Specific interference RNA (siRNA) sequences for PPARγ were selected based on a previous publication (30). The sense and antisense strands of siRNA (si-PPARγ) were 5′-GAC AUG AAU UCC UUA AUG AUU-3′ and 5′-UCA UUA AGG AAU UCA UGU CUU-3′, respectively, spanning 880–899 bp of mouse Pparg (GI:187960102). si-PPARγ was synthesized in 2′-deprotected, duplex, desalted, and purified form by Dharmacon Research, Inc. (Lafayette, CO). Nrf2^+/+ mice were anesthetized with isofluorane and si-PPARγ (2 mg/kg, in 50-μl phosphate-buffered saline for 25 g) or equivalent dose and volume of nonspecific control siRNA duplex (si-NS) (nontargeting siRNA #2, Dharmacon) was administered intranasally (31). Intranasal instillation was done daily for 3 consecutive days (1 d before, day 0, and at 1 d after hyperoxia or air exposure).

Vehicle (phosphate-buffered saline with 4% dimethyl sulfoxide) or 15d-PGJ₂ (30 μg/kg, EMD Chemicals, Inc., Gibbstown, NJ) in vehicle was administered intraperitoneally (100 μl) in Nrf2^+/+ and Nrf2^−/− mice daily starting 1 day before inhalation exposure. The dose was chosen based on previous publications in which PPARγ was efficiently induced in the lung (32, 33). The last injection was done 24 hours before the end of the designated inhalation exposure period.

Mice were placed individually in a Hazleton M60 battery inside a Hazleton 1000 chamber (Lab Products, Maywood, NJ). After acclimation in the chambers for 24 hours per day for 2–3 days, mice were exposed to more than 95% O₂ (UHP grade, Min. purity 99.994% O₂ tanks; National Welders, Durham, NC) 24 hours per day for 48 or 72 hours. Water and feed (modified AIN76/A) were provided ad libitum during the exposure. The temperature (72°F ± 3°F) and humidity (50 ± 15%) of the chambers were monitored. Mice for room-air exposure were placed in a Hazleton M60 battery placed on a countertop with food and water provided ad libitum for the exposure duration. The chambers were opened for approximately 30 minutes at the same time each morning to allow animal health checks (morbidity and mortality), water and feed checks, and excreta paper change. Immediately after the end of each exposure, mice were killed by sodium pentobarbital overdose (104 mg/kg). All animal use was approved by the NIEHS Animal Care and Use Committee.

The right lung of each mouse was lavaged in situ four consecutive times with Hanks' balanced salt solution (0.5 ml/25 g body weight) and analyzed for total protein content (a marker of lung permeability) and differential cell count (markers of lung cellular inflammation and epithelial cell exfoliation), as described previously (7).

The left lung from each mouse (n = 3/group) was perfused intratracheally in situ with zinc formalin, removed from the mouse, and fixed at a constant intraairway pressure of 25 cm fixative for 30 minutes The trachea was ligated and the inflated lung lobes were immersed in the same fixative for 1 to 3 days. The fixed left lung lobe was processed to be stained with hematoxylin and eosin (H&E) for histopathologic analysis. Immunohistological staining was done using an anti-PPARγ antibody (sc-7273X; Santa Cruz Biotechnology, Inc., Santa Cruz, CA).

Potential ARE sequences containing RTKAYnnnGCR for Nrf2 binding were determined in the 10 kb 5′ upstream region of murine Pparg (GI:83029857) using a position weight matrix (PWM) statistical model that was constructed based on a set of functional ARE sequences curated from published experimental studies (34).

Terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) assay was performed on lung tissue sections to determine the cellular DNA fragmentation in situ as a parameter of apoptotic cell death using DeadEnd Colorimetric TUNEL System (Promega, Madison, WI) according to the recommendations of the manufacturer. Briefly, deparaffinized lung sections were fixed in 4% paraformaldehyde, permeabilized with proteinase K, and incubated with biotinylated nucleotide mix (including terminal deoxynucleotidyl transferase) to incorporate biotinylated nucleotide at the fragmented 3′-OH DNA ends. After immunohistochemical procedures, TUNEL-positive nuclei were visualized under a light microscope. Negative control samples that were incubated without terminal deoxynucleotidyl transferase and a positive control sample that was treated with DNase I before incubation with biotinylated nucleotide mix were prepared for references.

