Progress in Assessing Air Pollutant Risks from In Vitro Exposures: Matching Ozone Dose and Effect in Human Airway Cells

Abstract

In vitro exposures to air pollutants could, in theory, facilitate a rapid and detailed assessment of molecular mechanisms of toxicity. However, it is difficult to ensure that the dose of a gaseous pollutant to cells in tissue culture is similar to that of the same cells during in vivo exposure of a living person. The goal of the present study was to compare the dose and effect of O₃ in airway cells of humans exposed in vivo to that of human cells exposed in vitro. Ten subjects breathed labeled O₃ (¹⁸O₃, 0.3 ppm, 2 h) while exercising intermittently. Bronchial brush biopsies and lung lavage fluids were collected 1 h post exposure for in vivo data whereas in vitro data were obtained from primary cultures of human bronchial epithelial cells exposed to 0.25–1.0 ppm ¹⁸O₃ for 2 h. The O₃ dose to the cells was defined as the level of ¹⁸O incorporation and the O₃ effect as the fold increase in expression of inflammatory marker genes (IL-8 and COX-2). Dose and effect in cells removed from in vivo exposed subjects were lower than in cells exposed to the same ¹⁸O₃ concentration in vitro suggesting upper airway O₃ scrubbing in vivo. Cells collected by lavage as well as previous studies in monkeys show that cells deeper in the lung receive a higher O₃ dose than cells in the bronchus. We conclude that the methods used herein show promise for replicating and comparing the in vivo dose and effect of O₃ in an in vitro system.

Recent proposals for identifying exposure hazards to toxic agents involve the identification of “toxicity pathways” in primary cultures of human cells. The possibility of extrapolating toxicity observed in vitro to human risk in vivo is attractive because it might allow a deeper understanding of the molecular basis of injury through use of “omics” techniques as well as the possibility of high throughput screening. Air pollutant gases and aerosols present a unique challenge to in vitro work because of difficulty replicating the dose and exposure conditions found in vivo. Few studies have attempted to re-create in vitro the dose and effect of an air pollutant observed in vivo. (National Research Council, 2007).

The goal of the present study was to contribute a proof-of-concept for direct extrapolation of air pollutant toxicity from in vitro cell cultures to in vivo human bronchial epithelial cells. Epithelial cells were chosen because they secrete mediators which initiate inflammatory events that are believed to be important in human risk assessments. They also can be expanded in air-liquid interface cultures and used in high throughput assays.

Ozone (O₃) was chosen as a prototype pollutant because it is one of the best-studied and ubiquitous of pure gaseous inhalants (U. S. E. P. A., 2013).

We showed previously that exposure to heavy oxygen-labeled ozone (¹⁸O₃) results in incorporation of ¹⁸O into cells and extracellular material obtained from bronchoalveolar lavage fluid (BALF) of humans and animals (Hatch et al., 1994, 2013; Plopper et al., 1998) creating a convenient micro-measurement of O₃ dose that is related to O₃ toxicity. We have also demonstrated that epithelial cells release mediators following O₃ exposure that are early markers of inflammation and that appear to have health consequences in the human population (Devlin et al., 1996; Devlin et al., 2012). The generation of these mediators can be quantified by gene expression assays. We here compare the ¹⁸O₃ dose and gene expression of two inflammatory mediators in cells obtained from human subjects exposed in vivo with the same measurements made in human primary cells expanded in culture and exposed in an air-liquid interface system in vitro.

MATERIALS AND METHODS

A flow chart depicting the study design is shown in Figure 1. In vivo exposures to ¹⁸O₃ (for dose measurement) and unlabeled O₃ for gene expression measurement were compared with in vitro exposures of the same length (2 h) that measured the same dose and effect endpoints. In vitro exposures to ¹⁸O₃ were performed on cells expanded from two subjects, and gene expression changes were evaluated in cells expanded from three subjects. Both the dose and effect in vitro studies employed the same range of ¹⁸O₃ or unlabeled O₃ concentrations. Cells were harvested at the termination of exposure in both the in vivo and in vitro studies.

FIG. 1.

Flow chart of the experimental design. The study involved both an in vivo and an in vitro exposure to ¹⁸O₃ of human epithelial cells. Bronchoalveolar lavage cells and constituents were also measured for comparison with the epithelial cells. Biological effects of O₃ consisted of measurements of IL-8 and COX-2 gene expression in cells harvested after exposure to unlabeled O₃. Cells were sampled 1 h after in vivo exposures and immediately following the in vitro exposures.

