Search tips
Search criteria 


Logo of plosonePLoS OneView this ArticleSubmit to PLoSGet E-mail AlertsContact UsPublic Library of Science (PLoS)
PLoS One. 2011; 6(10): e26590.
Published online 2011 October 21. doi:  10.1371/journal.pone.0026590
PMCID: PMC3198791

Targeted Deletion of Nrf2 Reduces Urethane-Induced Lung Tumor Development in Mice

Irina V. Lebedeva, Editor


Nrf2 is a key transcription factor that regulates cellular redox and defense responses. However, permanent Nrf2 activation in human lung carcinomas promotes pulmonary malignancy and chemoresistance. We tested the hypothesis that Nrf2 has cell survival properties and lack of Nrf2 suppresses chemically-induced pulmonary neoplasia by treating Nrf2+/+ and Nrf2-/- mice with urethane. Airway inflammation and injury were assessed by bronchoalveolar lavage analyses and histopathology, and lung tumors were analyzed by gross and histologic analysis. We used transcriptomics to assess Nrf2-dependent changes in pulmonary gene transcripts at multiple stages of neoplasia. Lung hyperpermeability, cell death and apoptosis, and inflammatory cell infiltration were significantly higher in Nrf2-/- mice compared to Nrf2+/+ mice 9 and 11 wk after urethane. Significantly fewer lung adenomas were found in Nrf2-/- mice than in Nrf2+/+ mice at 12 and 22 wk. Nrf2 modulated expression of genes involved cell-cell signaling, glutathione metabolism and oxidative stress response, and immune responses during early stage neoplasia. In lung tumors, Nrf2-altered genes had roles in transcriptional regulation of cell cycle and proliferation, carcinogenesis, organismal injury and abnormalities, xenobiotic metabolism, and cell-cell signaling genes. Collectively, Nrf2 deficiency decreased susceptibility to urethane-induced lung tumorigenesis in mice. Cell survival properties of Nrf2 were supported, at least in part, by reduced early death of initiated cells and heightened advantage for tumor cell expansion in Nrf2+/+ mice relative to Nrf2-/- mice. Our results were consistent with the concept that Nrf2 over-activation is an adaptive response of cancer conferring resistance to anti-cancer drugs and promoting malignancy.


Lung cancer is the leading cause of cancer mortality worldwide, thus creating an enormous public health burden [1]. Adenocarcinoma (AC), the predominant subtype of non-small cell lung carcinoma (NSCLC), is the most prevalent NSCLC among smokers and is the only lung cancer found in non-smokers [2]. Unfortunately due to the lack of useful biomarkers, AC is rarely detectable until advanced stages of the disease, which makes it one of the most clinically intractable of lung cancers [3].

In mice, urethane induces lung ACs and the time frame for development of hyperplasias, adenomas, and ACs is well characterized and accepted as a model for human lung AC [4], [5], [6]. Histopathologically, murine ACs are indistinguishable from human AC [4], [7]. The progenitor cells for AC development include type II pneumocytes and Clara cells [4], [7]. ACs are associated with mutations in Kras oncogene and tumor suppressor genes such as Tp53 in humans and mice.

Nuclear factor, erythroid derived 2, like 2 (Nrf2), is a ubiquitous key modulator of cellular defense against oxidative stress and inflammation. Nrf2 binds to a cis-acting antioxidant response element (ARE) to induce transcription of multiple cytoprotective proteins including phase 2 detoxifying enzymes, drug efflux pumps, and reactive oxygen species (ROS) scavengers [8]. Kelch-like ECH-associated protein 1 (Keap1) suppresses Nrf2 to maintain homeostasis of Nrf2-ARE responsiveness by sequestering and driving proteasomal degradation of cytoplasmic Nrf2 and by shuttling nuclear Nrf2 to the cytoplasm. Stimuli which dissociate Nrf2 from Keap1 to activate Nrf2 include pro-oxidants, antioxidants, and chemopreventive agents. In multiple models of pulmonary non-neoplastic inflammatory and oxidant diseases, exacerbated injury, inflammation, and oxidant stress were found in Nrf2 deficient mice (Nrf2-/-) compared to wild type (Nrf2+/+) mice [8], [9], [10].

Importantly, recent studies have revealed that somatic mutations in functional domains or epigenetic changes (methylation) in CpG islands of KEAP1 which decrease KEAP1 function are associated with increased risk of human lung neoplasia and chemoresistance [11], [12], [13], [14]. Further, siRNA silencing of NRF2 in these neoplastic cells inhibited their growth and recovered chemotherapeutic sensitivity [13], [15]. These studies suggested that lung cancer cells with ’permanent’ NRF2-ARE activation due to weakened KEAP1 acquired multiple advantages for proliferation and resistance to chemotherapy promoting malignancy. However, molecular mechanisms explaining this potential advantage for cancer cell growth have not been fully elucidated. In the current study, we tested the hypothesis that Nrf2 has a cell survival role in chemical-induced pulmonary neoplasia using Nrf2+/+ and Nrf2-/- mice. Using bioinformatic tools for transcriptome analysis, we defined potential downstream effector mechanisms of Nrf2-directed lung tumorigenesis.


Differential urethane sensitivity during pre-neoplastic stage

Pulmonary injury

Microscopic analysis of bronchoalveolar lavage (BAL) cells indicated pulmonary cytotoxicity including lysis at 9–11 wk after initial urethane injection in Nrf2-/- and Nrf2+/+ mice. Increases in mean numbers of BAL epithelial cells and leukocytes, as well as total protein concentration, were significantly greater in Nrf2-/- than Nrf2+/+ mice after 9–11 wk (Figure 1-A). By 11 wk, BAL cell aggregation was mostly resolved in both strains, but cell lysis remained in Nrf2-/- mice with concurrent manifestation of significantly higher large atypical macrophages showing phagocytosis or necrosis-like features than in Nrf2+/+ mice (Figure 1A; Figure S1). Significantly greater membrane rupture was also found in BAL cells from Nrf2-/- mice than Nrf2+/+ mice (Figure S1). At 9 wk, BAL cell clustering accompanied cytolysis, and this feature was markedly greater in Nrf2-/- mice compared to Nrf2+/+ mice (Figure S1). This observation is assumed to be associated with excess viscous secretion such as mucus as suggested by aggregation of cells with secreted mucus in Alcian Blue (pH 2.5)/periodic acid-Schiff (AB/PAS)-stained lungs (Figure S1), or by procoagulant activity due to fibrin deposition in mononuclear phagocytes which occurs during lung inflammation [16], [17]. Infiltrated neutrophils were obvious but not quantifiable due to extensive clustering and lysis of BAL cells in Nrf2-/- mice; consistent with this observation, neutrophil myeloperoxidase (MPO) activity was significantly higher in Nrf2-/- mice than in Nrf2+/+ mice treated with urethane (Figure 1A). Urethane-induced increase in lactate dehydrogenase (LDH) activity, a quantitative indicator of necrotic cell lysis, was also significantly greater in BAL returns from Nrf2-/- compared to Nrf2+/+ mice (Figure 1B).

