Effects of cellular differentiation in human primary bronchial epithelial cells: Metabolism of 4-(methylnitrosamine)-1-(3-pyridyl)-1-butanone

Abstract

Many of the toxicants in tobacco smoke undergo biotransformation in the lungs of smokers, both to reactive and to detoxified derivatives. Human air-liquid-interface (ALI) airway tissue models have emerged as an advanced in vitro model for evaluating the toxicity of inhaled substances; however, the metabolic potential of these cultures has not been evaluated extensively. In this study, we compared the metabolic activities of an ALI tissue model to the undifferentiated normal human primary bronchial epithelial (NHBE) cells from which it was derived. Measurement of the basal levels of gene expression for 84 phase I drug metabolism enzymes indicated that most genes were upregulated in ALI cultures compared to NHBE cells. Furthermore, the enzymatic activities of three cytochrome P450s involved in the bioactivation of tobacco-specific nitrosamines were higher in the ALI cultures, and the bioactivation of 4-(methylnitrosamine)-1-(3-pyridyl)-1-butanone (NNK), as measured by the formation of two of its major metabolites, i.e., keto acid and 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL), was significantly greater in the ALI cultures. Finally, NNK was a direct-acting genotoxicant in the ALI cultures, while the genotoxicity of NNK was detected in NHBE cells only in the presence of an exogenous liver S9 activation system. Taken together, our findings demonstrate the greater metabolic potential of well-differentiated ALI cultures than primary NHBE cells, supporting the potential use of ALI airway cultures as an alternative in vitro model for evaluating inhaled toxicants that require metabolic transformation.

1. Introduction

Animal models traditionally have been used for predicting human health risks from chemical exposures. Along with the cost and limitations of animal models, it would be useful to improve the scientific evidence supporting extrapolation from animal data to human data and consider developing alternatives that may supplement, and potentially reduce and replace animal testing (Hartung, 2008; Berube et al., 2010). Consistent with this thinking, FDA supports reducing the reliance on animal testing where adequate and scientifically valid non-animal alternatives are available. The well-differentiated human air-liquid-interface (ALI) airway tissue model, derived from normal human bronchial epithelial (NHBE) cells, has a close structural and functional resemblance to in vivo airway epithelium (Dvorak et al., 2011; Baxter et al., 2015). Such tissue models, therefore, represent a promising in vitro alternative testing system with desirable characteristics for evaluating the toxicity of inhaled chemicals.

Besides conducting gas exchange, the lung also has an important role in the biotransformation of inhaled substances (Tierney, 1974). For instance, drug metabolizing cytochrome P450 (CYP) enzymes are expressed in the lungs, although their expression patterns are substantially different from those in the liver and intestine (Olsson et al., 2011). Inhaled toxicants, including many environmental pollutants and harmful chemicals present in tobacco smoke, are either detoxified or bioactivated via various metabolic pathways, including those present in the lung (Moorthy et al., 2015; Anttila et al., 2011). Bioactivation of these toxicants is believed to contribute to the initiation and development of respiratory diseases. Thus, if ALI tissue cultures are to serve as an in vitro alternative for inhalation toxicity testing, their metabolic capability should be better understood.

Although the human ALI airway tissue models have been extensively characterized using both genomic and proteomic approaches (Dvorak et al., 2011; Baxter et al., 2015; Pezzulo et al., 2011), their metabolic potential remains largely unexplored. Recently, several studies have demonstrated that at least some CYPs are preserved in ALI airway cultures (Newland et al., 2011; Baxter et al., 2015; Boei et al., 2017). However, most existing studies using ALI airway tissue models have focused on gene expression and catalytic activities. Little direct testing of the effects of metabolic conversion on the toxicity of xenobiotics has been conducted.

4-(Methylnitrosamino)-1-(3-pyridyl]-1-butanone (NNK) is a potent procarcinogen present in tobacco smoke (IARC, 2012). NNK exposure correlates in a dose-dependent manner with the development of lung adenocarcinoma, a common type of lung cancer found in habitual smokers (Hecht, 1999a; Yuan et al., 2009). IARC has concluded that NNK is a human carcinogen (Group 1) based on mechanistic evidence, including the formation of DNA adducts in taobacco users. Bioactivation via carbonyl reductase 1-mediated reduction and CYP-catalyzed α-hydroxylation results in the formation of intermediate α-hydroxylated metabolites, which subsequently react with DNA to form mutagenic adducts (Morse et al., 1990; Maertens et al., 2010). CYP2A6, CYP2A13, and, to a lesser extent, CYP2B6 are the major CYPs involved in the metabolic transformation of NNK that contributes to its carcinogenicity, including tumorigenicity in the lung (Takeuchi et al., 2003; Weng et al., 2007; Zhang et al., 2007; Hecht, 1999b; Smith et al., 2003; Anntila et al., 2011).

In this study, we evaluated the effects of ALI tissue development on their metabolic potential. Basal gene expression profiles for 84 phase I drug metabolism enzymes were compared in well-differentiated ALI tissue models and the matching undifferentiated NHBE cells from which ALI cultures are derived. To further demonstrate the metabolic capability of these cultures, NNK was used as a model compound and the gene expression and metabolic activity of the CYPs implicated in NNK metabolism were measured. Metabolic transformation by carbonyl reductase I and NNK-related CYPs inferred from gene expression analysis, were confirmed by enzymatic activity assays and by quantifying the amount of 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL) and keto acid, two of the major metabolites of NNK, released into the apical washes and culture media. Furthermore, DNA damage in the ALI cultures caused by the active NNK metabolites was demonstrated using the alkaline comet assay.

2. Materials and methods

2.1. Chemicals and reagents

NNK (purity > 99%), nicotinamide adenine dinucleotide phosphate (NADP), glucose-6-phosphate dehydrogenase (G6PD), human placenta collagen Type IV, and SIGMAFAST™ Protease Inhibitor Cocktail were purchased from Sigma-Aldrich (St. Louis, MO). D3-NNK, D3-NNAL, NNAL, coumarin, 7-hydroxy coumarin, 7-hydroxy coumarin-D5, bupropion, 4-hydroxy bupropion, 4-hydroxy bupropion-D6, and keto acid were purchased from Toronto Research Chemicals (Ontario, Canada). β-Glucuronidase from Helix promatia was obtained from MP Biomedical (Solon, OH). KCl, MgCl₂, CaCl₂, Na₃PO₄, and Tris-base were purchased from J.T. Baker (Center Valley, PA). Comet lysis buffer and low-melting agarose were purchased from Trevigen (Gaithersburg, MD). Aroclor-1254-induced male Sprague-Dawley rat liver post-mitochondrial fraction (S9) was purchased from MOLTOX (Boone, NC). Rabbit anti-carbonyl reductase 1 antibody and mouse anti-GAPDH were purchased from Santa Cruz Biotechnology (Dallas, TX). Rabbit anti-CYP2A13 antibody was obtained from Cell Signaling Technology (Danvers, MA). The RNeasy Mini Kit, Omniscript^® Reverse Transcription kit, and OligodT primers were purchased from QIAGEN (Valencia, CA). Real Time-PCR (RT-PCR) primers were synthesized by Integrated DNA Technologies (Coralville, IA). PneumaCult™-Ex Medium Kit, PneumaCult™-ALI Medium Kit, 0.2% heparin solution, and 0.05% Trypsin-EDTA were purchased from STEMCELL Technologies (Vancouver, Canada). Airway Epithelial Cell Basal Medium and Bronchial Epithelial Cell Growth Kit were obtained from ATCC (Manassas, VA). M-PER Mammalian Protein Extraction Reagent and BCA Protein Assay were purchased from Pierce (Pittsburgh, PA). NuPage^® Novex^® 4–12% Bis-Tris gradient gels and SYBR^® Gold Nuclei Acid Gel Stain were purchased from Life Technologies (Carlsbad, CA). Odyssey Blocking Buffer and IRDye-conjugated secondary antibodies were obtained from LI-COR (Lincoln, NE).