In Vivo ¹⁸O₃ Exposures of Human Subjects

For the dosimetry of ten subjects, eight (seven Caucasian and one Hispanic) were male and two (one Caucasian and one Hispanic) were female, and the following data apply: age 26 ± 1 years (19–33 years), body weight 80.8 ± 2.8 kg, height 176.9 ± 3.0 cm, and resting forced vital capacity (FVC) 5.55 ± 0.27 l. For the biological effects study of nine subjects, all were males (one Black, one Hispanic and seven Caucasian), with the following characteristics: age 23.7 ± 0.9 (20–29 years), body weight 86.1 ± 2.6 kg, height 182.0 ± 2.0 cm, and resting FVC 5.88 ± 0.2 l. The study protocol was approved by the University of North Carolina Committee on the Rights of Human Subjects (Institutional Review Board) as well as the U.S. Environmental Protection Agency. All individuals provided written informed consent and were recruited under a contract to the Westat Corp. (Rockville, MD). All subjects underwent a physical examination, with a complete blood count and differential, serum electrolytes, glucose, and liver enzymes. Female subjects had a negative urine pregnancy test. Volunteers were excluded from the study if they had a smoking history within 2 years of the study, or were not able to refrain from taking over-the-counter anti-inflammatory agents, vitamins, or antioxidants for at least a week prior to exposure. Participants who reported a recent respiratory tract infection (or who had symptoms) were not studied for a period of at least 6 weeks after symptoms had disappeared.

Each volunteer was exposed on a single occasion for 2 h to ¹⁸O₃. Subjects were seated on a bicycle ergometer with their forehead touching a pad on the top of a 6 inch diameter cone that blew ¹⁸O₃ downward across the face at a flow rate of 500 l/min. The ozone concentration delivered did not deviate from the target concentration by more than ±5%. A sampling tube (with a flow rate of 2–2.5 l/min) similar to a head-mounted microphone collected breathing air 1–2 inches in front of the face of the subject. ¹⁸O₃ concentrations were thus monitored in the breathing zone every 10 s and stored as 2 min averages. Mean concentrations in the breathing zone ranged from 0.21 to 0.33 ppm and were 30–50% lower than in the incoming airstream.

Throughout the exposure, volunteers underwent moderate intermittent exercise (15 min intervals) on the bicycle ergometer. The level of exercise was targeted to a minute ventilation (V_e) of 25 l/min/m² body surface area (∼50 l/min/subject). Prior to the exposure day, all subjects participated in a training session to establish the workload that would be required to achieve this V_e and the same work load was targeted on the ergometer. Subjects breathed naturally by mouth or nose depending on the subjects' comfort level. On the study day, all subjects performed a pulmonary function test prior to the exposure using a Sensormedics Vmax 220 instrument and software (Sensormedics Corp., Yorba Linda, CA) according to American Thoracic Society guidelines. Spirometry was also conducted on all subjects immediately following the exposure. Decrements in FEV₁ were <20% for all subjects post- versus pre-exposure. All volunteers had forced expiratory volume in 1 s (FEV₁) and FVC baseline values of at least 80% of that predicted for their height, age, and gender. Subjects were outfitted with a telemetry unit for the duration of the ¹⁸O₃ exposure and bronchoscopy for continuous monitoring of heart rate and rhythm as well as finger oxidimetry for O₂ saturation. Each subject was exposed twice, once to air and once to O₃, separated by at least 2 weeks.

In vivo ¹⁸O₃ and O₃ generation

The ¹⁸O₃ and unlabeled O₃ exposures were conducted identically in a chamber (dimensions of 1.22 m wide × 2.44 m long × 1.83 m high) at the U.S. Environmental Protection Agency Human Studies Facility on the campus of the University of North Carolina, Chapel Hill, NC. Air from an oil-less compressor was passed through an air purifier (ZEKS Nomonox model CDP480CD00, West Chester, PA). It was then humidified by passage over heated deionized water. ¹⁸O₃ was generated by passing a 1–2% stream of ¹⁸O₂ (Isotec, Inc., Miamisburg, OH; 99 atom %) in argon through a silent electric discharge (model CMGK-F 0.5 O₃ generator; Innovatec Gerätetechnik GmbH, Rheinbach, Germany). The ¹⁸O₃ concentration was controlled by altering the flow of ¹⁸O₂ into a constant flow of argon that passed through the generator. Chamber relative humidity and temperature were measured once a second and stored as 2 min averages that yielded an overall mean and SD of 39.1 ± 0.2% and 22.5 ± 0.1°C, respectively.

