Transcriptomics Underlying Pulmonary Ozone Pathogenesis Regulated by Inflammatory Mediators in Mice

Ozone (O3) is the predominant oxidant air pollutant associated with airway inflammation, lung dysfunction, and the worsening of preexisting respiratory diseases. We previously demonstrated the injurious roles of pulmonary immune receptors, tumor necrosis factor receptor (TNFR), and toll-like receptor 4, as well as a transcription factor NF-κB, in response to O3 in mice. In the current study, we profiled time-dependent and TNFR- and NF-κB-regulated lung transcriptome changes by subacute O3 to illuminate the underlying molecular events and downstream targets. Mice lacking Tnfr1/Tnfr2 (Tnfr-/-) or Nfkb1 (Nfkb1-/-) were exposed to air or O3. Lung RNAs were prepared for cDNA microarray analyses, and downstream and upstream mechanisms were predicted by pathway analyses of the enriched genes. O3 significantly altered the genes involved in inflammation and redox (24 h), cholesterol biosynthesis and vaso-occlusion (48 h), and cell cycle and DNA repair (48–72 h). Transforming growth factor-β1 was a predicted upstream regulator. Lack of Tnfr suppressed the immune cell proliferation and lipid-related processes and heightened epithelial cell integrity, and Nfkb1 deficiency markedly suppressed lung cell cycle progress during O3 exposure. Common differentially regulated genes by TNFR and NF-κB1 (e.g., Casp8, Il6, and Edn1) were predicted to protect the lungs from cell death, connective tissue injury, and inflammation. Il6-deficient mice were susceptible to O3-induced protein hyperpermeability, indicating its defensive role, while Tnf-deficient mice were resistant to overall lung injury caused by O3. The results elucidated transcriptome dynamics and provided new insights into the molecular mechanisms regulated by TNFR and NF-κB1 in pulmonary subacute O3 pathogenesis.


Introduction
Ozone (O 3 ) is a highly reactive gaseous oxidant air pollutant. Elevated levels of ambient O 3 have been associated with increased hospital visits and respiratory symptoms, including chest discomfort, breathing difficulties, coughs, and lung function decrement [1][2][3]. Subjects with pre-existing diseases such as asthma, rhinitis, and chronic obstructive pulmonary disorder are known to be particularly vulnerable to O 3 and are at risk of hospitalization, exacerbations, or death [4][5][6].
Controlled O 3 exposure to healthy volunteers and experimental animals elicit a number of pathophysiological effects, which include airway inflammation accompanied by airway hyperresponsiveness, chemokine/cytokine production, mucus overproduction and hypersecretion, reactive oxygen species production, decrements in pulmonary function, altered immune status, and epithelial damage and compensatory proliferation predominantly in ciliated cells of the upper respiratory tract and club cells in terminal bronchioles [7]. Pulmonary O 3 responses were also augmented by metabolic disorders, including obesity and diabetes in humans, as well as in experimental animals [8][9][10], and association of air pollution and increased risk of diabetes was also reported in humans and mice [11,12]. Long term exposure to O 3 may cause lung tumors in certain strains of mice [13].
Studies have investigated the roles of various inflammatory mediators in the pathogenic airway response to O 3 . Signal transducers, including epidermal growth factor receptor, nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), mitogen activated kinases, and inflammasome proteins (e.g., Nlrp3) have been proposed to be downstream mechanisms of O 3 -induced airway inflammation [14][15][16]. Toll-like receptor 4 (Tlr4) has been identified as a O 3 -induced hyperpermeability susceptibility gene from murine genomewide linkage analysis of subacute O 3 -induced airway hyperpermeability and injury [17,18]. Furthermore, tumor necrosis factor (Tnf ) is a susceptibility gene for pulmonary inflammation induced by subacute O 3 [19].
The current study was designed to identify the transcriptome events underlying pulmonary O 3 pathogenesis and downstream targets of the TNFR and NF-κB signaling pathways. We determined time-dependent lung gene expression profiles changed by subacute O 3 in wild-type mice and in Tnfr-deficient mice. We also identified NF-κBdependent transcriptome changes using p50 NF-κB (NF-κB1)-deficient and -sufficient mice.

