Tumor Necrosis Factor- a Inhibits Aquaporin 5 Expression in Mouse Lung Epithelial Cells*

Aquaporin 5 (AQP5), the major water channel expressed in alveolar, tracheal, and upper bronchial epithelium, is significantly down-regulated during pulmonary inflammation and edema. The mechanisms that underlie this decrease in AQP5 levels are therefore of considerable interest. Here we show that AQP5 expression in cultured lung epithelial cells is decreased 2-fold at the mRNA level and 10-fold at the protein level by the proinflammatory cytokine tumor necrosis factor a (TNF- a ). Treatment of murine lung epithelial cells (MLE-12) with TNF- a results in a concentration- and time-dependent decrease in AQP5 mRNA and protein expression. Activation of the p55 TNF- a receptor (TNFR1) with an agonist antibody is sufficient to cause decreased AQP5 expression, demonstrating that the TNF- a effect is mediated through TNFR1. Inhibition of nuclear factor k B (NF- k B) translocation to the nucleus blocks the effect of TNF- a on AQP5 expression, indicat-ing that activation of NF- k B is required, whereas inhibition of extracellular signal-regulated or p38 mitogen-activated protein kinases showed no effect. These data show that TNF- a decreases AQP5 mRNA and protein expression and that the molecular pathway for this effect involves TNFR1 and activated anti-rabbit via chemiluminescence with variable film exposures. For quantification, films of Western blots were a Hewlett-Packard and Photoshop. Scan- ning was performed using chemiluminescence exposures that gave control bands in the lower gray scale. The labeling density was quantified using ImageQuant software. Values for AQP5 were corrected by quantification of the b -actin values and were expressed as an AQP5/ b -actin ratio. Densitometry are reported as volume-integrated values and expressed in percentages compared with the mean values in con- trols (vehicle-treated) (100%). Statistical Analysis— Statistical analysis of AQP5/L32 density ratios for RNA expression and AQP5/ b -actin density ratios for protein expression were performed using unpaired Student’s t test with equal vari- ance. Results are expressed as

Aquaporins (AQPs) 1 are water channel proteins that function to increase plasma membrane water permeability in secretory and absorptive cells in response to osmotic gradients (1). Aquaporins are found in tissues where the rapid and regulated transport of fluid is necessary such as in the kidney, salivary glands, and lung. Deficiency of AQPs results in human diseases such as nephrogenic diabetes insipidus due to mutations in AQP2 and cataract formation from AQP0 mutations (2,3). AQP5 is a mammalian water channel expressed on alveolar type I and II cells, tracheal and bronchial epithelium in the lung, in salivary and lacrimal gland epithelia, and in corneal epithelium (4,5). 2 Mice that are deficient in AQP5 have decreased production of saliva as well as altered saliva composition (6). 3 In addition, AQP5 knockout mice have a 90% decrease in airspace-capillary water permeability (7). These studies demonstrate the importance of AQP5 under normal conditions in both the salivary gland and lung.
Several AQPs have recently been demonstrated to undergo complex regulation; for instance, AQP2 is regulated in response to vasopressin both at the transcriptional and post-translational levels as well as through shuttling of the protein to the membrane (8). In addition, multiple AQPs are regulated under pathophysiological conditions such as altered expression of AQP1, AQP2, AQP3, and AQP4 in the kidney in a number of water balance disorders (9). Recently, through intratracheal infection of mice with adenovirus (10, 11), Towne et al. (12) showed that AQP5 mRNA and protein expression are decreased in a mouse model of pulmonary inflammation and edema. The decreased expression of AQP5 was found uniformly throughout the lung and was not restricted to regions of overt inflammation, suggesting local effects of a diffusible factor released in response to adenoviral infection. AQP5 was decreased both 7 and 14 days after adenoviral infection; however, the mechanism responsible for the regulation of AQP5 in inflammation is unknown.
