INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Substantial evidence suggests that cytokines are important mediators of the lung injury that follows infection or exposure to microbial products. Indeed, if endotoxin is given intravenously, intraperitoneally, or into the trachea of an experimental animal, this leads to edema and inflammatory cell infiltration of the lungs. This is a model often used for studying the adult respiratory distress syndrome (ARDS) (reviewed in Reference 1), an acute, highly lethal form of lung injury, usually secondary to one or more major predisposing conditions that include trauma, thermal injury, aspiration of fluids or gastric contents, acute pancreatitis, and infections. The syndrome is triggered by an increased permeability of the alveolar capillaries, which results in the pulmonary edema (reviewed in Reference 2).

The toxic effects of endotoxin seem largely to result from the cytokines induced, particularly interleukin 1 (IL-1) and tumor necrosis factor  (TNF-) (3). Their actions, which favor expression of endothelial and leukocyte adhesion molecules and increase migration of neutrophils and vascular permeability, are thought to have a key role in the pathogenesis of endotoxin-induced lung injury. Whether interferon- (IFN-), a cytokine produced mainly by stimulated T cells and natural killer (NK) cells, has a part in this situation has as yet been little studied. We and others have previously shown that IFN- is a prominent mediator of the systemic inflammatory responses to lipopolysaccharide (LPS) (4), and so might in various ways contribute to the lung injury caused by LPS. It influences both the production and the activity of other cytokines (e.g., TNF and IL-1) and of other mediators of inflammation (NO, reactive oxygen, platelet-activating factor [PAF], etc.) that are generated in response to endotoxin (reviewed in Reference 10). IFN- can reach and stimulate alveolar macrophages and, given intratracheally, directly activates these cells (11). It also increases the permeability of endothelial cell monolayers in vitro (12) and migration through them (13), and  augments expression of intercellular adhesion molecule 1 (ICAM-1) and similar molecules that lead to increased adhesion of leukocytes (14). Furthermore, IFN- stimulates the inducible enzyme NO synthase (iNOS), and so increases levels of NO: this itself has inflammatory and immunoregulatory effects that resemble those of IFN- (reviewed in Reference 15). LPS has been shown to lead to large amounts of NO in alveolar macrophages, lung epithelia, and endothelial and interstitial cells (reviewed in Reference 15), and the NO generated by iNOS is thought to have a key role in LPS-induced acute lung injury in animal models.

In this study, we examined the role of IFN- in the pulmonary edema, increased levels of cytokines and NO, and deaths resulting in mice challenged with LPS. Since it has been reported that monoclonal antibodies to IFN- and IL-12 protect against LPS toxicity (6, 16), we studied their effects on lung edema, and the endogenous production of cytokines and NO. We tested whether administration of IFN- or IL-12 affected the evolution of lung edema, or the resulting concentrations of cytokines and NO in the circulation, and bronchoalveolar lavage fluid (BALF). We gave LPS to IFN- receptor knockout (IFN-R KO) mice to see whether LPS still induced lung edema in these. Finally, we used aminoguanidine (AG), an inhibitor of iNOS, to see what part NO plays in lung edema and cytokine release.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Unless otherwise stated, experiments were carried out with 7- to 8-wk- old female NMRI mice bred under nonspecific pathogen-free (SPF) conditions (Experimental Animal Breeding Facility, University of Leuven, Leuven, Belgium). The generation and basic characteristics of the mutant mouse strain (129/Sv/Ev), which has a disruption in the gene encoding the  chain of the IFN- receptor (IFN-R KO mice), have been described (17); these mice were bred similarly. During experiments, mice were housed under conventional conditions and given regular animal food without additives. This laboratory animal study was approved by the local Animal Ethics Committee (Catholic University of Leuven), and conducted in conformity with the Belgian and European guidelines for the protection of animals used for scientific purposes (European Directive 86/609/EEC).