Bronchoalveolar Lavage (BAL) Analyses and Lung Histopathology
The right lungs from each mouse were lavaged in situ with HBSS, and the BAL returns were analyzed for the total protein content and cell differentials, as described previously [24]. Left lung tissues from each mouse were inflated gently with 10% neutrally buffered formalin, fixed under constant pressure for 30 min, and proximal (around generation 5) and distal (approximately generation 11) levels of the main axial airway were sectioned for paraffin embedding. Tissue sections (5-µm thick) were stained with hematoxylin and eosin (H&E). The tissues were also processed for immunohistochemical staining using a rat monoclonal (IgG 1 ) anti-macrophage receptor with collagenous structure (MARCO; 1:50 dilution of clone ED31, Hycult Biotech, Wayne, PA, U.S.A.). Briefly, deparaffinized and hydrated tissue sections on microscope slides were treated sequentially with antigen unmasking solution (Vector Laboratories, Burlingame, CA, USA), 0.1% proteinase K, and endogenous peroxidase quenching solution (5% H 2 O 2 ) before blocking with 1.5% serum (Vectastain ABC kits). Tissue sections were then incubated overnight at 4 • C with the anti-MARCO antibody. After incubation with biotinylated rat secondary antibody (1:200, Vectastain ABC kits) and Avidin/Biotin solution, the antigens were detected by a 3,3 -diaminobenzidine-peroxidase substrate solution (10 min), and the slides were mounted with cover glasses after dehydration.

Lung RNA Isolation and cDNA Microarray Analysis
Lung tissues from Tnfr +/+ and Tnfr -/mice were homogenized in 2 mL Trizol (Thermo Fisher Scientific, Waltham, MA, USA) and the isolated total lung RNA was processed for Affymetrix GeneChip array analyses using mouse MOE430A arrays (Affymetrix, Inc., Santa Clara, CA, U.S.A.) in George Washington University (Dr. Andrea De Biase), as described previously [28]. The total lung RNAs from the Nfkb1 +/+ and Nfkb1 -/mice were isolated using RNeasy Mini Kit (Qiagen Inc., Valencia, CA, USA) and cDNA microarray was performed on mouse 430 2.0 arrays (Affymetrix) in the NIEHS Microarray Core Facility, as indicated previously [29]. Array raw data were filtered by a lower expression percentile (at least 1 sample had values within the 20% cut-off rage) and the expression levels normalized to the mean value of the experimental control (wild-type mice/air) for each gene by the quantile algorithm were analyzed statistically using GeneSpring GX14 software (Agilent Technologies, Inc., Santa Clara, CA, USA). O 3 exposure time effects in Tnfr +/+ lungs (t-test, p < 0.01) and genotype effects in air exposure (t-test, p < 0.05) or O 3 exposure (two-way ANOVA, p < 0.05; Benjamin and Hochberg False Discovery Rate test for the multiple comparisons) were tested to identify the differentially expressed genes. Venn diagram analyses determined common genes varied by O 3 between the genotypes. Ingenuity pathway analysis (IPA, Qiagen Inc., Valencia, CA, USA) was used to identify the potential molecular interactions and functions, as well as the downstream and upstream pathways. Microarray data were deposited in the Gene Expression Omnibus (accession numbers: GSE166399 for Tnfr +/+ and Tnfr -/mice and GSE166398 for Nfkb1 +/+ and Nfkb1 -/mice).

Sandwich Enzyme-Linked Immunosorbent Assay (ELISA) for IL-6
Aliquots of the lung cytosolic proteins (90 µg) were used to determine IL-6 using a mouse-specific ELISA kit (R&D Systems, Minneapolis, MN, USA) according to the manufacturer instructions. The optical density was measured at 450 nm and the IL-6 concentrations were determined using a standard curve.