Exposures to unlabeled O₃ were performed by metering pure O₂, along with dilution air, into a silent arc O₃ generator. O₃ concentration was controlled by varying the power to the generator, but otherwise the exposures were done identically to the ¹⁸O₃ exposures.

Bronchoscopic sample collection

A bronchoscopy was conducted ∼45 min following the completion of the 2 h exposure to ¹⁸O₃ as described previously (Ghio et al., 1998) modified as follows. Immediately prior to the procedure, subjects were given intravenous fluids through a saline lock catheter placed in an antecubital vein to ensure they were hydrated. Primary bronchial epithelial cells were obtained by brush biopsy of both the right and left mainstem bronchus. Each bronchus was brushed six times in and out with a ∼5 cm excursion, and then repeated with a second brush. The two cytology brushes from each side were kept separate and the epithelial cells dislodged by agitation and brief vortexing in 0.1 ml of cold Dulbecco's phosphate buffered saline. The dislodged cells were pelleted by centrifugation at 400 × g at 4°C. For ¹⁸O measurements, the supernatant was removed and the remaining cell pellet was lysed by pipetting up and down in 0.1 ml of 0.05% SDS. This solution was then diluted by adding an additional 0.1 ml of water.

Bronchoalveolar lavage was conducted after the collection of the brush biopsies. Three 50 ml aliquots of sterile saline were sequentially injected into the right middle lung lobe. After each aliquot of saline was instilled it was gently aspirated back into the same syringe. The unfiltered lavage fluid was placed on ice in 50 ml conical tubes. Mean percent recovery volumes (range) of BALF from each lavage were as follows: wash no. 1; 29.4 ± 7.0% (14–38%), wash no. 2; 67.4 ± 9.2% (52–84%), wash no. 3; 77.8 ± 11.4% (58–92%). Samples were immediately centrifuged at 1000 × g for 10 min at 4°C to pellet the cell fraction. BALF supernatants from all three aliquots were pooled and stored at −80°C for subsequent analysis of ¹⁸O/¹⁶O content. Cell pellets from each aliquot were individually lysed by the addition of 0.1 ml of 0.05% SDS and the resulting cell lysates were dispersed by pipetting up and down and then diluted with an equal volume of water.

Primary Human Epithelial Cell Cultures and ¹⁸O₃ Exposures

Cell collection and culture

Two dose-response in vitro¹⁸O₃ exposure experiments were performed to measure ¹⁸O and three to measure mRNAs' coding for pro-inflammatory mediators. Each concentration had three technical replicates that were averaged. Each experiment used cells originating from bronchial brush biopsies from a separate volunteer that participated in the in vivo exposure experiments. Three wells (technical replicas) were exposed. Cells were expanded to passage three in bronchial epithelial growth medium (Lonza, San Diego, CA). This media contained a background level of 0.83 mg/ml protein, 0.16nM retinoic acid, and no ascorbic acid. Cells were plated on Costar Trans-Clear filter supports with a 0.2µm pore size (Costar Corp., Cambridge, MA) and inserted into 12-well tissue culture plates (22.1 mm diameter wells; Corning Inc., Wilkes-Barre, PA). Each filter support was pre-coated by addition of 0.2 ml of 40 ug/ml collagen solution. Culture media under the interface (1 ml volume) was enriched to 0.1uM retinoic acid after the cells reached 100% confluence. Air-liquid interface culture conditions were initiated 24–48 h after the addition of retinoic acid by removing the apical medium. The cells were maintained at air-liquid interface conditions by replacing the basolateral medium (containing 0.1uM retinoic acid) every 48 h for 4 days. The last medium exchange was always done immediately prior to the ¹⁸O₃ exposure. At this stage of passage, cells were not yet producing a measurable mucus layer on their surface.