2.2. Cell culture

Human ALI airway tissue models were established using the PneumaCult™-ALI Medium Kit as described previously (Cao et al., 2017). Cryopreserved primary normal human bronchial epithelial (NHBE) cells (Cystic Fibrosis Center Tissue Procurement and Cell Culture Core, University of North Carolina, Chapel Hill, NC) were expanded on bovine collagen Type I-coated tissue culture dishes in PneumaCult™-Ex Medium until the cells reached 90% confluence. The cells then were detached in 0.025% Trypsin-EDTA, pelleted by centrifugation at 1500 rpm for 5 min at 4 °C, and resuspended at a concentration of 4 × 10⁵ cell/mL. One hundred μL cell suspensions were seeded onto 24-well Transwell inserts (Corning, Tewksbury, MA) coated with human placenta collagen Type IV, with PneumaCult™-Ex expansion medium added to both the apical and basolateral compartments. The cells were grown submerged until they reached 100% confluence. Differentiation then was initiated by feeding the cultures from the basolateral chamber only with PneumaCult™-ALI Maintenance Medium. Medium was changed every other day for approximately 4 weeks, by which time the cultures became fully differentiated (Prytherch et al., 2011).

Primary NHBE cells used for making the monolayer cultures were maintained on 100-mm tissue culture dishes in ATCC Airway Epithelial Cell Basal Medium supplemented with the Bronchial Epithelial Cell Growth Kit. The cells were fed every other day and passaged when they reached approximately 90% confluence. For the NNK treatment, NHBE cells were seeded at a density of 3 × 10⁵ cells/well on a 6-well multiwell plate and cultured until confluence.

2.3. NNK treatment

A stock solution of 1M NNK was prepared in DMSO (neat). Treatment solutions were made by diluting the stock solution in Dulbecco's phosphate-buffered saline (DPBS) or ATCC Airway Epithelial Cell Growth Medium to the desired concentrations for treating the ALI cultures or NHBE cells, respectively. Cells were washed once with DPBS before the treatment. Thirty-five μL of the treatment solutions were added to the apical side of the ALI cultures and 1 mL treatment medium was added to the NHBE cells; cells were treated for 1 and 6 h. Concentrations of DMSO in the treatments were maintained at 0.1% for all groups. At the end of the treatment, the cells were washed once with DPBS; apical washes, basolateral media, and cell pellets were collected. The cell pellets were lysed in 10mM ammonium bicarbonate buffer. Samples were stored at −80 °C until chemistry analysis.

2.4. RNA isolation

Total RNAs from ALI cultures and NHBE cells were isolated using an RNeasy^® Mini Kit and by following the user's manual. The yield of RNA was determined spectrophotometrically by measuring its optical density at 260 nm; its purity and quality were evaluated using the Agilent RNA 6000 Nano kit on a 2100 Bioanalyzer (Santa Clara, CA). RNAs with integrity numbers > 9.5 were used for the study.

2.5. Human phase I drug metabolism enzyme RT² Profiler™ PCR array

cDNA was synthesized using 1 μg total RNA with random primers in a reaction volume of 20 μL at 42 °C for 15 min and subsequently inactivated at 95 °C for 5 min. The cDNA was diluted to 110 μL in nuclease-free water and stored at 20 °C. The human phase I enzyme RT² Profiler™ PCR array contains a total of 84 drug metabolizing genes and 5 housekeeping genes (http://www.sabiosciences.com/rt_pcr_product/HTML/PAHS-068A.html). For one 96-well PCR array, 2700 μL PCR master mix containing 1 × SYBR^® Green qPCR master mix were mixed with 102 μL diluted cDNA; 25 μL of the reaction mixture were added to each well. The qPCR was carried out using a ViiA™ 7 Real-time PCR System (Applied Biosystems, Carlsbad, CA) and the universal cycling conditions of 10 min at 95 °C, followed by 40 cycles of 15 s at 95 °C and 1 min at 60 °C.

Five housekeeping genes, i.e., β-2 microglobulin (B2M), hypoxanthine phosphoribosyltransferase 1 (HPRT1), ribosomal protein L13a (RPL13A), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and β-actin (ACTB), were used for normalization. The cycle threshold (Ct) of each gene was normalized to the average Ct of the 5 housekeeping genes to obtain a ΔCt. The comparative Ct method was used to calculate the relative expression of each gene. Fold change was calculated using the equation, 2^ΔΔCt (Livak and Schmittgen, 2001). Differentially expressed genes were identified using a two-tailed t-test. Statistical calculations were based on the ΔCt values. Hierarchical Cluster Analysis (HCA) was conducted using ΔCt values and the Array Track software (version 3.5.0; NCTR, Jefferson, AR) (http://www.fda.gov/nctr/science/centers/toxicoinformatics/ArrayTrack/).

2.6. Quantitative real-time PCR

The relative expression of 8 genes involved in NNK metabolism was evaluated using quantitative real-time PCR (qRT-PCR). Briefly, 1 μg total RNA was converted to cDNA using the Qiagen Omniscript^® Reverse Transcription kit. Target genes then were amplified using the QIAGEN RT² SYBR^® Green qPCR kit and the primer pairs listed in Table 1. β-Actin was used as the housekeeping gene for normalizing expression.

Table 1.

Primers for the qRT-PCR.

Gene Symbol	Forward primer (5′– 3′)	Reverse primer (5′ – 3′)	Amplicon size (bp)
CYP2A6	ccctcatgaagatcagtgagc	gcgctccccgttgctgaata	200
CYP2A13	caccctgcgctacggtttcc	gtcgatatccttaggcgactgagg	204
CYP2B6	atcgatctgacaccccagga	cctggaatcctttgaccccc	109
Carbonyl reductase 1	ctgatcccacaccctttcat	ttaagggctctgacgctcat	151
11-β HSD1	aactgaggaagttgacttcca	gaattcagaccagagatgctc	364
AKR1C1	gtaaagctttaaggccac	cacccatgcttcttctcgg	598
AKR1C2	gtaaagctctagaggccgt	cacccatggttcttctcga	598
AKR1C4	acagagctgtagaggtcac	cacccatagtttatgtcgt	598
β-Actin	acggctccggcatgtgcaag	tgacgatgccgtgctcgatg	198

qRT-PCR was performed on a ViiA™ 7 Real-time PCR System. The cycling protocol was a 10-min hold at 95 °C, followed by 40 cycles of 15 s at 95 °C, 15 s at 60 °C, and 30 s at 72 °C. Melting curve analysis was used to ensure amplification specificity. The expression of the genes was normalized to β-actin. The relative expression of the genes in ALI cultures to those in NHBE cells was calculated using the ΔΔCt method.

