p52 Overexpression Increases Epithelial Apoptosis, Enhances Lung Injury, and Reduces Survival after Lipopolysaccharide Treatment

Nuclear factor κB regulates a number of key genes involved in cellular processes such as proliferation, apoptosis, and inflammation. The NF-κB transcription factor family contains five members (p65/RelA, p100/p52, p105/p50, RelB, and c-Rel). Traditionally, NF-κB signaling is associated with activation through the canonical or noncanonical signaling pathway. In the noncanonical pathway, a heterodimer consisting of p100 and, most commonly, RelB remains sequestered in the cytoplasm as a result of the IκB-like inhibitory C terminus of p100. Upon activation, p100 is phosphorylated and undergoes partial proteolytic processing to p52, enabling the p52-containing heterodimer to translocate into the nucleus. Although many studies have identified crucial roles for canonical NF-κB signaling in inflammatory diseases, metabolic disorders, and cancer, few have investigated the involvement of noncanonical NF-κB signaling in these contexts. Global knockout of Relb or Nfkb2 (the genes for RelB and p100/p52) causes defects in secondary lymphoid organ development and impaired immune responses (1–3). Therefore, noncanonical NF-κB signaling has primarily been studied in hematopoietic cells, where it is an important pathway for regulating chemokine genes required for normal lymphoid organ development (4, 5). However, little is known about the function of noncanonical NF-κB signaling in nonimmune cell types.

Acute respiratory distress syndrome (ARDS) is a life-threatening form of hypoxemic respiratory failure that results in substantial morbidity and mortality. ARDS is characterized by an influx of inflammatory cells, epithelial apoptosis, and vascular permeability. Intratracheal (IT) treatment of mice with Escherichia coli LPS is commonly used as a model of ARDS. We showed previously that NF-κB signaling in the lung epithelium regulates the inflammatory response after LPS stimulation (6), suggesting that epithelial NF-κB signaling is a critical component of ARDS pathogenesis. Although the role of the noncanonical NF-κB pathway in LPS-induced inflammation is unknown, studies with lung epithelial cells in vitro showed that LPS stimulation induces noncanonical NF-κB activation with slower and more protracted kinetics compared with canonical NF-κB activation and that noncanonical NF-κB signaling may be important for regulation of proinflammatory cytokines (7).

To study the effects of noncanonical NF-κB signaling in vivo, we generated a novel transgenic mouse model with inducible expression of p52 specifically in airway epithelial cells. We found that p52 overexpression alone did not affect inflammatory cell recruitment or signaling; however, p52 overexpression in conjunction with LPS stimulation led to increased mortality, exaggerated lung injury, and enhanced epithelial cell apoptosis. Together, these studies demonstrate a novel role for noncanonical NF-κB signaling in cell survival/apoptosis during cellular stress and implicate p52 as a factor affecting the severity of ARDS pathogenesis.