Reagents

Phenol-extracted LPS from Serratia marcescens and aminoguanidine hemisulfate salt (AG) were purchased from Sigma (St. Louis, MO; Cat. No. L6136, lot 93F-4019). The rat monoclonal anti-murine IFN- antibody (F3) was prepared and purified as previously described (6). The C17.8 monoclonal antibody (from a cell line kindly provided by G. Trinchieri, Wistar Institute, Philadelphia, PA) is an IgG_2a of rat origin directed against mouse IL-12, which recognizes the p40 subunit of mouse IL-12, whether as a monomer, a homodimer, or as a part of the p70 heterodimer; it was purified by affinity chromatography on protein G (Pharmacia, Uppsala, Sweden). Rat monoclonal antibody to Escherichia coli -galactosidase from the GL117 cell line (courtesy of J. Abrams, DNAX, Palo Alto, CA) served as an isotype (IgG_2a) control; this was purified by anion-exchange chromatography on Hiload Q Sepharose (16/6; Pharmacia) and gel-filtration chromatography on Superdex 200 (16/6; Pharmacia). Cells were injected intraperitoneally into Pristane-primed athymic nude mice (nu/nu of NMRI background) to generate hybridomas. Recombinant mouse IFN- was obtained from the supernatant fluid of the CHO cell line Mick, developed in our laboratory (reference cited in Reference 6), which carries and expresses an amplified murine IFN- cDNA. It was concentrated and purified to a specific activity of 10^6.8 U/mg (range, 10^6.7-10^6.9 U/mg) as previously described (6). Recombinant mouse IL-12 (50% effective dose [ED₅₀], 0.05-0.2 ng/ml; endotoxin activity, < 0.1 ng/mg) was obtained from R&D Systems (Abingdon, UK). For in vivo administration, cytokine and antibodies were diluted in pyrogen-free saline.

Measurement of Pulmonary Edema

At chosen times, mice were killed by ether anesthesia and blood was collected by cardiac puncture. The lungs were excised, cleared of all extrapulmonary tissue, and weighed (total lung wet weight); they were then dried for 48 h at 80° C and weighed again (total dry weight). Pulmonary edema was expressed as the ratio of total wet weight to total dry weight. When data from separate experiments were pooled, the ratios in each experiment were standardized with a correction factor, c = (mean ratio from untreated mice in all experiments)/(mean ratio from untreated mice in the experiment concerned).

Bronchoalveolar Lavage Fluid

Mice were killed by ether anesthesia. The trachea was cannulated and the lungs infused with 0.7 ml of phosphate-buffered saline (PBS) supplemented with 0.01% Tween 20. BALF (average fluid recovery, 0.5 ml) was clarified by centrifugation at 3,000 rpm for 10 min and stored at 20° C until use.

Cytokine Assays

Plasma was obtained from blood collected from the orbital sinus and kept on ice for about 1 h. Plasma samples were clarified and stored at 20° C until titration. TNF levels were measured in the cytotoxicity assay on WEHI 164 cells, clone 13, as previously described (18). These cells are highly sensitive to both TNF- and TNF-. Murine TNF- (Innogenetics, Ghent, Belgium; specific activity, 10^7.25 U/mg protein) was included as an internal laboratory standard. IL-6 bioactivity was assayed by its growth-promoting effect on 7TD1 cells, cultured in flat-bottom microtiter plates (2,000 cells/well). After culture for 4 d, the number of viable cells was estimated by colorimetric assay of hexosaminidase levels, as described (19). IL-12 concentrations were determined by a sandwich enzyme-linked immunosorbent assay (ELISA) with two monoclonal antibodies, C15.6 and C17.8 (from cell lines kindly provided by G. Trinchieri), which recognize different epitopes of the p40 subunit of IL-12. The assay thus detects the p40 monomer, the p40 · p40 homodimer, and the p40·p35 heterodimer. Microtiter plates (Maxisorb; Nunc, Roskilde, Denmark) were coated at 4° C with antibody C15.6 (5 µg/ml) in 50 mM TRIS-HCl (pH 8.5), 154 mM NaCl (100 µl/well; 16 h). Plates were washed (0.05% Tween 20, 0.01% merthiolate in PBS), blocked (0.1% casein in 50 mM TRIS-HCl [pH 7.4], 154 mM NaCl, 0.01% merthiolate; 250 µl/well, for 5 h at room temperature), and again washed three times. Serial dilutions of samples in blocking solution supplemented with 0.05% Tween 20 were added (100 µl/well) and plates were incubated for 16 h at 4° C. After removal of the IL-12 samples, plates were washed five times, incubated with biotinylated anti-p40 antibody (C17.8; 3 µg/ml in blocking solution supplemented with 0.05% Tween 20; 100 µl/well) for 2 h at 37° C on a gyratory shaker (200 rpm), and washed. Streptavidin-peroxidase conjugate (Jackson Laboratories, Bar Harbor, ME; diluted 1:10,000) was added (100 µl/well) to the plates, which were incubated for 1 h at 37° C. The plates were washed, filled with substrate solution (0.002% H₂O₂, 0.42 mM 3,3',5,5-tetramethylbenzidine dihydrochloride [TMB; Sigma], 0.1 M sodium acetate, 0.01 M citric acid [pH 4.9]), and incubated for 30 min at room temperature. The resulting color was measured in an ELISA reader (Titertek; 450 nm; iEMS Reader Labsystem, Helsinki, Finland). The p40 concentrations in the samples were determined from the regression equation in comparison with that obtained with a murine IL-12 standard preparation (R&D Systems). The minimum concentration detectable was 100 pg/ml.

