Establishing an air-liquid interface exposure system for exposure of lung cells to gases

Abstract

Objective

Growing interest in non-animal-based models has led to the development of devices to expose cells to airborne substances. Cells/tissues grown at the air-liquid interface (ALI) are more representative of lung cells/tissues in vivo compared to submerged cell cultures. Additionally, airborne exposures should allow for closer modeling of human lung toxicity. However, such exposures present technical challenges, including maintaining optimal cell health, and establishing consistent exposure monitoring and control. We aimed to establish a reliable system and procedures for cell exposures to gases at the ALI.

Methods

We tested and adapted a horizontal-flow ALI-exposure system to verify and optimize temperature, humidity/condensation, and control of atmosphere delivery. We measured temperature and relative humidity (RH) throughout the system, including at the outlet (surrogate measures) and at the well, and evaluated viability of lung epithelial A549 cells under control conditions. Exposure stability, dosimetry, and toxicity were tested using ozone.

Results

Temperatures measured directly above wells vs. outflow differed; using above-well temperature enabled determination of near-well RH. Under optimized conditions, the viability of A549 cells exposed to clean air (2 h) in the ALI system was unchanged from incubator-grown cells. In-well ozone levels, determined through reaction with potassium indigotrisulfonate, confirmed dosing. Cells exposed to 200 ppb ozone at the ALI presented reduced viability, while submerged cells did not.

Conclusion

Our results emphasize the importance of monitoring near-well conditions rather than relying on surrogate measures. Rigorous assessment of ALI exposure conditions led to procedures for reproducible exposure of cells to gases.

Keywords:

• Air-liquid interface exposure
• ozone
• gases
• cytotoxicity
• in vitro

Introduction

Traditional toxicity testing of inhaled contaminants involves exposure of animals; growing interest in moving away from animal models – due to ethical concerns, questions regarding relevance to humans, high cost – has led to the development of air-liquid interface (ALI)-based cellular models and exposure systems to expose human cells and tissue samples to airborne test materials (Upadhyay & Palmberg 2018). These are intended to more closely approximate the interactions of inhaled test materials with pulmonary cells by simulating an inhalation-based exposure via the air-liquid interface. Cells/tissues grown at the air-liquid interface are more representative of lung cells/tissues compared to submerged cell cultures (Blank et al. 2006; Dvorak et al. 2011; Wu et al. 2017), and exposures occur by air, the medium through which humans are exposed. Interest in the development of advanced in vitro-based approaches for inhalation toxicology are aligned with recent efforts to develop novel in vitro technologies that could replace whole-animal testing approaches (Schmidt 2009). While many different ALI-based exposure approaches now exist, standardization and harmonization of these approaches will be required to improve the reliability and reproducibility of the results and conclusions and to gain regulatory acceptance (Lacroix et al. 2018).

Commercial devices for studying effects on cells of airborne exposures rely on distinct exposure approaches, i.e. cloud-, horizontal flow-, and perpendicular flow-type devices (Lenz et al. 2009; Aufderheide et al. 2013; Zavala et al. 2014). Horizontal flow systems direct airflow horizontally above the cells. The systems require an electrostatic deposition device to enhance particle deposition on the inserts containing cells. An example of this type of system is the Electrostatic Aerosol in vitro Exposure System (EAVES), which features a high-voltage corona wire to charge particles present in an airstream. The EAVES system was modified into the two-stage electrostatic precipitator, the Gillings Sampler (Zavala et al. 2014), which has subsequently been further adapted and released commercially as the CelTox Sampler (MedTec Biolab 2020). To date, the EAVES exposure system has been evaluated using exposures to diesel exhaust and concentrated coarse ambient particulate matter (de Bruijne et al. 2009; Volckens et al. 2009), while the Gillings Sampler was tested with p-tolualdehyde-coated mineral oil aerosol (Zavala et al. 2014).

ALI exposures deliver test gases or aerosols directly to the apical side of cells on inserts placed within the exposure chamber. The exposure chamber environment must therefore present conditions conducive to optimal cell health. These include continuous maintenance of temperature, high relative humidity (RH), and levels of airflow that enable the test material dilution to achieve the required dose without dehydration of the cells during the exposure. While an increasing number of studies on experimental applications of ALI exposures continue to be published, only a limited number provide thorough information on critical exposure parameters including temperature, RH, airflow rate, exposure duration, sensor locations, exposure controls, and dosimetric validations, among others. These parameters could affect cell viability, and are necessary to evaluate whether these studies have physiological and toxicological relevance (Lacroix et al. 2018; Leibrock et al. 2020).