The p52 cFlag pcDNA3 plasmid containing C-terminal FLAG-tagged murine p52 was a gift from Stephen Smale (University of California, Los Angeles; Addgene plasmid #20019). This plasmid was digested first with HindIII followed by SacI digestion to obtain a fragment containing the FLAG-p52. The ends of this fragment were filled in before ligation into the EcoRV site of a modified pBluescript II SK expression vector, which contains a (tet-O)₇-CMV promoter consisting of seven copies of the tetracycline (tet) operator DNA-binding sequence linked to a minimal CMV promoter together with bovine growth hormone polyadenylation sequences to ensure transcript termination. The final plasmid [(tet-O)₇-FLAG-p52-BGH.poly(A)] was verified by sequencing. To prevent basal leakiness, a construct expressing a tet-controlled transcriptional silencer (tTS) under control of the Clara cell secretory protein (CCSP) promoter [CCSP-tTS-hGH.poly(A)] was also included (6, 8). The (tet-O)₇-FLAG-p52 microinjection fragment was excised from the plasmid by digesting with AscI, resulting in a 2.1-kb fragment. The 5-kb CCSP-tTs fragment was excised using XhoI and NotI. Both constructs were purified with GELase Agarose Gel-Digesting Preparation (Epicentre) following the manufacturer’s instructions. These constructs were coinjected at the Vanderbilt Transgenic Mouse/Embryonic Stem Cell Shared Resource to generate transgenic lines of FVB background mice with cointegration of the CCSP-tTS and the (tet-O)₇-FLAG-p52 transgenes. Genotyping of the founder animals was performed by Southern blot, and all subsequent genotyping was performed by PCR analysis. Primers used for PCR of the (tet-O)₇-p52-FLAG are as follows: F: 5′-GACGCCATCCACGCTGTTTTG-3′ and R: 5′-AGGATAGGTCTTCCGGCCCTT-3′. The product size is 322 bp. Primers used for PCR of the CCSP-tTS are as follows: F: 5′-GAGTTGGCAGCAGTTTCTCC-3′ and R: 5′-GAGCACAGCCACATCTTCAA-3′. The product size is 472 bp. Four founder lines of CCSP-tTS/(tet-O)₇-FLAG-p52 mice were mated with CCSP-reverse tet-transactivator homozygous mice (a gift from Dr. J.A. Whitsett, University of Cincinnati, Cincinnati, OH) (9) to obtain triple transgenic mice, which were designated CCSP-rtTA CCSP-tTS/(tet-O)₇-FLAG-p52 (CCSP-p52). CCSP-p52 mice generated from two separate founder lines of CCSP-tTS/(tet-O)₇-FLAG-p52 mice were used for these studies.

All animal care and experimental procedures were approved and conducted according to the guidelines of the Vanderbilt University Institutional Animal Care and Use Committee. For experiments, age- and sex-matched mice were administered 2 g/l doxycycline (dox) in 2% sucrose drinking water to activate transgene expression. Water was replaced twice weekly. Mice were sacrificed at the indicated time points, and genotype-negative littermates (designated wild-type [WT]) were used as controls. Lungs were lavaged as previously described (10), and the left lung was tied off and flash-frozen in liquid nitrogen. The right lung was perfused and fixed by inflating with 10% neutral-buffered formalin. Total and differential bronchoalveolar lavage fluid (BALF) cell counts were determined as previously described (11).

To establish transgene expression, mice were placed on dox for 1 wk prior to IT LPS, bleomycin, or RelB adenovirus administration, and they remained on dox until sacrifice. For IT treatments, mice were anesthetized with isoflurane. E. coli LPS (serotype 055:B5; Sigma-Aldrich) was diluted in sterile PBS and delivered IT at a dose of 3 μg/g body weight. Bleomycin (0.08 U) diluted in sterile saline was administered IT. A total of 5 × 10⁸ PFU RelB-His adenovirus containing murine RelB with a His tag or control luciferase adenovirus (Ad-Luc; a gift from Dr. A. Powers, Vanderbilt University) was delivered IT. Inflammatory cell recruitment was assessed 96 h after adenoviral administration. For experiments with LPS stimulation after adenovirus administration, LPS was given IT 96 h after adenoviral delivery.

H&E staining was performed on 5-μm lung sections to assess lung histology. A pathologist scored lung fibrosis on H&E-stained sections, as previously described, using a scale of 0 to 4 (0 = normal lung architecture; 1 = increased thickness of ≤50% of interalveolar septa; 2 = thickening of >50% of interalveolar septa without fibrotic foci formation; 3 = thickening of the interalveolar septa with isolated fibrotic foci formation; 4 = formation of multiple fibrotic foci with distortion of parenchymal architecture) (12).