Mouse IFN- concentrations were determined by the sandwich ELISA described previously (20). Briefly, samples were incubated in microtiter plates coated with rat anti-mouse IFN--specific monoclonal antibody (DM1; gift of P. van der Meide, Institute for Applied Radiobiology and Immunobiology, Rijswijk, The Netherlands). The bound cytokine was detected by incubation in turn with antibody F1, used as secondary antibody, and a goat anti-rat immunoglobulin-peroxidase conjugate (Jackson Laboratories).

Plasma Levels of NO Derivatives

NO production was assessed as described previously (21) by measuring its stable degradation products, nitrate and nitrite. Nitrate was reduced to nitrite by means of nitrate reductase, and then total nitrite was determined spectrophotometrically by the Griess reaction.

Statistics

t Tests for multiple comparisons were carried out as described (22).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Mice Given LPS: Resulting Levels of Cytokines and NO Derivatives, Pulmonary Edema, and Deaths

The 50% lethal dose (LD₅₀) of S. marcescens LPS had previously been determined to be approximately 300 µg in NMRI mice; when mice were given twice this dose (600 µg), 50% died within 24 h, and all by 48 h (6). In the present study, we gave mice 300 µg intravenously. This led to significant lung edema between 6 and 32 h later, as detected by an increase in the wet/dry weight ratio (Figure 1). Histological examination of lungs removed at 6 h showed an increased blood supply in the alveolar septa, foci of vasodilatation with exudation of erythrocytes in the alveolar spaces, and mild perivascular accumulation of mononuclear cells and granulocytes. In lungs removed at 32 h, we saw areas with variable degrees of alveolar collapse accompanied by dense infiltrates of mononuclear cells and granulocytes.

View larger version (16K): [in this window] [in a new window]   Figure 1.   Induction of pulmonary edema by LPS in NMRI mice. Pulmonary edema was measured as wet/dry weight ratio (means ± SEM) in mice treated intravenously with 300 µg of LPS. Data are pooled from five experiments (three or four mice per group in each experiment). *p < 0.05, **p < 0.025, ***p < 0.001 for comparison with saline group (t test for multiple comparisons with a single control group).

Blood collected at different times was assayed for its content of TNF, IL-6, IL-12, and IFN-. As shown in Figure 2, TNF reached a distinct peak 1 h after LPS injection, but the levels returned to basal by 6 h. Concentrations of IL-6 began to increase by 1 h after LPS injection and reached their maximum between 3 and 6 h postinjection. Serum IL-12 peaked between 3 and 6 h, and IFN- appeared later (peak at 6 h). In BALF collected from mice challenged with LPS, we did not detect any increases in the levels of TNF, IL-12, or IFN-, but IL-6 levels were slightly raised (values [mean log₁₀ U/ml ± SE] at 1, 3, 6, and 24 h, respectively, were 0.74 ± 0.11, n = 10; 1.76 ± 0.13, n = 11; 1.39 ± 0.21, n = 11; and 0.82 ± 0.17, n = 9).

View larger version (20K): [in this window] [in a new window]   Figure 2.   Cytokine levels in sera of NMRI mice challenged with LPS. Serum levels (means ± SEM) were measured at the indicated times after treatment with LPS (300 µg, intravenous). Data are pooled from three to five experiments. The total number of mice is indicated in parentheses. *p < 0.001 for comparison with saline group (t test for multiple comparisons with a single control group).