Our objective in this work was to establish the means to verify and optimize exposure conditions for reliable and reproducible exposure of pulmonary cells to gaseous atmospheres, in particular to ozone, at the air-liquid interface. We established procedures to monitor and precisely control temperature, relative humidity, airflow parameters, and dose, and optimized these conditions for cell exposures in the ALI exposure system. We demonstrate that such steps are necessary to ensure that conditions are understood and optimized at the site of cell exposure.

Materials and methods

ALI exposure system

The CelTox Sampler (MedTec Biolob Inc.) performs in vitro exposure of cells at the ALI to gases and/or particles. It operates under negative pressure and includes on-board temperature and humidity regulation without diluting the test atmosphere. The Sampler includes a cell exposure chamber within a heated enclosure (Figure 1). The exposure chamber accommodates two custom reusable carrier plates constructed of anodized aluminum designed to accommodate 30 mm diameter 6-well Millicell-CM culture inserts. Whetted surfaces are largely stainless steel or anodized aluminum in order to minimize reactivity and/or wall losses. The exposure chamber includes a two-stage electrostatic precipitator for use with particle exposures (not employed in the present work). A replaceable foam gasket seals the sampler base to its lid. Tygon tubing and silicon tubing connect other components to the sampler or to each other. A battery-operated pump (Sensidyne model Gilian BDX-II) pulls air through the CelTox Sampler at the required 2.2 L/min; the manufacturer advises against departure from this flowrate. The user-calibrated pump includes a built-in rotameter and can experience non-condensing humidity up to 85% RH.

Figure 1. Modifications to the CelTox Sampler through comprehensive iterative testing and optimization stages, indicating the sampling locations for temperature (T), relative humidity (RH) and air flow (q).

Most of the system components sit in the heated enclosure. A Proportional-Integral-Derivative (PID) controller regulates the enclosure heater at the back of the enclosure. A fan circulates enclosure air inside the base of the CelTox Sampler; enclosure air does not contaminate the test atmosphere because the exposure chamber is a closed system. The diffusion humidifier, which uses a permeable non-reactive membrane to increase the humidity of the test atmosphere, sits on a heated bracket within the heated enclosure; whetted surfaces are stainless steel and Teflon. A second PID controller maintains the diffusion humidifier bracket temperature at 37 °C, although users can change this setting to modulate relative humidity. A factory-calibrated sensor mounted in a Delrin plastic and anodized aluminum housing measures temperature and relative humidity downstream of the CelTox Sampler. Third-party software (HS2000 V1100, provided by the manufacturer) records data from the sensor to a .txt file.

Modifications to the test system

Airflow and humidification

We installed a tee downstream of the air source (Oil-Free reciprocating Air Compressor, Ingersoll-Rand Model 2-OL5X5) to bleed off excess pressure. To address issues with kinked tubing resulting from moving the CelTox Sampler within its enclosure, we replaced most tubing with non-reactive Teflon tubing and stainless steel tubing. For tests requiring pre-humidified air, we directed compressed air to a bubble humidifier (Fideris Benchtop Model 909-151-1110) first. As the pump provided with the system included whetted surfaces not compatible with ozone, we replaced it with an ozone analyzer (Thermo-Environmental Model 49 C) that draws clean air at 1 L/min, supplemented with a small lab-grade pump to reach the recommended exhaust flow. Both were located outside of the heated enclosure.

As condensation from the humid exposure atmospheres (80%–90% RH at 37 °C) leaving the heated enclosure can interfere with measurement accuracy or cause damage to downstream equipment, we added a diffusion dryer (Permapure MD dryer, model MD-110-24F-4) after the exposure chamber to decrease the humidity to acceptable levels. The dryer directed the sample through a Teflon core; a dry air counterflow removed water quickly, driven by the partial pressure of water. We added a three-way valve downstream of the T/RH sensor to enable direction of heated enclosure air through downstream components to prevent any condensate created on opening the CelTox Sampler lid from being pulled into downstream tubes and equipment.