Total lung RNA was isolated from homogenates using RNeasy Mini Kit (Qiagen Inc., Valencia, CA). One μg of total lung RNA was reverse transcribed into cDNA in a 50-μl volume (35). For semiquantitative polymerase chain reaction (PCR), cDNA equivalent to 50 ng of RNA was amplified in a 25-μl reaction containing 240 nM of forward and reverse primers specific for mouse HO-1 (7), CD36 (5′-GCTTATTGGGAAGACAATCAA-3′, 5′-GCAAATGTCAGAGGAAAAGAA-3′), Nrf2 (7), NQO1 (7), GST-Ya (7), and 18s rRNA (5′-TACCTGGTTGATCCTGCCAG-3′, 5′-CCGTCGGCATGTATTAGCTC-3′) in a Gene Amp PCR System 9700 (Applied Biosystems, Foster City, CA), and expression levels were quantified with 18s ribosomal RNA as an internal control following the procedures described elsewhere (35).

Total lung proteins were prepared from lung homogenates in radioimmunoprecipitation buffer including phenylmethanesulfonyl fluoride (10 μg/ml) and protease/phosphatase inhibitor cocktail (Sigma-Aldrich, St. Louis, MO) using a polytron homogenizer. Nuclear proteins were prepared from pulverized lung tissues following procedures described previously (36). Proteins were quantified and stored in aliquots at −70°C.

Nuclear DNA binding activities were determined by gel shift analyses of nuclear protein aliquots (3–5 μg) on 3 × 10⁴ cpm (γ³²P) ATP end-labeled double-stranded oligonucleotides containing a consensus binding sequence (5′-AGGTCAAAGGTCA-3′) for PPAR/RXR consensus binding sequence (PPRE or direct repeat, sc-2587; Santa Cruz Biotechnology, Santa Cruz, CA), mutant PPRE oligonucleotide (sc-2588; Santa Cruz Biotechnology), mouse Nrf2 promoter-931 PPRE-like sequence (5′-TAATGCTGGTCAAAGGCCAAGGGGT-3′), ARE consensus sequence (35), -784 wild-type (Wt) mouse Pparg-ARE (5′-TCATTGTGACATAGCACTTAT-3′), or -784 mutant (Mt) mouse Pparg-ARE (5′-TCATTGGTACATAGCACTTAT-3′) following the procedure described previously (7). Specific binding activity was determined by preincubation of nuclear proteins with anti-Nrf2 antisera (8) or anti-RXRα (sc-553X, Santa Cruz Biotechnology) for 2 hours in ice followed by electrophoretic mobility shift assay. The gel was autoradiographed using an intensifying screen at −70°C, and bands were scanned using a Bio-Rad Gel Doc 2000 system (Bio-Rad Laboratories, Hercules, CA). Representative band images from independent analyses (n = 3/group) are presented.

To analyze specific binding activity of nuclear p65 nuclear factor kappa B (NF-κB) proteins, transcription factor ELISA was performed using nuclear extracts (2 μg) following the manufacturer's instruction (NF-κB TransAM kit; Active Motif, Carlsbad, CA) and procedures described previously (37).

Proteins (20–100 μg) were separated on SDS-phage gels and analyzed by Western blotting using specific antibodies against Nrf2 (sc-722X; Santa Cruz Biotechnology); PPARγ (sc-723X, Santa Cruz Biotechnology); RXRα (sc-553X; Santa Cruz Biotechnology); PTEN (sc-7974, Santa Cruz Biotechnology); bax (sc-7480; Santa Cruz Biotechnology); cleaved caspase-8 (#9748; Cell Signaling Technology Inc., Danvers, MA); lamin B (sc-6217, Santa Cruz Biotechnology); or pan actin (sc-1615, Santa Cruz Biotechnology). Representative band images from multiple independent blotting (n = 3/group) are presented. Specific protein bands were scanned and, if applicable, quantified using a Bio-Rad Gel Doc 2000 System.

The concentration of pulmonary proinflammatory cytokine IL-6 was determined in bronchoalveolar lavage (BAL) fluid (50 μl) using mouse-specific ELISA kits (R&D Systems, Minneapolis, MN) according to the manufacturer's instructions.