Figure 1
Bronchoalveolar lavage analysis (BAL) and body weight changes at a pre-neoplastic stage.

Body weight (BW) changes

Urethane restrained BW gain during the injection period (1–11 wk) in both genotypes relative to saline injection (Figure 1C). However, a lower BW was observed in Nrf2-/- mice than in Nrf2+/+ mice during the inflammatory period from 7–11 wk (Figure 1C).

Differential responses in early neoplastic stage

Apoptosis and proliferation

Immunohistochemical detection of proliferating cell nuclear antigen (PCNA, a cell proliferation marker [18]) at 12 wk demonstrated that PCNA-positive cells in G1/S phase were localized extensively in small adenomas, focal alveolar hyperplastic lesions, injured perivascular and peribronchial regions, and nodular lymphoid aggregates in adventitia of blood vessels. Compared to Nrf2-/-, more abundant and intense localization of PCNA were found in small adenomatous regions of Nrf2+/+ mice (Figure 2A). PCNA was sporadic in conducting airway epithelium of saline controls (Figure 2A). Consistent with differential BAL cell necrosis at 9–11 wk, the number of apoptotic lung cells determined by terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) in proximal pulmonary sections was more highly elevated in Nrf2-/- mice compared to Nrf2+/+ mice at 12 wk (Figure 2B). TUNEL-positive cells included epithelial, endothelial, and smooth muscle cells. Few apoptotic cells were found in lungs of vehicle control mice.

Figure 2
Early stage tumorigenesis at 12 wk.

Early adenoma formation

Lung tumors were evident and countable (>200 µm) from 12 wk following initiation of urethane treatment. The numbers of sporadic adenomas were significantly lower (2-fold) in Nrf2-/- mice compared to Nrf2+/+ mice (Figure 2C).

Differential tumorigenesis at 22 wk

Multiplicity (~43%) of small (≤1.6 mm) adenomas was significantly reduced in Nrf2-/- mice compared to Nrf2+/+ mice at 22 wk after urethane treatment (Figure 3A). However, no differences were found in overall tumor size or tumor morphology between genotypes. Large tumors were located primarily at the lung pleural surface, while small developing tumors were more frequent in internal regions. While no specific differences were observed in other BAL phenotypes, the numbers of neutrophils were significantly lower in Nrf2-/- mice compared to Nrf2+/+ mice at 22 wk (Figure 3B).

Figure 3
Adenoma development and persistent lung inflammation at 22 wk.

Nrf2 activation in tumor-bearing lungs

Compared to saline controls at 22 wk, lung Nrf2 protein was increased in tumor tissues and in remaining uninvolved lung tissues (UN: adjacent, non-tumor bearing regions) of Nrf2+/+ mice (Figure 4A). Nuclear levels of Nrf2 (Figure 4A) and total ARE binding activity of nuclear proteins (Figure 4B) were also higher in tumor-bearing lungs. Specific binding activities of Nrf2 and its dimerization partner small Maf on ARE were also heightened by urethane (Figure 4B). Abundant Nrf2 was located throughout the epithelium lining bronchial/bronchiolar and alveolar airways in control mice (Figure 4Ca). At 22 wk after urethane, the Nrf2-bearing cell populations were apparent in hyperplastic foci and growing adenomas in addition to their increase in the airway epithelium (Figure 4Cc).

Figure 4
Pulmonary Nrf2 expression and activation caused by urethane.

Gene expression profiling during urethane-induced tumorigenesis

Transcriptomics in early tumorigenesis of Nrf2+/+ mice (12 wks)

To identify potential mechanisms of urethane-induced tumorigenesis, we then used trancriptomics of tumor and uninvolved tissue to identify differentiating gene networks. Significant changes in 6519 lung gene transcripts were found in Nrf2+/+ mice relative to saline controls in 12 wk, and expression changes were mostly 2-fold or lower (Table S1; n = 3 mice/group). Functional classification by Ingenuity Pathway Analysis (IPA) determined that lung genes changed during early tumorigenesis have roles in cellular growth and proliferation, cell and tissue development, cell cycle and cellular assembly, small molecule biochemistry, lipid metabolism, and molecular transport (Figure 5A, top panel; Table S1).

Figure 5
Lung gene expression profiles during urethane-induced tumorigenesis.

Tumor gene expression profiles in Nrf2+/+ mice (22 wk)