Monitoring temperature and humidity

We assessed temperature using Resistance Temperature Detectors (RTDs; Cole-Parmer model HSRTD-3-100-A-300-E) at various sites (Figure 1). We measured relative humidity immediately downstream of the exposure chamber. As the Delrin housing of the standard CelTox Temp/RH Sensor would react with ozone, we replaced it with a thermohygrometer (Vaisala Model HMD-60Y) in a stainless steel housing. We devised a bypass for the diffusion humidifier, using a length of tubing of similar size. Labview-based software developed in-house collected data from probes and analyzers.

Generation and control of test atmosphere

We generated a test atmosphere of up to 1 ppm ozone using an ozone feedback control loop described elsewhere (Guénette et al. 1997). The analyzer measured the concentration of ozone exiting the CelTox system and adjusted the flow of ozone introduced before the tee, using the tee to bleed off any excess pressure. The analyzer requires non-condensing sample at room temperature and therefore must sample downstream of the dryer. Whetted surfaces between the cells and the analyzer were non-reactive (Teflon and stainless steel), and conditioned daily during an equilibration period.

Ozone delivery assessment

Confirmation of ozone delivery to the cells at the point of exposure was determined according to the method of Bader and Hoigné (1981). Briefly, Whatman filters were cut to fit within the inserts, impregnated with 40 µL potassium indigotrisulfonate, and placed in the ALI system for a 10 min exposure. Upon removal, filters were placed in 15 mL Falcon tubes containing 10 mL ddH₂O, and incubated at room temperature for 10 min. Following incubation, 100 µL was pipetted into a 96 well black plate in triplicate and the fluorescence was measured (Ex/Em: 250/400 nm).

Cell culture

The human lung epithelial (A549; ATCC, CCL-185) cell line (American Type Culture Collection, Manassas, VA, USA) was cultured in Dulbecco’s Modified Eagle’s Medium (DMEM; Hyclone, Fisher Scientific) in the presence of phenol red and 4.5 g/L glucose and 10% Fetal Bovine Serum (FBS; v/v/; heat-inactivated; Hyclone, Fisher Scientific) in 75 cm² tissue culture flasks (Corning, NY, USA) at 37 °C, 5% CO₂ in a humidified incubator (>90% RH). Cells were seeded on the apical surface of collagen-coated (A1048301, rat tail collagen, 10 µg/cm² density; Gibco, Fisher Scientific) hydrophilic polytetrafluoroethylene (PTFE) Millicell-CM inserts (PICMORG50; 4.2 cm² surface area, 0.4 µm pore size; MilliporeSigma, Fisher Scientific) at 4 × 10⁵ cells/insert. Inserts were placed in 6-well tissue culture plates (Costar, Corning, NY, USA) with 0.8 mL of DMEM media (phenol red-free, including 10% FBS and 25 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid, HEPES buffer) added to the apical compartment and 1.25 mL in the basolateral compartment. Cells were then incubated for 24 h (5% CO₂, 37 °C, >90% RH). The next day, the cell culture medium present in the apical compartment was aspirated and the inserts were transferred to the carrier plate containing fresh basolateral medium (1.25 mL, including 5% FBS, 25 mM HEPES), and the plate was transferred to the ALI exposure chamber to commence cell exposures. For exposures, HEPES was included to augment buffering capacity in order to maintain the pH of the culture medium outside of the CO₂ incubator conditions. HEPES was also present in the media of matching incubator controls.

Cell exposures

To assess the potential effects of ALI exposure chamber conditions on cell health, A549 cells were exposed to 80%–90% humidified clean air atmosphere at 37 °C for 2 h, followed by viability testing. In another set of experiments, cells were exposed to 200 ppb ozone at 80%–90% RH at 37 °C for 2 h in the ALI chamber to assess the effects of ozone on A549 cell viability. Prior to introduction or removal of the carrier plate containing the A549 cell inserts, the exposure chamber was briefly flushed with dry air to prevent condensation and nuisance precipitation. Concentration of ozone, temperature and RH were continuously monitored during exposures. For all experiments, cells were maintained in parallel in the incubator to serve as controls, with and without apical media. An additional control involved inclusion in adjacent wells within the ALI exposure system cells that were fully submerged in culture media (i.e. media present in apical and basolateral compartments).