Proinflammatory cytokines, such as tumor necrosis factor alpha (TNF-␣) and interferon gamma, are increased in expression shortly after infection with adenovirus (10,12) and may be potential mediators of the decrease in AQP5 expression. TNF-␣ is increased in mouse lungs as early as 6 h postinfection and remains elevated both 7 and 14 days after infection (12,13). The possibility that cytokines may be responsible for the decrease of AQP5 in pulmonary inflammation was therefore hypothesized. TNF-␣ has been implicated in many biological conditions, most notably autoimmune diseases (such as rheumatoid arthritis and inflammatory bowel disease), asthma, septic shock, and human immunodeficiency virus infection (14). Extensive studies show that TNF-␣ plays a pivotal role in inflammation including modulating the expression of many genes such as other proinflammatory cytokines, prostaglandins, major histocompatibility complex antigens, oncogenes, and transcription factors (14,15). The ability of TNF-␣ to induce such a wide variety of effects is likely because of its ability to activate multiple signal transduction pathways including mitogen-activated protein (MAP) kinases and nuclear factor B (NF-B) (16).
This study was therefore designed to assess directly the potential effects of TNF-␣ on AQP5 expression and to identify the signal transduction pathways mediating the response. Both AQP5 mRNA and protein expression are decreased in cultured murine lung epithelial cells in a time-and dose-dependent manner in response to TNF-␣ acting through the p55 type 1 TNF-␣ receptor 1 (TNFR1). This decrease in expression requires the nuclear translocation of NF-B. To our knowledge these studies provide the first example of regulation of an aquaporin by TNF-␣ and also provide the first evidence that aquaporin expression can be regulated by NF-B.

EXPERIMENTAL PROCEDURES
Experimental Reagents-Murine TNF-␣ (mTNF-␣) and human TNF-␣ (hTNF-␣) were obtained from Roche Biochemicals (Indianapolis). Monoclonal hamster anti-mouse TNFR1 (55R539) antibody was purchased from R & D Systems (Minneapolis). PD98059 was obtained from New England BioLabs (Beverly, MA) and was dissolved in dimethyl sulfoxide. U0126, SB205380, curcumin, MG-132, pyrrolidinedithiocarbonate (PDTC), and NF-B SN50 inhibitor peptide were purchased from Calbiochem. U0126, SB203580, curcumin, and MG-132 were dissolved in dimethyl sulfoxide. PDTC and SN50 were both dissolved in RPMI medium with 1% fetal bovine serum. Quercetin and dicoumarol were purchased from Sigma. Quercetin was dissolved in dimethyl sulfoxide. Dicoumarol was dissolved in ethanol. Electrophoresis reagents were from Bio-Rad. Reagents for enhanced chemiluminescence (SuperSignal) and the BCA protein assay kit were from Pierce. The antibody to ␤-actin was an anti-mouse monoclonal antibody purchased from Sigma. The rabbit, anti-mouse AQP5 antibody (LL639; Lofstrand Laboratories, Gaithersburg, MD) was generated against a synthetic peptide corresponding to the mouse AQP5 carboxyl terminus and affinity purified on a SulfoLink column (Pierce) conjugated with the immunizing peptide (17). The horseradish peroxidase-labeled antimouse and anti-rabbit IgG secondary antibodies were from Roche.
Cell Culture and Drug Treatments-Murine lung epithelial cells (MLE-12) were a gift from Dr. Jeffrey Whitsett (Children's Hospital Medical Center, Cincinnati, OH) (18). MLE-12 cells were propagated at 37°C with 5% CO 2 in RPMI 1640 medium (Life Technologies, Inc.) supplemented with L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 3% fetal bovine serum (Life Technologies, Inc.). Approximately 48 h before study, cells were seeded onto six-well tissue culture dishes at 5 ϫ 10 5 cells/well and were serum starved 24 h before study by replacing the medium with RPMI medium containing 1% fetal bovine serum. In the MAP kinase and NF-B inhibition studies, inhibitors were added in the specified concentrations for 1 h before the addition of 100 units/ml (1.5 ng/ml) mTNF-␣ as determined by a doseresponse curve (see Fig. 1). Cells were then maintained in the medium for the duration of the experiment. Each experiment was replicated in its entirety at least twice with at least three independent wells/experimental group.
Cell Viability-Effects of TNF-␣ and the inhibitors PD98059, SB203580, and SN50 on cell viability were measured with a standard trypan blue uptake assay after treatment with 100 units/ml mTNF-␣ for 8 and 24 h. Cell cultures were also examined morphologically via light microscopy.
Northern Analysis-Following the specified treatment, medium was aspirated, MLE cells were washed with phosphate-buffered saline, and 1 ml of TriReagent (Molecular Research Center Inc., Cincinnati, OH) was added for the isolation of total RNA as per the manufacturer's instructions. RNA was solubilized in Formazol (Molecular Research Center, Inc.), and RNA concentrations were determined via spectrophotometry and confirmed with agarose gel electrophoresis. Total RNA, 5 g/sample, was size fractionated by gel electrophoresis as described (12). The AQP5 cDNA probe was generated as described previously (12).