To evaluate transgene expression in CCSP-p52 mice, 5-μm lung sections were stained with an anti-FLAG Ab (600-403-383; Rockland). For TUNEL immunofluorescence staining, lung sections were stained using the fluorescein In Situ Cell Death Detection Kit (Roche), and TUNEL⁺ cells were counted in 15 60× fields using fluorescent confocal microscopy. Mean scores were calculated for each animal. For TUNEL coimmunofluorescence staining with CCSP or surfactant protein C (SPC), lung sections were stained with anti-CCSP (S-20; Santa Cruz) or anti-SPC Ab (Millipore), followed by the TUNEL staining protocol. SPC and TUNEL double-positive cells were enumerated in 10 20× fields, and total CCSP and TUNEL double-positive cells were counted on each lung section using fluorescent confocal microscopy. To assess nuclear p52 in human lungs, immunostaining using an anti-p52 Ab (C-5; Santa Cruz) was performed on normal lung sections from four life-long nonsmokers and on lung sections from four patients with ARDS. ARDS lung sections were a generous contribution from Dr. Lorraine Ware (Vanderbilt University). All patients had no known chronic lung disease.

Rat type II alveolar epithelial cell line RLE-6TN (American Type Culture Collection) was maintained at 37°C 5% CO₂ in DMEM (Invitrogen) with 4.5 g/l glucose and 2 mM l-glutamine, supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin. p52 expression construct p52 cFLAG pcDNA3 and control empty vector cFLAG pcDNA3 were gifts from Stephen Smale (Addgene plasmids #20019 and #20011). RLE-6TN parental cells (RLE-WT) were stably transfected with p52 cFLAG pcDNA3 (RLE-p52) or empty vector cFLAG pcDNA3 (RLE-EV) using Effectene transfection reagents (QIAGEN) according to the manufacturer’s protocol, and G418-selected clones were pooled. To confirm transgene expression, NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific) were used to prepare nuclear protein from RLE-EV and RLE-p52 transfected cells that had undergone selection; a Western blot was performed and probed with Abs, as described for Western blot analysis below.

Viability of RLE-WT, RLE-EV, and RLE-p52 cells was assessed using the CellTiter-Glo Luminescent Cell Viability assay (Promega), according to the manufacturer’s protocol. For serum-starvation experiments, an equal number of RLE-WT, RLE-EV, and RLE-p52 cells were plated in 96-well plates in 10% serum medium, as described above. Eight hours later, when cells had adhered to the wells, cells were switched to medium lacking serum, and cell viability was assessed after 12, 18, and 24 h in medium without serum. Percentage survival was determined by normalizing no-serum viability measurements to viability measurements in 10% serum medium for each cell line at each time point. For H₂O₂ treatment experiments, an equal number of RLE-EV and RLE-p52 cells was plated in 96-well plates. After overnight incubation, cells were treated with 100 μM H₂O₂, and cell viability was assessed at 24 h. Percentage survival was determined by normalizing H₂O₂-treated viability measurements to viability measurements in control (untreated) cells.

Nuclear protein was prepared from lung tissue and RLE-6TN cells using the NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific), separated by SDS-PAGE, transferred to nitrocellulose membranes, and probed using the following Abs: FLAG (F3165; Sigma), p100/p52 (4882; Cell Signaling), p65 (C-20; Santa Cruz), p50 (C-19; Santa Cruz), RelB (C-19; Santa Cruz), cRel (C; Santa Cruz), and TATA-binding protein (TBP) (N-12; Santa Cruz). Whole-cell lysates were prepared from RLE-6TN cells using RIPA buffer (Invitrogen) and probed with Abs for caspase-3 (9662; Cell Signaling) and actin (AC-74; Sigma). Immunodetection was performed using the corresponding Alexa Fluor–conjugated Abs and the Odyssey Infrared Imaging System (LI-COR Biosciences). All images were converted to grayscale.