Statistics
BAL, Western blotting, and qRT-PCR data are expressed as the group mean ± standard error of the mean (S.E.M.). Two-way ANOVA was used to evaluate the effects of exposure and genotype. The Student−Newman−Keuls test was used for a posteriori comparisons of the means for all multiple comparisons (p < 0.05). All of the statistical analyses were performed using SigmaPlot 13.0 program (Systat Software, San Jose, CA, USA)  Figure 1A), when most severe lung protein edema, inflammation, and histopathologic changes take places [24,30]. Venn diagram analyses determined that most of the significantly changed genes were unique at each time ( Figure 1A, and Tables 1 and S1), and the number of common O 3 responsive genes throughout the exposure (75 upregulated and 45 downregulated) were limited ( Figure 1A). Representative canonical pathways of the enriched genes also dissociated between 24 h and 48-72 h ( Figure 1B). Black bars = positive z-score (activation); gray bars = negative z-score (inhibition); white bars = no activity pattern available. (C) Pathway analysis determined that tumor necrosis factor (TNF) and nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) were potential key upstream regulators for the O3-responsive lung genes, which contribute to acute inflammatory and immune responses (e.g., migration of antigen presenting cells) at 24 h. (D) Cell cycle control of chromosome replication was the top canonical pathway of lung genes significantly upregulated by O3 (in red) at 48-72 h. (E) Top disease and biological functions of the genes altered by O3 included blood vessel lesion and vaso-occlusion at 48 h and lung tumor development at 72 h. (F) Transforming growth factor (TGF)-β1 and P53 are predicted to be one of the key upstream regulators orchestrating the lung transcriptome changes at later times (48-72 h) of O3 exposure. Gene or molecule colors indicate upregulation/activation (red/orange) or downregulation/inhibition (green/blue) after O3 exposure compared with air exposure. Analyses were done using ingenuity pathway analysis and GeneSpring software.  After 24 h of O 3 exposure, inflammatory mediators represented by TNF and NF-κB were predicted to be upstream regulators of O 3 -altered genes ( Figure 1C), and acute phase and inflammatory response (e.g., Alb, Saa3, Myd88, Socs3, Ccl17, and Cxcl14) and Nrf2-mediated oxidative stress response (e.g., Nrf2, Maff, Gclc, Gpx2, and Sod2) were predominantlyactivated pathways ( Figure 1B,C, and Tables 1 and S1). At later times of exposure (48 and 72 h), O 3 most markedly activated cell cycle and inhibited DNA damage repair responses (e.g., Ccnb1, Cdc6, Cdt1, Cdk1, Plk1, Mcm family, Cks2, and Pcna; Figure 1B,D and Table S1). Distinctively, after 48 h of O 3 , the enriched genes were predicted to activate cholesterol biosynthesis and affect leucocyte migration (e.g., Ccl17, Retnla, and Timp1) and blood vessel lesion/vaso-occlusion (e.g., Thbs1, Spp1, and Vldlr; Figure 1B,E). Transcriptomics changes at 72 h of O 3 may suppress xenobiotic degradation (e.g., Cyp4b1, Fmo3, and Aox3) and tissue repair (e.g., Timp1, Mmp12, and Sdc2) and induce lung tumorigenesis (e.g., Rrm2, Birc5, and Areg; Figure 1B,E). Upstream molecules including TGF-β1 and P53 were indicated to affect O 3 -induced transcriptome changes at later times ( Figure 1F). Among the upstream regulators of O 3 -responsive transcriptome (24-72 h), chemical drugs including simvastatin, acetaminophen, and sulforaphane (Table S2) were suggested as therapeutic intervention to reverse O 3 toxicity, such as reducing inflammation, reactive oxygen species, and lipids. Downregulated lung genes by O 3 included clusters of Mup; insulin-like growth factor binding protein 3 (Igfbp3); and xenobiotic metabolizing enzymes including cytochrome P450, family 1, subfamily a, polypeptide 1 (Cyp1a1), and aldehyde oxidase 3 (Tables 1 and S1). While Hspa1a and Hspa1b encoding TLR4-dependent heat shock protein 70 (HSP70) [28] were upregulated by O 3 , many other HSP genes (e.g., Hspb1, and Hsph1) were significantly decreased at 48 h of O 3 (Table S1).