2.7. Immunoblotting

Whole cell lysates were prepared using the M-PER Mammalian Protein Extraction Reagent supplemented with 1× SIGMAFAST™ Protease Inhibitor Cocktail. Protein concentrations were determined using the Pierce BCA Protein Assay. Sixty μg protein lysates were denatured in a loading buffer containing 1× LDS Sample Buffer and 1× dithiothreitol for 10 min at 90 °C. Proteins were separated on a NuPage^® Novex^® 4–12% Bis-Tris gradient gel and then were transferred onto a nitrocellulose membrane. The membrane was blocked with the Odyssey Blocking Buffer for 1 h at room temperature, followed by incubation first with primary antibodies overnight at 4 °C and then IRDye-conjugated secondary antibodies for 30 min at room temperature. Proteins were visualized using the LI-COR Odyssey CLx Imaging System. The intensity of the bands was quantified by densitometry using LI-COR Image Studio Software.

2.8. P450 CYP activity assay

The basal activities of CYP2A6/2A13 and CYP2B6 in ALI cultures and NHBE cells were measured by quantifying the formation rate of hydroxylated metabolites of coumarin and bupropion, respectively, in the apical washes and basolateral media for the ALI cultures and media for NHBE cells. Briefly, the ALI cultures were incubated with 100 μM coumarin or 50 μM bupropion from the apical side for 5 h; NHBE cells were treated with the same concentrations of coumarin or bupropion for 5 h. To collect apical washes, the apical side of the ALI cultures was washed twice with 100 μL DPBS each time. The washes were pooled and centrifuged at 600 ×g for 10 min at 4 °C; the supernatant then was transferred to a microcentrifuge tube containing the corresponding basolateral media and stored at −80 °C until the chemistry analysis. Media from the NHBE cells were transferred to a microcentrifuge tube and centrifuged at 600 ×g for 10 min at 4 °C to remove the cell debris. Concentrations of the 7-hydroxy coumarin and 4-hydroxy bupropion were measured by UPLC-MS, as described previously (Newland et al., 2011). The cell lysate from both ALI cultures and NHBE cells was extracted using 10mM ammonium bicarbonate buffer. Protein concentrations were determined using the BCA Protein Assay and used for calculating the metabolite formation rate. To determine the concentration of 7-hydroxy coumarin, the pH of the basolateral medium first was adjusted to 5.5 using HCl and the acidified medium then was incubated with 7000 units/mL β-glucuronidase for 16 h at 37 °C. Following the enzymatic digestion, 10 volumes of dichloromethane were added to each sample. The reactions were air-dried and reconstituted with 1 volume of acetonitrile before the LC/MS analysis. To determine the concentration of 4-hydroxy bupropion, the medium (apical washes and basolateral medium for ALI cultures and media for NHBE cells) was used directly without enzymatic digestion. One hundred ng/mL 7-hydroxy coumarin-D5 or 4-hydroxy bupropion-D6 internal standards were spiked into each sample, for detecting 7-hydroxy coumarin or 4-hydroxy bupropion, respectively. Both metabolites were quantified using a Waters e2695 Alliance UPLC/MS (Milford, MA).

2.9. Keto acid analysis by LC-MS/MS

The levels of keto acid were quantified by LC-MS/MS, as previously described with minor modifications (Lee et al., 2007). Briefly, apical washes and basolateral media were collected separately, each added into 4 volumes of acetonitrile containing 0.1% formic acid, vortex mixed, and centrifuged at 14,000 ×g for 5 min. Ten μL of the supernatant then were injected onto a Shimadzu Prominence UFLC coupled with an AB Sciex 3200 QTRAP mass spectrometer. The analyte was eluted at a flow rate of 0.5 mL/min on a Waters Atlantis T3 C18 column (4.6 × 150 mm, 5 μm) at 40 °C using a gradient mobile phase containing solvent A (10 mM ammonium acetate) and solvent B (acetonitrile), both containing 0.1% formic acid. Elution started with 10% solvent B for 0.5 min, followed by a linear gradient of 10% to 90% solvent B in 9.5 min, returning to 5% solvent B in 0.5 min; this solvent mixture was maintained for 4.5 min to re-equilibrate the column. The eluates were monitored by QTRAP mass spectrometry with the positive electrospray mode (ESI+) using multiple reaction monitoring (MRM). The MRM transition m/z 180.147 to 134.200 was used for quantitation of keto acid. One additional transition, m/z 180.147 to 80.000, was used for analyte confirmation. The experimental parameters were optimized as follows: Curtain gas (CUR): 50; IonSpray (IS) Voltage: 4000 V; Temperature (TEM): 600 °C; Ion Source Gas 1 (GS1): 60; Ion Source Gas 2 (GS2): 80; CAD gas: medium; Declustering Potential (DP): 30; Entrance Potential (EP): 7; Collision Energy (CE): 30 V; and Collision Cell Exit Potential (CXP): 4. Keto acid was quantified using AB Sciex Analyst 1.6 software and a linear calibration curve of 0.625 μg/mL to 40 μg/mL keto acid.

2.10. NNK and NNAL analysis by HPLC-MS

Concentrations of NNAL were quantified by HPLC-MS as previously described (Lee et al., 2007), with minor modifications. Fifty μL samples or calibration standards were added to 200 μL acetonitrile (with 0.1% formic acid) containing 100 ng/mL D3-NNK and 50 ng/mL D3-NNAL internal standards. The samples were vortex mixed and centrifuged at 14,000 ×g for 5 min. Ten μL supernatant were injected into a Waters e2695 Alliance HPLC System coupled with an ACQUITY QDa Mass Detector. The analytes were eluted on a Waters Atlantis T3 C18 column (4.6 × 150 mm, 5 μm) at 40 °C using gradient elution at a flow rate of 0.5 mL/min with a mobile phase composed of 0.1% formic acid (solvent A) and HPLC-grade acetonitrile, 0.1% formic acid (solvent B). The mobile phase was initially 10% solvent B, followed by a 15-min linear gradient increase to 90% solvent B, then returned to 10% solvent B using a 0.5-min linear gradient, and equilibrated for 4.5 min. The eluate was monitored by mass spectrometry with the electrospray ion source operating in the positive ion mode (ESI+) and Select Ion Recording (SIR). The monitored (M + H)⁺ ions were m/z 208.2 for NNK, m/z 211.2 for D3-NNK, m/z 210.2 for NNAL, and m/z 213.2 for D3-NNAL. The levels of NNK and NNAL were quantified by comparing the peak intensity of the analytes with their deuterated internal standards using a calibration curve generated by the Waters Empower 3 software.