Total mRNA was isolated using the RNeasy Mini kit (QIAGEN), according to the manufacturer’s instructions. Quantitative real-time PCR was performed using SYBR Green PCR Master Mix (Applied Biosystems) and the following primer sets: Cxcl12 (F: 5′-CCCATTCTCCTCATCCTCAT-3′; R: 5′-ACTCTGCTCTGGTGGAAGGT-3′) (13), Cxcl13 (F: 5′-AACTCCACCTCCAGGCAGAATG-3′; R: 5′-TGTGTAATGGGCTTCCAGAATACC-3′) (14), Ccl19 (F: 5′-GGCCTGCCTCAGATTATCTGCCAT-3′; R: 5′-GGAAGGCTTTCACGATGTTCC-3′), Ccl21 (F: 5′-GGACCCAAGGCAGTGATGGAG-3′; R: 5′-CTTCCTCAGGGTTTGCACATAG-3′) (15), RelB (F: 5′-GGGCATCCAGTGTGTTAGGAAGAA-3′; R: 5′-GGAAGCAGATCCTGACGACATTCA-3′), and GAPDH (F: 5′-TGAGGACCAGGTTGTCTCCT-3′; R: 5′-CCCTGTTGCTGTAGCCGTAT-3′). Expression values were normalized to GAPDH using the ΔCT method.

For the quantitative real-time PCR array analysis, RLE-WT, RLE-EV, and RLE-p52 cells, plated in 10-cm plates, underwent serum starvation for 18 h, as described for cell-viability measurements. RNA was isolated from cells using the RNeasy Mini Kit (QIAGEN), according to the manufacturer’s protocol. RNA was converted to cDNA using the RT² First Strand Synthesis Kit (QIAGEN). Apoptosis genes were assessed using the Rat Apoptosis RT² Profiler PCR array (QIAGEN), according to the manufacturer’s instructions. Four biological replicates were run for each cell type, and threshold cycle values > 32 were discarded from the analysis. Expression values for each sample were normalized to the mean of five measured housekeeping genes using the ΔCT method.

Levels of IFN-γ, IL-1β, MCP-1, G-CSF, IL-6, KC, and MIP-1α were measured in BALF and lung lysates as part of a multiplex mouse cytokine magnetic bead array panel (EMD Millipore, Darmstadt, Germany), using a Luminex 100 analyzer. Assays were performed according to the manufacturer’s instructions with assistance from the Vanderbilt University Medical Center Hormone Assay and Analytical Services Core. Other analytes in the panel that were below the limit of detection included GM-CSF, IL-4, IL-10, and IL-12p40.

Lung nuclear protein was isolated as described for Western blot analysis, and p52 was detected using the TransAM NF-κB ELISA kit (Active Motif). Albumin was measured in cell-free BALF supernatants using mouse albumin ELISA (GenWay Biotech). G-CSF, Mip-1α, KC (R&D Systems), MCP-1, IL-1β, and IFN-γ (BioLegend) were measured in whole-lung lysates. All ELISAs were performed according to the manufacturer’s instructions.

Data were analyzed using GraphPad Prism 5.0 software (GraphPad Software), and all values are presented as the mean ± SEM. Unpaired Student t tests were performed for comparisons between two groups. To analyze differences among more than two groups, one-way ANOVA, followed by a Tukey's posttest, was used. For the Kaplan–Meier survival analysis, a log-rank test was used. The p values ≤0.05 were considered statistically significant.

To investigate whether noncanonical NF-κB signaling is activated in the lungs of ARDS patients, we performed p100/p52 immunostaining on lung sections from patients with ARDS and control subjects. Faint cytoplasmic staining, but no nuclear staining, was observed in airways from normal lungs (Fig. 1). In contrast, ARDS airways and parenchyma had a marked increase in nuclear and cytoplasmic staining (Fig. 1). This finding of increased nuclear p52 staining in airways and lung parenchyma of ARDS patients suggests that p52 activation could influence ARDS pathogenesis.

To investigate the role of p52 activation in ARDS, we developed a novel mouse model of p52 expression in the airway epithelium using the tet-on system, placing a C-terminal FLAG-tagged p52 under the control of a (tet-O)₇-CMV promoter. Four founder lines were generated and mated to transgenic mice expressing reverse tet-transactivator under the control of the airway epithelial–specific CCSP promoter, generating mice designated CCSP-p52.