In vitro ¹⁸O₃ exposure of epithelial cells

Cells were exposed to four ¹⁸O₃ concentrations and an air control (0.0, 0.25, 0.50, 0.75, or 1.0 ppm ¹⁸O₃) for 2 h on the same day. No lid was present and atmospheres consisted of filtered air or ¹⁸O₃ in 5% CO₂ and maintained at 37°C and 88% relative humidity. Previous studies by us and others have shown that these concentrations do not cause more than minimal (5% or less) decreases in cell viability as measured with lactate dehydrogenase release assay. The exposure chambers were 30 × 30 × 40 cm stainless steel boxes with airflow passing from top to bottom through porous shelves. They were housed inside Nuaire Incubators (Plymouth, MN) that kept the temperature constant and provided the necessary CO₂ analyzer. Air flowing through the chambers originated from an oil-less compressor that was passed through a purifier (Aacdo Medical Inc., Cleves, OH) and a humidifier (Pure Steam, Chaska, MN) and pumped from the chamber using a Fox Venture Eductor (Dover, NJ). ¹⁸O₃ was generated similar to the in vivo exposures except that the ¹⁸O₃ generator was taken from a Bendix NO/NOx analyzer (Lewisburg, WV). Chamber concentrations of ¹⁸O₃ were monitored using TECO 49i O₃ analyzers (Thermo Science, Franklin, MA) and controlled by altering the flow rate of an argon/¹⁸O₂ mixture passing through the O₃ generator.

In both in vivo and in vitro exposures, there was a small excess (100–200 ppm) of ¹⁸O₂ resulting from incomplete conversion of ¹⁸O₂ to ¹⁸O₃ (conversion efficiency is 1–2%). We have shown that ¹⁸O₂ exposure at 21% results in ¹⁸O incorporation, but that ¹⁸O₃ incorporation is 142,000- to 210,000-fold higher on a per ppm basis (Supplementary table 1). This difference in incorporation appears to be the result of higher chemical reactivity of the ¹⁸O₃ and is evident even though the incorporation of ¹⁸O₂ into cells also includes uptake resulting from normal metabolic processes.

Tissue and BALF Fraction Preparation for ¹⁸O Analysis

BALF cell pellets were resuspended in 0.2 ml of 0.025% SDS. BALF supernatants were combined and centrifuged for 1 h at 22,000 × g to obtain a white “surfactant pellet.” The resulting dried supernatant contained proteins and lipids mixed in a high NaCl matrix that could interfere with measurement of ¹⁸O. For this reason, we precipitated the protein by adding trichloroacetic acid (TCA) to a final concentration of 10% and centrifuged again (40,000 × g for 1 h at 4°C). The surfactant and TCA pellets were resuspended in 0.4 ml of the 0.025% SDS solution.

Immediately following the in vitro exposures, cells were analyzed for ¹⁸O content. This time point was chosen to minimize metabolism of the ¹⁸O signal by cells and to coincide with the same time point in which ozone-induced changes in mRNAs coding for pro-inflammatory proteins have been observed. Cells were lysed by the addition of 0.1 ml of 0.05% SDS to the apical surface. The cells were dislodged from the filter support using a pipette tip and the lysate transferred to a cryovial. The filter support was rinsed with an additional 0.1 ml of water and this was pooled with the original 0.1 ml of cell lysate for a final SDS concentration of 0.025% in the final sample. Epithelial cell SDS lysates from both the in vivo and in vitro¹⁸O₃ exposures and BALF cells from the in vivo exposures were stored at −80°C, then lyophilized and stored with desiccant at 4°C.

Analysis of ¹⁸O in lyophilized samples has been described previously (Hatch et al., 1994; Santrock et al., 1989). Briefly, lyophilized samples weighing 0.5–1.5 mg were placed into silver cups (3 × 5 mm), crimped, and then placed into an elemental analyzer (Carlo Erba Instruments, Italy) where all oxygen in the samples was converted to CO under a stream of pure helium. The elemental analyzer quantified the CO which yielded a percentage (per dry weight) of total elemental oxygen in each sample. The helium/CO effluent of the elemental analyzer was captured and passed by continuous flow through a heated (110°C) column of granular I₂O₅ (to convert CO to CO₂), then a cryogenic trap (−60°C to remove produced I₂), and then bled by capillary into the vacuum of an isotope ratio mass spectrometer (SIRA 10, VG instruments) where the ratios between CO₂ masses were determined. ¹⁸O₃ exposed samples were compared with their natural abundance controls inserted into each run.