Air-Exposed Lungs
Lung genes basally expressed lower in Tnfr -/mice than in Tnfr +/+ mice (Tables 2 and S3) were represented by Mup clusters, Pttg1, HSPs (Hsph1 and Hspa4l), S100 calcium binding proteins (S100a8 and S100a9), chemokines (Ccl5 and Ccl17), cytochrome P450 family (Cyp1a1 and Cyp3a11), and apolipoproteins (Apoa2 and Apoa1). In contrast, basally overexpressed lung genes in Tnfr -/mice compared with Tnfr +/+ mice (Tables 2 and S3) included Dbp and nicotineamide nucleotide transhydrogenase (Nnt). Basally Tnf -dependently expressed genes were related to vascular disorders, as well as to the inhibition of inflammatory cell response ( Figure 2A,B and Table 3).  In air-exposed basal lungs, the inhibition of interleukins (ILs) 1A and 17A were predicted to suppress TNFR-dependent lung genes (e.g., Ccl5, S100a8, S100a9, and Serpina1), leading to the inhibition of inflammatory cell chemotaxis. (C) After 24 h O3 exposure, a compensatory increase of the genes involved in immune cell activation and movement were manifest in Tnfr -/-lungs compared with Tnfr +/+ mice. (D) After 48 h of O3 exposure, when lung injury and inflammation are greatest, modulation of potential upstream regulators including transforming growth factor (TGF)-β1 may change transcriptomes to suppress lymphocyte proliferation and eicosanoid synthesis and activate epithelial cell spreading/integrity in Tnfr -/-lungs. (E) Tnfr -/-mouse lungs after 72 h of O3 exposure had transcriptome changes to suppress the release of neurotransmitters and inhibit neurodegeneration. Gene or molecule colors indicate upregulation/activation (red or orange) or downregulation/inhibition (green or blue) in Tnfr -/-mice compared with Tnfr +/+ mice after air or O3 exposure. Analyses were done using Ingenuity Pathway Analysis and GeneSpring software. In air-exposed basal lungs, the inhibition of interleukins (ILs) 1A and 17A were predicted to suppress TNFR-dependent lung genes (e.g., Ccl5, S100a8, S100a9, and Serpina1), leading to the inhibition of inflammatory cell chemotaxis. (C) After 24 h O 3 exposure, a compensatory increase of the genes involved in immune cell activation and movement were manifest in Tnfr -/lungs compared with Tnfr +/+ mice. (D) After 48 h of O 3 exposure, when lung injury and inflammation are greatest, modulation of potential upstream regulators including transforming growth factor (TGF)-β1 may change transcriptomes to suppress lymphocyte proliferation and eicosanoid synthesis and activate epithelial cell spreading/integrity in Tnfr -/lungs. (E) Tnfr -/mouse lungs after 72 h of O 3 exposure had transcriptome changes to suppress the release of neurotransmitters and inhibit neurodegeneration. Gene or molecule colors indicate upregulation/activation (red or orange) or downregulation/inhibition (green or blue) in Tnfr -/mice compared with Tnfr +/+ mice after air or O 3 exposure. Analyses were done using Ingenuity Pathway Analysis and GeneSpring software.