Cell viability

Following A549 cell exposure, the inserts were transferred back to the original 6-well plate kept in the incubator for the duration of the experiment. The DMEM media (phenol red-free) with 10% FBS and resazurin reagent (AlamarBlue^TM, 10% vol/vol; Invitrogen, Fisher Scientific) was replaced in the apical compartment, and cell viability was assessed (up to 4 h) by reading fluorescence in aliquots obtained from this compartment. Note that while we have detected AlamarBlue in the basal compartment, indicating some leakage, sampling from the apical compartment yields samples that exhibit much higher intensity and lower variability than those obtained from the basolateral compartment, and are representative of the whole. Reduction of the resazurin reagent to fluorescent resorufin product was measured by top-reading of fluorescence at λEx = 540 nm and λEm = 600 nm using a Synergy^TM two multi-mode microplate reader (BioTek, Winooski, VT, USA). The assay measures reductive metabolism of the cells resulting from cytosolic, mitochondrial and endoplasmic reticulum enzyme activity, with decreased fluorescence indicative of impaired cell health (Gonzalez and Tarloff 2001).

Calculations and statistical analyses

We calculated dew point using the August–Roche–Magnus formula (Alduchov & Eskridge 1997). Resazurin reduction assay data from the 2 h clean air and ozone exposure experiments were analyzed by two-way ANOVA with Treatment and Time as factors, followed by the Holm-Sidak multiple comparison procedure to elucidate the pattern of significant effects (α=0.05; Sigma-Plot 13, Systat Software Inc., San Jose, CA, USA). The data were transformed as required to meet the assumptions of normality and homoscedasticity. The figures represent non-transformed data. Data are shown as mean ± standard error of the mean (S.E.M.).

Results

Exposure conditions

To characterize and optimize environmental conditions that cells would experience during exposure, we evaluated relative humidity and temperature at the site of exposure. In order to do so, we threaded a small solid-state RTD through the exposure chamber gasket and above one of the holes of the sample cover plate. The exposure chamber reached steady temperature and humidity after an equilibration period of 75 to 90 minutes (Figures 2(A,B)). Removal of the exposure chamber lid caused an expected drop in temperature and relative humidity, which lasted as long as the chamber remained unsealed. Conditions at the well recovered quickly after adding cells: within 4 minutes of resealing the exposure chamber, temperature and relative humidity returned to their equilibration values. Environmental conditions remained stable at the well for the 2 h of exposure. RTDs simultaneously sampling elsewhere (in the base of the CelTox Sampler; in opposing corners of the heated enclosure; in the test atmosphere entering and exiting the heated enclosure) indicated a 10 °C temperature gradient within the enclosure (data not shown).

Figure 2. Temperature and relative humidity (RH) in the CelTox Sampler ALI exposure system during a typical cell exposure. (A) Continuous measurements of temperature using an RTD located above a well and a thermohygrometer located downstream of the exposure chamber established the equilibration period and recovery after adding cells. (B) Comparison of relative humidity measurements downstream of the exposure chamber and in-well % RH calculated from downstream dew point. Similar profiles occurred when the water bath was excluded (data not shown). (C) Comparison of temperature monitored at the chamber exhaust using a thermohygrometer to simultaneous measurements directly above the cell insert using a solid-state resistance temperature detector (RTD).

Testing various means of humidifying the test atmosphere demonstrated that the inline diffusion humidifier achieved the target humidity (Supplementary Figure S1). In initial tests, we heated the enclosure for 30 minutes before adding the diffusion humidifier filled with water, both pre-heated overnight in a 37 °C incubator. After the CelTox Sampler evolved to include a heated bracket for the diffusion humidifier, we relied solely on the heated enclosure and the heated bracket to produce humidity from the diffusion humidifier filled with un-heated water. Operators typically observed condensation in the exposure chamber when adding cells. Introduction of a bypass that allowed for flushing the exposure chamber with dry air briefly before opening resolved this issue. Within six minutes, the downstream RH dropped below 30% (Figure 2(B)) and we observed a significant decrease in visible condensate.

