INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Acute lung injury, associated with sepsis and other critical illness, substantially increases mortality rates of the underlying critical illness (1). Modulating the inflammatory response to diminish adverse consequences of acute lung injury may be of therapeutic benefit, but understanding is incomplete (4).

Inhibition of endogenous NO production (5, 6) or administration of exogenous NO (7, 8) are interventions that are currently being investigated in critically ill patients with sepsis and acute lung injury. This apparent contradiction between blocking endogenous NO in severe sepsis while administering exogenous NO in acute lung injury highlights our incomplete understanding of these interventions. Altering endogenous or exogenous NO may have unplanned or unanticipated effects because NO has a wide array of actions. For example, NO alters cytokine expression in some cell lines (9), inhaled NO alters bronchoalveolar lavage concentrations of proinflammatory cytokines in human adult respiratory distress syndrome (ARDS) (12), and NO can diminish lung injury (13). These observations suggest that NO may be an important immune modulator in the proinflammatory cytokine response associated with acute lung injury, although the mechanism is unknown. Whether reduction of endogenous NO production by NO synthase inhibition has opposite and adverse consequences on pulmonary inflammation is also not known.

Our goal was to determine whether inhibition of NO synthase alters proinflammatory cytokine expression in vivo during acute lung injury, and to understand better the mechanism of this effect. Accordingly, we studied acute lung injury induced by intratracheal endotoxin installation in mice. Mice were pretreated with either l-nitro-arginine methyl ester (l-NAME), an NO synthase inhibitor (decreases endogenous NO), or d-NAME as a control. We measured the effect of these interventions on tumor necrosis factor  (TNF-) and interleukin 6 (IL-6) expression. These were chosen as representative proinflammatory cytokines because of their prominent roles in the acute inflammatory response (16) and because of their clinical importance as indicated by the correlation of TNF- and IL-6 serum levels with outcome in critically ill patients (17, 18). To understand further the mechanism of altered cytokine expression due to NO we studied the effect of NO on NF-B in isolated mouse peritoneal macrophages, chosen because this transcription factor is an important nuclear factor binding to the promoter region of a number of proinflammatory cytokine genes.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal Model

These experiments were approved by the University of British Columbia Animal Care Committee and conform to Canadian and National Institutes of Health (NIH) guidelines regarding animal experimentation.

All experiments were conducted in outbred 5-wk-old female CD-1 mice weighing approximately 25 g. In the first set of experiments l-NAME (250 µl of a 5 mM solution or 13.5 mg/kg) (19, 20) was injected intraperitoneally. This dose is approximately 10 times the 50% effective dose (ED₅₀) for arterial pressure effect (19) and more than twice the ED₅₀ for plasma nitrite and nitrate concentrations after endotoxin infusion in mice (20). In one control group an equivalent volume and concentration of d-NAME was injected intraperitoneally while a second control group did not receive any pretreatment. These groups controlled for the specific effects of NO synthase inhibition and any nonspecific effects of the injection of these drugs. Sixty minutes later mice were anesthetized by inhalation of 3% halothane and 40 µl of endotoxin (lipopolysaccharide, 1 mg/ml; Sigma, St. Louis, MO) in phosphate-buffered saline (PBS) was injected intratracheally. In the control group endotoxin-free PBS alone was injected intratracheally. The mice were allowed to recover in room air. Six hours later the mice were anesthetized by inhalation of 5% halothane and sacrificed. Bronchoalveolar lavage (BAL) was performed with 1 ml of PBS and BAL fluid white blood cell counts were determined (Coulter S8-80; Coulter Electronics, Hialeah, FL). BAL fluid cytospins were Wright stained and the polymorphonuclear and mononuclear cells in randomly selected fields (100 cells in total) were counted at ×400 magnification. The right lung was excised and frozen in liquid nitrogen, and stored at 70° C for subsequent cytokine enzyme-linked immunosorbent assays (ELISAs). Half of the left lungs were frozen at 70° C for subsequent RNA extraction and reverse transcriptase-polymerase chain reaction (RT-PCR) and half were fixed in 4% paraformaldehyde for paraffin embedding and serial sectioning and staining. The number of polymorphonuclear and mononuclear cells in 10 to 15 randomly related fields of hematoxylin and eosin (H&S)-stained lung sections was determined.