To test the induction of p52 expression, CCSP-p52 mice were administered dox in drinking water (2 g/l) for 1 wk. High levels of FLAG-p52 expression were detected in two lines of CCSP-p52 mice by Western blotting of lung nuclear protein (Fig. 2A), demonstrating successful induction of FLAG-p52 expression and translocation of transgenic p52 into the nucleus. FLAG-p52 expression was also detectable in CCSP-p52 mice administered dox for 1 mo (data not shown). No leaky FLAG-p52 expression was detected in other organs after dox treatment or in the lungs in the absence of dox (data not shown). FLAG immunostaining demonstrated FLAG-p52 expression in airway epithelium (Fig. 2B). Together, these data show that CCSP-p52 mice administered dox express p52 in airway epithelium and that transgene expression results in nuclear p52 accumulation.

We investigated whether p52 overexpression in vivo results in an inflammatory phenotype similar to canonical NF-κB pathway activation (6). However, no significant difference in inflammatory cells was observed in lungs of CCSP-p52 mice compared with WT mice, and mice exhibited normal lung histology after dox treatment for 1 wk or 1 mo (Fig. 2D, 2E). We also measured activation of other NF-κB family members by Western blot using lung nuclear protein extracts from mice on dox for 1 wk. As shown in Fig. 2C, no difference was found in the activation of other NF-κB family members. These data suggest that p52 overexpression in epithelial cells alone does not alter inflammatory signaling or inflammatory cell recruitment.

In contrast to canonical NF-κB signaling, which is increased in the lungs within the first 24 h following IT injection of E. coli LPS (16), we found that treatment of WT mice with IT LPS (3 μg/g body weight) led to increased nuclear p52 in the lungs that peaked at 48 h (Supplemental Fig. 1). To investigate the function of p52 in the context of LPS stimulation in vivo, IT LPS was administered to dox-treated CCSP-p52 and WT mice. A significant increase in mortality was observed in CCSP-p52 mice compared with WT mice beginning 72 h after LPS (Fig. 3A). We measured albumin in BALF as a marker of lung injury and found that albumin levels were significantly higher in BALF from CCSP-p52 mice 48 h after LPS administration (Fig. 3B). Total inflammatory cell numbers in BALF were also increased in CCSP-p52 mice compared with WT mice at 48 h after LPS (Fig. 3C), and this inflammatory infiltrate was characterized by increased numbers of macrophages at 48 and 72 h and neutrophils at 48 h (Fig. 3D–F). Nuclear p65, p50, RelB, and c-Rel were assessed by Western blotting of lung nuclear protein from mice 24, 48, and 72 h after LPS to determine whether altered activation of other NF-κB family members was associated with the enhanced inflammatory influx in CCSP-p52 mice. Compared with WT lungs, no differences in activation of other NF-κB family members in CCSP-p52 lungs were observed at any time point (Fig. 3G, 3H, data not shown).

Because we observed increased inflammatory cells in BALF from CCSP-p52 mice following LPS treatment, we expected to find increased levels of cytokines/chemokines in the lungs. Therefore, we measured a panel of mediators in BALF at 24 h after LPS treatment and in lung lysates at 24, 48, and 72 h after LPS. Surprisingly, no differences were found in the levels of G-CSF, IL-1β, MCP-1, MIP-1α, IFN-γ, IL-6, or KC between CCSP-p52 and WT mice at any time point (Supplemental Fig. 2A–D). We also measured expression of previously described noncanonical NF-κB signaling targets Cxcl12, Cxcl13, Ccl19, and Ccl21 (4) in lung lysates from CCSP-p52 and WT mice and found no differences between groups (Supplemental Fig. 2E). To determine whether inflammatory cells from CCSP-p52 mice produced more inflammatory cytokines, we isolated macrophage and neutrophil populations from the lungs of WT and CCSP-p52 mice 48 h after LPS and measured the expression of KC, TNF-α, IL-6, and IL-1β by quantitative real-time PCR. However, no increases in cytokine production by inflammatory cells from CCSP-p52 mice were observed (data not shown). Together, these data show that overexpression of p52 in the airway epithelium does not increase cytokine expression in this model.