To determine whether downstream measurements were a reasonable surrogate for measurements at the well, we recorded temperature simultaneously at the well and downstream of the exposure chamber. Temperature reported by the downstream sensor often did not match temperature measured at the well; larger discrepancies occurred primarily during the equilibration phase and when opening the exposure chamber. Figure 2(A) illustrates a typical cell exposure: 30 min into the equilibration period, the downstream sensor reported a temperature reading of 36 °C while the well sensor read 34.5 °C and continued to climb (Figure 2(A)). During exposures, when the exposure chamber and its enclosure remained undisturbed for 2 h, the difference between the measurements decreased; even so, Figure 2(A) illustrates that while the well sensor reported a stable average temperature of 37 °C (SD = 0.3) throughout the 2 h cell exposure period, the downstream sensor reported an average temperature of 35.9 °C (SD = 0.4). During this exposure, the difference between the two sensors was sometimes as high as 2 °C. On one occasion, the downstream sensor reported a stable temperature at the desired value, while the well sensor detected temperatures that continued to climb to 41 °C (Figure 2(C)). While this technical issue was rectified by the manufacturer, it further demonstrated that measurements taken at the outlet may differ from those taken at the well.

In order to determine to what extent downstream RH measurements were a reasonable approximation for measurements at the well, we calculated the dew point of the test atmosphere using temperature and relative humidity measurements at the outlet of the exposure chamber, and then inferred relative humidity at the well, given the temperature measured at the well. As seen in Figure 2(B), calculations indicated a difference on the order of 5%–10% RH between the estimated in-well values compared to values measured at the outlet during exposures.

Viability of A549 cells at the air-liquid interface

To determine whether the ALI exposure conditions maintain the viability of A549 lung epithelial cells during exposure, we conducted the resazurin reduction assay on cells exposed for 2 h to clean air at the ALI (Figure 3). The assay showed that A549 cells cultured at the ALI were metabolically active (two-way ANOVA, Time point x Treatment interaction, p< 0.001; ALI versus blank no cells, p< 0.001), as revealed by the cellular reduction of resazurin. The metabolic activity of the cells at the ALI was comparable to the incubator with no significant difference between the groups (Figure 3), indicating maintenance of A549 cell viability during clean air exposure at the ALI, despite the brief unavoidable exposure to dry air while loading or extracting cells.

Figure 3. Viability (metabolic activity) of A549 cells exposed to clean air at the ALI for 2 h. Resazurin reduction by A549 cells was measured over 4 h immediately after exposure. “Submerged” refers to cells covered by media rather than cultured at the ALI. Mean values with SEM; n = 3–6 independent experiments are presented. *ALI (clean air) significantly different from the blank control, p < 0.001, two-way ANOVA, Holm-Sidak post-hoc test, with no significant difference between ALI (clean air) and other control groups.

Control of ozone exposures

Feedback control maintained steady ozone concentration in the exposure chamber at target concentrations over a range of exposure levels (ppb) by measuring ozone concentration just after the cells and adjusting itself to the desired level (Figure 4(A)). Since ozone is highly reactive, levels in the exposure chamber may dissipate through reactions with chamber components and may be affected by environmental factors such as humidity. Therefore, it is critical to additionally quantify ozone in the proximity of the cells as well as at optimal (high humidity) and suboptimal (no humidity) RH levels. Delivery of ozone to the cell surface was verified by measuring the fluorescent reactive product of indigo trisulfonate and ozone at the insert membrane. Results confirm linearity of the ozone dose response over a range of ozone concentration independent of humidity level, demonstrating effective performance of the feedback control loop (Figure 4(B)).

Figure 4. Control of ozone delivery and exposure of A549 cells at the ALI. (A) Ozone levels in the exposure chamber measured at various target concentrations (ppb) over time, using ozone analyzer. (B) Dosimetry of ozone delivered to the cell culture insert, measured using indigo trisulfonate. (C) Viability (metabolic activity) of A549 cells exposed to 200 ppb ozone at the ALI for 2 h, as measured over a 3-hour time period immediately post-exposure. “Submerged” refers to cells covered by media rather than cultured at the ALI; “Incub no media” refers to cells maintained in the incubator with no apical media. Mean values with SEM; n = 4 independent experiments are presented. *ALI (ozone) significantly different from the incubator (no apical media) control, p < 0.05, two-way ANOVA on ranks, Holm-Sidak post-hoc test.

Exposure of A549 cells to ozone at the air-liquid interface

After optimizing and validating ozone delivery, A549 cells on inserts were exposed to ozone diluted in clean air at 80%–90% humidity level to assess whether the reactive gas can be delivered to cells at the ALI to produce detectable biological effects. A 2 h exposure to 200 ppb ozone at the ALI reduced metabolic activity in A549 cells (two-way ANOVA on ranks, Time point x Treatment interaction, p< 0.001; ALI versus Incubator, no media; p< 0.05; Figure 4(C)). Both the cells submerged in culture media exposed simultaneously to ozone within the ALI exposure system and the incubator controls, which were not exposed to ozone, exhibited the same lack of detectable biological effects, demonstrating that the presence of media above the cell monolayer protected the cells from the direct effects of ozone.

