A multiplex inhalation platform to model in situ like aerosol delivery in a breathing lung-on-chip

Prolonged exposure to environmental respirable toxicants can lead to the development and worsening of severe respiratory diseases such as asthma, chronic obstructive pulmonary disease (COPD) and fibrosis. The limited number of FDA-approved inhaled drugs for these serious lung conditions has led to a shift from in vivo towards the use of alternative in vitro human-relevant models to better predict the toxicity of inhaled particles in preclinical research. While there are several inhalation exposure models for the upper airways, the fragile and dynamic nature of the alveolar microenvironment has limited the development of reproducible exposure models for the distal lung. Here, we present a mechanistic approach using a new generation of exposure systems, the Cloud α AX12. This novel in vitro inhalation tool consists of a cloud-based exposure chamber (VITROCELL) that integrates the breathing AXLung-on-chip system (AlveoliX). The ultrathin and porous membrane of the AX12 plate was used to create a complex multicellular model that enables key physiological culture conditions: the air-liquid interface (ALI) and the three-dimensional cyclic stretch (CS). Human-relevant cellular models were established for a) the distal alveolar-capillary interface using primary cell-derived immortalized alveolar epithelial cells (AXiAECs), macrophages (THP-1) and endothelial (HLMVEC) cells, and b) the upper-airways using Calu3 cells. Primary human alveolar epithelial cells (AXhAEpCs) were used to validate the toxicity results obtained from the immortalized cell lines. To mimic in vivo relevant aerosol exposures with the Cloud α AX12, three different models were established using: a) titanium dioxide (TiO2) and zinc oxide nanoparticles b) polyhexamethylene guanidine a toxic chemical and c) an anti-inflammatory inhaled corticosteroid, fluticasone propionate (FL). Our results suggest an important synergistic effect on the air-blood barrier sensitivity, cytotoxicity and inflammation, when air-liquid interface and cyclic stretch culture conditions are combined. To the best of our knowledge, this is the first time that an in vitro inhalation exposure system for the distal lung has been described with a breathing lung-on-chip technology. The Cloud α AX12 model thus represents a state-of-the-art pre-clinical tool to study inhalation toxicity risks, drug safety and efficacy.

normalizing the mean-fluorescent intensity (MFI) of Phalloidin (staining for F-actin fibers, in red) with Hoechst nuclei intensity (in blue). Data are represented as mean ± SEM (N=3; ROI = 4 per conditions per N). ( D) TER was measured from untreated and low concentration TiO2 NPs (0.021µg/cm2) exposed cells under ALI and ALI+CS conditions at 0h before exposure, 24h and 48h after TiO2 NPs exposure (N=2; n=4/timepoint). (E) mRNA was harvested from control and TiO2 high concentration exposed cells (ALI and ALI+CS) at 48h exposure timepoint. qPCR studies were performed with N=2; n=2/conditions and exposure significance were measured in relation to CTRL cell values. Data are represented here as mean ± SEM.