2.11. Alkaline comet assay

DNA damage was measured as strand-breaks using the alkaline comet assay. Cells were washed with DPBS, dislodged by trypsinization, and resuspended in ice-cold DPBS to obtain a single cell suspension. The cell suspension then was mixed with low-melting agarose at a ratio of 1:10 (v/v) and the mixture was evenly spread onto a Trevigen CometSlide™ (Gaithersburg, MD). Single cells were lysed by immersing the slides in the lysis solution for 2 h at 4 °C. Following electrophoresis at 1 V/cm for 20 min in freshly prepared electrophoresis buffer (200 mM NaOH and 1 mM EDTA, pH > 13.0), the slides were neutralized in 0.4 M Tris-HCl (pH 7.5), fixed in 70% ethanol, and stained with SYBR^® Gold Nucleic Acid Gel Stain. Air-dried slides then were stored at 4 °C in the dark until scoring. Comet images were analyzed using a Leica fluorescence microscope (Buffalo Grove, IL) and the Comet Assay IV software (Perceptive Inc., UK). Percent DNA in the tail was used as the metric for quantifying DNA damage. One hundred and fifty randomly selected cells were analyzed for each sample.

An S9 metabolizing system was prepared by diluting the S9 liver fraction from Aroclor-1254-treated male rats in a reduced NADPH-generating mix (50 mM Na₃PO₄, pH 8.0, 30 mM KCl, 10 mM MgCl₂, 4 mM NADP, 5 mM glucose-6-phosphate, 10 mM CaCl₂) at a ratio of 1:4. The final concentration of S9 protein was 2 mg/mL.

2.12. Statistical analysis

Results are expressed as the mean ± standard error of the mean (SEM). All analyses were conducted using GraphPad Prism (La Jolla, CA). Statistical significance was determined using one-way ANOVA followed by the Dunnett's test. p ≤ .05 was considered statistically significant.

3. Results

3.1. Differentially expressed phase I drug metabolism genes in NHBE cells and ALI cultures

A qRT-PCR-based array was used to examine the basal expression of 84 phase I drug metabolism genes in undifferentiated NHBE cells and well-differentiated ALI cultures. The NHBE cells and ALI cultures were derived from the same donor to eliminate donor-specific effects. Relative fold-changes were calculated by comparing the expression of the genes in ALI cultures to those in NHBE cells; this information, along with the average Ct values from the experimental replicates and p values, are tabulated in Table 2. Genes with fold-changes equal to or > 2 and p values < .05 were considered differentially expressed between NHBE cells and ALI cultures.

Table 2.

Relative abundance of the phase I drug metabolism genes between ALI cultures and NHBE cells.