As controls for loading of total RNA, the 28 S and 18 S ribosomal RNA bands were examined with ultraviolet exposure of the ethidium bromide-stained gel, and subsequent to probing with AQP5, blots were stripped in boiling 0.5% SDS for 30 min and reprobed with a mouse L32 ribosomal protein mRNA probe as described (12,19). Northern blots probed with either the AQP5 or L32 were quantified by exposure of a phosphor screen, scanned by means of a Storm 840 scanner, and analyzed using ImageQuant software (all from Molecular Dynamics, Sunnyvale, CA). RNA values are reported as the AQP5/L32 ratio for each sample. Phosphorimaging results are reported as volume-integrated values and are expressed in percentages compared with the mean values in controls (vehicle-treated) (100%).
Western Analysis-Following treatment of cells in the specified medium for the indicated amount of time, the medium was aspirated, and the cells were washed with ice-cold phosphate-buffered saline. Ice-cold isolation solution (250 mM sucrose, 10 mM triethanolamine, 1 g/ml leupeptin, and 0.1 mg/ml phenylmethylsulfonyl fluoride, adjusted to pH 7.6) was added to each well for scraping and collecting of the cells. Cells were then lysed via three successive freezing and thawing cycles in dry ice and a 37°C water bath, respectively. Total protein content was determined by the BCA assay with bovine serum albumin as the standard. Cell homogenates, 5 g/sample, were solubilized in Laemmli sample buffer and boiled for 5 min. SDS-polyacrylamide gel electrophoresis and Western blotting were carried out as described previously (12). Membranes were incubated overnight at 4°C with anti-AQP5 antibody at a dilution of 0.5 g/ml and anti-␤-actin antibody at a dilution of 1:50,000 in 0.5% blocking solution (Roche). After washing, the membranes were incubated with 100 milliunits/ml horseradish peroxidaselabeled anti-rabbit secondary antibody and 50 milliunits/ml peroxidaselabeled anti-mouse secondary antibody for 1 h at room temperature, washed again, and visualized via enhanced chemiluminescence with variable film exposures. For quantification, films of Western blots were scanned using a Hewlett-Packard scanner and Adobe Photoshop. Scanning was performed using chemiluminescence exposures that gave control bands in the lower gray scale. The labeling density was quantified using ImageQuant software. Values for AQP5 were corrected by quantification of the ␤-actin values and were expressed as an AQP5/␤-actin ratio. Densitometry results are reported as volume-integrated values and expressed in percentages compared with the mean values in controls (vehicle-treated) (100%).
Statistical Analysis-Statistical analysis of AQP5/L32 density ratios for RNA expression and AQP5/␤-actin density ratios for protein expression were performed using unpaired Student's t test with equal variance. Results are expressed as means Ϯ S.E. A p value of Ͻ0.05 was considered statistically significant.

TNF-␣ Reduces AQP5 mRNA and Protein Expression in MLE-12
Cells-MLE-12 cells were utilized to examine the effects of TNF-␣ on AQP5 mRNA and protein expression. MLE-12 cells were treated with various concentrations of mTNF-␣ for various time points to determine whether the response to TNF-␣ was dose-and time-dependent. Cells were incubated in media supplemented with 10, 50, 100, 500, or 1,000 units/ml mTNF-␣ for 8 h before isolation of total RNA for Northern blot analysis. AQP5 mRNA levels were decreased significantly with 50, 100, and 500 units/ml mTNF-␣ treatment (Fig. 1, A and B). AQP5 protein levels were decreased significantly after 24 h of treatment with 50, 100, 500, and 1,000 units/ml mTNF-␣ ( Fig. 1, C and D) in a dose-dependent manner. AQP5 mRNA and protein were decreased maximally after treatment with 100 units/ml mTNF-␣, therefore this concentration was used for subsequent experiments.
