A three-dimensional intestinal tissue model reveals factors and small regulatory RNAs important for colonization with Campylobacter jejuni

Abstract

The Gram-negative Epsilonproteobacterium Campylobacter jejuni is currently the most prevalent bacterial foodborne pathogen. Like for many other human pathogens, infection studies with C. jejuni mainly employ artificial animal or cell culture models that can be limited in their ability to reflect the in-vivo environment within the human host. Here, we report the development and application of a human three-dimensional (3D) infection model based on tissue engineering to study host-pathogen interactions. Our intestinal 3D tissue model is built on a decellularized extracellular matrix scaffold, which is reseeded with human Caco-2 cells. Dynamic culture conditions enable the formation of a polarized mucosal epithelial barrier reminiscent of the 3D microarchitecture of the human small intestine. Infection with C. jejuni demonstrates that the 3D tissue model can reveal isolate-dependent colonization and barrier disruption phenotypes accompanied by perturbed localization of cell-cell junctions. Pathogenesis-related phenotypes of C. jejuni mutant strains in the 3D model deviated from those obtained with 2D-monolayers, but recapitulated phenotypes previously observed in animal models. Moreover, we demonstrate the involvement of a small regulatory RNA pair, CJnc180/190, during infections and observe different phenotypes of CJnc180/190 mutant strains in 2D vs. 3D infection models. Hereby, the CJnc190 sRNA exerts its pathogenic influence, at least in part, via repression of PtmG, which is involved in flagellin modification. Our results suggest that the Caco-2 cell-based 3D tissue model is a valuable and biologically relevant tool between in-vitro and in-vivo infection models to study virulence of C. jejuni and other gastrointestinal pathogens.

Fig 1. Set-up of a dynamically cultured Caco-2 cell-based 3D small intestine model.

(A) The ECM SISmuc scaffold is generated by decellularization of a porcine jejunal segment with sodium deoxycholate, DNase digestion, and gamma radiation. (B) The SISmuc is fixed in between two metal rings (cell crown). The former mucosal side faces the apical compartment and is reseeded with Caco-2 cells and cultured for up to 28 days either statically or dynamically. (C) Simulation of fluid dynamics over the surface of a cell crown during orbital shaking with 65 rpm (clockwise rotation). (D) FITC-dextran permeability assays (FDPA) of Caco-2 cells reseeded on 2D-Transwells (0.4 μm pore size) or on SISmuc over 28 days of culture under static and dynamic conditions. FDPA values represent the percentage of input FITC-dextran that has diffused to the basolateral compartment. Error bars indicate standard deviations (SDs) of three independent biological replicates. Statistical analysis is indicated for permeability differences between the statically and dynamically cultured tissue models. **: p < 0.01, *: p < 0.05, using Student’s t-test.

Fig 2. Dynamic culture enhances architecture and localization of cell-cell junction proteins of the Caco-2 tissue model.

(A) Hematoxylin and Eosin (H&E) staining of paraffin sections of human small intestinal tissue (left panel) or Caco-2 cells on either 2D-Transwell (TW) inserts (middle panel) or SISmuc after 21 days in static or dynamic culture (right panel). Images below are higher magnifications of the center micrographs. Scale bars: 200 μm. (B, C) Confocal microscopy images of human small intestine (left panel), Caco-2 cells on TW (middle panel), as well as statically or dynamically cultured 3D tissue models (right panel). Tissues were stained with DAPI (nuclei, blue) and antibodies against KRT18 (cytokeratin 18, magenta), E-cadherin (AJ, green) (B), and occludin (TJ, green) (C). Scale bars: 10 μm.

Fig 3. The 3D tissue model is colonized and disrupted by C. jejuni in a strain-specific manner.