Gene Bank ID	Description	Gene Symbol	Average Ct ALI	Average Ct NHBE	Fold Change ALI/NHBE	p Value
Up-regulated genes
NM_000783	Cytochrome P450, family 26, subfamily A, polypeptide 1	CYP26A1	20.9 ± 0.5	33.6 ± 0.4	10964.3 ± 3955.6	2.18E-08
NM_000774	Cytochrome P450, family 2, subfamily F, polypeptide 1	CYP2F1	22.2 ± 0.5	33.2 ± 0.6	3114.5 ± 590.3	5.32E-08
NM_000779	Cytochrome P450, family 4, subfamily B, polypeptide 1	CYP4B1	19.7 ± 0.5	29.7 ± 0.5	1620.4 ± 363.0	3.82E-08
NM_000689	Aldehyde dehydrogenase 1 family, member A1	ALDH1A1	18.1 ± 0.6	27.6 ± 0.4	1101.6 ± 167.0	8.49E-10
NM_000669	Alcohol dehydrogenase 1C (class I), gamma polypeptide	ADH1C	21.6 ± 1.0	30.8 ± 0.6	918.5 ± 203.5	7.45E-08
NM_001460	Flavin containing monooxygenase 2	FMO2	25.3 ± 0.6	33.9 ± 0.2	600.2 ± 76.4	5.16E-11
NM_000766	Cytochrome P450, family 2, subfamily A, polypeptide 13	CYP2A13	25.9 ± 0.9	34.0 ± 0.0	441.2 ± 49.9	2.75E-10
NM_006894	Flavin containing monooxygenase 3	FMO3	25.4 ± 1.0	33.3 ± 1.0	377.1 ± 52.2	9.81E-07
NM_000898	Monoamine oxidase B	MAOB	24.6 ± 0.8	31.9 ± 0.7	252.6 ± 22.9	6.19E-07
NM_000767	Cytochrome P450, family 2, subfamily B, polypeptide 6	CYP2B6	27.1 ± 0.3	33.9 ± 0.2	187.5 ± 74.9	4.14E-07
NM_000772	Cytochrome P450, family 2, subfamily C, polypeptide 18	CYP2C18	25.0 ± 1.0	31.3 ± 1.0	123.8 ± 20.8	6.66E-06
NM_000769	Cytochrome P450, family 2, subfamily C, polypeptide 19	CYP2C19	27.1 ± 0.7	32.7 ± 0.9	74.7 ± 7.1	3.15E-06
NM_000694	Aldehyde dehydrogenase 3 family, member B1	ALDH3B1	20.4 ± 0.5	25.5 ± 0.2	55.3 ± 12.5	3.30E-07
NM_023944	Cytochrome P450, family 4, subfamily F, polypeptide 12	CYP4F12	25.5 ± 0.3	30.2 ± 0.4	44.8 ± 22.2	1.46E-05
NM_001080	Aldehyde dehydrogenase 5 family, member A1	ALDH5A1	24.8 ± 0.6	29.6 ± 0.3	44.2 ± 9.6	4.53E-08
NM_000673	Alcohol dehydrogenase 7 (class IV), mu or sigma polypeptide	ADH7	20.3 ± 0.8	24.9 ± 0.2	38.4 ± 4.8	1.23E-08
NM_000771	Cytochrome P450, family 2, subfamily C, polypeptide 9	CYP2C9	29.7 ± 0.6	34.0 ± 0.0	33.1 ± 18.5	1.66E-05
NM_001461	Flavin containing monooxygenase 5	FMO5	23.8 ± 1.0	28.1 ± 0.6	29.7 ± 5.6	2.02E-05
NM_000770	Cytochrome P450, family 2, subfamily C, polypeptide 8	CYP2C8	28.2 ± 0.7	32.2 ± 1.1	24.5 ± 3.5	9.07E-05
NM_000695	Aldehyde dehydrogenase 3 family, member B2	ALDH3B2	23.5 ± 0.5	27.4 ± 0.2	24.3 ± 5.6	2.50E-07
NM_000784	Cytochrome P450, family 27, subfamily A, polypeptide 1	CYP27A1	25.7 ± 1.0	27.9 ± 0.4	7.3 ± 1.1	1.48E-06
NM_002022	Flavin containing monooxygenase 4	FMO4	24.8 ± 1.0	26.5 ± 0.2	5.1 ± 0.8	2.31E-05
NM_000110	Dihydropyrimidine dehydrogenase	DPYD	24.3 ± 0.4	25.9 ± 0.6	4.9 ± 1.7	2.34E-03
NM_004820	Cytochrome P450, family 7, subfamily B, polypeptide 1	CYP7B1	25.5 ± 0.8	27.1 ± 0.3	4.8 ± 0.2	3.23E-06
NM_183374	Cytochrome P450, family 26, subfamily C, polypeptide 1	CYP26C1	31.9 ± 0.5	33.0 ± 0.7	4.3 ± 3.1	3.91E-02
NM_017781	Cytochrome P450, family 2, subfamily W, polypeptide 1	CYP2W1	26.8 ± 0.4	27.9 ± 0.7	3.5 ± 1.0	1.06E-02
NM_000896	Cytochrome P450, family 4, subfamily F, polypeptide 3	CYP4F3	26.2 ± 0.4	27.1 ± 0.6	3.0 ± 0.9	1.32E-02
NM_000693	Aldehyde dehydrogenase 1 family, member A3	ALDH1A3	21.1 ± 0.4	21.7 ± 0.4	2.5 ± 0.8	1.05E-02
NM_001086	Arylacetamide deacetylase (esterase)	AADAC	27.5 ± 0.5	28.1 ± 0.2	2.4 ± 0.9	7.52E-03
NM_000104	Cytochrome P450, family 1, subfamily B, polypeptide 1	CYP1B1	24.5 ± 0.4	25.0 ± 0.2	2.3 ± 1.0	1.11E-02
NM_000691	Aldehyde dehydrogenase 3 family, member A1	ALDH3A1	19.9 ± 0.3	20.3 ± 0.4	2.2 ± 0.7	1.36E-02
NM_001082	Cytochrome P450, family 4, subfamily F, polypeptide 2	CYP4F2	31.1 ± 0.4	31.4 ± 0.2	2.0 ± 0.6	6.88E-03
Down-regulated genes
NM_000962	Prostaglandin-endoperoxide synthase 1 (prostaglandin G/H synthase and cyclooxygenase)	PTGS1	32.6 ± 0.8	23.9 ± 0.5	−250.0 ± 0.0	3.58E-06
NM_004181	Ubiquitin carboxyl-terminal esterase L1 (ubiquitin thiolesterase)	UCHL1	28.8 ± 0.8	20.3 ± 0.2	−250.0 ± 0.0	1.00E-09
NM_000499	Cytochrome P450, family 1, subfamily A, polypeptide 1	CYP1A1	34.0 ± 0.1	26.8 ± 0.5	−100.0 ± 0.01	3.01E-06
NM_000963	Prostaglandin-endoperoxide synthase 2 (prostaglandin G/H synthase and cyclooxygenase)	PTGS2	28.2 ± 1.0	21.4 ± 0.1	−100.0 ± 0.0	1.37E-07
NM_019885	Cytochrome P450, family 26, subfamily B, polypeptide 1	CYP26B1	32.5 ± 1.0	26.8 ± 0.1	−25.0 ± 0.0	3.44E-04
NM_000379	Xanthine dehydrogenase	XDH	28.8 ± 0.6	23.9 ± 0.2	−16.7 ± 0.0	5.03E-06
NM_000785	Cytochrome P450, family 27, subfamily B, polypeptide 1	CYP27B1	29.3 ± 0.3	25.0 ± 0.6	−12.5 ± 0.0	2.40E-04
NM_021187	Cytochrome P450, family 4, subfamily F, polypeptide 11	CYP4F11	25.3 ± 0.4	21.4 ± 0.4	−9.1 ± 0.0	7.30E-06
NM_022568	Aldehyde dehydrogenase 8 family, member A1	ALDH8A1	34.0 ± 0.0	30.7 ± 0.3	−5.9 ± 0.1	5.40E-04
NM_024514	Cytochrome P450, family 2, subfamily R, polypeptide 1	CYP2R1	26.4 ± 0.8	23.2 ± 0.2	−5.9 ± 0.0	7.01E-07
NM_000106	Cytochrome P450, family 2, subfamily D, polypeptide 6	CYP2D6	28.0 ± 1.0	24.5 ± 0.2	−4.8 ± 0.2	1.30E-02
NM_000500	Cytochrome P450, family 21, subfamily A, polypeptide 2	CYP21A2	30.8 ± 0.5	28.1 ± 0.3	−4.0 ± 0.1	3.52E-04
NM_003748	Aldehyde dehydrogenase 4 family, member A1	ALDH4A1	24.9 ± 0.5	22.3 ± 0.3	−3.7 ± 0.1	1.83E-04
NM_030622	Cytochrome P450, family 2, subfamily S, polypeptide 1	CYP2S1	22.8 ± 0.4	20.5 ± 0.3	−3.0 ± 0.1	3.76E-04
NM_001182	Aldehyde dehydrogenase 7 family, member A1	ALDH7A1	22.9 ± 0.8	20.7 ± 0.2	−2.9 ± 0.0	9.39E-05

Forty-seven out of the 84 genes were differentially expressed; relative to their expression in NHBE cells, 32 genes were up-regulated and 15 genes down-regulated in the differentiated ALI cultures. Compared to undifferentiated NHBE cells, the expression of CYP26A1, CYP2F1, CYP4B1, ALDH1A1, ADH1C, FMO2, FMO3, MAOB, CYP2B6, CYP2A13, and CYP2C18 increased by > 100-fold in differentiated ALI cultures, while the expression of PTGS1, UCHL1, CYP1A1, and PTGS2 decreased by > 100-fold. Several enzymes that are known to be involved in xenobiotic metabolism, such as CYP3A4, CYP1A2, and CYP2E1 remained unchanged.

3.2. Analysis of gene expression profiles between NHBE cells and ALI cultures

Differences in the profiles of the phase I enzyme gene expression between NHBE cells and ALI cultures were analyzed using an unsupervised Hierarchical Cluster Analysis (HCA) with Ward's distance metric (Fig. 1). Two large clusters, one with four mRNAs from NHBE cells (black) and the other with four mRNAs from ALI cultures (red), were clearly separated. A greater number of genes were expressed in higher abundance in differentiated ALI cultures than in undifferentiated NHBE cells. Pearson's correlation analysis further demonstrated that there were differences between the gene expression of phase I enzymes in NHBE cells and ALI cultures (Fig. 2), with correlation coefficients ranging between 0.64 and 0.67 (Table 3).

Fig. 1.

Hierarchical clustering analysis of gene expression profiles. Hierarchical clustering analysis of gene expression profiles in 4 NHBE cultures (black bracket) and 4 ALI cultures (red bracket), all derived from the same donor. The clustering analysis was based on the ΔCt values of 84 phase I drug-metabolizing enzyme genes. A Ct number of 34 was used in the analysis for genes with low abundance or not expressed. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2.