Effects of Tnf and Il6 on O 3 -Induced Lung Injury
As seen in Tnfr -/mice [24], the mice deficient in Tnf cluster genes (Tnf, Lta, and Ltb) [34], and the mice treated with the TNF antibody [19], BAL fluids from Tnf -/mice had significantly reduced numbers of lung neutrophils and epithelial cells and amounts of proteins compared with those from Tnf +/+ mice at 48 h of O 3 ( Figure 4A). The histopathologic analysis indicated that O 3 -induced centriacinal proliferation indicated by thickened bronchiolar and terminal bronchiolar epithelium (arrows) were also less marked in Tnf -/mice compared with Tnf +/+ mice ( Figure 4B). The current microarray analysis and a previous study [14] demonstrated that the abundance of Il6 mRNA was higher in both Tnfr -/and Nfkb1 -/mice compared with their corresponding wild-type mice after O 3 (Tables 2, S4 and S6). In Il6 -/mice, lung protein hyperpermeability determined by the BAL protein concentration was significantly higher than that in Il6 +/+ mice ( Figure 4C). However, the numbers of O 3 -induced neutrophils or epithelial cells in BAL fluids were not significantly different between the two genotypes (data not shown). Consistent with the heightened BAL protein level in Il6 -/mice, H&E-stained lung tissue sections depicted more marked edema and permeability in the perivascular region (arrows), which accompanied protein exudation (pink staining) and congestion (red blood cells) into the alveolar air space in Il6 -/mice compared with Il6 +/+ mice after O 3 ( Figure 4D). Gene expression data and BAL analysis suggested a potential protective role for IL-6 in this model. ELISA determined significantly increased levels of IL-6 in Tnf +/+ (48 h) and Nfkb1 +/+ (24 and 48 h) mouse lungs after O 3 exposure ( Figure 4E). The O 3 -enhanced IL-6 protein amounts were significantly higher in Tnfr -/and Nfkb1 -/mice compared with their corresponding wild-type mice ( Figure 4E), which supported TNFR-and NF-κB1-dependent Il6 mRNA abundance. Representative light photomicrographs of H&E-stained lung sections from Tnf +/+ and Tnfr -/-mice, (B) and Il6 +/+ and Il6 -/-mice (D) exposed to air or O3 (48 h). Black arrows depict bronchiolar/terminal bronchiolar epithelium under proliferation. Blue arrows depict protein exudation in perivascular regions. AV = alveoli. BV = blood vessel. BR = bronchiole or terminal bronchiole. Bars = 100 μm. TNFR-and NF-κB1-dependent level of lung IL-6 proteins determined by enzyme-linked immunosorbent assay (E). Data are presented as means + S.E.M (n = 3/group). *, significantly different from genotype-matched air control mice (p < 0.05). +, significantly different from O3-exposed corresponding wild-type (Tnf +/+ , Il6 +/+ , Tnfr +/+ , or Nfkb1 +/+ ) mice (p < 0.05).

Validation of Microarray Results
qRT-PCR determined TNFR-dependently increased tissue inhibitor of metalloproteinase (Timp1) and Il33 or decreased Mup1 after air and O 3 exposure (Supplemental Figure S1A). Timp1 and Il33 mRNAs were significantly upregulated in O 3 -resistant Tnfr -/mice compared with susceptible Tnfr +/+ mice. A significant decline of Mup1 mRNA abundance by O 3 was greater in Tnfr -/mice than in Tnfr +/+ mice. Differential expression of NF-κB1-dependent genes, Jchain, Dbp, and Saa3, were also significantly different between two genotypes at baseline and/or after 48 h O 3 ( Figure S1B). The mRNA expressions of common differentially regulated genes Pttg1 and Il6 were significantly lower or higher, respectively, in both Tnfr -/and Nfkb1 -/mice compared with their corresponding wild-type mice ( Figure S1C). Western blot analyses found TNFR-dependent variations of the total TGF-β1 and MUP1 proteins and NF-κB1-dependent level of nuclear CCNB1 and STAT1 proteins in the lungs exposed to O 3 ( Figure 5A). The amount of total MARCO and nuclear c-Fos proteins, common differentially regulated gene products, were also varied similarly in Tnfr -/and Nfkb1 -/mice compared with their corresponding wild-type mice ( Figure 5A). The total lung protein levels of c-Fos were time-dependently increased by O 3 in all mice, while the cytoplasmic c-Fos abundances were marginally changed or decreased by O 3 ( Figure 5A). MARCO was detected in alveolar macrophages