To analyze the time course of TNF-␣-mediated inhibition of AQP5 expression, MLE-12 cells were treated with mTNF-␣ for various time points before isolation of total RNA. AQP5 mRNA was decreased significantly to about 50% of control (medium alone) levels after 4, 8, and 24 h of treatment with mTNF-␣ (Fig. 2, A and B). However, after 48 h of treatment with TNF-␣, AQP5 mRNA returned to control levels (Fig. 2, A and B). AQP5 protein was decreased dramatically to about 10% of control levels after 24, 48, and 72 h of treatment with mTNF-␣ (Fig. 2, C and D). Therefore, although AQP5 mRNA returns to base- The L32 ribosomal protein mRNA probe was used to control for equal loading. B, Northern blots were quantified by phosphorimaging, and the AQP5/L32 ratio was calculated for each sample. Data are plotted as a percentage of control, vehicle-treated, samples (n ϭ 6, mean Ϯ S.E.; * ϭ p Ͻ 0.01, ** ϭ p Ͻ 0.05 versus control). C, Western blot analysis of cell homogenates (5 g/lane) isolated after treatment as in A for 24 h. Immunoblotting was performed with an affinity-purified anti-AQP5 antibody, and a ␤-actin-specific antibody was utilized to control for equal loading. D, immunoblots were quantified by densitometry and expressed as the AQP5/␤-actin ratio for each sample. Data are plotted as a percentage of control (n ϭ 6, mean Ϯ S.E.; * ϭ p Ͻ 0.01, ** ϭ p Ͻ 0.05 versus control). The L32 ribosomal protein mRNA probe was used as a loading control. B, quantification of AQP5 and L32 signals by phosphorimaging is expressed as the AQP5/L32 ratio for each sample and is plotted as a percentage of control, vehicle alone, treated samples (n ϭ 6, mean Ϯ S.E., * ϭ p Ͻ 0.01, ** ϭ p Ͻ 0.05 versus control). C, MLE-12 cells were treated with 100 units/ml mTNF-␣ for 24, 48, or 72 h before the isolation of cell homogenates. Cell homogenates (5 g/lane) were subjected to Western blot analysis with AQP5 and ␤-actin-specific antibodies. D, signals on immunoblots were quantified using densitometry and expressed as the AQP5/␤-actin ratio for each sample. Data are plotted as a percentage of control, vehicle alone, treated samples (n ϭ 6, mean Ϯ S.E., * ϭ p Ͻ 0.05 compared with control). line levels by 48 h of treatment with mTNF-␣, AQP5 protein levels do not return to base line even by 72 h, suggesting that the regulation of this gene is accomplished at both the transcriptional and the post-transcriptional levels.
The decrease in AQP5 mRNA and protein expression was not the result of apoptosis or the general inhibition of cell metabolism because the number of live cells (trypan blue exclusion) and the protein content after 24 and 48 h of incubation with 100 units/ml mTNF-␣ were the same in control and TNF-␣treated cells (data not shown). Examination of cells by light microscopy demonstrated that cells treated with mTNF-␣ remained adherent and morphologically similar to control cells. The quality and quantity of total RNA recovered and L32 mRNA contents visualized on the same blots as AQP5 mRNA were not influenced by TNF-␣. These data demonstrate that AQP5 mRNA and protein are decreased in MLE-12 cells after treatment with TNF-␣ in a time-and dose-dependent manner.
Inhibition of AQP5 Expression Is Signaled through the p55 TNF-␣ Receptor-The first step in TNF action is binding to specific receptors that are expressed on the plasma membrane of virtually all cells except erythrocytes (20,21). Two distinct receptors for TNF-␣ have been identified, the p55 type 1 receptor (TNFR1) and the p75 type 2 receptor (TNFR2) (21,22). These receptors share structural homology in the extracellular TNF-␣ binding domains and exhibit similar binding affinities for TNF-␣, but they induce separate cytoplasmic signaling pathways after receptor-ligand binding (21,22). Mouse TNF-␣, which was utilized in all previous experiments, is capable of binding to and signaling through both TNF-␣ receptors on mouse cells (23). The monoclonal TNFR1 agonist antibody 55R539 specifically binds to and activates signaling through TNFR1 and does not cross-react with TNFR2 (22,24). To determine which TNF-␣ receptor signals the decrease in AQP5, we treated MLE-12 cells with the TNFR1 agonist antibody 55R539 and measured AQP5 mRNA and protein levels. MLE-12 cells were treated with the TNFR1 agonist antibody at 0.33, 1, or 3 g/ml or with mTNF-␣ for 8 h, and AQP5 mRNA expression was evaluated. Northern blot analysis demonstrated that activation of TNFR1 by treatment of MLE-12 cells with all three concentrations of the TNFR1 agonist antibody reduced AQP5 mRNA expression to an extent similar to that obtained by the simultaneous triggering of both receptors by mouse TNF-␣ (Fig. 3, A and B). In addition, human TNF-␣, which specifically binds and activates TNFR1 on murine cells (20,25), also promotes decreased AQP5 expression (data not shown). Treatment of MLE-12 cells with either mouse TNF-␣ or 1 or 3 g/ml agonist TNFR1 antibody for 24 h also resulted in a similar decrease in AQP5 protein (Fig. 3, C and D). Therefore, stimulation of TNFR1 alone decreases both AQP5 mRNA and protein expression to a level equal to that seen with mouse TNF-␣, suggesting that decreased AQP5 expression in response to TNF-␣ is mediated principally through TNFR1.