Cytokine ELISAs

The entire right lung was homogenized in 1 ml of ice-cold PBS and then centrifuged at 1,500 rpm for 10 min at 4° C, followed by collection and storage of supernatant at 20° C. Measurements of antigenic TNF- and IL-6 concentrations in lung homogenates were made using a sandwich ELISA. Because we used the entire lung as the sample, our measurements reflect cytokine content of the entire lung. The selected antibodies were chosen on the basis of their ability to be paired (TNF- [G281-2626 and MP6-XT3; PharMingen, San Diego, CA]; IL-6 [MP5-20F3 and MP5 32C11; PharMingen]). ELISA plates were prepared by incubation at 4° C overnight with 50 µl per well of either anti-IL-6 (1 µg/ml) or anti-TNF- (2 µg/ml). Plates were washed four times and nonspecific binding was blocked with 200 µl of PBS with 2% bovine serum albumin (BSA) per well for 90 min. Diluted cell-free supernatants (50 µl) were added and incubated for 3 h. The sample was replaced with 50 µl (1 µg/ml) of the paired biotinylated antibody and incubated for 60 min. Subsequently, avidin-peroxidase conjugate was added (Bio-Rad Laboratories, Hercules, CA) followed by chromogen substrate (o-phenylenediamine [OPD]; Dako, Carpinteria, CA). Plates were read at 490 nm using an ELISA plate reader (Rainbow reader; SLT Laboratory Instruments, Salzburg, Austria). Sensitivities of the TNF- and IL-6 ELISAs were 40 and 50 pg/ml, respectively.

RT-PCR

Total cellular RNA was isolated from snap-frozen lung lobes by phenol-chloroform extraction. RNA was ethanol precipitated and dissolved in diethyl pyrocarbonate-treated water and total RNA concen- tration was determined by spectrophotometry. Five micrograms of RNA was reversed transcribed (SuperScript II reverse transcriptase; GIBCO-BRL, Gaithersburg, MD) using oligo(dT)_12-18 primers (GIBCO-BRL). The cDNA was amplified by PCR using specific prim-ers for TNF-, IL-6, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) according to optimized protocols. Primers for IL-6 were 5' GAT GCT ACC AAA CTG GAT ATA ATC 3' and 5' GGT CCT TAG CCA CTC CTT CTG TG 3' (21). Primers for TNF- were 5' AGG GGC CAC CAC GCT CTT C 3' and 5' TAG TCG GGG CAG CCT TGT CC 3' (Oligo 5.0 primer analysis software; National Bioscience, Plymouth, MN). Primers for GAPDH, a reporter mRNA, were 5' CCC ATC ACC ATC TTC CAG 3' and 5' ATG ACC TTG CCC ACA GCC 3'. The reverse-transcribed cDNA (0.5 µg in 2 µl) was added with specific cytokine primer pairs to a PCR mix with 1 U of Taq DNA polymerase (GIBCO-BRL) in a 20-µl reaction volume. The PCR for IL-6 and GAPDH included 38 cycles of 95° C for 30 s, 60° C for 45 s, and 72° C for 30 s followed by 1 cycle of 72° C for 6 min. This number of cycles was chosen because all three PCRs were in their exponential phase of amplification. The PCR for TNF- and GAPDH was identical to that for IL-6, except that the annealing temperature at 67.5° C. PCR products were identified by electrophoresis on a 2% agarose gel containing 0.2 µg of ethidium bromide per milliliter. The resulting image was captured (Eagle Eye; Stratagene, La Jolla, CA) and densitometry was performed using an automated gel-imaging system (Image PC; Scion, Frederick, MD).