Quantitative RT-PCR for Detection of Gene Expression Changes

Gene expression changes of the inflammatory markers IL-8 and COX-2 in human airway epithelial cells were quantified using reverse transcription polymerase chain reaction (RT-PCR). Total RNA was isolated using an RNeasy Mini Kit according to manufacturer's instructions (Qiagen, Valencia, CA). RNA quantity was determined on a Nanodrop ND-1000 and RNA integrity was assessed using RNA6000 Nano chips on a Bioanalyzer 2100 (Agilent). Total RNA (100–200 ng) was reverse transcribed to generate cDNA using the High Capacity cDNA Reverse Transcription kit (Life Technologies Applied Biosystems, Foster City, CA). Quantitative fluorogenic amplification of cDNA was performed using the ABI StepOnePlusTM Real-Time PCR System (Applied Biosystems) with iTaq Universal Probes Supermix (Bio-Rad), primer/probe sets for the target gene of interest multiplexed with b-actin (ACTB) primer/probe sets as the normalizing housekeeping gene (20× VIC/MGB, Primer Limited, Applied Biosystems). Values from triplicate technical replicates for each sample were averaged and data analyzed using the DDCt method to obtain fold change values following ozone exposure with respect to air exposure values.

Data Analysis

The “incorporation of ¹⁸O” is defined here as the measure of O₃ dose. A¹⁸O/¹⁶O ratio was derived from the relative CO₂ masses obtained from the mass spectrometer as a unitless “delta value.” The enrichment in ¹⁸O in the ¹⁸O₃ exposed samples was quantified and units changed to umoles of ¹⁸O/mole of total oxygen by subtracting the mean of the background natural abundance of ¹⁸O from all samples. This background, consisting of ∼0.2 atom % of all oxygen, was obtained from analysis of air-exposed cells in the case of in vitro data or from blood plasma samples of the subjects in the case of in vivo exposures. The denominator for the enrichment in ¹⁸O was then converted to per gram dry weight by use of the mean elemental percentage of oxygen (obtained as output from the elemental analyzer). This value was obtained from 10 subjects, each exposed to both clean air and ozone, and paired t-tests were used to determine statistical significance (p < 0.05). For in vitro exposure experiments, one-way ANOVA was used to calculate significance.

Analysis of gene expression performed on in vivo exposed samples involved paired t-test with each subject serving as his/her own control. Analysis of in vitro gene expression data was performed by ANOVA followed by Dunnett's test of individual means.

RESULTS

Methodology Findings

The cells obtained by brush biopsy were estimated to be greater than 95% epithelial cells. We originally explored trying to determine the percentage of different types of epithelial cells in the biopsies but they come off the brush in clumps making this impossible. The only cells that grow from these biopsies are the basal cells. Ciliated cells, mucus cells, clara cells, etc. are all terminally differentiated and do not divide in culture. Cell counts and differentials obtained from BALF are shown in Supplementary table 2. Percentage of polymorphonuclear leukocytes (PMNs) was increased by O₃ exposure at both the 1 h and 24 h BALF collection times.

We were able to obtain sufficient sample from a single culture well of confluent epithelial cells, or from the brush biopsy from one main bronchus. BALF cells and surfactant pellets yielded larger samples than were obtained from TCA pellets of BALF. Our ¹⁸O analysis required 0.5–1.5 mg of dry weight with a total oxygen percentage >10%. The bronchial biopsies barely met these requirements and for this reason the gene expression changes could not have been measured in the same samples as the ¹⁸O dose. The detection limit for the ¹⁸O assay depends on (1) the quantity of elemental oxygen in the sample (accuracy improves with higher oxygen/sample) and (2) number of replicate samples (accuracy improves with higher number of samples analyzed due to both “memory” from the previous samples and increased statistical power with larger sample numbers). Dry weights of brush biopsy samples are shown in Supplementary table 3. The ¹⁸O values for the left and right bronchial brush biopsies were averaged because there was no correlation between the two measurements. The use of SDS for lysis and dispersion of samples prevented sticking of dry sample to the walls of the tubes that facilitated weighing the samples into silver sampling cups after lyophilization. The contribution of SDS to the dry weight of the samples was ∼10% for the BALF cell and surfactant fractions, and 5% for the other samples.

Elemental Oxygen Percentages

Oxygen percentages that were used in converting the units of ¹⁸O dose as described above were as follows (mean ± SE): in vitro epithelial cells expt. 1, 17.0 ± 0.7; in vitro epithelial cells expt. 2, 17.2 ± 1.2; in vivo bronchial brush biopsies, 12.7 ± 0.9; BALF cells, 16.2 ± 1.1; BALF supernatant TCA pellet, 11.9 ± 0.8; BALF surfactant pellet, 18.0 ± 0.2; plasma 1, 23.8 ± 1.7; Plasma 2, 21.4 ± 1.1.