Decreased AQP5 Expression by TNF-␣ Does Not Require Activation of the ERK or p38 MAP Kinase Pathway-Signaling through TNFR1 leads to distinct effector functions, including MAP kinase activation and the activation of NF-B (14,26). TNF-␣ activates signaling through three MAP kinase pathways: extracellular signal-regulated kinase (ERK), Jun N-terminal kinase (JNK), and p38 (16,22), and all three pathways are activated by mTNF-␣ treatment of MLE-12 cells (27). To investigate the possible involvement of the ERK cascade in decreased AQP5 expression in response to TNF-␣, the effects of the MEK1/2 (the upstream kinase of ERK1/2) inhibitor PD98059 (28, 29) were examined. PD98059 inhibits MEK1/2 by blocking activation of MEK1/2 by Raf kinase (28). 20 M PD98059 was added to MLE cells for 1 h before the addition of mTNF-␣, and cells were collected for RNA isolation 8 h after the addition of TNF-␣. Northern blot analysis demonstrated that the addition of PD98059 alone did not alter AQP5 mRNA expression significantly. AQP5 mRNA expression was decreased to the same extent with the addition of PD98059 and TNF-␣ as with TNF-␣ alone (Fig. 4, A and B). Therefore, decreased AQP5 mRNA expression in response to TNF-␣ does not require signaling through ERK MAP kinase.
To examine the effect of ERK inhibition on AQP5 protein expression in response to TNF-␣, MLE-12 cells were treated with 20 M PD98059 for 1 h before the addition of TNF-␣, and cell homogenates were collected 24 h after the addition of TNF-␣. Western blot analysis demonstrated that AQP5 protein was unaltered with the addition of PD98059 alone; when PD98059 was incubated in combination with TNF-␣, AQP5 expression was decreased to the same extent as with TNF-␣ alone (Fig. 4, C and D). Together these data demonstrate that the MEK inhibitor PD98059 has no effect on either the decrease in AQP5 mRNA or protein expression seen in response to TNF-␣. Therefore, decreased AQP5 expression in response to TNF-␣ does not require activation of the ERK MAP kinase pathway.
Similar results were obtained with the use of SB203580, a drug that specifically inhibits p38 kinase activity (30,31). MLE-12 cells were treated with 10 M SB203580 for 1 h before the addition of TNF-␣. Cells were then isolated 8 and 24 h after the addition of TNF-␣ for RNA and protein isolation, respectively. SB203580 alone did not alter AQP5 mRNA or protein expression significantly (Fig. 5, A-D), and addition of the inhibitor plus TNF-␣ resulted in a decrease in AQP5 expression which was similar to that with TNF-␣ alone (Fig. 5, A-D). These data show that the p38 inhibitor SB203580 has no effect on the decrease in AQP5 mRNA and protein expression seen in response to TNF-␣, suggesting that decreased AQP5 expression in response to TNF-␣ does not require activation of the p38 MAP kinase pathway.
Several compounds have been demonstrated to inhibit activation of JNK by TNF-␣; however, these inhibitors are fairly nonspecific for JNK and affect other processes in the cell. Dicoumarol is a quinone reductase inhibitor (32) that has been demonstrated to inhibit JNK activation; however, use of this compound in combination with TNF-␣ resulted in rapid cell death, thus the effect of dicoumarol on AQP5 expression could not be assessed. Quercetin and curcumin are a plant flavanoid and pigment, respectively, and previously they have been demonstrated to inhibit JNK activation (33)(34)(35). Use of both inhibitors individually resulted in decreased AQP5 mRNA expression with inhibitor alone (data not shown). Expression was not decreased further with inhibitor plus TNF-␣. However, curcumin has been demonstrated to decrease NF-B activation as well as JNK activation (34), so the effects of this inhibitor could be the result of inhibition of either pathway. These results could suggest that JNK and/or NF-B activation is necessary for both basal and decreased AQP5 expression in response to TNF-␣.