Pearson's correlation analysis. The log₁₀ transformed ΔCt values of 84 phase I drug-metabolizing genes were used for the scatter plot. Pearson's correlation coefficients between NHBE cells and ALI cultures were calculated.

Table 3.

Pearson's correlation coefficient between ALI cultures and NHBE cells. The correlation matrix was calculated based on ΔCt numbers of individual array. The numbers represent the pairwise Pearson's correlation coefficient r value.

	NHBE-1	NHBE-2	NHBE-3	NHBE-4	ALI-1	ALI-2	ALI-3	ALI-4
NHBE-1	1.00
NHBE-2	1.00	1.00
NHBE-3	0.99	0.99	1.00
NHBE-4	0.99	0.98	0.99	1.00
ALI-1	0.66	0.67	0.65	0.67	1.00
ALI-2	0.65	0.66	0.65	0.66	1.00	1.00
ALI-3	0.65	0.66	0.64	0.66	0.98	0.98	1.00
ALI-4	0.65	0.66	0.64	0.66	0.98	0.98	1.00	1.00

3.3. Expression of genes implicated in NNK metabolism in NHBE cells and ALI cultures

The expression of genes encoding 8 enzymes implicated in NNK metabolism, including 3 CYPs (CYP2A6, CYP2A13, and CYP2B6) and 5 reductases (carbonyl reductase 1, 11β-HSD1, AKR1C1, AKR1C2, and AKR1C4) (Hecht, 1998; Chiang et al., 2011; Atalla et al., 2000), were further compared in NHBE cells and ALI cultures using the single-gene qRT-PCR. CYP2A6 and CYP2A13 were expressed in ALI cultures, but barely detectable in NHBE cells. CYP2B6 was expressed over 100-times higher in ALI cultures than in NHBE cells. Genes encoding two of the reductases, i.e., 11β-HSD1 and AKR1C4, were not expressed in either cell system. Expression of the genes encoding the remaining three reductases, i.e., carbonyl reductase 1, AKR1C1, and AKR1C2, was detected in both NHBE cells and ALI cultures, but at significantly higher levels in ALI cultures with fold changes (ALI/NHBE) of 3.5, 8.0, and 12.0, respectively.

3.4. Protein expression of the major NNK metabolism enzymes

Immunoblotting was used to quantify the expression of three key enzymes involved in NNK metabolism, i.e., CYP2A6, CYP2A13, and carbonyl reductase 1, in whole cell lysates made from NHBE cells and ALI cultures. Preliminary experiments demonstrated that the CYP2A13 antibody lacked specificity and detected both CYP2A6 and CYP2A13, possibly due to the high structural homogeneity between these two enzymes. As shown in Fig. 3, the intensity of the bands corresponding to the molecular weight of CYP2A6/2A13 was significantly stronger in ALI cultures than in NHBE cells (top panel). Similarly, the expression of carbonyl reductase 1 also was higher in differentiated ALI cultures (Fig. 3, middle panel). These observations are consistent with the expression pattern of the genes encoding these enzymes.

Fig. 3.

Protein expression of the major metabolic enzymes implicated in NNK metabolism. Protein expressions of CYP2A6/2A13 and carbonyl reductase 1 were analyzed using immunoblotting in NHBE cells and ALI cultures. β-Actin was used as the loading control.

3.5. Basal metabolic activity of CYP2A6/2A13 and CYP2B6 in NHBE cells and ALI cultures

The basal catalytic activities of CYP2A6/2A13 and CYP2B6 were measured in NHBE cells and ALI cultures using two enzyme-specific substrates, i.e., coumarin and bupropion, respectively. Release of the hydroxylated substrates into the media (apical washes and basolateral media for ALI cultures) was quantified after a 5-h incubation and the rates of metabolite formation were calculated (Table 4). In the ALI cultures, 7-hydroxy coumarin and 4-hydroxy bupropion were formed at rates of 0.955 ± 0.087 and 0.053 ± 0.008 pmol/mg protein/min, respectively. In contrast, neither of the metabolites was detected in medium from NHBE cells, suggesting that the undifferentiated cells fail to metabolize these two substrates.

Table 4.

Metabolite formation rate by CYP2A6/2A13 and CYP2B6 in ALI cultures and NHBE cells after a 5-h treatment with CYP-specific substrates.

In vitro model	Metabolite formation (pmol/mg protein/min)
	7-Hydroxy Coumarin	4-Hydroxy Bupropion
ALIa	0.955 ± 0.087	0.053 ± 0.008
NHBE	< LOD	< LOD

LODs for 7-hydroxy coumarin and 4-hydroxy bupropion are 1.9 ng/mL and 1.5 ng/mL, respectively.

^aApical washes and basolateral media were combined for the metabolite measurement.

3.6. α-Hydroxylation of NNK in NHBE cells and ALI cultures

The bioactivation of NNK in ALI cultures was demonstrated by quantifying the formation of one of its key end metabolites, keto acid, in both the apical wash and basolateral medium after 1- and 6-h treatments with 100 and 500 μM NNK (Table 5). The time-course comet assay revealed that NNK induced the greatest level of DNA strand-breaks after a 1-h treatment and its effect on DNA damage diminished after a 6-h treatment (Fig. 4A). The 1-h and 6-h treatment durations, therefore, were selected to explore the association between NNK-induced DNA damage and the levels of its metabolites. These two concentrations of NNK were non-cytotoxic in both cell systems based on the MTS cell viability assay (data not shown). Within 1 h of treating the ALI cultures, NNK was converted into keto acid in a concentration-dependent manner. The amount of keto acid was decreased after a 6-h incubation. Consistent with its molecular size and hydrophilicity, keto acid was found mainly in the apical washes. NNK also was metabolized to keto acid in NHBE cells, but only at the higher concentration, and the level of metabolite was approx. 4-fold lower compared to that generated by the ALI cultures.

Table 5.

Keto acid formation in ALI cultures and NHBE cells treated with NNK.

NNK Doses (μM)	ALI cultures (pmol)		NHBE cells (pmol)
	Apical wash	Basolateral medium	medium
1 h Treatment
0	< LOD	< LOD	< LOD
100	3.07 ± 0.54	< LOD	< LOD
500	7.05 ± 0.84	< LOD	< LOD
6 h Treatment
0	< LOD	< LOD	< LOD
100	0.74 ± 0.18	< LOD	< LOD
500	3.65 ± 0.20	< LOD	< LOD

LOD: Limit of Detection. LOD for keto acid detection is 0.7 ng/mL.

Fig. 4.

DNA damages induced by NNK in the ALI cultures. (A). NNK-induced DNA strand-breaks peaked at 1 h and decreased in a time-dependent manner in ALI cultures. (B). DNA damages in response to a single 1-h NNK treatment were increased in a concentration-dependent manner in ALI cultures. (C). Sub-categorization of cell populations damaged at different levels of severity based on tail intensity in ALI cultures exposed to NNK for 1 h. Low damage: tail intensity < 8%; Moderate damage: tail intensity ≥ 8% and < 15%; Severe damage: tail intensity ≥ 15%. *, Ɨ, # p < .05 was considered statistically significant compared to the vehicle-treated control in the groups with low, moderate, and severe DNA damage, respectively.