The p52 precursor p100 preferentially binds to RelB in the cytoplasm, inhibiting its nuclear translocation (17). When noncanonical signaling is activated, processing of p100 to p52 induces translocation of RelB/p52 heterodimers into the nucleus (17, 18). To investigate the effect of RelB overexpression in combination with p52 overexpression on inflammatory cell recruitment, dox-treated WT and CCSP-p52 mice were administered adenoviral RelB or control adenovirus (Ad-Luc) by IT injection. RelB overexpression in the lungs was confirmed by quantitative real-time PCR (Supplemental Fig. 3A), and inflammatory cell recruitment was assessed 96 h after adenoviral administration. RelB overexpression caused a significant increase in inflammatory cell recruitment, which was characterized by increased macrophages, lymphocytes, and neutrophils in WT and CCSP-p52 mice compared with mice treated with Ad-Luc (Supplemental Fig. 3B–E). However, in the context of LPS stimulation, RelB overexpression did not affect inflammatory cell influx into the lungs of CCSP-p52 mice after LPS treatment (Supplemental Fig. 3D, 3E). Together, these data indicate that RelB overexpression causes an inflammatory response independent of p52. However, RelB overexpression did not augment the effects of p52 overexpression in the context of LPS stimulation, indicating that p52 is the most important component of the noncanonical NF-κB signaling pathway responsible for altering the response to LPS.

To determine whether CCSP-p52 mice are more susceptible to lung injury, we tested their response to a different insult by administering IT bleomycin (0.08 U), a chemotherapeutic agent that causes lung inflammation, injury, and fibrosis in mice. Three weeks after bleomycin administration, no differences in BALF inflammatory cells or lung fibrosis were observed between dox-treated WT and CCSP-p52 mice (Supplemental Fig. 4), suggesting that p52-mediated exacerbation of lung injury may be dependent on the stimulus.

Because no differences in cytokine expression were observed between LPS-treated WT and CCSP-p52 mice, we wondered whether p52 overexpression could contribute to lung injury by altering epithelial cell survival. Therefore, we measured apoptosis by TUNEL staining in WT and CCSP-p52 lung sections at 24 and 48 h after LPS. Although no significant differences in apoptosis were observed at 24 h after LPS, a trend toward increased apoptosis in CCSP-p52 lungs was present at 48 h after LPS (Fig. 4A). By coimmunofluorescence staining of TUNEL and CCSP, we looked more specifically at apoptosis in the CCSP⁺ airway epithelial cells where p52 is expressed. At 48 h after LPS, a significant increase in the number of CCSP⁺TUNEL⁺ cells was detected in the airways of CCSP-p52 mice compared with WT mice (Fig. 4B, 4C). In the absence of LPS stimulation, no significant difference in apoptotic CCSP⁺ cells was observed in CCSP-p52 mice compared with WT mice (Fig. 4B). To determine whether apoptosis was restricted to the p52 transgene–expressing CCSP⁺ airway epithelial cells, we evaluated apoptosis of type II alveolar epithelial cells by coimmunofluorescence staining of TUNEL and the type II cell–specific marker SPC. We found no differences in SPC⁺TUNEL⁺ type II cells between WT and CCSP-p52 mice after LPS (Fig. 4D). Together, these data suggest that p52 directly regulates epithelial cell survival during conditions of inflammation or cellular stress.

To explore the mechanism by which p52 promotes apoptosis of epithelial cells, we generated lung epithelial cells with stable p52 overexpression using the RLE-6TN rat lung epithelial cell line (designated RLE-p52). p52 overexpression and nuclear localization in these cells were confirmed by Western blotting of nuclear protein (Fig. 5A). To investigate whether p52 overexpression resulted in differential apoptosis in RLE-6TN cells, we first tested inflammatory stimuli (LPS, TNF-α, and IL-1β); however, none of these treatments resulted in substantial apoptosis in RLE-EV or p52-overexpressing cells, indicating that these cells are resistant to apoptosis in response to inflammatory stimuli. Next, we evaluated the effect of reactive oxygen species stimulation by treating cells with H₂O₂ (100 μM). At 24 h after H₂O₂ treatment, RLE-p52 cells demonstrated a significant decrease in cell viability compared to RLE-EV cells (Fig. 5B). Increased apoptosis of RLE-p52 cells was supported by Western blotting for cleaved caspase-3 at 18 h of exposure to H₂O₂ (Fig. 5C).