Discussion

To ensure that ALI system conditions are physiologically relevant, the exposure system needs to approximate the in vivo conditions in the lungs, i.e. 37 °C and 100% RH (Déry, 1973). However, maintaining these conditions stably and consistently for an extended time-period is technically challenging. Our objective was to establish a reliable system and procedures that enable reproducible cell exposures to gases at the ALI. We optimized an ALI-exposure system for controlled delivery of air and ozone to lung epithelial cells (A549) grown at the ALI, with a focus on the fundamentals required for optimal cell viability, including temperature, humidity/condensation, and control of atmosphere delivery. This was achieved through system-wide measurements of temperature and humidity, and evaluations of cell viability under control conditions, followed by controlled 2 h exposure of cells to the criteria air pollutant ozone.

Optimal cell viability requires physiological conditions for relevant exposure duration

Successful exposure of cells at the ALI requires equilibration and stabilization of the environment for optimal cell function. This includes meeting the thermal, humidity and airflow requirements that promote cell viability and minimize condensation/evaporation. Due to the great variety in exposure system designs and the different exposure scenarios performed, this can be a challenging task. Consequently, exposures with early ALI systems relied on short-term exposures (in minutes) and low airflow rates to ensure optimal cell viability (Ritter et al. 2001; Lenz et al. 2009; Persoz et al. 2010; Xie et al. 2012). However, exposures of such short duration may not sufficiently capture the biological changes that may develop in acute/sub-acute exposure scenarios in vivo and may not provide sufficient dose in exposures that involve particulate matter.

Recent work underscores the importance of exposure system optimization. A number of studies document significant decreases in cell viability associated with the exposure chamber conditions in exposures to air for 1 h (Steinritz et al. 2013; Frijns et al. 2017; Zavala et al. 2017). For example, conditions of low temperature (22 °C) and low humidity (18 or 55% RH) resulted in >40% reduction in the viability of BEAS-2B cells exposed to air for 1 h at the ALI compared to 37 °C and 75% RH; humidity and temperature optimization required extensive adaptation of the manufacturer-supplied system (Zavala et al. 2017). Researchers conducting an inter-laboratory reproducibility study using the Cultex RFS system have also observed unexplained declines in cell viability from ALI exposures to pure air (up to 1 h) across laboratories, effects attributed to impurities in their air source (Steinritz et al 2013). Leibrock and colleagues (2020) have recently highlighted the benefits of optimizing an ALI exposure system based on the perpendicular flow Vitrocell 12/3 CF module (Leibrock et al 2020). The authors indicated improvements in viability of A549 cells exposed to clean air, from 45% (with respect to incubator control) observed with the unoptimized module, and up to 75%−90% when various exposure parameters were adjusted. These included a heated chamber lid (38 or 45 °C) to prevent condensation, >90% RH, adjusted airflow in inlet tubes, and 5% CO₂ added to the airstream. Unlike perpendicular flow systems such as the Vitrocell and Cultex RFS, which require an adjustment of needle valves and individual aerosol inlet tubes to reduce the cell-directed airflow and prevent desiccation and mechanical damage to cells (Pariselli et al 2006; Anderson et al 2013; Leibrock et al 2020), horizontal-type flow systems direct air flow horizontally to the cells, and allow for diffusion of gases or vaporized test materials throughout the chamber. Under the conditions established in the present work, which include inline humidification, excess evaporation or damage to cells was avoided. This is critical: in the presence of undesired effects on cell viability resulting from poorly optimized exposures, the true effects of the test substance on cells could be underestimated, as these are typically derived in relation to controls. Including multiple types of controls, as shown in the present work, is desirable in order to account for an array of factors that could bias the observations. As shown in our work, by optimizing the parameters of temperature and humidity at the near-well level (37 °C and 80%–90% RH), with constant airflow rate of 2.2 L/min, we achieved A549 cell viability similar to incubator and submerged control cells within the ALI exposure system.