and was localized mostly in their plasma membranes and/or cytoplasm ( Figure 5B). Consistent with the differential protein levels detected by Western blotting, lower levels of MARCO localization were found in Tnfr -/and Nfkb1 -/mouse lungs relative to their wild-type mice after 48 h of O 3 exposure ( Figure 5B). c-Fos proteins, common differentially regulated gene products, were also varied similarly in Tnfr -/-and Nfkb1 -/-mice compared with their corresponding wild-type mice ( Figure 5A). The total lung protein levels of c-Fos were time-dependently increased by O3 in all mice, while the cytoplasmic c-Fos abundances were marginally changed or decreased by O3 ( Figure 5A). MARCO was detected in alveolar macrophages and was localized mostly in their plasma membranes and/or cytoplasm ( Figure 5B). Consistent with the differential protein levels detected by Western blotting, lower levels of MARCO localization were found in Tnfr -/-and Nfkb1 -/-mouse lungs relative to their wild-type mice after 48 h of O3 exposure ( Figure 5B).  , transforming growth factor (TGF)-β1, major urinary protein 1 (MUP1), nuclear c-Fos, nuclear G2/mitotic-specific cyclin-B1 (CCNB1), and nuclear signal transducer and activator of transcription 1 (STAT1). β-Actin (for total and cytosol) and Lamin-B1 (for nuclear) levels detected as Ccl17, Timp1, Saa3, Lcn2, and Mmp14) [41]. Transcriptomes of tracheobronchial epithelium and parenchyma, on the other hand, had predominant changes in cell cycle and DNA repair genes (e.g., Cdc20b, Cdk1, and Retnla) in response to subchronic O 3 [41]. In this study, most transcriptome changes by subchronic O 3 were common in male and female mice, while female mice were more susceptible to inflammatory cell influx, epithelial loss, and compensatory proliferation, which was supported by the more robust changes of the gene expression in females mice [41]. In Fischer rat lungs, acute O 3 (5 ppm, 2 h) altered genes with similar functions reported in mouse studies, and upregulation of inflammatory and redox (e.g., Jun, Cxcl2, Nos2, Hsp27, and Nfkb1), cell cycle and DNA repair (Ccne1, Cdc2, and Arrb15b), and lipid metabolism (e.g., Faah and Plaa) genes were evident [42]. In developmental mouse lung at 3 days postnatal (transition from saccular to alveolar stage), transcriptome changes by acute O 3 (1 ppm, 3 h) exposure were less robust than those seen in adult mice, and the global suppression of cell cycle-related lung genes (e.g., Cenpf, Cdca8, Cdk1, Cntn14, Cdc45, Mki67, and Pcna) was rather marked until 24 h postexposure [43]. These results indicate that acute O 3 exposure disturbs cellular proliferation and differentiation of lungs undergoing development.
Although the current study demonstrated whole lung transcription profiles without dissection of compartment-or cell-specific transcriptomics, our results characterized multiphasic transcriptome changes by O 3 exposure time, and only 18% genes were altered commonly in more than two time points. Subacute O 3 responsive genes likely have roles in oxidative injury and antioxidant induction, chemotaxis, and immune cell development during early exposure (at 24 h); cell cycle progress, blood vessel lesions, and cholesterol biosynthesis during peak lung injury (at 48 h); and xenobiotic metabolic process, tumorigenesis, and tissue injury/repair at a later time (72 h). After comparison with compartmental transcriptome studies, we predicted the cellular or tissue origin for reproducible O 3 responsive genes across multiple transcriptome studies; for example, increased lipocalin 2 (Lcn2) and small proline-rich protein 1A (Sprr1a) and decreased Igfbp3 may be mainly from the parenchyma [41]; increased matrix metalloproteinase 14 (Mmp14) from the macrophages [41]; increased resistin like alpha (Rentnla), leucine-rich alpha-2-glycoprotein 1 (Lrg1), and Timp1 from both the macrophages and parenchyma [40,41]; increased Saa3 from all compartments of airways [40,41]; and increased chromatin licensing and DNA replication factor 1 (Cdk1) and ubiquitin-conjugating enzyme E2C (Ube2c), as well as decreased Cyp1a1, Mup family, and serine (or cysteine) peptidase inhibitor, clade A, member 3K (Serpina3k) from the conducting airways [40,41].