Plasma NO derivatives (Figure 3) were detectable 3 h after LPS injection and rose to maximal values between 6 and 24 h. In BALF, the levels were lower, but followed a similar pattern.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Nitrite levels in plasma and bronchoalveolar fluid of NMRI mice challenged with LPS. NO2 levels (means ± SD) were measured at the indicated times after intravenous treatment with 300 µg of LPS. Data represent means of three independent experiments (three mice per group in each experiment). *p < 0.01, **p < 0.001 for comparison with saline group (t test for multiple comparisons with a single control group).

On the basis of these observations, we chose for further study the time points when TNF, IL-6, IL-12, IFN- and NO derivative levels and lung edema were expected to be at their peak after LPS injection. Survival was monitored for 7 d.

Role of Endogenous IFN- in LPS-induced Lung Edema

To study the role of IFN- in the effects of LPS, mice were treated with 0.5 mg of monoclonal anti-IFN- antibody 24 h before LPS challenge (300 µg, intravenous). The control group received saline. The results of several experiments are summarized in Table 1. It can be seen that, in accordance with previous reports (6, 8, 9), administration of the antibody effectively neutralized the appearance of IFN- in the blood, and completely protected the mice against the lethal effects of LPS. Nevertheless, the antibody did not influence TNF, IL-6, or IL-12 levels in serum or BALF, and completely failed to moderate the LPS-induced lung edema. There was significant reduction in the levels of NO derivatives in BALF 24 h after LPS challenge, but not in the circulation (data not shown).

To investigate further the involvement of endogenous IFN-, we compared the induction by LPS of lung edema in wild-type and IFN-R KO mice of the 129 strain. In wild-type mice, 300 µg of LPS was needed to induce edema at 6 h (Figure 4), but IFN-R KO mice were more susceptible, and approximately six times less LPS was sufficient to induce edema. Nevertheless, when 300 µg of LPS was injected, the extent of edema was no greater in the IFN-R KO mice than in wild-type mice.

View larger version (11K): [in this window] [in a new window]   Figure 4.   LPS-induced lung edema in wild-type (WT) and IFN-R KO mice. Pulmonary edema was measured as the wet/dry weight ratio (means ± SEM) in mice treated with varying doses of LPS. Data are pooled from two experiments (three or four mice per group in each experiment). *p < 0.001 (t test for multiple comparisons with a single control group).

Role of Exogenously Administered IFN- in Pulmonary Edema in Response to LPS

Our finding that lung edema develops in IFN-R KO mice in response to much less LPS than in wild-type mice suggests that endogenously formed IFN- might provide some degree of protection against its development. To investigate this possibility, groups of mice were given intraperitoneal injections of   recombinant murine IFN- (100,000 units) 2 h before LPS challenge; control groups received saline. The results summarized in Table 2 show that systemic administration of IFN- did indeed reduce lung edema induced by LPS. Histological examination showed that lower fluid content of the lungs was associated with reduced bleeding into the interstitial and alveolar spaces (data not shown). In sharp contrast, whereas all mice given LPS alone were alive 24 h later, those pretreated with IFN- before LPS challenge had all died by this time. IFN- pretreatment also led to increased serum levels not only of IFN- but also of IL-6, IL-12, and TNF in response to LPS (Table 2); TNF levels also stayed higher for longer than 6 h, unlike in mice not pretreated with IFN-. In BALF, LPS- induced levels of IL-6 and IL-12(p40) were greater in mice pretreated with IFN- [at 3 h, IL-12(p40) levels were 0.51 ± 0.09 (n = 10) in saline-injected control mice, versus 2.74 ± 0.9 (n = 10) ng/ml in those given IFN-; IL-6 levels were 1.69 ± 0.08 (n = 10) and 2.62 ± 0.17 (n = 9) log₁₀ U/ml, respectively]. TNF, however, remained undetectable in BALF.

Although IFN- pretreatment resulted in higher serum levels of IFN- at all times than in controls, it did not significantly augment LPS-induced levels of NO derivatives, except in the serum at 1 h and in BALF at 3 h after LPS challenge (Table 2). It should be mentioned that IFN- given without LPS led to high serum levels of NO derivatives (mean 236.3 ± 32.15 mM at 4 h postinjection in two mice pretreated with IFN-; 62.45 ± 6.85 mM in two saline-treated mice; p = 0.03).