The 5′-upstream region of Pparg (GI:83029857) was amplified by PCR amplification using oligonucleotide primers bearing restriction enzyme sites (SacI or NheI), 5′-GCGAGCTCCACTGAATTATATTAGGTCA-3′, 5′-GCGAGCTCGTTGCTATTGATAGATAAAC-3′, and 5′-GCGCTAGCGTTTTGTCTATGTCTTGCAA-3′, and Pparg-ARE segment (−1000 to +1), which contains a putative -784 ARE sequence (5′- TCATTGTGACATAGCACTTAT-3′, from -784 to -764), or Pparg-ΔARE segment (-730 to +1), which has deletion of the -784 ARE, were generated. The promoter segments were processed for cloning into SacI and NheI sites of a pGL3-basic vector (Promega) containing a heterologous SV40 promoter and a firefly luciferase reporter gene using a Rapid DNA and Ligation Kit (Roche Applied Science, Indianapolis, IN) to create pPparg-ARE and pPparg-ΔARE. A mammalian overexpression vector for murine Nrf2 was generated by PCR amplification of the DBA/J mouse cDNA using oligonucleotide primers 5′-GGTACCCATGATGGACTTGGGAGTTGCCA-3′ and 5′-TCTAGACTCCATCCTCCCGAACCTAGT-3′. cDNA encoding Nrf2 were then cloned into the KpnI and XbaI sites of p3XFLAG-CMV (Sigma-Aldrich) for subsequent bacterial expression (pFlag-Nrf2). All plasmid sequences were confirmed by sequencing analysis performed in the NIEHS Sequencing Core facility.

A point mutation was introduced into the Pparg -784 ARE using a kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions to create mutated AREs (pPparg-ARE_mt; T-785A) in pPparg-ARE (or pPparg-ARE_wt). Oligonucleotide primers used for PCR amplification of 1,000 bp upstream containing the mutated ARE were 5′-GAGGGCTCCAATTTTTTCATTGAGACATAGCACTTATCACTTAAA-3′ (forward) and 5′-TTTAAGTGATAAGTGCTATGTCTCAATGAAAAAATTGGAGCCCTC-3′ (reverse). The mutations in the promoter sequences were confirmed after cloning.

Transient transfection was performed in BEAS-2B cells (ATCC, Manassas, VA) with 60 to 70% confluency using Effectene Transfection Reagent (Qiagen) following standard procedures. In brief, cells were transfected with 100 ng of each vector, pPparg-ARE or pPparg-ΔARE together with 1 ng of Renilla luciferase plasmid (pRL-TK, Promega). In a separate experiment, pPparg-ARE or pPparg-ΔARE were cotransfected with 100 ng of Nrf2 overexpressing plasmid (pFlag-Nrf2). Immediately after the end of exposure to air or hyperoxia (85 ± 5%; 4, 6, or 12 h), aliquots of cell extracts were assayed for firefly and Renilla luciferase activities using a dual luciferase kit (Promega) with a microplate luminometer (Fluoroskan Ascent FL; Thermo Scientific Inc., Waltham, MA). Background firefly luciferase activity from mock-transfected cell extracts was subtracted from each experimental measurement. Firefly luciferase activity was then normalized to that of Renilla luciferase to correct for variation in transfection efficiency in each cell extract, and the normalized average activity for pPparg-ARE in the air exposure group in the absence of Nrf2 overexpression was assigned a value of 1. For the site-directed mutagenesis assay, pPparg-ARE (pPparg-ARE_wt) or pPparg-ARE_mt was transfected into BEAS-2B cells with or without pFlag-Nrf2. All transfections were performed in triplicate, and each experiment was repeated (n = 6–9/group).

PPARγ was detected largely in alveolar macrophages and in epithelium lining conducting airways in air-exposed mice, and the populations of PPARγ-positive cells were smaller in Nrf2^−/− mice than in Nrf2^+/+ mice (Figure 2A). After hyperoxia exposure, PPARγ-bearing cells were markedly increased throughout the lungs of Nrf2^+/+ mice. PPARγ was detected predominantly in infiltrating inflammatory cells, bronchial epithelial cells, endothelial cells, and smooth muscle cells of injured regions with perivascular and peribronchial edema in (Figure 2B). PPARγ-positive cells were infrequent in the lungs from Nrf2^−/− compared with Nrf2^+/+ mice after O₂ (Figure 2B).