Among genes significantly changed over saline controls at 22 wk in Nrf2+/+ mice (8683 genes), 83% were changed significantly only in tumors (Figure 5-B; n = 3 mice/group). Expression kinetic profiles of genes modulated in tumor tissues (22 wk) were similar to their kinetics during early tumorigenesis (12 wk) and in the microenvironment of tumors (i.e., UN) at 22 wk (Figure 5B). Genes changed by urethane (mostly ≤2-fold) in UN tissues played roles mainly in small molecule biochemistry, immune cell trafficking, and cellular movement (Figure 5A, middle panel). A small number of genes (n = 78) associated with signal transduction, transcriptional regulation, ubiquitin-dependent protein catabolism, and muscle contraction were changed in UN, but not in tumors at 22 wk. Individual genes that demonstrated the greatest fold induction in the tumors included transmembrane protein 27 (Tmem27, 51-fold), BCL12-like 15 (Bcl2l15, 48-fold), and claudin 2 (Cldn2, 43-fold). In contrast, genes suppressed the most in the tumors included carbonic anhydrase 3 (Car3, -92-fold), myosin light polypeptide 7, regulatory (Myl7, -87-fold), troponin T2, cardiac (Tnnt2, -85-fold), actin, alpha, cardiac muscle 1 (Actc1, -52-fold), and cytochrome P450, family 2, subfamily a, polypeptide 4 (Cyp2a4, -36-fold). Also changes in numerous families of gene clusters were evident, which included ankyrin and ankyrin repeat domain, Rho GTPase activating proteins, chemokine (C-C motif) or (C-X-C motif) ligand and receptors, cell cycle protein families, C-type lectin domain family, Cyp family, fibroblast growth factors and receptors, gap junction protein (Gj) family, and G protein-coupled receptors. Genes changed more than 2-fold in tumors (n = 3461, up-regulated 1233 and down-regulated 2228 genes) were associated functionally in networks of genetic/skeletal and muscular/developmental disorders, cellular and hematological system development, hematopoiesis, cell cycle/growth, and cancer (Figure S2; Table S2). Genes involved in cellular growth and proliferation were modulated significantly by urethane regardless of the times (12 and 22 wk) and tissues (UN, tumors) during carcinogenesis, while genes associated with cancer (e.g., Bcl6, Peg3), cellular movement (e.g., Prdx1, Cadm1), neurological disease (e.g., Cxcl1. Aff2), and inflammatory response (e.g., Itga4, Pla2g7) were significantly changed in all the tissue types after tumor development at 22 wk (Figure 5A, lower panel). Genes for small molecular biochemistry (oxidation, metabolism, and transport reactions of acids, vitamins, and hormones) and lipid and nucleic acid metabolisms (e.g., Cyp1a1, Slco1a5, Fabp1) and cell morphology (e.g., Cdkn1a, Cxcl9) were modulated in common during the early tumorigenesis stage (12 wk) and in UN tissues at 22 wk (Figure 5A) suggesting similar molecular events are taking place in early neoplasia and in the tumor microenvironment.

Nrf2-dependent transcriptomics in early tumorigenesis (12 wk)

Because we found Nrf2-dependent differences in tumorigenesis, we used trancriptomic profiling of tumors and uninvolved tissues to identify gene networks that provide potential mechanistic insight to Nrf2-mediated effects. Gene transcripts (n = 118) that were significantly different between Nrf2+/+ and Nrf2-/- mice 12 wk after urethane treatment (Table 1; n = 3 mice/group) were categorized primarily in cell-to-cell signaling and interaction, connective tissue development and function, cellular movement, tissue/cell morphology and cellular development, and inflammatory response and immune cell trafficking processes (Figure 6A; Figure S3A). Expression profiles were classified into 4 cluster sets (Figure 6B, Dataset S1). Transcripts highly up-regulated by urethane in Nrf2-/- mice relative to Nrf2+/+ mice (set 0) included melanoma antigen (Mela), growth arrest specific 6 (Gas6), integrin alpha 6 (Itga6), matrix metalloproteinase 2 (Mmp2), CD34 antigen (Cd34), D site albumin promoter binding protein (Dbp), and transforming growth factor, beta induced (Tgfbi). Other gene clusters (sets 1–3) that were relatively suppressed in Nrf2-/- mice compared to Nrf2+/+ mice (Figure 6B, Table 1, Dataset S1) included genes for ARE-responsive antioxidants, solute carrier family, chemokines, arginase type II (Arg2), and attractin like 1 (Atrnl1).

Figure 6
Effect of Nrf2 deletion on gene expression during lung tumorigenesis.
Table 1
Representative lung genes from a total of 118 that were significantly (p<0.05) different between Nrf2+/+ and Nrf2-/- mice 12 wk after beginning urethane treatment.

Nrf2-dependent transcriptomics in tumors (22 wk)

Urethane caused Nrf2-dependent changes in a small number of the gene transcripts (n = 11) in the tumor microenvironment (UN) at 22 wk relative to controls. Transcripts included polymerase, delta (Poldip2), sonic hedgehog (Shh), and ribonuclease T2 (Rnaset2). IPA indicated that tumor microenvironment participates in a network of cellular growth and proliferation, tumor morphology, and cellular development (Figure S3B; n = 3 mice/group). A number of genes (n = 376) were significantly changed in tumor tissues relative to vehicle-treated tissues in a Nrf2-dependent manner at 22 wk (Table 2, Dataset S2). The largest family of genes encodes thiol metabolism enzymes (e.g. glutathione-S-transferase family), solute carrier family members, and transmembrane proteins. Predominant functional categories of Nrf2-dependently changed tumor genes at 22 wk (Figure 6A, bottom panel) were largely similar to those of Nrf2-dependent lung genes determined at the 12 wk early neoplastic stage (Figure 6A, top panel). However, Nrf2 effector genes in tumor tissue were more closely associated with drug metabolism, cell cycle and death and organismal survival, and tumor morphology (Figure 6A, Figure S3C). Canonical pathway analysis by IPA determined that Nrf2 effector genes in lung tumors overlaid multiple signal transduction mechanisms including predicted glutathione metabolism (12 out of 98 pathway molecules match, 12/98), oxidative stress response (14/183), aryl hydrocarbon receptor signaling (12/154), LPS/IL-1 mediated inhibition of RXR function (14/213), glioblastoma multiforme signaling (12/163), and others (Dataset S3). Expression profiles of these 376 Nrf2-dependent tumor transcripts were classified (Figure 6C, Dataset S2); subsets of tumor genes that were relatively less-responsive to urethane (set 0) or more suppressed (set 1) in Nrf2-/- than in Nrf2+/+ mice involved cell cycle, antioxidant regulation, cell-cell interaction, and metabolism. Some antioxidants and solute carrier family genes were differentially down-regulated in tumors (set 2), while vasculature, immunity, and other antioxidant genes were differentially up-regulated in tumors (set 3). Although major functional categories of Nrf2-dependent transcripts were similar in early neoplastic lungs and in tumors (see Figure 6B), only 21 individual genes were found to be common at these two stages (Figure 6D). Among them, Cd34, protein tyrosine phosphatase, receptor type, F (Ptprf), Txnrd1, Pgd, Ugt1a1, Gpr137b, and dynein light chain Tctex-type 1 (Dynlt1) were connected in a molecular network of cancer-cell cycle-cell death (Figure S3C).

Table 2
Representative lung tumor gene transcripts from a total of 376 that were significantly (p<0.01).