In another model of acute cellular stress, serum starvation, we found that RLE-p52 cells had significantly reduced cell viability compared with RLE-EV cells and LE-WT cells (Fig. 5D). Similar to H₂O₂ treatment, we observed increased apoptosis of RLE-p52 cells by Western blotting for cleaved caspase-3 after 18 h of serum starvation (Fig. 5E). To identify specific mediators of apoptosis differentially upregulated in RLE-p52 cells, we measured apoptotic factors in RLE-p52, RLE-EV, and RLE-WT cells after 18 h of serum starvation using a quantitative real-time PCR apoptosis array. Compared with RLE-EV and RLE-WT cells, 10 proapoptotic factors were significantly upregulated in RLE-p52 cells, including Bcl10, Bid, Bcl2l11 (Bim), Bok, caspase-4, caspase-6, caspase-7, Fas, Gadd45α, and Pycard (Fig. 6), suggesting that p52 plays a role in regulating the expression of these genes in stressed cells and that these genes, in turn, promote apoptosis.

These studies describe the generation of a novel transgenic mouse model that enables expression of the NF-κB family member p52 in specific cell populations and the characterization of the model in airway epithelial cells. Using this model, we identified an important and unexpected role for p52 in the airway epithelium in the setting of acute lung injury. In CCSP-p52 mice, p52 overexpression did not induce inflammatory cell recruitment or activation of other NF-κB family members. However, in combination with LPS stimulation, p52 overexpression augmented epithelial apoptosis, enhanced lung inflammation without an increase in cytokine production, and enhanced lung injury and mortality. These effects appear to be driven directly by p52, because no changes in other NF-κB family members were observed after LPS stimulation, and addition of RelB did not augment the phenotype. In addition, apoptosis of p52-expressing cells was specific to settings of acute cellular stress, because apoptosis was not observed in CCSP-p52 mice without LPS treatment or in RLE-p52 cells under normal culture conditions. In patients with ARDS, increased nuclear p52 staining was observed in the airways and lung parenchyma, suggesting that p52 could impact lung injury. Taken together, our studies implicate p52 as a potential factor in determining ARDS severity through regulation of epithelial survival/apoptosis.

Our findings, in conjunction with studies published by other investigators, suggest that noncanonical and canonical NF-κB activation in the lung epithelium have different effects on inflammatory signaling and inflammatory cell recruitment. Overexpression of constitutively active IκB kinase IKKβ in the lung epithelium drives canonical NF-κB pathway activation, causing a profound inflammatory response and elevated chemokine production (6). In this study, we show that overexpression of the noncanonical pathway effector p52 does not cause inflammation on its own; however, in the setting of LPS treatment, it can enhance inflammatory cell influx into the lungs, although it is likely that the inflammatory cell infiltrate is an indirect effect of epithelial cell death and altered barrier function. Although other studies suggested that p52 cooperates in regulating inflammatory cytokines in airway epithelial cells (7), we observed no differences in inflammatory cytokine production. In contrast, RelB overexpression led to increased inflammatory cell recruitment, which was not enhanced by p52 overexpression. However, in the setting of LPS stimulation, we found that RelB overexpression did not augment lung inflammation and injury, indicating that p52 is the functional effector of the noncanonical NF-κB pathway in this model. In a different inflammatory context, RelB was shown to limit cigarette smoke–induced inflammatory cell recruitment and cytokine production (19). Collectively, these data imply that different NF-κB pathway components are uniquely involved in regulating lung epithelial inflammatory signaling, and their effects on inflammation may depend on the specific inflammatory signaling environment.