Decreased AQP5 Expression in Response to TNF-␣ May Require the Nuclear Translocation of NF-B-NF-B is a ubiquitous transcription factor that is activated by proinflammatory cytokines such as TNF-␣ and interleukin-1 (14,26,36). In addition, studies have demonstrated that TNF-␣ activates NF-B by signaling through TNFR1 (36,37). NF-B is a homoor heterodimer of DNA binding subunits whose activity is regulated by the IB proteins. Under unstimulated conditions, NF-B dimers are retained in an inactive form in the cytoplasm through association with one of the IB proteins. Upon stimulation, IB molecules are rapidly phosphorylated and degraded  isolated from MLE-12 cells treated with 100 units/ml mTNF-␣ or 0.33, 1, or 3 g/ml monoclonal TNFR1 agonist antibody 55R539 for 8 h. B, quantification of AQP5 signals by phosphorimaging is expressed as the AQP5/L32 ratio for each sample. Data are plotted as a percentage of control, vehicle alone, treated samples (n ϭ 6, mean Ϯ S.E., * ϭ p Ͻ 0.01 compared with control). C, MLE-12 cells were treated with mTNF-␣ or 1 or 3 g/ml TNFR1 agonist antibody for 24 h before isolation of cell homogenates and subsequent Western blot analysis (5 g/lane). D, immunoblotting was quantified by densitometry and is expressed as the AQP5/␤-actin ratio for each sample. Data are plotted as a percentage of control, vehicle alone, treated samples (n ϭ 6, mean Ϯ S.E., * ϭ p Ͻ 0.01 compared with control). through a ubiquitin/proteasome pathway, thus unmasking a nuclear localization signal in NF-B. Free NF-B then translocates from the cytoplasm to the nucleus where it can regulate transcription of genes with a B site (16,36,37). To examine the possible involvement of NF-B activation in the TNF-␣mediated down-regulation of AQP5, several inhibitors of NF-B were utilized. Multiple compounds have been reported to inhibit NF-B activity through a variety of mechanisms, the majority of which are not specific for NF-B. Inhibitors of the proteasome, such as MG-132, inhibit activation of NF-B by blocking degradation of IB (38). PDTC inhibits NF-B by suppressing the release of the inhibitor subunit IB from the latent cytoplasmic form of NF-B (39) and thus blocks the activation of NF-B; however, PDTC is also a potent antioxidant and affects many other processes in the cell. Treatment of MLE-12 cells with 10 M MG-132 or 100 mM PDTC, concentrations that were shown previously to inhibit NF-B activity in response to TNF-␣ (38,39), resulted in decreased AQP5 expression with inhibitor alone (data not shown). No difference was seen between cells treated with inhibitor or inhibitor plus TNF-␣; however, AQP5 expression was already decreased to the same extent with inhibitor alone as with TNF-␣ alone (data not shown). Therefore, no direct conclusions can be drawn about the importance of NF-B signaling on AQP5 expression with the use of these inhibitors.