Intraperitoneal Macrophage Isolation

Five-week-old female CD-1 mice were anesthetized with 3% halothane and injected intraperitoneally with 1 ml of 10% sodium caseinate in 0.9% NaCl, and were allowed to recover. Three days later the mice were sacrificed after inhalation of 5% halothane. The peritoneal cavity of each mouse was opened and washed three times with 3 ml of heparinized saline. The lavage fluid was filtered through sterile gauze, pooled, and centrifuged at 1,700 rpm for 6 min. After discarding the supernatant, macrophages were resuspended in 10 ml of RPMI 1640 supplemented with penicillin (100 U/ml), streptomycin (100 (µg/ml), and 5% fetal bovine serum. Macrophages were counted using a Coulter counter S8-80 (Coulter Electronics). Macrophages were diluted to 1 × 10⁶ per ml and plated at 7 × 10⁶ cells per 10-cm tissue culture dish for an electrophoretic mobility shift assay (EMSA) (see below) and were allowed to recover overnight in an incubator (37° C with 95% O₂ and 5% CO₂). These cells were then challenged for 3 h with one of the following in RPMI 1640 supplemented with penicillin and streptomycin: lipopolysaccharide (LPS, 200 ng/ml), LPS (200 ng/ ml) plus l-NAME (100 µM), LPS (200 ng/ml) plus d-NAME (100 µM), or LPS (200 ng/ml) plus sodium nitroprusside (1 mM). To determine if the effect of NO was dependent on guanylate cyclase the final experiment with LPS and nitroprusside was repeated after pretreating cells for 1 h with 10 µM 1H [1,2,4]oxadiazolo[4,3-alpha]quinoxalin-1-one (ODQ), a selective guanylate cyclase inhibitor (22). Controls were unchallenged macrophages.

Nuclear Protein Extraction

Nuclear protein was isolated from the challenged macrophages for use in the EMSA according to Dignam and coworkers (23). Briefly, the cells were transferred into a hypotonic solution, causing the release of cytoplasmic protein. Cytoplasmic protein was discarded and the remaining intact nuclei were subjected to further hypotonic conditions. Nuclear protein was collected and immediately stored at 70° C. The protein concentration was determined using a Bio-Rad protein assay kit.

Electrophoretic Mobility Shift Assay

The double-stranded DNA probe used to assay binding of NF-B in the EMSA was 5' AGT TGA GGG GAC TTT CCC AGC C 3' (Santa Cruz Biotechnology, Santa Cruz, CA). The underlined consensus sequence represents the NF-B protein-binding site. Briefly, this double-stranded DNA was radiolabeled with [-³²P]ATP. Ten micrograms of protein from each treatment group was preincubated with poly[dI-dC] to bind any nonspecific proteins, followed by a 20-min incubation with or without excess cold NF-B oligonucleotide (controls) or an antibody to one of the NF-B subunits (p65 or p50; Santa Cruz Biotechnology) for the supershift assay. Radiolabeled NF-B was then incubated with each reaction and the DNA-protein complexes were separated by polyacrylamide gel electrophoresis. The resulting gel was dried and autoradiographed. Supershifts with specific antibodies confirmed the identity of the NF-B homodimer and heterodimer bands.

Immunohistochemistry

Three-micron sections of paraffin-embedded lungs were deparaffinized and rehydrated. Sections were blocked with 5% normal goat serum diluted in 0.05 M Tris-buffered saline, pH 7.6 (TBS; GIBCO, Grand Island, NY), plus 1% BSA (fraction V; Sigma) for 20 min to prevent nonspecific binding. Excess blocking solution was removed and rabbit anti-mouse TNF- (polyclonal; Genzyme, Cambridge, MA) or rabbit IgG (negative control) diluted to 1:200 in TBS-BSA was added and incubated for 60 min at room temperature. The section was washed for 20 min, three times with TBS, followed by the addition of biotinylated goat anti-rabbit IgG (Dako) for 30 min and again washed with TBS. ABC-AP complex (Dako) was added and incubated for 30 min. The slides were then washed and the substrate added for 20 min. Naphthol AS-B1 phosphate and new fuchsin substrate were used, resulting in the formation of a red, insoluble precipitate whenever the antigen was present. Levamisole was added to the substrate to block endogenous alkaline phosphatase. Sections were rinsed in water, counterstained with hematoxylin, dehydrated, cleared, and mounted.