Comparison of In Vitro and In Vivo ¹⁸O₃ Incorporation

¹⁸O₃ exposure resulted in significant incorporation of ¹⁸O following the in vivo study and each ¹⁸O₃ concentration in the in vitro study (Fig. 2). The mean ± SE ¹⁸O values of bronchial brush biopsies taken after in vivo¹⁸O₃ exposure are plotted along with the concentration response of ¹⁸O data from the in vitro exposure. In vivo exposed cells showed lower ¹⁸O incorporation than in vitro exposed cells at the same ¹⁸O₃ concentration. For example, the ¹⁸O incorporation in vivo was 64% lower than that seen at the same ¹⁸O₃ level (0.3 ppm) on the dose response curve of the in vitro exposed cells. We were unable with the present sample size to observe any relationships between ¹⁸O incorporation and subject age, sex, or race. Variability observed in the in vitro data could have been the result of (1) differences between the subjects from which cells were obtained, (2) variation in the two ¹⁸O analysis runs, or (3) between the replicate culture wells exposed at each ¹⁸O₃ concentration. The variability in natural abundance of ¹⁸O (zero ppm ¹⁸O₃) was small compared with variability among ¹⁸O₃ exposed cells.

FIG. 2.

In vitro and in vivo dose comparison: ¹⁸O incorporation into bronchial brush biopsies collected from human subjects exposed for 2 h to 0.3 ppm ¹⁸O₃ while exercising intermittently or human primary epithelial cells exposed in vitro to a range of ¹⁸O₃ concentrations (mean ± SE). Standard error bars are within the data point for the in vivo data (N = 10 human subjects for in vivo data and 5 culture wells derived from two replicate in vitro exposures to each ¹⁸O₃ concentration for in vitro data). Each mean shown is significantly elevated (* denotes p < 0.05) relative to natural background ¹⁸O (zero ¹⁸O₃).

Comparison of In Vitro and In Vivo O₃ Biological Effects

Figure 3 shows the fold increases (±SE) in gene expression of IL-8 and COX-2 in bronchial brush biopsies from in vivo O₃ exposures and primary epithelial cell cultures exposed in vitro to the same range of O₃ concentrations as was used for the ¹⁸O₃ study. Similar to the ¹⁸O data, the in vivo gene expression was lower than that observed following in vitro exposure. The IL-8 increase in expression in the in vivo exposed cells fell about 30% below that observed in vitro and the in vivo COX-2 response fell about 17% below the response observed in vitro. The in vitro expression of IL-8 was higher than the expression of COX-2 but both genes were induced in a concentration-related manner that leveled off at 0.75 and 1.0 ppm O₃. Although epithelial cells obtained from subjects exposed in vivo showed elevations in gene expression smaller than were observed at the lowest dose (0.25 ppm) used in the in vitro study, the increases in both genes were significant by paired t-test of the nine subjects studied.

FIG. 3.

In vitro and in vivo biological effects comparison of O₃ exposures under conditions similar to the dose comparison in Figure 2. Shown are the fold increases in expression of IL-8 or COX-2 genes (means ± SE) in bronchial brush biopsies obtained from nine human subjects exposed in vivo to O₃ compared with their response following exposure to clean air. Also shown are fold increases of primary epithelial cells exposed in vitro to each O₃ concentration (means ± SE for three subjects and three replicates per subject) compared with their response following exposure to clean air. Asterisks denote significance (p < 0.05).

Comparison of ¹⁸O in Bronchial Brush Biopsies to Three BALF Fractions

¹⁸O incorporation measured after in vivo exposure to ¹⁸O₃ resulted in higher ¹⁸O incorporation into cell and surfactant fractions of BALF than bronchial brush biopsies from the same subjects (Fig. 4). The BALF surfactant incorporated the highest levels of ¹⁸O, which had a mean value slightly higher than the BALF cells, fourfold higher than BALF TCA pellet, and about eightfold higher than the mean of the left and right bronchial brush biopsies. A subject-by-subject plot of the relationship between the four tissue pools is shown in Supplementary figure 1.

FIG. 4.

Comparison of ¹⁸O incorporation into epithelial cells (shown in Fig. 2) and BALF constituents of the same subjects following exposure to ¹⁸O₃ (0.3 ppm, 2 h with intermittent exercise). Means ± SE are shown for 10 subjects. All ¹⁸O₃ exposed tissues had significantly elevated ¹⁸O when compared with the natural abundance ¹⁸O concentration of blood plasma.