Validation of Nrf2-dependent transcripts identified at early and late stages of carcinogenesis

Differential mRNA expression of multiple ARE-responsive genes that were increased (aldo-keto reductase family 1, member B8, Akr1b8; aldehyde oxidase 3, Aox3; Prdx1) or decreased (aldehyde dehydrogenase 1 family, member A1, Aldh1a1; Gstm; glutathione peroxidase 2, Gpx2) in tumors were confirmed using qRT-PCR (Figure S4A). Total tissue glutathione (GSH) levels were also compared between the two genotypes to support Nrf2-dependent changes of many genes engaged in GSH synthesis mechanisms (e.g., glutathione synthetase, Gss; Gclc; glucose-6-phosphate dehydrogenase X-linked, G6pdx) after urethane treatment (Figure S4A). Non-antioxidant genes with Nrf2-dependent expression patterns were confirmed by qRT-PCR in 12 wk (Mela, Cd34, Areg, Itga6, Arg2, Nek6) and in 22 wk (Cxcl1, Cks1b, Cdkn2c, Gjb1,Cd14, prostaglandin I2 synthase, Ptgis), most of which were involved in the primary functional networks identified using the pathway analysis tool (Table 2, Figure S3).


We have found that murine Nrf2 has a cell survival role in lung cancer. That is, Nrf2 deficiency reduced sensitivity to urethane-induced lung tumorigenesis. Importantly, Nrf2 deficiency significantly enhanced lung inflammatory and cell injury responses during early pre-neoplastic stages. Transcriptomic analysis of the Nrf2-dependent gene transcript expression during early tumorigenesis (12 wk after urethane) identified specific pathways including cell-to-cell signaling and interaction, connective tissue development and function, glutathione metabolism and oxidative stress, and immune/inflammatory responses. Nrf2-dependent gene expression in the late stages of tumorigenesis was associated with cell cycle regulation, organismal injury/survival, xenobiotic/thiol metabolism, and cell-to-cell signaling. Many similarities were found in these functional categories between the early and late phases of tumorigenesis, while immunological responses early and cell cycle/growth and organismal survival in the more advanced stages were unique. Results suggest that reductions of ARE-responsive cytoprotective genes in tumor cells of Nrf2-/- mice render disadvantage to their survival during urethane-induced tumorigenesis and that other, non-antioxidant/defense pathways observed such as cell cycle and cell-to-cell signaling are essential components to this biological consequence. Together, the current findings support observations that activated Nrf2 is associated with an increased risk of human lung cancer [13].

Our results deviated from other experimental tumorigenesis models in non-pulmonary tissues (e.g., gallbladder, liver, stomach, colon, esophagus, skin, head/neck, prostate, bladder, mammary) where tumors or cancers were increased in Nrf2-/- mice relative to Nrf2+/+ mice [19], [20], [21], [22], [23], [24], [25], [26], [27], [28]. However, a recent study found that experimentally implanted lung cancer cells (Lewis lung carcinomas) produced significantly more pulmonary metastatic nodules in Nrf2-/- than in Nrf2+/+mice, while the metastatic nodule formation was suppressed in Keap1-knockdown mice compared to Keap1 normal mice [29]. The authors suggested that Nrf2 in host lung tissues prevents pulmonary metastasis through the stabilization of the redox balance in the hematopoietic and immune cells by increasing immunosuppressive cells which can lead to a decrease in CD8+ T-cell immunity. We also tested the role of Nrf2 in a 2-stage tumorigenesis model using 3-methylcholanthrene (3-MCA) as the initiator followed by butylated hydroxytoluene (BHT) as the promoter. Interestingly, no differences in tumor multiplicity were found between Nrf2-/- and Nrf2+/+ mice (data not shown), which suggested that the molecular pathways required for urethane-induced tumorigenesis are different from the 2-stage model. Given the potential role for Nrf2 in protection from cytotoxicity during the urethane-induced tumor initiation stage, it is possible that the cell survival or beneficial nature of Nrf2 may depend on chemical carcinogens and on the stages of carcinogenesis (i.e., initiation, promotion, or metastasis) in mice. Therefore, assessment of other models, specifically those bypassing inflammation and injury phase during initiation or those focused on the advanced stages, well past tumor promotion, may provide additional insight to the characteristics of pulmonary Nrf2.

Promotion of cell survival is a phenotype that has been linked to Nrf2. In embryonic fibroblasts and HepG2 cells, cell survival involves adaptation to deal with the increased ROS called "an adaptive survival response" and in some cases involves an Nrf2 effector, heme oxygenase-1 [30]. In the current study we found that not only well-known ARE responsive genes (e.g., Prdx1, Ugt) but other novel Nrf2-dependent redox genes such as Akr1b8, NADPH oxidase organizer 1 (Noxo1), and solute carrier genes (e.g., Slc13a2) may have a role in cellular survival during lung tumorigenesis. Recent studies determined that permanent or constitutive NRF2 activation in human primary lung cancer cells and lung cancer cell lines (e.g., A549, H460) with KEAP1 and/or NRF2 somatic mutations was associated with enhanced chemoresistance and radioresistance, and siRNA inhibition of NRF2 restored the chemosensitivity of these cells [15], [25], [28], [31]. Chemoresistance observed in these studies resulted from the overexpression of ARE-responsive genes encoding drug efflux pumps including multidrug resistant–associated protein (Mrp). This indicates that permanent NRF2 activation may be a novel biomarker in lung cancers and that Nrf2 inhibitors may provide increased efficacy for the treatment of cancers in specific patients with abnormally high NRF2-ARE activation.

The current study also further supports a role of inflammation in the urethane-induced carcinogenesis model [32], and suggests the association of multiple inflammatory mediators, particularly Ptgis and Cxcl1, with Nrf2-dependent tumorigenesis. PTGIS, the enzyme that synthesizes prostacyclin, is anti-inflammatory in the lung [33]. Importantly, a 92% reduction in tumor multiplicity was found in Ptgis-overexpressing mice [33] and prostacyclin analogs are currently in use for human NSCLC clinical trials in the National Cancer Institute Lung Cancer Biomarker and Chemoprevention Consortium. Heightened Ptgis expression in tumors of Nrf2-/- mice suggested that tumor microenvironment suppressed the normal inflammatory response in these mice. We also found persistent neutrophilic infiltration and concurrent induction of Cxcl1 (a neutrophil chemoattractant) in Nrf2+/+ mice by 22 wk, but not in Nrf2-/- mice. Neutrophils are often observed during the more advanced stages of lung tumor development, specifically urethane-induced [32], and are thought to be involved in the production of ROS as indicated by skin carcinogenesis models [34].