Section 1: in vitro dose correlation
The translation from in vitro dosing of inhaled corticosteroids used in this study to the daily inhaled dosing of a patient with a respiratory condition was approached by a mathematical approximation. Thorrson et al. in 2001 demonstrated the relative deposition difference of FL in pressurized metered-dose Inhaler (pMDI) and a Diskus inhaler. A pMDI is an inhaler in which a powdered drug is pressurized together with a propellant gas inside a container. By actuation of the inhaler the drug is released aerosolized. A Diskus inhaler on the other hand is a non-pressurized inhaler which is actuated by breath. The inhalation flow aerosolizes the metered drug dose into the respiratory tract. Calculations of lung deposition were performed by considering the ratio of substance remaining in the mouth after the use of the inhaler, systemic bioavailability, plasma cortisol concentration, clearance, halftime, mean residence time, mean absorption time and volume of distribution at steady state (Thorsson et al., 2001). It was concluded that on average 20% and 12% of the nominal dose per inhalation was deposited in the lung for pMDI and Diskus drug delivery respectively. By reverse calculation of the dosage used on the lung-on-chip, assuming an average alveolar surface in the distal lung is about 70-100m 2 (Fröhlich et al., 2016) using the manufacturer (Vitrocell) provided deposition calculations per cm 2 of nebulized substance, an in vivo estimation of the dosage was performed. A factor of 0.85 was added to the calculation to account for the 15% deviation in mass deposition within the Cloud α AX12. The area within one cellular well of the AX12 plate was nebulized with 300μL of a Fluticasone propionate concentration , this was extrapolated to an adult healthy average lung surface area to determine the estimated mass deposited. Depending on the inhaler device the nominal dose to deposited mass ratio of the inhaled corticosteroid (Fluticasone propionate) with concentrations of 100nM and 500nM was obtained as: Global initiative for asthma has recommended the use of 100-250nM as total daily low dose, >250-500 as medium and above 500nM as high total daily dose (Global Initiative for Asthma 2022). Therefore, our 100nM FL dose correlates to medium recommended daily dosing in adults, whereas 500nM corresponds to high daily dose for an adult.
Next, to approximate the deposition of NPs in vivo from in vitro gathered data, a similar approach was taken.  from healthy Calu3 cells on-chip from day 2 until day 14 (N=3; n=6/timepoint). (C) TER (Ohm-cm2) was measured from ZnO NPs nebulized Calu3 cells at 0h (before exposure) and 6h and 24h after exposure under ALI and ALI+CS culture conditions (N=2; n=3-5/timepoint/conditions). (D) LDH measured under ALI and ALI+CS Calu3 cells exposed to ZnO NPs in relative to CTRL cells (N=2; n=4). (E) mRNA was harvested from ALI+CS conditions at 24h timepoint and measured for inflammatory markers like IL2, IL6, IL8 and TNFα. Significance was calculated in relation to CTRL cell express ion values (N=2; n=4) and (F) mRNA was harvested from ALI+CS Calu3 cells on-chip at 24h timepoint and measured for epithelial related genes like TJP1 and ECAD (N=2; n=4). Significance was calculated in relation to CTRL healthy cell expression levels. (G) mRNA isolated from CTRL and ZnO exposed AX iAECs mono-cell culture on-chip (in ALI and ALI+CS) at 48h exposure timepoint. qPCR studies were performed with N=2; n=2/conditions and exposure significance were measured in relation to CTRL samples. (H) TER readings were recorded from AX iAECs/d-THP1s/HLMVECs tri-cell culture on-chip before (0h) and 24h and 48h after exposure with TiO2+ZnO NPs mixture (n=4 -6/conditions/time-point). (I) Gene expression was analyzed for WIF1, IL1b and IFNg from both ALI and ALI+CS samples at 48h (n=4 /conditions). Data are shown as mean ± SEM.  Control and PHMG exposed cells on-chip were stained with ZO1 (in green), CD163 (in red) and nuclei with Hoechst (in blue). ( B) TER (Ohm-cm2) was measured before (0h) and 24h after PHMG exposure to Calu3 cells on-chip in both ALI and ALI+CS conditions (N=1; n=5-6/conditions/time-point). Data are represented as mean ± SEM.

Fig. S6. Nebulized Fluticasone (FL) treatment effect on barrier recovery and cytotoxicity triggered by PHMG induction on-chip (A)
Overall timeline for cell seeding starting at Day 0 (D0) to PHMG exposure on D21 and FL treatment on D22 in the AX12 plate. (B) Time-series TER profile (Ohm-cm2) was measured for cells in ALI and (C) ALI+CS culture conditions, including before induction (0h) to 6h and 24h after PHMG exposure and until 24 h after FL treatment (N=2; n=6). (D) Cytotoxicity was calculated from LDH levels for ALI and ALI+CS samples (N=1; n=3/condition). Data are represented as mean ± SEM.

IL2
TAC AAG AAC CCG AAA CTG ACT CG ACA TGA AGG TAG TCT CAC TGC C

IL1a
AGA TGC CTG AGA TAC CCA AAA CC CCA AAG CAC ACC CAG TAG TCT