SN50 is a cell-permeable synthetic peptide that specifically competes with the nuclear localization sequence of the p50 subunit of NF-B (40). SN50 has been demonstrated to penetrate cells rapidly (within 15 min) and inhibit NF-B translocation to the nucleus, thereby inhibiting NF-B DNA binding (41). In addition, SN50 is specific for the nuclear localization signal of NF-B and therefore is a more specific inhibitor of NF-B activation. MLE-12 cells were treated with 50 g/ml SN50 for 1 h before the addition of TNF-␣ for 8 h and the subsequent isolation of RNA. Treatment with SN50 alone caused a modest decrease in AQP5 mRNA; however, pretreatment with SN50 prevented further inhibition of AQP5 mRNA expression by TNF-␣ (Fig. 6, A and B). AQP5 mRNA expression with inhibitor or inhibitor plus TNF-␣ was significantly greater than with TNF-␣ alone, showing that inhibition of NF-B blocked the decrease in AQP5 expression in response to TNF-␣. Similar results were obtained when cells were treated with SN50 for 1 h and then treated with TNF-␣ for 24 h followed by protein isolation and Western blot analysis. The decrease in AQP5 protein expression was inhibited by pretreatment with SN50 (Fig. 6, C and D). Therefore, NF-B translocation to the nucleus is likely necessary for decreased AQP5 expression in response to TNF-␣. DISCUSSION Pulmonary inflammation is characterized by increased cytokine expression, inflammatory cell infiltration, and excess fluid accumulation or pulmonary edema (42). Recently, aquaporins in the lung were demonstrated to be down-regulated in a mouse model of pulmonary inflammation (12). Through the use of knockout mice, AQP5 was shown to be required for the majority of water transport in the lung (7). Therefore, altered expression of AQP5 in inflammation may play a significant role in the edema seen in pulmonary infection. Although the regulation of AQP5 is of considerable interest, the mechanisms regulating AQP expression remain poorly understood. Here we demonstrate that AQP5 is down-regulated in a time-and dose-dependent manner by TNF-␣ signaling through TNFR1. Inhibition of the activation of the MAP kinases, ERK and p38, demonstrated activation of ERK, and p38 is not necessary for the effect of TNF-␣ on AQP5 expression. However, inhibition of the nuclear translocation of NF-B showed that the decrease of AQP5 mRNA and protein expression in response to TNF-␣ is dependent upon the activation of NF-B. To our knowledge, these studies provide the first example of down-regulation of an aquaporin by a proinflammatory cytokine as well as the first demonstration that aquaporin expression can be regulated by NF-B.
TNF-␣ is a pivotal mediator of inflammation and has been demonstrated to regulate numerous genes essential to the inflammatory process, including other cytokines and cell adhesion molecules (21,22,43). Pulmonary inflammation is accompanied by edema, and administration of TNF-␣ alone has been demonstrated to result in pulmonary edema (21). Given the connection among TNF-␣, inflammation, and edema, we explored whether AQP expression might also be regulated by TNF-␣. AQP5 mRNA and protein were reduced by TNF-␣ in MLE-12 cells. AQP5 mRNA was decreased maximally within 4 h of treatment with TNF-␣ and returned to normal levels by 48 h. AQP5 protein, on the other hand, was decreased to a much greater extent than AQP5 mRNA and did not return to normal levels by 72 h of TNF-␣ treatment. The expression of several proteins, such as CTP:phosphocholine cytidylyltransferase and IB, is decreased in response to TNF-␣ through increased protein degradation irrespective of changes in mRNA levels (36,44). TNF-␣ also inhibits skeletal muscle protein synthesis and thus affects translational efficiency in a number of genes (45). The observed inhibition of AQP5 protein expression is consistent with regulation at both pre-and post-translational levels. Further experiments are required to determine whether mechanisms that regulate protein level such as translational efficiency, protein degradation, or other intracellular signaling pathway(s) are involved in the TNF-␣-induced downregulation of AQP5 protein.
The biological actions of TNF-␣ are initiated by its binding to a 55-kDa receptor (TNFR1) and/or to a 75-kDa receptor (TNFR2) (22,43). Although these two receptors induce both distinct and overlapping responses, the majority of effects of TNF-␣ studied occur through signaling through TNFR1 (36,37). However, some responses initiated by TNF-␣ require the combined activation of both receptors or occur by signaling through TNFR2 alone (22,37). Through the use of a TNFR1 agonist antibody we demonstrated that activation of TNFR1 alone reduced AQP5 mRNA and protein expression to the same extent as with mouse TNF-␣, which signals through both receptors. Thus, selective signaling through TNFR1 results in the reduction of AQP5 gene expression. Our results do not preclude the possibility that signaling through TNFR2 might also decrease AQP5 expression but that signaling through this receptor is not required.