The NF-κB family of proteins has an important role in inflammatory responses initiated by TNF [44]. Despite the identification of a few well-accepted NF-κB target genes in humans and mice (e.g., NFKBIA, TNFAIP3, and MYC), transcriptional outputs through NF-kB are not well understood due to the complexity of NF-κB dynamics and the NF-κB-binding landscape in the gene expression. NF-κB responses in gene transcription are known to vary depending on the cell type as well as the initiating stimulus [45]. In addition, p50 and p52, among five NF-kB family proteins, do not have a transactivation function, and they can activate transcription through heterodimerization with p65 or others [46]. Importantly, the p50−p50 homodimer binding to NF-κB motif inhibits other NF-κB dimer complex binding, and thus it often, but not always, serves as a transcriptional suppressor for NF-κB target genes [47]. The p50−p50 homodimer has thus been shown to have anti-inflammatory functions through repression of proinflammatory genes and enhancement of anti-inflammatory genes [48,49]. TNF I triggered a strong, sustained p65−p50 activation with a relatively lower level of p50−p50 [47]. Therefore, common differentially regulated genes by TNFR and NF-κB1 in the current study may include NF-κB target genes inducible by the TNFR-NF-κB (p50−p65) axis, as well as those suppressible by the p50−p50 homodimer. That is, the genes suppressed in Tnfr -/and Nfkb1 -/mice (e.g., Pttg1, Mmp3, and Marco) are likely p50−p65-inducible genes. Genes downregulated in Tnfr -/mice but overexpressed in Nfkb1 -/mice after O 3 exposure (e.g., Gzma, Cyp1a1, Nkg7, Il6, Ccl20, and Kit) are possibly p50−p50-repressed genes. Functional NF-κB motifs have been discovered in the murine Il6 promoter [50]. Therefore, together with augmented pulmonary protein hyperpermeability in Il6-deficient mice, IL-6 was predicted as an antiinflammatory cytokine in the current subacute O 3 pathogenesis and p50−p50 homodimer may modulate its transcription. We elucidated the potential NF-κB binding motifs from several common differentially regulated genes (e.g., Psca and Edn1), and these genes are postulated as direct downstream targets of the TNFR-NF-κB signaling pathways.
One of the genes modulated by both TNFR and NF-κB1 is Marco. MARCO expressed in alveolar macrophages recognizes oxidized lipids and provides innate defense against inhaled pathogens [51]. As a downstream effector of TLR4, which is a murine susceptibility gene for subacute O 3 -induced pulmonary hyperpermeability [17,28], MARCO plays a protective role in subacute O 3 -exposed mouse lungs through the inhibition of oxidized surfactant lipid production and inflammation [52]. As TLR4 and TNFR are key immune receptors in subacute O 3 pathogenesis, and the NF-κB pathway is also known to play an important role in the TLR4-mediated immune responses [53,54], we compared the current transcriptome profiles with TLR4-dependent O 3 transcriptomics (GEO accession # GSE20715, [28]). Commonly regulated genes by TLR4 and NF-κB1 were enriched in lipid derangement, including the disruption of membrane phospholipids, reaction with unsaturated fatty acids in airway lining fluids, interruption of fatty acid/steroid metabolism (e.g., Dbp and Cpt1a), as well as in cell-mediated immunity and lymphoid tissue hyperplasia (e.g., Cxcl1, Ccl20, and Ptpn2; Figure S2). Common gene transcripts regulated by TNFR and TLR4 were enriched in the engulfment of phagocytes (e.g., Marco, Icam1, Lcn2, and Il33), protein ubiquitination (e.g., Dnaja1, Hspd1, and Psma3), fatty acid metabolism (e.g., Acot7, Ptges, Elovl1, and Lpin1), and glutathione homeostasis/redox (e.g., Gsr, Gstm1, and Gltx; Figure S3). Overall, the TNFR-NF-κB and TLR4-NF-κB pathways or crosstalk modulated distinct transcriptomes during the development of O 3 -induced lung injury in mice. Further studies are warranted for these common genes regulated by these three critical immune and inflammatory mediators (Table S7).