3.7. Carbonyl reduction of NNK in NHBE cells and ALI cultures

Formation of another major NNK metabolite, NNAL, was quantified after 1- and 6-h treatments with 100 and 500 μM NNK in both NHBE cells and ALI cultures. NNAL was barely detected in cell lysates from both NHBE cells and ALI cultures (data not shown). We, therefore, estimated the total NNAL formation based on measurements made on the apical washes and basolateral medium (Table 6). Formation of NNAL was increased in a concentration- and time-dependent manner in both NHBE cells and ALI cultures. The uptake of NNK also was measured and found to be much higher in the monolayer NHBE cells than in ALI cultures. Given the difference in the total moles of administered and uptake of NNK, the conversion rate of NNK to NNAL presumably better reflects the difference in the efficiency of carbonyl reduction between these two cell systems. NNK was more efficiently reduced to NNAL in ALI cultures than NHBE cells after a 1-h incubation, with conversion rates above 53% at both concentrations in the ALI cultures. Interestingly, the conversion rates were slightly decreased at the higher dose in both NHBE cells and ALI cultures. Biotransformation to NNAL appeared to be saturated after a 6-h incubation in ALI cultures and close to saturation in NHBE cells at the same sampling time point.

Table 6.

NNAL conversion rate in ALI cultures and NHBE cells after NNK treatment.

NNK doses (μM)	ALI cultures			NHBE cells
	NNAL (nmol)	NNK uptakea (nmol)	Conversion rateb (100%)	NNAL (nmol)	NNK uptakea (nmol)	Conversion rateb (100%)
1-h Treatment
0	< LOD	< LOD	n.a	< LOD	< LOD	n.a
100	0.80 ± 0.02	1.11 ± 0.02	72.18 ± 2.03	4.19 ± 0.10	25.91 ± 1.39	16.31 ± 1.28
500	1.74 ± 0.19	3.28 ± 0.14	53.19 ± 6.93	16.64 ± 0.68	134.53 ± 13.11	12.74 ± 1.87
6-h Treatment
0	< LOD	< LOD	n.a	< LOD	< LOD	n.a
100	2.12 ± 0.02	1.94 ± 0.01	109.67 ± 1.28	20.60 ± 0.23	24.39 ± 1.08	84.79 ± 4.09
500	11.39 ± 0.61	10.02 ± 0.35	114.26 ± 9.17	93.26 ± 0.92	105.58 ± 5.94	88.85 ± 4.62

LOD: limit of detection. LODs for NNK and NNAL are 13.6 ng/mL and 76.7 ng/mL, respectively.

n.a: not applicable.

^aNNK update = total administered NNK - NNK measured in the medium (and apical washes for the ALI cultures).

^bConversion rate = 100% × NNAL formed/NNK uptake.

3.8. DNA strand-breaks induced by NNK

The genotoxicity of NNK was assessed using the alkaline comet assay, which detects both direct DNA strand-breaks and alkali-labile DNA damage (Fairbairn et al., 1995). To determine the optimal manifestation time of NNK-induced DNA damage, we treated ALI cultures with 500 μM NNK for 1, 6, and 24 h (Fig. 4A). The 1-h treatment was found to induce a significantly higher level of DNA damage compared to the other two time points. The subsequent experiments, therefore, were conducted with 1-h treatments. The raw percent DNA in the tail data are provided in Supplementary Table 1. Exposing ALI cultures to 100 and 500 μM NNK for 1 h resulted in concentration-dependent increases in DNA damage, with a significant response detected at 100 μM (Fig. 4B). When the cells were categorized into sub-populations based on the severity of DNA damage, the number of cells with low levels of damage were decreased in a concentration-dependent manner and, accordingly, the sizes of the sub-populations with moderate and severe damage were increased (Fig. 4C). In contrast, NNK failed to induce DNA damage in NHBE cells under the same treatment conditions (Fig. 5, the group on the left), presumably due to the lack of metabolic activation of NNK. To confirm this hypothesis, we co-treated NHBE cells with NNK and a liver S9 system, which supplies metabolism enzymes exogenously. A significant concentration-dependent increase in DNA damage was observed in NHBE cells treated with NNK in the presence of S9 (Fig. 5, the group on the right).

Fig. 5.

The effect of NNK on DNA damage in the NHBE cells. Metabolic activation by exogenous enzymes was required for NNK to induce DNA strand-breaks in NHBE cells. A 1-h single exposure to NNK induced DNA strand-breaks in a concentration-dependent manner only in the presence of S9 mixture in NHBE cells. *p < .05 is considered statistically significant compared to the vehicle-treated control.

4. Discussion

Exposure to environmental pollutants, including second-hand tobacco smoke and deliberately inhaled tobacco smoke is known to contribute to permanent damage to the respiratory tract, such as chronic inflammatory diseases and lung cancer (Crotty et al., 2015; Pfeifer et al., 2002). The metabolic activity of the lung plays a protective role by detoxifying and potentiating the elimination of inhaled chemicals. In some instances, however, chemicals, such as polycyclic aromatic hydrocarbons and tobacco-specific nitrosamines, are biotransformed into reactive metabolites, leading to the formation of mutagenic DNA-adducts (Castell et al., 2005; Hecht, 1998). In vitro lung models with metabolic capabilities that replicate what occurs in vivo, therefore, may be useful for evaluating the toxicity of inhaled substances.

Cell lines derived from normal or tumor tissues of the lung, such as lung carcinoma A549 cells, mucoepidermoid carcinoma H292 cells, and normal BEAS-2B cells, have been used for understanding the mechanisms of respiratory toxicity and assessing the adverse health effects caused by inhaled toxicants. Several studies, however, have reported that these cells have limited metabolic activity (Garcia-Canton et al., 2013) and, therefore, may not be optimal for assessing the risks associated with chemicals that undergo metabolic transformation.

It has been shown that well-differentiated ALI airway cultures derived from normal human primary bronchial epithelial cells not only resemble the structure of the in vivo airway epithelium (Dvorak et al., 2011; Pezzulo et al., 2011; Cao et al., 2015), but also preserve some of the enzymatic activities of airway epithelium, including those of several CYP enzymes (Baxter et al., 2015; Newland et al., 2011). Studies also have reported that the global expression of many genes, including those involved in metabolic transformation, is markedly enhanced during differentiation (Pezzulo et al., 2011; Boei et al., 2017). In this study, we further evaluated the metabolic capability of the ALI airway tissue models. We compared the basal gene expression for 84 phase I metabolism enzymes and the catalytic activities of the major NNK-related CYPs in undifferentiated NHBE cells and differentiated ALI cultures. Metabolic activation of a tobacco-specific nitrosamine, NNK, also was assessed by quantifying the formation of two of its metabolites, NNAL and keto acid.

