Experimental timeline showing the time point and site of LPS administration. b) TLR4 expression in IECs on day 7 i) without and ii) with LPS exposure (10 ng mL⁻¹ LPS), shown by z‐stack projection. Scale bars: 100 µm. c) Internalization of FITC‐labeled LPS by IECs (arrows) shown by xy‐ (largest insert), xz‐, and yz‐ cross‐sections of a z‐stack. Cell borders were visualized by F‐actin staining. Scale bars: 50 µm. d) IF staining of the IEB model at day 7 under different conditions: i) control IEB model without apical LPS with 3D cellular organization (arrow) and well‐established ZO‐1, ii) IEB with 10 ng mL⁻¹ LPS on the apical side. The LPS‐treated IEB model was underdeveloped in 3D, and giant, multinuclear cells were observed within the barrier while IECs within the control barrier appeared to have small and even cell size (dashes boxes). Scale bars: 100 µm. e) IEB model permeability for 70 kDa RITC‐dextran at day 7, shown as Papp values (n  =  3 basal compartments of 3 individual IEB models). Values are means ± SD. Statistical analysis by one‐way ANOVA with Tukey–Kramer post hoc test (**: p < 0.01).

Figure 5

Versatile experimental setup using the MultiU‐Int microfluidic chip. a) Different immune cell populations (i.e., MNs, MFs, and iDCs) were co‐cultured with a single on‐chip IEB model at separate locations, allowing for side‐by‐side comparison of cell type‐specific responses to the same stimuli. b) Similar IEB model‐immune cell co‐cultures were exposed to different stimuli on the same chip, allowing for the scrutiny of distinct cellular behaviors in response to different stimuli. c) Experimental timeline showing the time point and site of immune cell loading, triggering of inflammation, and therapy administration.

Figure 6

a) Histogram plot showing relative changes in CD64 expression of i) MNs, ii) MFs, and iii) iDCs in on‐chip IEB model‐MNP co‐cultures. b‐i) MN‐derived, ii) MF‐derived, and iii) iDC‐derived cytokine profiles at day 7 (n  =  3). Data is represented as x‐fold change with respect to the baseline of the “no‐MNP” control. Data was acquired from three individual basal compartments (n = 3) of one representative experiment out of a total of three independent experiments. c) Cell compositions (shown as the proportion of cells that were positive for CD4, CD14, and CD209, respectively) within PBMC populations in on‐chip IEB model‐PBMC co‐cultures under different conditions (n  =  4). Data was acquired from four individual basal compartments of the same IEB model (n = 4). d) Changes in levels of representative cytokines (TNF‐ α, GM‐CSF, and IL‐6) detected in on‐chip IEB model‐PBMC co‐cultures (n  =  4). Values shown in b) and d) are mean ± SD and statistically compared by two‐way and one‐way ANOVA with Tukey‐Kramer post‐hoc test, respectively (ns: not significant, *: p < 0.05, **: p < 0.01, ***: p < 0.001, and ****: p < 0.0001). e) Schematic illustration summarizing key inflammatory chemo/cytokine signaling of different immune cell subsets on chip and the proposed mechanism of action of Infliximab (Solid lines indicate direct effects, while dashed lines indicate an indirect effect of Infliximab.).