Increasing evidence and TNFR-/TLR4-enriched pathways indicate an association of airway O 3 responses with extracellular and/or cellular lipid biology. On airway epithelium lining fluids rich in surfactant, inhaled O 3 chemically reacts with cholesterol or phospholipids and generates cytotoxic ozonolysis products represented by 5β,6β-epoxycholesterol (β-epoxide) [55,56]. These lipid-ozonized products are proinflammatory and are known to contribute to O 3 -induced airway inflammation [57,58]. Eicosanoids (e.g., prostaglandins, leukotrienes, and thromboxanes) synthesized by peroxidation of arachidonic acid by lipoxygenases, cyclooxygenases, and cytochrome P450 are also inflammatory mediators increased by O 3 leading to airway hyperresponsiveness and extrapulmonary outcomes, including vasoconstriction [59,60]. O 3 exposure-released adrenal-derived stress hormones (e.g., epinephrine and corticosterone) disrupted lipid and carbohydrate metabolism, leading to hyperglycemia, glucose intolerance, and lung injury in rats [61,62]. Further rodent studies demonstrated that obesity augmented acute O 3 -induced airway hyperresponsiveness and inflammation [63][64][65], and diabetes caused early and exacerbated lung inflammation and fibrotic changes in response to subchronic O 3 (0.5 ppm, 4 h/day for 13 weeks) [10]. Epidemiological studies also showed a positive association between O 3 exposure and adult insulin resistance and preexisting lipid disorders and metabolic conditions (e.g., obesity and diabetes) [66][67][68]. Metabolomic analysis of human serum revealed that acute O 3 exposure markedly increased lipid mobilization and catabolism products (e.g., monoacylglycerol and medium-and long-chain free fatty acids) [69]. Interestingly, human population studies indicated an association of gain-of-function TNF −304G/A polymorphism with obesityrelated airway hyperresponsiveness in asthmatics [70]. In obese mice, Tnfr2 deficiency reduced body weight and acute O 3 -induced inflammation and obesity-related airway hyperresponsiveness [64,71]. Overall, these studies suggested a role of lipid derangement in airway and extrapulmonary O 3 pathogenesis.
Our transcriptomic and pathway analyses results suggested direct effects of known or potential NF-κB motif-bearing genes in O 3 -induced pulmonary edema (e.g., Il6), T cell immunity (e.g., Ccl17, Ccl22, and Il27ra), cardiac mortality and vasoconstriction (e.g., Edn1), extracellular matrix degeneration (e.g., Col1a2 and Mmp9), and interruption of lipid metabolism (e.g., Dbp and Tef ) via the TNFR-NF-κB signaling axis. However, these TNFR-or NF-κB-dependent genes may be affected by multiple transcription factors or be regulated indirectly by other intracellular signaling pathways during O 3 pathogenesis. We previously demonstrated that AP-1 and c-Jun NH2-terminal kinase 1 MAPK are associated with TNFR signaling [14]. The presence of functional AP-1 binding sites in many of the TNFR-or NF-κB1-dependent genes determined in the current study, such as chemokines, cyclins, and E2F transcription factors [72][73][74], supports indirect effects or complex interplays. p38 MAPK and its upstream epidermal growth factor receptor are also known to play key roles in transcriptional activity directly and/or via crosstalk with NF-κB for inflammation and airway hyperreactivity response by O 3 [75].
In summary, the time-dependent gene expression and pathway analyses in the current study provided important insight into the downstream molecular events during the development of multi-phasic lung injury by subacute O 3 . Comparative transcriptome analyses defined common transcriptional profiles and potential cross talk between critical O 3 -related inflammatory mediators, TNFR and NF-κB, as well as TLR4. Our results increase the understanding of the molecular mechanisms of pulmonary O 3 toxicity for further research.