Consistent with previous reports, the basal expression of many phase I drug metabolism genes was significantly higher in ALI cultures than in NHBE cells (Dvorak et al., 2011; Boei et al., 2017). More than half of the metabolic enzymes were differentially expressed between ALI cultures and NHBE cells; and 68% of the differentially expressed genes were significantly up-regulated in ALI cultures (Table 2). A significant number of these genes are present in relatively low abundance in NHBE cells with Ct numbers above 30. Considering the inducibility of certain metabolic enzymes, lower abundance in basal gene expression may not necessarily result in a lack of metabolic transformation in NHBE cells, at least for extended treatments. Inducers specific to the CYPs in question should be studied in order to evaluate this possibility.

Approximately 32% of the differentially expressed genes were down-regulated in ALI cultures. In contrast to previous reports (Baxter et al., 2015; Newland et al., 2011), CYP1A1 was expressed at very low level in ALI cultures (Table 2); its expression, however, was significantly higher in NHBE cells. As the expression of CYP1A1 is inducible, it may be that the conditions that we employed for our ALI cultures did not supply sufficient stimulation for its induction. In fact, analysis of normal pulmonary tissues from non-smokers and smokers revealed that the expression of CYP1A1 was only detected in tissues from smokers (McLemore et al., 1990), suggesting that not only CYP1A1 is inducible, but it normally is expressed at low levels in differentiated airway tissues. It has been reported that cell culture conditions, such as decreased attachment, hydrodynamic shear, and morphological changes of the cells, can induce CYP1A1 gene expression (Sadek and Allen-Hoffmann, 1994a; Sadek and Allen-Hoffmann, 1994b). It is also possible, therefore, that the relatively high expression of CYP1A1 in NHBE cells may be due to how the cells are maintained in culture.

Evidence from animal studies indicates that pulmonary bioactivation of NNK is required for its mutagenicity and carcinogenicity (Hecht, 2012). NNK is converted to its active metabolites via two major metabolic pathways, i.e., carbonyl reductase-mediated reduction to form a potent lung carcinogen, NNAL, and CYP-mediated α-hydroxylation of both NNK and NNAL to form keto acid and hydroxy acid, respectively (Weng et al., 2007; Li et al., 2014; Hecht, 2003). Carbonyl reductase I, AKR1C1/1C2/1C4, and 11β-HSD1 are the known reductases that mediate the carbonyl reduction of NNK, with carbonyl reductase I accounting for approximately 60% of NNAL formation in the cytosolic fraction of the lung (Atalla et al., 2000). CYP2A6/2A13 and CYP2B6 are involved in NNK α-hydroxylation; lung-specific CYP2A13 is believed to be the major phase I enzyme catalyzing these reactions (Su et al., 2000; Smith et al., 2003; Megaraj et al., 2014). Single gene qRT-PCR confirmed that the expression of carbonyl reductase I and AKR1C1/1C2 was up-regulated in ALI cultures compared to NHBE cells, while 11β-HSD1 or AKR1C4 expression was not detected in either cell system. Higher expression of all three CYPs was observed in the ALI cultures than in NHBE cells, with CYP2A13 being found in much higher abundance in ALI cultures. We further demonstrated a correlation between gene expression and the catalytic activity of these CYPs by measuring the hydroxylation products of two CYP-specific substrates, coumarin and bupropion (Table 4). Both 7-hydroxy coumarin and 4-hydroxy bupropion were detected in assays with ALI cultures, but not with NHBE cells, confirming the basal activities of these CYPs in the differentiated ALI cultures.

The metabolic potential of the ALI cultures was further demonstrated by measuring the formation of NNAL and keto acid in cultures treated with NNK. After a 1-h treatment with 100 and 500 μM NNK, the conversion rates from NNK to NNAL were approximately 4-fold higher in ALI cultures than in NHBE cells (Table 6). This observation is consistent with the findings on the differential expression of carbonyl reductase I in NHBE cells and ALI cultures at both gene and protein levels. Formation of keto acid was detected in ALI cultures, but not in NHBE cells, after a 1-h treatment with NNK and decreased after a 6-h treatment, suggesting a rapid bioactivation of NNK in ALI cultures (Table 5). These findings further corroborate our observations on the gene expression and catalytic activities of the CYP enzymes implicated in NNK metabolism.

Besides demonstrating the metabolic conversion of NNK in ALI cultures, we also evaluated the DNA damage caused by NNK using the alkaline comet assay. Numerous in vivo and in vitro studies indicate that NNK exposure results in DNA strand-breaks (Ibuki et al., 2015; Jorquera et al., 1994; Weitberg and Corvese, 1997). As expected, we also found that acute exposure to NNK induced transient DNA strand-breaks in ALI cultures, presumably due to the formation of reactive NNK metabolites. DNA damage, however, was not detected in NNK-treated NHBE cells (Fig. 5). When exogenous metabolism enzymes were added to the NNK treatments of NHBE cells, percent DNA in the tail was increased in a concentration-dependent manner (Fig. 5). This observation further suggests the indispensable role of metabolic activation in NNK-induced DNA damage, and indicates that ALI cultures have this metabolic activity, while NHBE cells do not.

By categorizing cells from the ALI cultures based on the severity of DNA damage (i.e., low, moderate, and severe), we found that the size of the populations with moderate and severe DNA damage was increased in NNK-treated groups (Fig. 4C). These observations indicate that NNK may selectively induce DNA damage in a sub-population of cells in the ALI cultures. Considering the heterogeneous composition of the cell types in ALI cultures, it is highly likely that the metabolic activity profiles differ between cell types, resulting in different levels of sensitivity to the DNA damage caused by protoxicants. In fact, in vivo studies have demonstrated that NNK is differentially bioactivated by different cell types in the lung (Schrader et al., 2000; Smith et al., 1992; Tjalve, 1991). Evaluating NNK-induced DNA damage in cells sorted, for instance, by cell type, therefore, may provide information to make more accurate predictions on the genotoxicity of NNK.

In summary, our findings demonstrate the differential expression of phase I metabolism enzymes in NHBE cells and in ALI airway cultures. By comparing the DNA damage caused by NNK in these two cell models, we demonstrated that the different metabolic capabilities of these cell models correlate with differential DNA strand-break formation in response to NNK exposure. We conclude that the metabolic characteristics of the testing systems should be an important consideration when selecting models for in vitro testing. The evidence presented in this study suggests that ALI airway cultures have the potential to bioactivate protoxicants, such as NNK, and thus may be a useful alternative in vitro model for evaluating inhaled toxicants that require metabolic transformation. There are improvements that can be made on these ALI cultures to further increase the resemblance between the in vitro and in vivo models. The most obvious one is to increase its complexity by adding additional cell types, such as fibroblasts or macrophages, which may strongly influence tissue responses in the ALI cultures. However, studies have demonstrated that the source of fibroblast affects the differentiation of the airway epithelial cells (Pageau et al., 2011), along with a number of other technical challenges. Extensive characterization, therefore, needs to be carried out before adopting in vitro models with greater complexity for toxicity testing.