Data Analysis

Analysis of variance was used to test for differences in measured variables between groups, choosing p < 0.05 as significant. When significant differences were found, specific differences were identified using a sequentially rejective Bonferroni test procedure. Data are reported as means ± standard error throughout.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

TNF- and IL-6 expression in lungs increased 6 h after endotoxin was intratracheally instilled (p < 0.001, LPS groups versus PBS groups in Figures 1 and 2). Pretreatment with the NO synthase inhibitor, l-NAME, increased TNF- protein expression by 92% (p < 0.05) (Figure 1). NO synthase inhibition, in mice that did not receive endotoxin, did not alter lung TNF- and IL-6 concentrations. NO synthase inhibition also increased IL-6 expression by 70% (p < 0.05) (Figure 2). TNF- (Figure 3) and IL-6 (Figure 4) mRNA expression increased by 29% (p < 0.05) and 288% (p < 0.05), respectively, in mice pretreated with the NO synthase inhibitor, l-NAME, compared with those pretreated with d-NAME.

View larger version (16K): [in this window] [in a new window]   Figure 1.   TNF- concentrations in lung homogenates from mice. Compared with the PBS groups, TNF- concentrations increased in the LPS groups (p < 0.001). NO synthase inhibition by l-NAME further increases the TNF- concentration compared with either LPS control group (*p < 0.05).

View larger version (16K): [in this window] [in a new window]   Figure 2.   IL-6 concentrations in lung homogenates from mice. Compared with the PBS groups, IL-6 concentrations increased in the LPS groups (p < 0.001). NO synthase inhibition by l-NAME further increases the IL-6 concentration (*p < 0.05).

View larger version (21K): [in this window] [in a new window]   Figure 3.   TNF- RT-PCR of RNA extracted from lungs of mice. (Top) Results shown as raw data. (Bottom) The same data shown as average densitometry. l-NAME increases TNF- mRNA expression in LPS-treated mice (*p < 0.05).

View larger version (24K): [in this window] [in a new window]   Figure 4.   IL-6 RT-PCR of RNA extracted from lungs of mice. (Top) Results shown as raw data. (Bottom) The same data shown as average densitometry. l-NAME increases IL-6 mRNA expression in LPS-treated mice (*p < 0.05).

Endotoxin instilled intratracheally substantially increased the neutrophil content from 0.8 ± 0.5 (n = 4) to 21.9 ± 5.6 (n = 10) neutrophils per ×100 field (p < 0.001) on lung sections. Endotoxin alone did not alter mononuclear leukocyte content (control, 6.8 ± 1.4 [n = 4]; endotoxin, 7.0 ± 1.4 [n = 10] mononuclear leukocytes per ×100 field on lung sections). NO synthase inhibition after endotoxin did not significantly alter neutrophil or mononuclear leukocyte content in the lung sections. Similarly, endotoxin administered intratracheally increased BAL neutrophil concentration from < 1 (n = 10) to 53 ± 18 (n = 8) neutrophils/mm³ but did not alter mononuclear leukocyte concentrations in BAL fluid (control, 57 ± 7/ mm³ [n = 10]; LPS, 56 ± 14/mm³ [n = 8]). NO synthase inhibition did not significantly alter the neutrophil content (41 ± 13/ mm³, n = 10, p = 0.57) or mononuclear leukocyte content (70 ± 24/mm³, n = 10, p = 0.63) in BAL fluid. Thus, changes in lung leukocyte concentrations do not fully account for changes in lung proinflammatory cytokine expression.

Immunohistochemical staining identified pulmonary macrophages as the primary source of TNF- in this murine model (Figure 5). To understand further the mechanism of alterations of proinflammatory cytokine expression by NO we then studied murine peritoneal macrophages.

View larger version (39K): [in this window] [in a new window]   Figure 5.   Representative immunostained sections of LPS-treated mouse lungs. (A) Section is immunostained for TNF- and shows that antigenic TNF- is predominantly expressed by macrophages (arrows) in the lung after LPS stimulation. (B) Section is treated with nonspecific IgG for control comparison and shows no specific staining of macrophages (arrows).

Extraction of nuclear proteins and subsequent EMSA demonstrated that endotoxin application increases the availability of NF-B for binding to the labeled consensus sequence (Figure 6). That the band on the autoradiogram is specific for NF-B binding is demonstrated by its absence in the presence of excess cold NF-B oligonucleotide (Figure 6). A supershift assay using antibodies to the p50 and p65 subunits identified this band as the NF-B p50/p65 heterodimer.