Role of IL-12 in Lung Edema Induced by LPS

Previous work has shown that the first step in the chain of events that leads to endotoxic shock (5, 6, 23) is production of IFN- by NK cells. Since IL-12 is a potent stimulator of IFN- production by both these cells and T lymphocytes, we looked to see whether endogenous IL-12 production affected the early lung edema resulting from LPS. Mice were injected with monoclonal anti-IL-12 antibody (0.5 or 1 mg, intraperitoneal) or saline 24 h before they were challenged with LPS. This pretreatment abolished the lethal effects of LPS, but not the resulting edema, measured from the pulmonary fluid content at 6 h (Table 3), or at 32 or 48 h. Other groups of mice were killed at different times, and the amounts of cytokines and NO derivatives in the plasma and BALF were measured. After anti-IL-12 antibody administration, circulating IL-12(p40) levels between 3 and 6 h after LPS challenge were less than in controls (Table 3). There was no IFN- in the plasma at 6 h; serum TNF and IL-6 levels were not influenced, but levels of NO derivatives were significantly reduced 24 h after LPS challenge.

To study further the role of IL-12, mice were given exogenous recombinant murine IL-12 (0.1 µg) 24 and 1 h before LPS challenge. As shown in Table 4, IL-12 increased the number of deaths, and all mice were dead within 24 h. As expected, IL-12 levels in sera and BALF were increased, but also IFN- and NO derivatives were detected sooner in the serum than in controls, and reached significantly higher levels. There was no effect on the extent of pulmonary edema at 6 h.

Effect of Aminoguanidine on LPS-induced Lung Edema

To see what part was played by NO in LPS-induced lung injury, we blocked NO synthase with aminoguanidine (AG). Given intraperitoneally at a dose of 4 mg/mouse, 1 h before LPS challenge, this led to 45-50% lower levels of NO derivatives in the serum and BALF (Table 5); TNF production in the serum was also reduced, but there was no effect on the development of lung edema unless the dose of AG was increased to 8 or 10 mg.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Histological examination confirmed that mice given an intravenous injection of a 50% lethal dose of LPS quickly developed lung edema, which could be quantified in terms of the increase in the wet/dry weight ratio of the lungs. The edema reached a maximum 6 h after LPS injection,then decreased to some extent, but finally tended to rise again at about the time the animals started to die. Nevertheless, throughout the period of observation, the extent of the edema (increase in wet/ dry weight ratio from 4.5 to 4.8) was relatively small in comparison with a massive degree of fluid accumulation seen in clinical ARDS. In a previous study, Faggiona and co-workers (24), who challenged mice with LPS given intraperitoneally, found that pretreatment with a neutralizing anti-IFN- antibody inhibited the late (24 h) edema and lethal effects. In the present study, we focused attention on the early pulmonary reaction that develops after intravenous LPS. Somewhat to our surprise, we found IFN-R KO mice to be more sensitive, responding to much smaller doses of LPS. Moreover, mice able to respond to IFN- had less edema in response to LPS when pretreated with exogenous IFN-, even though this potentiated the lethal effects of LPS. On the other hand, pretreatment with a neutralizing anti-IFN- antibody had no effect on the development of edema, even though it prevented the lethal effects of LPS, as reported previously (4, 8, 9). A possible explanation for these results is that any IFN- present in the body and thus active before LPS is injected can inhibit the development of edema, but additional IFN- generated by LPS is formed too late to influence the extent of the edema. Indeed, we found that IFN- was the last of the four cytokines we studied to reach its peak level in serum; this was at 6 h, the time also of the peak of pulmonary edema. Whatever the explanation, the early lung edema seems not to be related, or rather to be inversely related, to later events that lead to death of the animals.

We found that systemic production of IFN- was preceded by production of IL-12(p40), with a peak serum level 3 h after LPS injection, and of TNF (peak at 1 h). This led us to wonder if the edema resulted from the effects of either of these cytokines. Production of IFN- in LPS-treated mice is known to depend critically on preceding production of IL-12 (25), and the lethal effect of LPS in BCG-primed mice was found to depend on the presence of endogenous IL-12 (26). Furthermore, in IFN-KO mice, a 4-d treatment course with IL-12 resulted in severe pulmonary edema and infiltration (27). In our model system, treatment with anti-IL-12 antibody at a dose that prevented induction of IFN- and death in response to LPS had no effect on early pulmonary edema. Concordantly, exogenous IL-12 boosted cytokine induction and the lethal effects of LPS, but had no effect on the development of pulmonary edema. These data show that the induction of IL-12 and IFN- in response to LPS is not causally associated with the early edema, and that this edema is not directly related to the events that later led to the death of the mice.