Signaling through TNFR1 leads to alterations in gene expression via activation of multiple signal transduction pathways including activation of the MAP kinase family, ERK1/2, p38, and JNK (22,26). Two principal transcription factors activated by TNF-␣ are NF-B and activating protein 1 (14,22). The AQP5 5Ј-flanking region contains consensus binding sites for both activating protein 1 and NF-B (17). It has been demonstrated recently that AQP5 expression is increased in MLE cells in response to hypertonic stress through an ERKdependent mechanism (29). In the present study, the decrease in AQP5 mRNA and protein expression in response to TNF-␣ was not affected by the MEK inhibitor PD98059, suggesting that signaling through ERK MAP kinase was not required for the effect. Therefore, although TNF-␣ activates ERK in MLE cells, AQP5 expression was decreased, and this response was independent of ERK activation. This result does not conflict with the previous study because although ERK activation was necessary for the hypertonic induction of AQP5, it was not sufficient as activation of ERK by TPA did not induce AQP5 FIG. 6. Decreased AQP5 mRNA and protein expression in response to TNF-␣ may require the nuclear translocation of NF-B. A, Northern blot analysis of 5 g/lane RNA isolated from MLE-12 cells treated with mTNF-␣, the NF-B-specific inhibitor, 50 g/ml SN50 alone, SN50 plus TNF-␣, or vehicle alone for 8 h. B, AQP5 mRNA levels were quantified by phosphorimaging and expressed as the AQP5/L32 ratio for each sample. Data are plotted as a percentage of control, vehicle alone, treated cells (n ϭ 6, mean Ϯ S.E., * ϭ p Ͻ 0.05 compared with control, § ϭ p Ͻ 0.01 compared with cells treated with TNF-␣ alone). C, cells were treated with vehicle alone, mTNF-␣, 50 g/ml SN50, or SN50 plus TNF-␣ for 24 h before the isolation of cell homogenates and subsequent Western blot analysis (5 g/lane) with AQP5 and ␤-actin-specific antibodies. D, densitometry of Western blot results expressed as the AQP5/␤-actin ratio for each sample. Data are plotted as a percentage of control, vehicle alone, treated cells (n ϭ 6, mean Ϯ S.E., * ϭ p Ͻ 0.05 compared with control, § ϭ p Ͻ 0.01 compared with cells treated with TNF-␣ alone). (29). Likewise, inhibition of the activation of p38 MAP kinase by SB203580 demonstrated that p38 does not likely have a role in the reduction of AQP5 in response to TNF-␣.
TNF-␣ has both stimulatory and inhibitory effects on gene expression, and its effects are often mediated through the activation of the transcription factor NF-B (26,36,43). This study provides the first link between aquaporin expression and NF-B activation. SN50, which specifically inhibits entry of NF-B into cell nuclei, inhibited the decrease of AQP5 mRNA and protein in response to TNF-␣, suggesting that the TNF-␣ effect requires the nuclear translocation of NF-B. The mechanism by which NF-B activation regulates AQP5 expression is not clear, but it could involve several mechanisms including direct interaction of NF-B with one of the putative B binding sequences present in the mouse AQP5 gene (17). Studies have shown that when NF-B binds to a consensus NF-B binding site as a heterodimer of p50/p65, it acts as a transcription activator; yet when the NF-B p50/p50 homodimer binds to promoters, it functions as a transcriptional repressor (46,47). In addition, both p50 and p65 subunits have been demonstrated to decrease expression of genes, such as the ␣1(I) collagen gene, through direct interaction with other transcription factors such as SP1 and CCAAT/enhancer-binding protein (48,49). Alternatively, NF-B could serve to activate a repressor that inhibits transcription of AQP5. Further studies are required to dissect the mechanism behind NF-B regulation of AQP5 expression.
Inflammatory lung diseases, such as chronic bronchitis, adult respiratory distress syndrome, cystic fibrosis, and asthma, are associated with elevated levels of TNF-␣ in lung fluids (43). Therapeutic approaches that inhibit the action of TNF-␣ are a focus of intense research and have recently been employed in the treatment of rheumatoid arthritis and inflammatory bowel disease (14,36). In this regard it is essential to understand the downstream effectors of TNF-␣ to predict the efficacy of therapy as well as potential side effects. TNF-mediated inhibition of AQP5 may help explain why pulmonary inflammation is accompanied by pulmonary edema and could implicate potential therapy aimed at TNF and/or AQP5.
In summary, TNF-␣ decreased the level of AQP5 mRNA and protein in MLE-12 cells. Decreased AQP5 expression in response to TNF-␣ occurs through activation of TNFR1 and does not require activation of ERK or p38 MAP kinases. However, translocation of NF-B to the nucleus is likely required for regulation of AQP5 by TNF-␣. This is the first report of a proinflammatory cytokine decreasing the expression of an aquaporin and provides information that may begin to explain the relationship between inflammation and edema in the lung.