Our transcriptome analysis also identified numerous urethane-dependent genes in lung tumors of wild-type mice (BALB/cCR), and the functional categories of these neoplastic genes (e.g., genetic disorder, cardiovascular system function, cell death, immunological disease) were largely not concordant with those of genes changed at the early-neoplastic stage (e.g., small molecular biochemistry, nucleic acid metabolism, cell morphology, lipid metabolism, cell cycle, molecular transport, protein trafficking) and in uninvolved tissues of tumor-bearing lungs (e.g., small molecular biochemistry, immune cell trafficking, hematopoiesis, nucleic acid metabolism, drug metabolism, nerve system function). These findings indicate distinct biochemical and molecular changes are occurring in tumor tissues, while transcriptomics in the pre/early neoplastic stage and in uninvolved tumor microenvironments are in general similar. However, transcriptome changes in cellular growth and proliferation-associated genes were common throughout all tissues and neoplastic stages. Genes such as claudin 3 (Cldn3), thymidylate synthetase (Tyms), and thymidine kinase 1 (Tk1) in these tumors were previously found in a single injection urethane model (24–42 wk following urethane) in tumors of A/J mice [35]. In addition, comparison with toll-like receptor 4 (Tlr4) dependent genes in a 2-stage (MCA/BHT) mouse pulmonary carcinogenesis model [36] identified many genes (e.g. angiotensin I converting enzyme 2, Ace2; Bcl2-like 14,Bcl2l14; claudin 2, Cldn2) also to be Nrf2-dependent. Genes changed by urethane in the early neoplastic stage that also differed in the MCA-BHT model [36] included amphiregulin (Areg), arginase 1 (Arg1), chemokine receptor 2 (Ccr2), cyclin-dependent kinase inhibitor 1A (Cdkn1a), epiregulin (Ereg), gap junction protein, alpha 1 (Gja1), and secreted phosphoprotein 1 (Spp1). Overall from previous and current tumor analyses using different dosing regimens, carcinogens, and strains of mice, a panel of common transcripts (e.g. Arg1, Areg, Ereg, Ccr2, Cdkn1a, Gja1, Ptgis, Spp1) may provide a useful tool for early diagnosis of lung neoplasm.

In conclusion, our novel results provide evidence that Nrf2 contributes to lung tumorigenesis susceptibility. We propose that during pulmonary inflammation and injury in the early stages of tumor development induced by urethane, lung cells in Nrf2-/- mice had reduced cell survival factors such as cellular redox and drug metabolism enzymes and cell maintenance system, relative to those in Nrf2+/+ cells (see hypothetical schematic, Figure 7). Consequently, Nrf2-/- mice had decreased cytoprotection and massive death of initiated cells that would normally develop into adenomas (Figure 7). Nrf2-dependent genes associated in cell cycle and cell death also affect differential proliferation of these initiated tumor progenitor cells between Nrf2+/+ and Nrf2-/- mice (Figure 7). Overall, increased susceptibility to urethane-induced pre-neoplastic injury leading to net cell loss due to lack of cell survival signals in Nrf2-/- mice is paradoxically beneficial to their tumor suppression. Results indicate the importance of understanding the pre-neoplastic pathologic events as well as understanding the mechanisms through which cancer chemotherapy using supplemental antioxidants modulates multi-stage carcinogenesis. In addition to this unique notion, our transcriptomal analysis provides new insights into the downstream molecular mechanisms of Nrf2 in a pulmonary neoplastic microenvironment.

Figure 7
Proposed role for Nrf2 in urethane-induced lung tumorigenesis.

Materials and Methods

Ethics Statement

All animal use was approved by the National Institute of Environmental Health Sciences Animal Care and Use Committee under protocol number P03-05 and follows the Helsinki convention for the use and care of animals.

Animals and carcinogen treatment

Female wild-type mice (BALB/cCR(Japan)-Nrf2+/+) and Nrf2 deficient (BALB/cCR(Japan)-Nrf2-/-) littermates were generated from a breeding colony at the National Institute of Environmental Health Sciences (NIEHS) animal facility. Animals were housed in a virus- and antigen-free room. Water and mouse chow (modified AIN-76A, Harlan Teklad, Madison, WI) were provided ad libitum. Mice (4–6 wk) were injected with urethane (1 mg per g BW in saline; Sigma, St. Louis, MO) or with vehicle (saline) i.p. weekly for 7 wk. Mice were sacrificed at 9, 11, 12, or 22 wk after the first injection by Nembutal overdose (104 g/Kg BW).

Lung fixation, tumor analysis, histopathology and immunohistopathology

Lung tissues were fixed with Tellyesniczky's fixative (22 wk post injection) or 10% neutral buffered formalin (other times) for histopathology and tumor analysis as described previously [37]. Apoptotic cell death was determined on lung tissue sections in situ using a TUNEL kit (Promega, Madison, WI) as described previously [38]. See Methods S1 for additional details.

BAL cell analyses and myeloperoxidase assay

The right lung of each mouse was lavaged in situ, four consecutive times with HBSS (0.5 ml/25 g BW). BAL fluid was then analyzed for total protein content (a marker of lung permeability change) and numbers of epithelial (a marker of epithelial cell injury) and inflammatory cells following published procedures using a modified Wright’s stain (Diff-Quik; Baxter Scientific Products, McGaw Park, IL) [36]. Due to cellular lysis in BAL which prohibited neutrophil enumeration, neutrophil abundance in BAL was quantified through measurement of neutrophil myeloperoxidase (MPO) using a colorimetric analysis kit (Cytostore, Canada) which assessed conversion of hydrogen peroxide to oxygen radicals by MPO. Cell viability and cytotoxicity were determined by the amount of released LDH in BAL aliquots (50 µl) using a colorimetric assay (Sigma-Aldrich, St. Louis, MO).

Tissue protein isolation

Tumors (involved regions) were micro-dissected from lungs and remaining uninvolved tissue (UN) was used as a control in urethane-treated lungs at 22 wk. Snap frozen tumor, UN, or whole lungs (saline-treated controls and urethane-treated lungs at 12 wk) were homogenized (n = 3/group) in RIPA buffer. Nuclear proteins were isolated from pulverized lung samples following procedures described elsewhere [39]. Proteins were quantified and stored in aliquots at −70°C.

Electrophoretic mobility shift assay (EMSA)

Nuclear DNA binding activity of Nrf2 was determined by EMSA analyses of nuclear protein aliquots (5 µg) on 3×104 cpm [γ32P] ATP end-labeled ARE consensus sequence following procedures described previously [40]. Specific binding activity for Nrf2 and small Maf protein was determined by pre-incubation of nuclear proteins with anti-Nrf2 (sc-722x, Santa Cruz Biotechnology, Santa Cruz, CA) and anti-small Maf (F/K/G) antibody (sc-22831x, Santa Cruz ), respectively, followed by EMSA.