Might NO, then, be responsible for LPS-induced lung edema? We measured progressively increasing levels of NO derivatives in serum and BALF during the 24 h after LPS injection. Treatment with the NO synthase inhibitor AG prevented the early (3 to 6 h) rise in levels of NO derivatives, but did not prevent the appearance of high levels at 24 h. The treatment also prevented te appearance of early pulmonary edema but not the fatal outcome. This suggests that early NO production is causally associated with the pulmonary edema, but again that this edema does not play a part in its later lethal effects. Note, however, that the AG treatment regimen in our experiments was not intended to achieve the long-term suppression of NO production that might have affected mortality. Other studies have shown that NO induced by LPS does play a part in various aspects of LPS toxicity. Just as in our mouse model system, Numata and coworkers (28) showed that inhibition of iNOS reduced neutrophil accumulation and prevented edema in the lungs of LPS-injected dogs. However, others have reported data supporting beneficial effects of LPS-induced NO. Thus, administration of L-NMMA, another inhibitor of NO synthase, was found to potentiate LPS-induced intestinal and liver damage in a sepsis model and to abrogate the protective effect of a low LPS dose against subsequent challenge with a lethal dose (29). Also, endogenous NO, generated as a result of activation of iNOS, prevented tissue damage in the lungs and livers of mice challenged with a combination of LPS and formyl-norleucyl-phenylalanine (30). In contrast, results in iNOS KO mice indicated that NO induced by LPS via the iNOS system had no effect on lung injury (31) or indeed potentiated both it (32) and the lethal effect of LPS (33).

Interestingly, treatment with AG also inhibited LPS-induced increases in TNF, which suggests that NO is somehow involved in the regulation of TNF production. In vitro studies have indicated that NO increases the production of TNF by neutrophils and monocytic cells in response to LPS (34, 35).

AG inhibits both constitutive and inducible NO synthase. Therefore, our finding that pulmonary edema is reduced by AG does not necessarily mean then that induced NO is involved. Indeed, we found that treatment with anti-IFN- antibody did not significantly affect levels of NO derivatives during the 24 h after LPS injection. Anti-IL-12 antibody decreased the LPS-induced circulating IFN- at 6 h, but nitrite levels only as late as at 24 h. Taken together, these data support the view that endogenous LPS-induced IFN- does not enhance early NO release, and so has no effect on the pulmonary edema controlled by NO.

Our observations raise the question of how a early pulmonary edema can be reduced as the result of lower availability of NO in the early hours after exposure to LPS, on the one hand, and by previous exposure to IFN- on the other hand. One possibility is that lower levels of NO may result in less vasodilation and hence in less or later fluid transudation. As to the mechanism underlying the inhibitory effect of IFN- pretreatment, one of several possible mechanisms is that, by differentially affecting LPS receptor expression in various tissues, IFN- may divert LPS to other sites in the body and thereby reduce its effects in the lungs.

In summary, our results suggest that in our LPS-based model, (1) by some unidentified mechanism, preexisting IFN- makes lung tissue less likely to develop early edema; (2) NO production contributes to the development of this initial edema; (3) LPS leads to production of IFN- that plays little part in the early edema, but does contribute to later lethal events; and hence (4) there is no direct or simple relation between the early pulmonary response and these later lethal events.

The mild pulmonary edema that follows acute exposure of mice to LPS should not be considered to provide a comprehensive model for the study of ARDS. Nevertheless, the mechanisms involved may reflect an early component in the complex pathogenesis of this condition, especially if it results from gram-negative sepsis. In this context, both our present and previous results (24) suggest that endogenous IFN- may both protect against ARDS and also increase its severity. In human ARDS, the steps in the chain of events that lead up to the early stages in the syndrome will often proceed more slowly that in the LPS-based animal models, which mostly are designed to be hyperacute. Also, microbial proliferation is often an important component of clinically occurring ARDS. Therefore, it seems possible that the IFN- production that occurs in the early stages of clinical ARDS may have sufficient time to exert a protective effect, both by helping the host to control microbial proliferation and perhaps by inhibiting the early pulmonary inflammation; in later stages of the syndrome, the influence of any IFN- formed is likely to be detrimental.