To investigate whether Nrf2 directly or indirectly regulates PPARγ expression, potential ARE sequences for Nrf2 binding in the 5′ upstream region of Pparg were analyzed. We used a PWM statistical model (34) that was constructed based on a set of functional ARE sequences curated from published experimental studies. Theoretical PWM score is proportional to binding energy (38), and as the PWM value increases, so does the likelihood that the ARE is bona fide. Our computational method identified nine putative ARE or ARE-like sequences (Table 1) from 10 kb of the 5′-upstream promoter sequence of the Pparg (GI:83029857). Among nine putative AREs, the most proximal upstream sequence spanning -784 and -764 bp matched the ARE consensus sequence (RTKAYnnnGCR; R = A or G, K = G or T, Y = C or T, n = A, C, G, or T) with the highest PWM score (14.2). This PWM score is compatible with that of validated AREs in pulmonary Nrf2 target genes including Nqo1 (ch16:68318405–68318425, PWM 16.4), Txnrd1 (ch12:103183169–103183189, PWM 17.9), and Gclm (ch1:94087027–94087047, PWM 12.2) (34). This potential ARE (hereafter designated -784) was pursued for further functional analysis.

To determine functional relevance of the -784 ARE in regulation of Pparg expression, the 1-kb upstream region of the Pparg promoter bearing the -784 ARE sequence was cloned in a luciferase reporter vector (pPparg-ARE) and the promoter activity was analyzed in airway epithelial cells by transient transfections. A plasmid containing the Pparg promoter constructs with deletion of the -784 ARE region (pPparg-ΔARE) was also generated and used in transfections. In airway epithelial cells overexpressing Nrf2 by transfection with Flag-tagged Nrf2 expression vector (pFlag-Nrf2), hyperoxia exposure significantly enhanced the luciferase activity (at 6 or 12 h) after transfection with pPparg-ARE (Figure 3A). Promoter activity of cells transfected with pPparg-ΔARE was not significantly changed by O₂, and was significantly lower after either air or O₂ exposure relative to the pPparg-ARE transfection (Figure 3A). Further, site-directed mutagenesis analysis determined significantly lower luciferase activity in cells transfected with the vector containing a point mutation in the ARE (GAGACATAGC, pPparg-ARE_mt) than in cells transfected with the wild type ARE (GTGACATAGC, pPparg-ARE_wt) after O₂ exposure (Figure 3B). Together, these results consistently demonstrated that Nrf2 binding on Pparg -784 ARE is critical for hyperoxia-induced PPARγ expression.

Nrf2 binding to the -784 ARE was further analyzed by gel shift assays using lung nuclear proteins from Nrf2^+/+ and Nrf2^−/− mice. Total DNA binding activity (SB) of nuclear proteins on the Wt -784 ARE probe was higher at baseline (air) in Nrf2^+/+ mice than in Nrf2^−/− mice (Figure 3C). The Wt -784 ARE binding activity was increased by O₂ in Nrf2^+/+ mice but not in Nrf2^−/− mice (Figure 3C). Total DNA binding activity of nuclear proteins on a mutated -784 ARE probe (Mt) demonstrated loss of DNA binding activity in Nrf2^+/+ and Nrf2^−/− mice exposed to air and O₂ (Figure 3C). Specific Nrf2 binding activity (SSB) on the Wt probe as determined by addition of anti-Nrf2 antisera (8) was detected only in Nrf2^+/+ mice, and was greatly increased by O₂ (Figure 3C). No specific Nrf2 binding activity was found in any of the experimental groups when Mt -784 ARE probe was used for the assay (Figure 3C).