Relationships Between Different In Vivo Sample ¹⁸O Measurements

No relationship was observed between the ¹⁸O incorporation in the left and right bronchial biopsies. A search for relationships across the 10 human subjects in the measured values for ¹⁸O incorporation in different tissue compartments yielded the result that BALF surfactant ¹⁸O increased in proportion to BALF cell ¹⁸O (R² = 0.73; see Supplementary fig. 2) but not between other BALF constituents (R² < 0.5). The concentration of ¹⁸O₃ measured at the mouthpiece of the subjects (which ranged from 0.21 to 0.33 ppm) did not appear to be correlated with the ¹⁸O accumulation in any of the individual tissue or BALF constituents or any combination of the sum of the ¹⁸O concentrations in BALF.

DISCUSSION

We achieved a proof-of-concept for obtaining comparable measurements of O₃ dose and effect between in vivo exposed epithelial cells and in vitro exposed primary epithelial cells exposed in an air-liquid interface culture. Prior to this, few, if any, equivalent measurements for an air pollutant gas had been identified and compared under both in vivo and in vitro conditions. Problems have included finding equivalent denominators applicable to both types of exposure. For example, measurements of released cytokines or LDH depend on equivalent volumes of extracellular fluid that cannot be controlled equally under in vivo and in vitro conditions. Although the main goal of the study was to compare the dose and response in epithelial cells, some valuable insights were obtained about the relative O₃ dose in different cell types. Epithelial cell production of inflammatory mediators appears to be easily measured by gene expression assays, and primary human epithelial cells appear to be a good model that could be expanded in air-liquid interface cultures used in high throughput assays.

This is the first report of ¹⁸O incorporation into human bronchial epithelial cells in vivo or into any type of cell exposed to ¹⁸O₃ in vitro. Our previous studies examined BALF constituents of humans and rats exposed in vivo to ¹⁸O₃ but did not include epithelial cells (Hatch et al., 1994). It was interesting that both the O₃ dose and effect observed in vivo always fell below that observed in the in vitro exposures at the same air concentration of O₃. Previous studies that measured O₃ air concentrations in the posterior pharynx of exercising human subjects found that fractional removal of O₃ in the nose is ∼40% (Gerrity et al., 1994). Thus, scrubbing of O₃ in the upper airways might explain the lower dose and effect of O₃ in vivo compared with in vitro. Other factors that could have influenced the relative dose and effect between in vivo and in vitro exposures include the following. (1) Brush biopsy of cells in vivo probably removed some underlying cells that would have a lower exposure to O₃ than the surface cells, (2) there may have been differences in the competition of the ¹⁸O₃ reaction with antioxidants that do not form addition products with ¹⁸O₃ (see below), or (3) a longer (1 h vs. minutes) delay following in vivo exposure compared with in vitro exposure could have resulted in some clearance of ¹⁸O or a return to normal gene expression.

It is likely that the bronchial brush biopsy samples measured here gave a low approximation of epithelial cell dose compared with the distal airways that appear to be the main targets for O₃ in the human lung. A prior study of resting Rhesus monkeys showed that in vivo exposure (0.4 or 1.0 ppm ¹⁸O₃ for 2 h) resulted in bronchial cell levels of ¹⁸O similar to those observed here for bronchial cells of exercising humans (Plopper et al., 1998): the bronchial cells of monkeys incorporated 4–7 ug ¹⁸O/g dry while the human bronchial cells incorporated ∼10 ug ¹⁸O/g dry. The smallest airways of the monkey (respiratory bronchioles) had a 2–4-fold higher ¹⁸O incorporation than cells of the main bronchus suggesting that human terminal airways would be higher as well. This higher dose deeper in the lung is supported by the 4–8-fold higher ¹⁸O incorporation observed here in the BALF constituents. Histopathology data on O₃ injury to epithelium and O₃ dosimetry uptake models also predict the highest O₃ dose and effect in the small terminal airways of all species (Barry et al., 1985; Miller et al, 1978). It would not be possible to biopsy tissues of these small airways in human subjects. The above-mentioned references also demonstrate that although we examined only one dose of ¹⁸O₃ in the present in vivo study, a dose response behavior should be expected. Taken together, our results suggest that the dose response we observed here in the in vitro exposed epithelial cells covered a reasonable range of doses for human airway epithelial cells exposed in vivo.

