Comparison of air-liquid interface transwell and airway organoid models for human respiratory virus infection studies

Abstract

Introduction: Complex in vitro respiratory models, including air-liquid interface (ALI) transwell cultures and airway organoids, have emerged as promising tools for studying human respiratory virus infections. These models address several limitations of conventional two-dimensional cell line and animal models. However, the lack of standardized protocols for the application of these models in infection studies limits the possibilities for comparing results across different research groups. Therefore, we applied a collaborative approach to harmonize several aspects of experimental methodology between different research laboratories, aiming to assess the comparability of different models of human airway epithelium in the context of respiratory viral infections.

Methods: In this study, we compared three different models of human respiratory epithelium: a primary human bronchial epithelial cell-derived ALI transwell model, and two airway organoid models established from human airway- and lung-derived adult stem cells. We first assessed the presence of various differentiated cell types using immunofluorescence microscopy. Using a shared stock of influenza A virus, we then assessed viral growth kinetics, epithelial cytokine responses, and serum-mediated inhibition of infection.

Results: The presence of club, goblet, and ciliated cells was confirmed in all models. We observed similar viral replication kinetics with a >4-log increase in virus titre across all models using a TCID50 assay. Following infection, a reproducible antiviral cytokine response, including a consistent increase in CXCL10, IL-6, IFN-λ1, IFN-λ2/3, and IFN-β, was detected across all models. Finally, neutralization was assessed by pre-incubation of virus with human serum. Reduced viral replication was observed across all models, resulting in a 3- to 6-log decrease in virus titres as quantified by TCID50.

Discussion: In conclusion, all three models produced consistent results regardless of the varying cell sources, culturing approaches, and infection methods. Our collaborative efforts to harmonize infection experiments and compare ALI transwell and airway organoid models described here aid in advancing our understanding and improving the standardization of these complex in vitro respiratory models for future studies.

Keywords: complex in vitro models; harmonization; influenza virus; mucosal models; respiratory tract.

Figure 1

Schematic overview of culturing procedures for the ALI transwell and airway organoid models. Created in BioRender (https://BioRender.com/e98t944).

Figure 2

Characterization of histoarchitecture and cell type composition of air-liquid interface (ALI) transwell and airway organoid models. (A) Representative images from one donor of ALI transwell cultures displaying immunofluorescent staining of tight junctions (ZO-1), ciliated cells (α tubulin), secretory goblet cells (MUC5AC), and club cells (CC10). The maximum projection image shown is composed of 15 confocal images acquired with a Z-step of 0.22 μm, giving a total thickness of 3.3 μm. Scale bars equal 50 μm. (B, C) Representative immunofluorescent images from well-differentiated airway organoids cultured in Matrigel (B) or BME2 (C). Immunofluorescent labelling of stained formalin-fixed and paraffin-embedded sections of organoids showing the expression of markers for basal cells (KRT5/P63), cilia (ARL13B), secretory goblet cells (MUC5AC), and club cells (CC10). Scale bars equal 100 μm.

Figure 3

Comparison of influenza A/H1N1 virus infection susceptibility and infection kinetics of airway air-liquid interface (ALI) transwell and organoid models. (A) At 24 hpi, infected ALI transwell cultures were immunostained to detect viral nucleoprotein (NP, orange), ciliated cells (α-tubulin, green), and actin filaments (phalloidin, purple). Scale bar equals 50 μm. (B, C) At 24 hpi, infected organoids cultured in Matrigel (B) or BME2 (C) were visualized by immunostaining of formalin-fixed and paraffin-embedded sections (for Matrigel) and non-paraffin-embedded organoids (for BME2) to detect viral nucleoprotein (NP, red), actin filaments were stained with pan-cytokeratin (PAN-CK, green) or phalloidin (green), and nuclei were stained by DAPI (blue). Scale bars equal 100 μm. (D–F) Virus quantification was performed by TCID50 assay on Madin-Darby Canine Kidney (MDCK) cells for all 3 models. TCID50/ml for the time points 2, 24, 48, and 72 hpi are shown. (D) At the indicated time points, apical washes were collected from infected ALI transwell cultures (MOI=0.01) from 3 donors (A/B/C). The data shown are from 3 technical replicates (inserts) per donor. (E) Human airway organoids derived from 3 donors (D/E/F) were cultured in Matrigel, infected following enzymatic dissociation (MOI=0.1), and sampled at the indicated hpi. The data shown are from 2 technical replicates (wells) per donor. (F) Organoids derived from 3 donors (G/H/I) were cultured in BME2, infected with influenza virus following mechanical disruption (MOI=0.01), and sampled at the indicated hpi. The data shown are from 1 (I) or 2 (G/H) technical replicates (wells) per donor. Replicates were obtained in 2 independent experiments. The limit of detection was 2.89 TCID50/mL for panel (D) and 3.16 TCID50/mL for panels (E, F). The geometric mean with geometric standard deviation (SD) is shown.

Figure 4

Dynamics of host responses to A/H1N1 influenza virus infection in air-liquid interface (ALI) transwell and airway organoid models, measured by cytokine production. Concentrations of the cytokines CXCL10, IL-6, IL-8, IFN-λ1, IFN-λ2/3, and IFN-β were measured using LEGENDplex assay. (A) Basolateral medium samples from (mock-)infected ALI transwell cultures of 3 donors (A/B/C) were collected at the indicated time points and assessed for cytokine production. The data shown are from 3 technical replicates (inserts) per donor. (B) Cytokine concentrations were measured in the supernatant of (mock-)infected dissociated airway organoids cultured in Matrigel, derived from 3 donors (D/E/F). The data shown are from 2 technical replicates (wells) per donor. (C) Cytokine concentrations were measured in the supernatant of (mock-)infected dissociated airway organoids cultured in BME2 from 3 donors (G/H/I). The data shown are from 1 (I) or 2 (G/H) technical replicates (wells) per donor. Replicates were obtained in 2 independent experiments. The geometric mean with geometric SD is shown. For visualization purposes, cytokine concentrations below the limit of quantification (LOQ) were assigned values of 0.5 times the LOQ. Dotted lines and grey shading indicate the LOQ for each cytokine.

Figure 5

Inhibition of viral replication by pooled human serum containing influenza virus-specific neutralizing antibodies. TCID50/mL values at 24 hours post infection are shown for all models for three conditions: no serum, 1:400 dilution of serum, and 1:80 dilution of serum. (A) Values are from ALI transwell cultures from 3 donors (A/B/C). The data shown are from 3 technical replicates (inserts) per donor. (B) Values from airway organoids cultured in Matrigel derived from 3 donors (D/E/F) are shown. The data shown are from 2 technical replicates (wells) per donor. (C) Values from airway organoids cultured in BME2 derived from 3 donors (G/H/I), without technical replicates. For all 3 models, virus quantification was performed by TCID50 assay on Madin-Darby Canine Kidney cells, where the limit of detection was 2.89 TCID50/mL for panel (A) and 3.16 TCID50/mL for panels (B, C). The geometric mean with geometric standard deviation (SD) is shown (A, B).