Importance of temperature measurement at the well

While a number of commercially available ALI exposure systems contain external control modules for the control of temperature, flowrate, and humidity, the in-chamber characterization of the physiologically critical exposure parameters is user-dependent. Frequently, system design limitations may prevent optimal placement of the sensors; for example, many systems rely on sensors placed before the exposure chamber or at the exhaust of the chamber or in a culture media supply chamber (Savi et al. 2008; Aufderheide et al. 2016; Latvala et al. 2016; Leibrock et al. 2020) as a surrogate for environmental measurements at the well. The observed discrepancy between temperature measurements taken simultaneously at the well and downstream of the exposure chamber illustrates the need to measure temperature at the well rather than relying on measurements at the outlet, both in terms of accuracy with respect to cell exposure conditions and as means to identify and troubleshoot issues. While the temperature overshoot issue was rectified by the manufacturer, we would not have been aware of it without including a sensor to measure temperature at the well. As the existing (original) sensor is located within the heated enclosure, it reflects to some extent the temperature of the heated enclosure, which can differ significantly from the target temperature of 37 °C, notably during the equilibration period. Enclosure temperature fluctuations occur when the controller triggers the heater to turn on (for example, when the lid of the heated enclosure opens). The documented temperature gradient within the enclosure suggests that the impact on the sensor may vary according to its location within the heated enclosure. Although not tested in the present study, probe housing materials (for example, plastic versus stainless steel) could also affect sensitivity and responsiveness of the sensor.

Our results also demonstrate the relevance of obtaining real-time “in-well” temperature measurements for reliable determination of RH conditions at the cells. Relative humidity is a non-linear function of temperature, and when there is a difference between downstream temperature and well temperature, downstream relative humidity will not accurately reflect relative humidity at the wells. The quantity of water in the test atmosphere does not change between the downstream sensor and the wells, making dew point a reliable indicator of humidity experienced by the cells.

Equilibration period promotes stable exposure conditions

A cell exposure whose goal is to reproduce conditions experienced by lung cells in vivo would ideally provide unfailingly steady temperature and RH at lung physiological conditions. In practice, the goal must be to minimize cell exposure time to suboptimal conditions. Some deviation is unavoidable: opening the exposure chamber to add the cells causes a sudden drop in temperature and RH, but it is of short duration in the context of a 2 h exposure. Minimizing the effect of suboptimal conditions in the CelTox Sampler requires an equilibration period of approximately 75 to 90 minutes, during which the rate of change of temperature and relative humidity level off as they attain targeted set points. Following equilibration, the viability of cells exposed to clean air for 2 h at the ALI was comparable to that observed for incubator controls with and without apical media present. In a number of studies, adverse effects of ALI system conditions on cell viability required subsequent limiting of exposure duration (Persoz et al. 2010; Tang et al. 2012; Xie et al. 2012). The capacity to prolong duration of ALI exposures while maintaining cell health, as demonstrated by our work and others, should enable more relevant comparisons with in vivo exposure scenarios. For example, distinct cytokine gene expression profiles were observed in A549 cells exposed to TiO₂ nanoparticles for 30 min versus 4 h, reflecting a difference in the deposited dose over exposure time (Diabaté et al, 2020).

Handling test atmospheres to avoid condensation

Unless the test atmosphere is already at 80%–90% RH, additional humidity is necessary. Earlier CelTox models relied on a heated water bath and could increase relative humidity by only 25%–50% RH; optimal performance required external pre-heating of water before placing it in the water bath (data not shown) to accelerate the required equilibration period. Replacement with a diffusion humidifier provided reliable inline humidification at the required level, even when using dry air.

However, humid test atmospheres consequently require special handling to protect analytical equipment and to prevent undesirable condensation. When handling a test atmosphere with high relative humidity, even a slight drop in temperature (such as might be experienced when the test atmosphere exits the heated enclosure) may cause the relative humidity to reach 100%, leading to the formation of water droplets. The droplets may cause issues for downstream equipment; in our case, the analyzer measuring ozone concentration required non-condensing test atmosphere. Addition of a non-diluting and non-reactive dryer after the exposure chamber protected the downstream analyzer from condensation.