Immunoblot analyses and GSH measurement

Lung total (30–50 µg) or nuclear (10 µg) proteins were separated on appropriate percentage Tris-HCl SDS-PAGE gels (Bio-Rad, Hercules, CA) and analyzed by routine immunoblotting using specific antibodies against Ki67 (Abcam Inc., Cambridge, MA), Nrf2 (Santa Cruz), lamin B1 (sc-20682, Santa Cruz), or actin (sc-1615, Santa Cruz). Representative protein blot images from multiple analyses were scanned by a Bio-Rad Gel Doc system (Hercules, CA). Total GSH level was determined in lung RIPA homogenates by a colorimetric method following manufacturer instructions (Northwest Life Science Specialties, LLC, Vancouver, WA).

Affymetrix cDNA Microarray Analysis

Total RNA was isolated from homogenates of total lung, UN or TU tissues (n = 3/group) using Qiagen RNeasy Mini kits (Qiagen Inc., Valencia, CA). Microarray processing was done by the NIEHS Microarray Core Facility as described previously [36] with Affymetrix mouse genome 430_2.0 microarrays using Affymetrix reagents and protocols (Santa Clara, CA). GeneSpring Expression Analysis (Agilent Technologies, Inc., Santa Clara, CA) was used for statistical analyses and characterization of data and Ingenuity Pathway Analysis (IPA) software (, Ingenuity Systems, Inc., Redwood City, CA) identified potentially significant functional connections and mechanistic pathways. Characterization of the microarray data is described in Methods S1.

Quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR)

An aliquot (1 µg) of total lung RNA was reverse transcribed as previously published [40] and gene specific primers used in quantitative RT-PCR reactions, similar to previous publications [33]. For additional details, see Methods S1.

Statistical analysis of non-microarray data

Data were expressed as group mean±standard deviation. Two-way ANOVA was used to evaluate effects of treatment (vehicle, urethane) between two genotypes (Nrf2+/+, Nrf2-/-). The Student-Newman-Keuls test was used for a posteriori comparisons of means (p<0.05). All statistical analyses were performed using the SigmaStat 3.0 software program (SPSS Science Inc., Chicago, IL).

Supporting Information

Figure S1

Airway cytotoxicity and secretion in pre-neoplastic stage. Giemsa-stained cytocentrifuge slides of bronchoalveolar lavage (BAL) fluid show increased BAL cell populations and greater cell clustering and lysis caused by urethane in Nrf2-/- than in Nrf2+/+ mice before tumor development at 9 wk (inset: higher magnification of BAL cells). Enhanced airway secretion (black arrow heads) and epithelial mucous production (black arrows) as indicated by AB/PAS-stained bronchial airway sections at 9 wk was consistent with clustering of the BAL cells in Nrf2-/- mice (inset: lower magnification of proximal airway and airspace). At 11 wk, clustering was resolved but greater cell lysis and cell debris in Nrf2-/- mice were accompanied by appearance of macrophages with phagocytosis-like features (red arrows) Insets present higher magnification of phagocytic macrophages. Representative slides showing intermediate magnitude of pathology for each treatment group are presented. AV, alveoli; BR, bronchi; PA, pulmonary artery. Bars indicate 100 µm.


Figure S2

Top functional networks of significantly changed gene transcripts in lung tumors of Nrf2+/+ mice at 22 wk. Ingenuity Pathway Analysis (IPA) generated essential functional networks of the genes significantly (≥2-fold, n = 3461) changed in Nrf2+/+ tumors. Highest association score (40) was for the genetic disorder-skeletal and muscular disorders- developmental disorder network (A) where the genes such as coiled-coil domain containing 85A, (Ccdc85a), dystrophin, muscular dystrophy (Dmd), and fyn proto-oncogene (Fyn) were mapped. Many genes (e.g., CCAAT/enhancer binding protein (C/EBP) alpha, Cebpa; lymphotoxin B, Ltb) were identified for cellular development-cellular growth and proliferation-hematological system development and function-hematopoiesis networks (B and C, Scores 34 and 33, respectively) and others related in cell cycle- cellular movement-cancer network (D, score 33, e.g., p21, Cdkn1a; cell division cycle 20, Cdc20) or in amino acid metabolism-molecular transport-small molecular biochemistry (E, Score 33, e.g., platelet derived growth factor receptor, beta polypeptide Pdgfrb; platelet-activating factor acetylhydrolase, isoform 1b, subunit 3, Pafah1b3) were also closely associated during tumorigenesis. (.ppt).


Figure S3

Top functional networks of Nrf2-dependent genes changed during urethane-induced tumorigenesis. (A) Ingenuity Pathway Analysis (IPA) generated a key network (Score 54) of cell-to-cell signaling and interaction-connective tissue development and function-ophthalmic disease with Nrf2-dependently regulated genes in pre-/early-neoplastic microenvironment (12 wk), in which genes encoding matrix metalloproteinase 2 (Mmp2) and D site albumin promoter binding protein (Dbp) were mapped as core molecules. (B) Nrf2-dependent lung tumor genes were highly associated in the networks of cell cycle-cancer-connective tissue development and function (Score 43, e.g., cyclin D1, Ccnd1; E2F transcription factor 3, E2f3), cell-to-cell signaling and interaction-tissue development-cell function (Score 38, e.g., chemokine (C-X-C motif) ligand 1, Cxcl1; integrin alpha 4, Itga4), and tumor morphology-cancer-dermatological disease and conditions (Score 38, e.g., G protein-coupled 56, Gpr56; glutathione-S-transferase, alpha 3, Gsta3) as generated by IPA. (C) Nrf2-dependently modulated genes in common (n = 21 genes) in early-neoplastic microenvironment (12 wk) and in lung tumors (22 wk) were functionally associated in cancer-cell cycle-cell death network (Score 23). They included genes encoding CD34, UGT1a1, fetuin beta (Fetub), G protein-coupled receptor 137B (Gpr137b), ATP binding cassette, subfamily C, member 4 (Abcc4), phosphogluconate dehydrogenase (Pgd), etc. (.ppt).