View larger version (32K): [in this window] [in a new window]   Figure 6.   (Top) A representative autoradiogram of a polyacrylamide gel demonstrating results of the EMSA of an oligonucleotide with an NF-B-binding site. (Bottom) Graph representing the average densitometry of NF-B bands for n repetitions of this experiment. Each repetition of this experiment pooled macrophages from 3 to 6 mice, so that the final results represent an average effect for 9 to 20 mice. Compared with unstimulated macrophages, LPS increases NF-B induction. l-NAME added to LPS further increases NF-B induction compared with LPS alone or compared with LPS with d-NAME treatment. Excess cold NF-B oligonucleotide eliminates the binding of the labeled probe.

Treatment of peritoneal macrophages with the NO synthase inhibitor l-NAME increased the availability of NF-B for binding to the consensus oligonucleotide sequence (Figure 6) whereas an NO donor, sodium nitroprusside, decreased binding (Figure 7). Guanylate cyclase inhibition using ODQ did not reverse the effects of the NO donor (Figure 7). Therefore this immunomodulatory effect of NO does not appear to operate via NO stimulation of guanylate cyclase.

View larger version (31K): [in this window] [in a new window]   Figure 7.   (Top) A representative autoradiogram of a polyacrylamide gel demonstrating results of the EMSA of an oligonucleotide with an NF-B-binding site. (Bottom) Graph representing the average densitometry of NF-B bands for n repetitions of this experiment. Each repetition of this experiment pooled macrophages from 3 to 6 mice, so that the final results represent an average effect for 9 to 20 mice. LPS treatment increases NF-B induction. The NO donor sodium nitroprusside (SNP) added to LPS decreases NF-B induction. This decrease is not prevented by pretreatment with the guanylate cyclase inhibitor ODQ (LPS/SNP/ODQ).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Taken together, these results suggest that NO downregulates proinflammatory protein and mRNA expression during acute lung injury by an effect upstream of the transcription factor NF-B, which binds to the promoter region of proinflammatory cytokine genes. However, this immunomodulatory effect of NO does not depend on NO stimulation of guanylate cyclase. These results suggest that NO synthase inhibition or exogenous NO administration could play a role in modulating the pulmonary inflammatory response in human acute lung injury.

NO synthase inhibition has been used in critically ill septic patients to reverse the vasodilating effects of endogenous NO production in vascular walls (5, 6). These studies show potentially beneficial increases in mean arterial pressure but detrimental effects including a decrease in cardiac output. Although nonspecific NO synthase inhibition may be problematic (24), specific inhibition of inducible NO synthase may show more benefit and fewer detrimental side effects (25). In studies of NO synthase inhibitors in septic humans other aspects of NO physiology, including effects on the inflammatory response, have not been fully investigated. Here we investigate the effects of NO synthase inhibition on the pulmonary proinflammatory cytokine response.

In addition to its role in mediating vasodilatation, it is clear that NO also is a key inflammatory mediator. NO is important in mediating killing by macrophages (26) and NO can contribute to lung injury in immune complex disease (27). Yet, NO may also have antiinflammatory properties. This has been best studied in endothelial cells. For example, NO decreases monocyte chemoattractant protein 1 (9), selectively reduces expression of VCAM-1 (10), and inhibits macrophage colony-stimulating factor gene transcription (28) in cultured human endothelial cells. NO also has immunomodulatory effects in lymphocytes (29, 30) and likely in other cell lines. In contrast to its effect on endothelial cells, NO-releasing agents increased cytokine-induced TNF synthesis in human mononuclear cells (31). Thus, the effect of NO on the pulmonary cytokine response to acute lung injury was uncertain.

Our results extend these in vitro findings to demonstrate an effect on the immune response in the whole lung in vivo. This extension is important, because the phenotypes of cultured cells and cells under in vivo conditions can differ significantly. We found that NO alters NF-B inducibility in murine macrophages and NO synthase inhibition increases lung TNF- and IL-6 concentrations and corresponding mRNA in an endotoxin model of acute lung injury. Whole-lung observations in murine pulmonary granulomatous lesions demonstrate that NO has a complex effect, decreasing IL-12 and IFN- and increasing IL-4 and IL-10 production (11). Thus, altering endogenous NO (e.g., NO synthase inhibition) has important immunomodulatory effects on pulmonary inflammation.