Intranasal administration of si-PPARγ suppressed total (40%) expression of lung PPARγ proteins relative to si-NS treatment in Nrf2^+/+ mice exposed to air (Figure 4A). This reduction level is similar to the degree of PPARγ suppression in various tissues of Pparg^+/− mice (39, 40). After 72 hours O₂, significant elevation of total PPARγ levels (45%) was found in si-NS–treated mouse lungs (Figure 4A). Total PPARγ level in mice treated with si-PPARγ was not significantly changed after hyperoxia (Figure 4A). The nuclear level of PPARγ did not differ significantly between treatments in air control mice (Figure 4A). Nuclear translocation of lung PPARγ was significantly increased by O₂ in mice treated with si-NS, whereas it remained unchanged in mice treated with si-PPARγ (Figure 4A). No hyperoxia-enhanced PPRE binding was found in mice treated with si-PPARγ (Figure 4B).

Relative to si-NS controls, si-PPARγ treatment significantly exacerbated (10-fold) hyperoxia-induced pulmonary neutrophilic infiltration in Nrf2^+/+ mice (Figure 5A). si-PPARγ also caused a slight, but statistically significant, increase in mean numbers of BAL neutrophils after air exposure. Significant hyperoxia-induced increases in the number of BAL lymphocytes (Figure 5A) and eosinophils (Figure 5A) were observed only in mice treated with si-PPARγ. In addition to the cellular inflammation, O₂ caused significantly greater increase in pulmonary NF-κB p65 activity (Figure 5B) and BAL IL-6 concentrations (Figure 5C) in mice treated with si-PPARγ than in mice treated with si-NS.

TUNEL-positive cells were detected mainly in epithelium lining bronchi, bronchioles, and terminal bronchioles in all air-exposed mice (Figure 6A). Hyperoxia exposure enhanced TUNEL staining of the airway epithelium, and caused marked apoptosis in inflammatory cells (macrophages and neutrophils), alveolar and bronchial/bronchiolar endothelial cells, and endothelium and smooth muscles of perivascular and peribronchial regions (Figure 6A). Hyperoxia-induced increases in TUNEL-positive cells were more obvious in mice treated with si-NS relative to mice treated with si-PPARγ (Figure 6A). Furthermore, O₂ caused greater protein levels of signal transducers known to be associated with PPARγ-induced cellular apoptosis, phosphatase and tensin homolog deleted on chromosome 10 (PTEN), caspase-8 (active form), and bax in the lungs of mice treated with si-NS relative to the si-PPARγ–treated lungs (Figure 6B).

Recent studies indicate that PPARγ may control the expression of cytoprotective genes including GST-α, HO-1, and CD36 directly through their PPRE or indirectly through Nrf2 (41–45). To determine whether PPARγ modulates these cytoprotective genes and Nrf2 in the lung, their expression was compared in wild type mice treated with si-PPARγ or si-NS (Figure 7A). Constitutive mRNA expression of HO-1 was significantly suppressed in the lungs of Nrf2^+/+ mice after si-PPARγ treatment. Significant up-regulation of HO-1 mRNA by O₂ was inhibited in mice treated with si-PPARγ. Message level of CD36 was significantly elevated in si-NS–treated mice, but not in si-PPARγ–treated mice after O₂. Similarly, O₂ significantly induced mRNA expression of Nrf2, NQO1, and GST-Ya in si-NS–treated mice, but their expression levels were not significantly affected by O₂ in si-PPARγ–treated mice. Gel shift analysis determined that total ARE binding activity of pulmonary nuclear proteins (SB) was suppressed at baseline and after hyperoxia (72 h) in mice treated with si-PPARγ compared with mice treated with si-NS (Figure 7B). We applied the computational PWM statistical method that has been used to search AREs, and identified two potential sequences for PPAR binding (PPRE) in the 5′ upstream region of murine Nrf2 (Table 2). Gel shift analysis of lung nuclear proteins on the Nrf2 promoter -2931 PPRE-like sequence identified enhanced binding activity (SB) after O₂ (72 h) with a similar pattern of band shift as shown on PPRE consensus sequence (Figure 7C). Addition of anti-RXRα antibody identified elevated specific PPRE-PPARγ-RXRα binding activity (SSB) after O₂. These findings suggested involvement of the potential PPRE in transcriptional activation of Nrf2, and warrant further functional study. Overall, attenuated expression of Nrf2 and ARE- or PPRE-bearing genes and suppressed nuclear ARE binding activity in the lung concurrently with suppressed PPARγ level suggests that PPARγ could serve not only as a Nrf2 downstream effector but may also modulate Nrf2-ARE pathways in the pathogenesis of hyperoxia.