The discovery that p52 overexpression promotes apoptosis of airway epithelial cells after LPS stimulation was unexpected, because many studies have suggested that p52 is prosurvival/proliferative. In fibroblasts, p52 and RelB protect against reactive oxygen species–induced senescence by regulating CDK4 and CDK6 expression and antagonizing p53 function (20); in prostate cancer cells, p52 promotes proliferation through regulation of cyclin D1 (21). In vivo studies demonstrated that expression of p100 in the mammary gland leads to hyperplasia (22), and global p52 expression caused by deletion of the C-terminal inhibitory domain causes hyperplasia of the gastric epithelium (23). In our studies, we found that p52 promoted epithelial cell apoptosis only in the context of acute cellular stressors, including LPS treatment in vivo. In contrast, we observed no differences in bleomycin-induced fibrosis in CCSP-p52 mice. In this model, bleomycin causes direct DNA damage leading to cell-cycle arrest and apoptosis of airway and alveolar epithelial cells, followed by inflammation and fibrosis. Although it is possible that apoptosis of CCSP-expressing cells does not impact the degree of bleomycin-induced inflammation and fibrosis, the lack of phenotypic differences between CCSP-p52 mice and WT mice suggests that p52 may have a different functional role in this direct epithelial injury model compared with LPS treatment, which can induce epithelial cell death through extrinsic and intrinsic apoptosis pathways, as well as activation of effector cells, particularly neutrophils (24).

Interestingly, of the 10 proapoptotic genes that we identified with increased expression in p52-overexpressing cells during serum starvation, 9 are known to be regulated by NF-κB signaling (Bcl2l11, Casp4, Fas, and Pycard) or contain NF-κB consensus sequence-binding sites in their promoters (Bcl10, Bid, Bok, Casp7, and Gadd45α), indicating that p52 may preferentially bind these gene promoters under conditions of cellular stress. Structurally, p52 lacks a transactivation domain, suggesting that interactions with other NF-κB family members or transcriptional cofactors are necessary for regulation of target genes. In the setting of cellular stress, the pool of available NF-κB–binding partners may be altered or additional transcriptional cofactor(s) that cooperate with p52 may be activated, affecting p52 binding site preferences. Additionally, epigenetic changes may occur as a result of stress signals that alter accessibility of regulatory regions of proapoptotic genes. Future studies are necessary to identify factors that modulate the transcriptional outcome of p52 activation in different contexts.

Prior studies demonstrated that LPS stimulation causes apoptosis of airway epithelial cells in murine models (25, 26), and our studies suggest that p52 augments this apoptotic response. Although we cannot conclude that enhanced epithelial cell apoptosis is directly responsible for the increased mortality of CCSP-p52 mice following LPS treatment, epithelial cell apoptosis was shown to impact the degree of lung injury and survival of rodents after LPS administration (26). Similarly, epithelial cell apoptosis is a prominent feature in the lungs of humans with ARDS (27–31), and the degree of epithelial injury is an important factor predicting patient outcome (30, 32). The identification of increased nuclear p52 in lungs of ARDS patients, combined with the effects of p52 expression in the LPS model of ARDS/acute lung injury, indicates that noncanonical NF-κB pathway signaling could be an important factor in regulating epithelial cell survival/apoptosis, barrier integrity, and disease severity in patients with ARDS. Although more investigation is required to better understand the function of p52 in the setting of ARDS, our studies indicate that p52 interactions and/or target genes may serve as potential therapeutic targets for patients with ARDS.

We thank the Vanderbilt Transgenic Mouse/Embryonic Stem Cell Shared Resource, supported by National Institutes of Health Grants DK020593 and CA68485, for technical assistance in generating the CCSP-tTS (tet-O)₇-FLAG-p52 transgenic mice and the Vanderbilt University Medical Center Hormone Assay and Analytical Services Core, supported by National Institutes of Health Grants DK059637 and DK020593, for assistance in running and analyzing the multiplex cytokine bead array.

The authors have no financial conflicts of interest.