NO has demonstrated effects on a number of pulmonary cell lines. However, we think that most of the NO synthase inhibitor mediated increases in TNF- and IL-6 were likely due to increased production in pulmonary macrophages. Our immunohistochemical staining shows macrophages to be the predominant cell expressing TNF- in this murine model of acute lung injury. This observation is consistent with previous studies that demonstrate the importance of macrophages as a source of proinflammatory cytokines in the lung (31).

Our results are consistent with the observation in patients with ARDS that inhaled NO decreased proinflammatory cytokine expression in BAL fluid (12). In NO-treated patients, IL-8 and IL-6 concentrations in BAL fluid supernatants were reduced (12). This was associated with decreased neutrophil production of H₂O₂ and CD11b/CD18 expression. Our observation that NO synthase inhibition has the converse effect suggests that therapeutic use of NO synthase inhibitors in septic, critically ill patients may have unanticipated, and possibly undesirable, effects on the pulmonary inflammatory response. Whether NO synthase inhibition in these individuals increases proinflammatory cytokine expression in the lung has not yet been tested. Extrapolating our results to the human condition of acute lung injury should be done cautiously. Important differences between humans and mice in terms of NO production by pulmonary macrophages exist. Human acute lung injury is a much more diverse condition than exists in our uniform model of endotoxin administration to mice. Human acute lung injury often involves participation by bacteria. NO appears to play an important role in cell killing and bacterial clearance (24, 26).

Our results suggest that NO acts by decreasing NF-B but this antiinflammatory effect of NO does not appear to involve guanylate cyclase. Two lines of evidence support this conclusion. First, our results are similar to those of Zeiher and co-workers, who demonstrated that endothelial cGMP levels did not alter MCP-1 mRNA expression whereas NO did decrease MCP-1 expression and also decreased NF-B-like binding activity during stimulation with TNF- (9). Similarly, the reduction in endothelial cell expression of VCAM-1 by NO also appears to be independent of guanylate cyclase (10). Furthermore, Peng and colleagues show that in human vascular endothelial cells stimulated with TNF-, NO induces and stabilizes I-B, resulting in inhibition of NF-B; this effect is also independent of guanylate cyclase (32). The second line of evidence is that NO classically stimulates cyclic GMP, which results in a decrease in intracellular calcium. Yet, increases in intracellular calcium downregulate proinflammatory cytokine gene expression (33). Therefore, an NO-induced decrease in proinflammatory cytokines must involve additional regulatory pathways independent of guanylate cyclase. Other intracellular signaling pathways that may be involved include reactive oxygen intermediates, possibly in combination with NO to form peroxynitrite (34, 35).

Macrophages produce NO via both constitutive NO synthase (eNOS) and inducible NOS (iNOS) (36). Constitutive NO expression by macrophages is much less than that possible with maximum iNOS activity, yet the constitutive expression may be important in intracellular signal transduction. Furthermore, early (within the first hour after endotoxin or other stimulus) NO effects appear to be more important than later NO effects on cytokine production (37), so that the effect of NO production from iNOS, which takes several hours to induce, may be less important than the NO expressed before significant iNOS induction. Thus, we chose a nonspecific NOS inhibitor for this study and we chose a 3-h time point to measure NF-B because this time point is associated with peak expression of a number of proinflammatory cytokine genes that are regulated by NF-B. Halothane could conceivably have an impact on the pulmonary inflammatory response (38). However, because all groups received exactly the same anesthetic we do not think that this effect alters the primary conclusion of our study.

In summary, NO inhibits proinflammatory cytokine expression by downregulating nuclear factors that bind to the promoter region of these proinflammatory cytokine genes, notably NF-B. NO does not produce this effect via its well known stimulation of cyclic GMP. Instead, NO appears to be acting via an as yet not fully identified pathway. Conceivably, just as reactive oxygen intermediates interact with tyrosine kinases and other molecules, NO or NO derivatives may act directly on intracellular signaling pathways.