Preliminary tests uncovered a potential issue with condensation triggered by cell handling: on opening the lid post-equilibration, the 85% RH, 37 °C atmosphere suddenly experienced ambient conditions, causing condensation notably on the surfaces near the exposure chamber inlet and above the cells. Condensation in the exposure chamber can be problematic as it may introduce water droplets onto cells, compromising the ALI and potentially affecting cell viability. In addition, the presence of water droplets could affect dosimetry and reactivity of the test atmospheres. For example, water droplets can serve as a reactive surface for ozone, or they can capture particles from the airstream. When condensation occurred on the lid above the cells, it was unevenly distributed, potentially affecting the wells differentially across the carrier plate. Condensation in tubing also represents a reactive surface and a possible flow obstruction, of particular concern when downstream instruments are sensitive to condensate. It may also compromise the validity of using downstream analyses as a measure of cell dose. We addressed this issue by routinely removing the diffusion humidifier and flushing with dry air using the bypass, just before opening the CelTox Sampler lid. This terminated generation of humidity; the bypass also allowed continued airflow through the exposure chamber, rapidly flushing the system of humidity.

The 3-way valve added downstream of the temp/RH sensor proved a useful addition for several reasons. If condensation appeared on opening the exposure chamber, the valve ensured that condensate would not reach downstream equipment. The valve made it possible to flush downstream tubing and equipment using warm dry air from the heated enclosure. Finally, the valve was a convenient location to measure airflow through the exposure chamber.

Ozone exposure at the air-liquid interface

Ozone is a highly reactive gas that can readily interact with the surrounding surfaces, causing some materials to age or corrode. This interaction would also result in ozone quenching, modifying the concentration of ozone encountered by cells. Replacement of the supplied temp/RH probe with a temperature/RH sensor in stainless steel housing addressed the potential for interaction with probe materials.

Two-hour exposure to 200 ppb of ozone resulted in decreased A549 cell viability. The ozone dose applied in our study approximates high episodic 1 h concentrations of ground-level ozone that have been encountered in major urban centers such as Houston or Los Angeles (160–180 ppb) and which have exceeded the US National Ambient Air Quality Standard (NAAQS) concentration of 120 ppb (NRC 2008). The benefits of direct cell exposure to ozone have previously been demonstrated using an ALI system prototype (Tarkington et al. 1994), as well as with an early generation of the Cultex system (Ritter et al. 2001). Both studies indicated cytotoxicity of ozone to lung-derived cell lines after short-term exposure (ca. 60–120 min) to levels ca. 200–500 ppb, and an inverse relationship between cytotoxicity and the amount of culture media overlaying the cells. Similarly, we have shown that the presence of culture media above the cells was protective, with media acting to quench ozone and eliminating effects of ozone on cell viability. It is important to consider that factors that influence cell phenotype, including the length of time cells are cultured at the ALI prior to exposure, may contribute to their sensitivity to exposure; examining the impact of such factors was beyond the scope of the present study. As indicated by the present work, the issues of poor sensitivity of submerged model systems and their low in vivo relevance can be overcome by application of an ALI-based approach. Because exposures are conducted under conditions that more closely model real-life exposures, relevant biological responses can be assessed across multiple domains (e.g. functional, morphological, proteomic, transcriptomic) with greater confidence.

Conclusion

Through a comprehensive optimization of ALI exposure parameters, we established an ALI exposure system for reproducible exposure of cells to gases. Optimization of the exposure conditions (temperature, humidity and airflow) resulted in stable and reproducible delivery and reliable monitoring of the test atmospheres to A549 cells at the ALI. Allowing the chamber environment to reach temperature and relative humidity targets was critical for ensuring that cell viability remained optimal over the 2 h exposure to clean air, as there was no loss of viability due to non-optimal temperature or relative humidity conditions. Some alterations of the exposure system were required to ensure materials compatibility with the reactive gas. Furthermore, optimal placement of temperature sensors was necessary for reliable determination of the exposure conditions with respect to temperature and RH. Once the system and procedures were optimized, reproducible environmental conditions were maintained during exposure to air or ozone, enabling assessment of toxicologically-relevant biological responses. Our results show how such optimization and verification of conditions is critical to ensuring that toxicity can be attributed to action of the test agent, rather than to suboptimal exposure conditions.

Acknowledgments

The authors gratefully acknowledge Kevin Curtin, Marjolaine Godbout-Cheliak and Alain Filiatreault for technical assistance. The work was supported by Health Canada through the Addressing Air Pollution Horizontal Initiative.

Disclosure statement

No competing interests to declare.