In addition to the epithelial cell data, we learned that BALF constituents experience a high dose of ¹⁸O₃ compared with bronchial epithelial cells. This difference might be due to the larger surface area exposed or biochemical differences in reactivity of the ¹⁸O₃. We have observed in other studies that low ascorbate is related to high ¹⁸O₃ levels in BALF (Gunnison and Hatch, 1999; Kari et al., 1997). The lack of ascorbate in the culture media used here would have resulted in a loss of ascorbate from cells during culture (Lane et al., 2013) which would have increased the incorporation of ¹⁸O₃ in in vitro cells compared with the cells exposed in vivo that are known to be bathed in extracellular ascorbate (Slade et al., 1993; van der Vliet et al., 1999). In an earlier study, we reported a 30–50% lower ¹⁸O incorporation into human BALF cells and surfactant pellets than we observed here. This was in spite of the fact that the earlier study involved a higher ¹⁸O₃ concentration (0.4 ppm, 2 h rather than 0.3 ppm, 2 h) and higher targeted minute ventilation (60 l/min/subject vs. 50 l/min/subject in the present study). One reason for this difference might be that the earlier study measured the ¹⁸O₃ concentration at a site distant from the breathing zone of the subjects so that surfaces for reaction with ¹⁸O₃ within the exposure chamber could have reduced the actual concentrations inhaled. In the present study, ¹⁸O₃ was measured in the breathing zone to better ensure that losses of ¹⁸O₃ on surfaces would not occur. The in vitro exposures might also be affected by this phenomenon, however, there appeared to be fewer surfaces for reaction with ¹⁸O₃ in the in vitro chambers than in the in vivo chambers. Another finding is that our earlier study employed dialysis of the BALF supernatant to separate the protein/lipid fraction from the NaCl present in the saline lavage—a necessity because high levels of NaCl can interfere with the ¹⁸O assay. The earlier reported ¹⁸O levels in the dialyzed BALF high speed supernatant were higher than the present results found in the TCA-precipitated BALF high speed supernatant. It appears that TCA treatment may have dissociated the ¹⁸O adduct, or failed to concentrate the same extracellular materials in BALF as did dialysis.

Limitations of the present study include the small number of in vitro replicate exposures as well as of human subjects contributing to the cultures. In addition, we observed significant intra- and inter-subject variations in ¹⁸O in cells removed by brush biopsy. Part of this variation may be attributable to the fact that not all the inhaled ozone will react with the surface of airway epithelial cells that are then removed by brush biopsy. Some will react with compounds present in lung lining fluid such as mucin or antioxidants and these compounds may differ in concentration and among people or even in the same person on a day-to-day basis. This is also true to some extent within a subject. It is not known if mucin or antioxidants are distributed evenly throughout the airways. Another factor that may be contributing to variability is that airway cells are not a monolayer and there is a possibility that some of the cells not directly opening into the airway and therefore not receiving a dose of ozone (e.g., basal cells) would be removed during brush scraping to varying degrees from one individual to another or even from different locations within an individual. It is not known if the makeup of cell composition (e.g., % of ciliated cells, mucin secreting cells, clara cells, basal cells, etc.) is homogenous throughout the airway. Finally, some variability was probably due to the limited mass of sample that made it impossible to obtain more than one ¹⁸O analysis for each bronchus sampled. Although the BALF cells sampled after in vivo exposure were obtained in a higher sample mass than the tissue mass of bronchial brush biopsies, we believe that the overall contribution of epithelial cells to the inflammatory response would be greater because of a larger surface area exposed to O₃ and a higher secretion of inflammatory cytokines in epithelial cells than other cells (Devlin et al., 1994). Future studies should compare the sensitivity to injury of alveolar macrophages and epithelial cells and also determine the contribution of mucus or epithelial lining fluid covering the cells to the resulting dose and effect of O₃. Another potential limitation is the possibility that the subjects studied had been exposed previously to ozone and as such might have experienced epigenetic effects that could alter the cell responses following in vitro ozone exposure. No attempt was made to control for this possibility because there are so many other outside influences that could also induce epigenetic effects, and because we presently have no evidence for this type of change in the scenario we have studied here.

In summary, our results suggest that in vitro exposure conditions can be made to reasonably match the in vivo dose of O₃ to epithelial cells exposed in vivo and that comparable measures of dose and effect can be obtained. These studies add confidence to the possibility of extrapolating effects observed in vitro to effects that might occur in human airways in vivo.

SUPPLEMENTARY DATA

Supplementary data are available online at http://toxsci.oxfordjournals.org/.

FUNDING

United States Environmental Protection Agency intramural funds and a cooperative agreement (CR833463-01) between the U.S. EPA and the Center for Environmental Medicine, Asthma and Lung Biology at the University of North Carolina at Chapel Hill.