Figure S4

Confirmation of microarray transcript profiles. (A) Expression of selected antioxidant/defense gene transcripts that were found by microarray analysis to be significantly increased or decreased by urethane in tumor tissues (22 wk): aldo-keto reductase family 1, member B8 (Akr1b8), aldehyde oxidase 3 (Aox3), peroxiredoxin 1 (Prdx1), glutathione peroxidase 2 (Gpx2), glutathione-S-transferase-µ (Gstm), and aldehyde dehydrogenase 1 family, member A1 (Aldh1a1). Relative suppression of these cytoprotective genes was evident in Nrf2-/- mice. Total glutathione (GSH) was determined in whole lung homogenates using a colorimetric kinetic analysis to support the differential induction of key GSH synthesis enzymes (e.g., glutathione synthetase, Gss; glutamate-cystein ligase, catalytic subunit, Gclc; glucose-6-phosphate dehydrogenase X-linked, G6pdx) between Nrf2+/+ and Nrf2-/- mice during the tumorigenesis. Mean±SEM are presented (n = 3/group). *, p<0.05 vs. genotype-matched saline control mice. +, p<0.05 vs. treatment-matched Nrf2+/+ mice. Expression of selected lung genes significantly varied between Nrf2+/+ and Nrf2-/- mice at 12 wk (B) and at 22 wk (C) were determined by qRT-PCR. All qRT-PCR graphs depicted fold differences relative to Nrf2+/+ saline level of 18s-normalized data. All qRT-PCR data are represented as group mean±SEM (n = 3/group).


Table S1

Representative lung genes significantly changed by urethane at 12 wk in Nrf2+/+ mice (n = 6519, p<0.05).


Table S2

Top functional networks and involved genes significantly (p<0.05) changed in uninvolved tissues and in tumors of Nrf2+/+ mice at 22 wk.


Dataset S1

12 wk Saline vs. Urethane Nrf2-WT vs. Nrf2-KO.


Dataset S2

Genes differentially changed between Nrf2-WT and Nrf2-KO mice by urethane in tumors at 22 wk (n = 376).


Dataset S3

Canonical pathways for 376 tumor genes significantly varied between Nrf2+/+ and Nrf2-/- mice at 22 wk.


Methods S1



The authors thank Drs. Donald Cook and Michael Fessler at the NIEHS for their excellent critical review of the manuscript. Microarray analysis was performed at the NIEHS Microarray Core.


Competing Interests: The authors have declared that no competing interests exist.

Funding: This research was supported by the Intramural Research Program of the National Institute of Environmental Health Sciences and Michigan State University internal funds. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.


1. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin. 2005;55:74–108. [PubMed]
2. Schottenfeld D. Etiology and Epidemiology of Lung Cancer. In: Pass HI, Mitchell JB, Johnson DH, Turrisi AT, Minna JD, editors. Lung Cancer- Principles and Practice. 3rd ed. Philadelphia: Lippincott Williams and Wilkins; 2005. pp. 3–24.
3. Colby TV, Noguchi M, Henschke C, Vazquez MF, Geisinger K, Yokose T, et al. Adenocarcinoma. In: Travis WD, Brambilla, E, Muller-Hermelink, H.K, Harris, C.C, editors. World Health Organization classification of tumours: pathology and genetics of tumours of the lung, pleura, thymus, and heart. Lyon, France: IARC Press; 2004. pp. 35–44.
4. Malkinson AM. Primary lung tumors in mice: an experimentally manipulable model of human adenocarcinoma. Cancer Res. 1992;52:2670s–2676s. [PubMed]
5. Mason RJ, Kalina M, Nielsen LD, Malkinson AM, Shannon JM. Surfactant protein C expression in urethane-induced murine pulmonary tumors. Am J Pathol. 2000;156:175–182. [PubMed]
6. Stathopoulos GT, Sherrill TP, Cheng DS, Scoggins RM, Han W, et al. Epithelial NF-kappaB activation promotes urethane-induced lung carcinogenesis. Proc Natl Acad Sci U S A. 2007;104:18514–18519. [PubMed]
7. Malkinson AM. Molecular comparison of human and mouse pulmonary adenocarcinomas. ExpLung Res. 1998;24:541–555. [PubMed]
8. Cho HY, Kleeberger SR. Nrf2 protects against airway disorders. Toxicol Appl Pharmacol. 2009;244:43–56. [PubMed]
9. Harvey CJ, Thimmulappa RK, Sethi S, Kong X, Yarmus L, et al. Targeting Nrf2 signaling improves bacterial clearance by alveolar macrophages in patients with COPD and in a mouse model. Sci Transl Med. 3:78ra32. [PubMed]
10. Lugade AA, Vethanayagam RR, Nasirikenari M, Bogner PN, Segal BH, et al. Am J Respir Cell Mol Biol; 2011. Nrf2 Regulates Chronic Lung Inflammation and B cell Responses to non-typeable Haemophilus influenzae. [PMC free article] [PubMed]
11. Ohta T, Iijima K, Miyamoto M, Nakahara I, Tanaka H, et al. Loss of Keap1 function activates Nrf2 and provides advantages for lung cancer cell growth. Cancer Res. 2008;68:1303–1309. [PubMed]
12. Padmanabhan B, Tong KI, Ohta T, Nakamura Y, Scharlock M, et al. Structural basis for defects of Keap1 activity provoked by its point mutations in lung cancer. Mol Cell. 2006;21:689–700. [PubMed]
13. Singh A, Misra V, Thimmulappa RK, Lee H, Ames S, et al. Dysfunctional KEAP1-NRF2 interaction in non-small-cell lung cancer. PLoS Med. 2006;3:e420. [PMC free article] [PubMed]
14. Wang R, An J, Ji F, Jiao H, Sun H, et al. Hypermethylation of the Keap1 gene in human lung cancer cell lines and lung cancer tissues. Biochem Biophys Res Commun. 2008;373:151–154. [PubMed]
15. Homma S, Ishii Y, Morishima Y, Yamadori T, Matsuno Y, et al. Nrf2 enhances cell proliferation and resistance to anticancer drugs in human lung cancer. Clin Cancer Res. 2009;15:3423–3432. [PubMed]
16. Schultz MJ, Dixon B, Levi M. The pulmonary protein C system: preventive or therapeutic target in acute lung injury? Thorax. 2009;64:95–97. [PubMed]
17. Selvaraj P, Venkataprasad N, Vijayan VK, Prabhakar R, Narayanan PR. Procoagulant activity of bronchoalveolar lavage fluids taken from the site of tuberculous lesions. Eur Respir J. 1994;7:1227–1232. [PubMed]
18. Moldovan GL, Pfander B, Jentsch S. PCNA, the maestro of the replication fork. Cell. 2007;129:665–679. [PubMed]
19. Becks L, Prince M, Burson H, Christophe C, Broadway M, et al. Aggressive mammary carcinoma progression in Nrf2 knockout mice treated with 7,12-dimethylbenz[a]anthracene. BMC Cancer. 2010;10:540. [PMC free article] [PubMed]
20. Iida K, Itoh K, Kumagai Y, Oyasu R, Hattori K, et al. Nrf2 is essential for the chemopreventive efficacy of oltipraz against urinary bladder carcinogenesis. Cancer Res. 2004;64:6424–6431. [PubMed]
21. Khor TO, Huang MT, Prawan A, Liu Y, Hao X, et al. Increased susceptibility of Nrf2 knockout mice to colitis-associated colorectal cancer. Cancer Prev Res (Phila) 2008;1:187–191. [PMC free article] [PubMed]
22. Kim YR, Oh JE, Kim MS, Kang MR, Park SW, et al. Oncogenic NRF2 mutations in squamous cell carcinomas of oesophagus and skin. J Pathol. 2010;220:446–451. [PubMed]
23. Osburn WO, Karim B, Dolan PM, Liu G, Yamamoto M, et al. Increased colonic inflammatory injury and formation of aberrant crypt foci in Nrf2-deficient mice upon dextran sulfate treatment. Int J Cancer. 2007;121:1883–1891. [PubMed]
24. Ramos-Gomez M, Kwak MK, Dolan PM, Itoh K, Yamamoto M, et al. Sensitivity to carcinogenesis is increased and chemoprotective efficacy of enzyme inducers is lost in nrf2 transcription factor-deficient mice. Proc Natl Acad Sci U S A. 2001;98:3410–3415. [PubMed]
25. Shibata T, Ohta T, Tong KI, Kokubu A, Odogawa R, et al. Cancer related mutations in NRF2 impair its recognition by Keap1-Cul3 E3 ligase and promote malignancy. Proc Natl Acad Sci U S A. 2008;105:13568–13573. [PubMed]
26. Stacy DR, Ely K, Massion PP, Yarbrough WG, Hallahan DE, et al. Increased expression of nuclear factor E2 p45-related factor 2 (NRF2) in head and neck squamous cell carcinomas. Head Neck. 2006;28:813–818. [PubMed]
27. Xu C, Huang MT, Shen G, Yuan X, Lin W, et al. Inhibition of 7,12-dimethylbenz(a)anthracene-induced skin tumorigenesis in C57BL/6 mice by sulforaphane is mediated by nuclear factor E2-related factor 2. Cancer Res. 2006;66:8293–8296. [PubMed]
28. Zhang P, Singh A, Yegnasubramanian S, Esopi D, Kombairaju P, et al. Loss of Kelch-like ECH-associated protein 1 function in prostate cancer cells causes chemoresistance and radioresistance and promotes tumor growth. Mol Cancer Ther. 2010;9:336–346. [PMC free article] [PubMed]
29. Satoh H, Moriguchi T, Taguchi K, Takai J, Maher JM, et al. Nrf2-deficiency creates a responsive microenvironment for metastasis to the lung. Carcinogenesis. [PubMed]
30. Surh YJ, Lee HG, Li MH, Joung EJ, Na HK, et al. Antioxid Redox Signal; 2010. Nrf2-Mediated Heme Oxygenase-1 Upregulation as Adaptive Survival Response to Glucose Deprivation-Induced Apoptosis in HepG2 Cells. [PubMed]
31. Singh A, Bodas M, Wakabayashi N, Bunz F, Biswal S. Gain of Nrf2 function in Non-Small-Cell Lung Cancer Cells Confers Radioresistance. Antioxid Redox Signal. 2010;13:1627–1637. [PMC free article] [PubMed]
32. Redente EF, Orlicky DJ, Bouchard RJ, Malkinson AM. Tumor signaling to the bone marrow changes the phenotype of monocytes and pulmonary macrophages during urethane-induced primary lung tumorigenesis in A/J mice. Am J Pathol. 2007;170:693–708. [PubMed]
33. Keith RL, Miller YE, Hoshikawa Y, Moore MD, Gesell TL, et al. Manipulation of pulmonary prostacyclin synthase expression prevents murine lung cancer. Cancer Res. 2002;62:734–740. [PubMed]
34. Wei H, Frenkel K. In vivo formation of oxidized DNA bases in tumor promoter-treated mouse skin. Cancer Res. 1991;51:4443–4449. [PubMed]
35. Stearman RS, Dwyer-Nield L, Zerbe L, Blaine SA, Chan Z, et al. Analysis of orthologous gene expression between human pulmonary adenocarcinoma and a carcinogen-induced murine model. Am J Pathol. 2005;167:1763–1775. [PubMed]
36. Bauer AK, Fostel J, Degraff LM, Rondini EA, Walker C, et al. Transcriptomic analysis of pathways regulated by toll-like receptor 4 in a murine model of chronic pulmonary inflammation and carcinogenesis. Mol Cancer. 2009;8:107. [PMC free article] [PubMed]
37. Bauer AK, Dixon D, DeGraff LM, Cho HY, Walker CR, et al. Toll-like receptor 4 in butylated hydroxytoluene-induced mouse pulmonary inflammation and tumorigenesis. J Natl Cancer Inst. 2005;97:1778–1781. [PubMed]
38. Cho HY, Gladwell W, Wang X, Chorley B, Bell D, et al. Nrf2-regulated PPAR{gamma} expression is critical to protection against acute lung injury in mice. Am J Respir Crit Care Med. 2010;182:170–182. [PMC free article] [PubMed]
39. Cho HY, Morgan DL, Bauer AK, Kleeberger SR. Signal transduction pathways of tumor necrosis factor–mediated lung injury induced by ozone in mice. Am J Respir Crit Care Med. 2007;175:829–839. [PMC free article] [PubMed]
40. Cho HY, Jedlicka AE, Reddy SP, Kensler TW, Yamamoto M, et al. Role of NRF2 in protection against hyperoxic lung injury in mice. Am J Respir Cell Mol Biol. 2002;26:175–182. [PubMed]

Articles from PLoS ONE are provided here courtesy of Public Library of Science