(A) CFU quantification (as percent of input) of C. jejuni strains NCTC11168 and 81–176 from 24–120 hours p.i. in the static (upper panel) and the dynamic (lower panel) 3D tissue model. Error bars indicate SDs of four biological replicates. (B) FDPA-based measurements of epithelial barrier disruption during C. jejuni infections of 3D tissue models cultured statically (upper panel) or dynamically (lower panel). Depicted is the mean of four independent experiments with corresponding SDs. Mock indicates non-infected controls. FDPA values are depicted as fold changes relative to the value at time point zero. Based on these fold changes, statistical significance was calculated between the two wild-type strains for each time point, as well as between NCTC11168/81-176 and the non-infected control at 24 hrs p.i. ****: p < 0.0001, **: p < 0.01, *: p < 0.05, ns: not significant, using Student’s t-test. (C) Confocal microscopy images of paraffin sections of the 3D tissue model cultured dynamically during infection with C. jejuni strain 81–176 (upper panel) or NCTC11168 (lower panel) and respective non-infected controls. Bacteria were visualized with an anti-C. jejuni antibody (green), nuclei were stained with DAPI (blue), and AJs were imaged using an anti-E-cadherin antibody (adherens junctions, magenta). Scale bars: 10 μm.

Fig 4. C. jejuni transmigration across the 3D tissue model is delayed compared to 2D-Transwell infections.

(A) Caco-2 cells were grown statically on 2D-Transwells (pore size 3 μm) (upper panel) or dynamically on SISmuc scaffold (lower panel). Transmigration of C. jejuni wild-type strains 81–176 and NCTC11168 and their respective ΔflaA deletion mutant strains (ΔflaA) was determined by isolating CFUs up to six hours p.i. from the basolateral compartments. Experiments were conducted four times and CFUs are depicted as the percentage of input CFUs. Graphs represent the mean value with corresponding SDs, where significance was calculated for the recovered CFUs between the two wild-type strains at each time point. (B) CFUs from the basolateral compartment of non-reseeded 2D-Transwell (TW) and SISmuc from 10 min to 180 min p.i. (upper panel) with C. jejuni wild-type strains NCTC11168 and 81–176. For either wild-type strain, CFUs from the basolateral compartment of the 2D-Transwell were compared to those recovered from the basolateral compartment of the 3D tissue model. Asterisks at the early time points (10 min and 20 min p.i.) indicate a significant difference in transmigration between 2D-Transwell and 3D tissue model for both NCTC11168 and 81–176 (upper panel), as well as between NCTC11168 and 81–176 in the 3D tissue model only (lower panel). Early time points (10 min and 20 min p.i.) are depicted separately as a bar graph (lower panel) for better resolution. Experiments were conducted four times and graphs represent the mean value with corresponding SDs. ***: p < 0.001, **: p < 0.01, *: p < 0.05, ns: not significant, using Student’s t-test.

Fig 5. Adherence and internalization of C. jejuni is impeded in the 3D tissue versus 2D-monolayer environment.

(A, B) Adherence (upper panels) and internalization (lower panels) of C. jejuni strains 81–176 (A) and NCTC11168 (B) were examined at four hrs p.i. in 2D-monolayers and 4–48 hrs p.i. in the 3D tissue model. Bars represent the mean of four independent experiments with respective SDs. ****: p < 0.0001, ***: p < 0.001, **: p < 0.01, *: p < 0.05, ns: not significant, using Student’s t-test.

Fig 6. Infection with C. jejuni NCTC11168 deletion mutants differs between 2D-monolayer and 3D tissue model infections.

(Ai-v) Schematic comparison between wild-type and respective phenotypic manifestation by deletion of flaA (i), kpsMT (ii), cas9 (iii), csrA (iv), and ptmG (v). (C) The sRNA pair CJnc180/190 post-transcriptionally represses ptmG mRNA, which in turn is involved in the legionaminic acid flagellin glycosylation pathway in C. jejuni. (B, D) Isolation of CFUs from 2D-monolayers (4 hrs p.i.) or tissue models (24 hrs p.i.) for C. jejuni NCTC11168 wildtype (WT), deletion mutants (ΔflaA, ΔkpsMT, Δcas9, ΔcsrA), and their respective complementation strains (C flaA, C kpsMT, C cas9) (B), as well as for ΔptmG, ΔCJnc180/190 in addition to their complementation (C ptmG, C CJnc180/190) and overexpression (OE ptmG) strains (D). CFUs are depicted as the percentage of their respective input CFUs and represent the mean of three biological replicates with corresponding SDs. Statistical significance was calculated between wild-type CFUs (2D/3D) and those recovered for each mutant strain (2D/3D). Thus, asterisks or ns above each bar indicate the significance of the tested mutant compared to their respective wildtype in 2D or 3D. ****: p < 0.0001, ***: p < 0.001, **: p < 0.01, *: p < 0.05, ns: not significant, using Student’s t-test.