We have previously demonstrated that mice exposed to sublethal hyperoxia (an atmosphere of >95% oxygen for 4 days, followed by return to room air) have significantly impaired pulmonary innate immune response. Alveolar macrophages (AM) from hyperoxia-exposed mice exhibit significantly diminished antimicrobial activity and markedly reduced production of inflammatory cytokines in response to stimulation with LPS compared with AM from control mice in normoxia. As a consequence of these defects, mice exposed to sublethal hyperoxia are more susceptible to lethal pneumonia with Klebsiella pneumoniae than control mice. Granulocyte/macrophage colony-stimulating factor (GM-CSF) is a growth factor produced by normal pulmonary alveolar epithelial cells that is critically involved in maintenance of normal AM function. We now report that sublethal hyperoxia in vivo leads to greatly reduced alveolar epithelial cell GM-CSF expression. Systemic treatment of mice with recombinant murine GM-CSF during hyperoxia exposure preserved AM function, as indicated by cell surface Toll-like receptor 4 expression and by inflammatory cytokine secretion following stimulation with LPS ex vivo. Treatment of hyperoxic mice with GM-CSF significantly reduced lung bacterial burden following intratracheal inoculation with K. pneumoniae, returning lung bacterial colony-forming units to the level of normoxic controls. These data point to a critical role for continuous GM-CSF activity in the lung in maintenance of normal AM function and demonstrate that lung injury due to hyperoxic stress results in significant impairment in pulmonary innate immunity through suppression of alveolar epithelial cell GM-CSF expression.
- alveolar epithelial cell
- acute respiratory distress syndrome
- granulocyte/macrophage colony-stimulating factor
acute lung injury (ALI) and the acute respiratory distress syndrome (ARDS) together represent a continuum of acute pulmonary disease that causes enormous morbidity and mortality throughout the United States (35). Critically ill individuals with acute respiratory failure are at significantly increased risk for developing nosocomial pneumonia. Those patients who develop nosocomial pneumonia have greatly prolonged duration of mechanical ventilation, length of stay in the intensive care unit, and mortality compared with similarly ill individuals who avoid this complication (7). A number of factors may contribute to the increased risk of pneumonia in patients receiving mechanical ventilation (7, 8, 10). Much of the focus in studies of the pathophysiology of ventilator-associated pneumonia has been on the role of the endotracheal tube or on the alteration of gastric and oropharyngeal flora in association with increased risk of aspiration of secretions (12). Recently, a number of investigators have described impaired pulmonary innate immune responses as a consequence of sepsis syndrome in animal models (9, 11, 33).
Previously, we have found that breathing high concentrations of supplemental oxygen may itself contribute to impaired pulmonary host defense and increased susceptibility to pneumonia (1). When mice were exposed to hyperoxia for a period that typically did not cause lethal lung injury (sublethal hyperoxia), alveolar macrophage (AM) function became significantly impaired. There was decreased expression of Toll-like receptors (TLR) on the cell surface, significantly decreased release of early response cytokines TNF-α (TNF) and IL-6 and impaired phagocytosis and killing of bacteria. In a model of pneumonia due to Klebsiella pneumoniae, these defects were associated with increased growth of bacteria in the lungs, increased systemic spread of infection, and significantly increased mortality in mice previously exposed to sublethal hyperoxia, compared with normoxic controls. The abnormalities we identified in AM harvested from mice following exposure to sublethal hyperoxia resembled in many respects the defects in AM activity associated with deletion of the gene for granulocyte/macrophage colony-stimulating factor (GM-CSF) or its receptor (25, 31). Therefore, the present study was undertaken to determine the potential role of GM-CSF in the innate immune response in the lung following sublethal hyperoxia. We now report that GM-CSF production by alveolar epithelial cells (AEC) was diminished after in vivo hyperoxia. Furthermore, pharmacological treatment with GM-CSF reversed many of the defects in macrophage function.
MATERIALS AND METHODS
Specific pathogen-free, 6- to 8-wk-old, wild-type C57BL/6 mice were obtained from Jackson Laboratories (Bar Harbor, ME). Mice were housed in isolator cages within the Animal Care Facilities at the Veterans Affairs Research Laboratories until the day of experimentation. The animal care committee at the Ann Arbor Department of Veterans Affairs Medical Center approved the experimental protocols. Mice received food and water ad libitum.
Sublethal hyperoxia and treatment with GM-CSF.
The model of sublethal hyperoxia involved exposure to high concentrations of oxygen for 4 days, as in our previous work (1). C57BL/6 mice were placed in a Plexiglas chamber attached to a compressed oxygen source, with a continuous oxygen monitor (Proox model 110; Biospherix, Lacona, NY) adjusted to maintain an atmosphere of 95% oxygen. Mice continued to receive food and water as above and were checked daily for any evidence of disease. CO2 in the chamber was monitored and remained <5%. Under these conditions, C57BL/6 mice exposed to 95% oxygen for 4 days experienced little or no mortality, and, if allowed to recover in room air, were indistinguishable from their matched controls. In experiments to determine the effects of short-term treatment with GM-CSF, mice were inoculated with GM-CSF (9 μg/kg subcutaneous) at entry and after 48 h in hyperoxia. Control mice received subcutaneous inoculations with saline vehicle alone.
Laser capture microdissection and real-time PCR.
Lungs from hyperoxia and control animals were distended with a 1:3 mix of optimum cutting temperature compound and PBS via a tracheal catheter, placed in tissue cassettes, and snap-frozen on dry ice. Six-micrometer frozen sections were stained using the HistoGene LCM Frozen Section Staining Kit (Arcturus) according to the manufacturer’s instructions. Alveolar wall cells were identified at ×20 magnification and captured using CapSure LCM caps (Arcturus). Total RNA was prepared from the caps using Absolutely RNA RT-PCR microprep kit (Stratagene, La Jolla, CA) following the manufacturer’s instructions. In separate experiments, total lung RNA was obtained from left lungs of mice using an RNeasy Protect Mini Kit (Qiagen Sciences, Valencia, CA) according to the manufacturer’s instructions. Samples were DNase treated (Qiagen) before real-time PCR.
DNase-treated total RNA was converted to cDNA, and specific PCR products were generated using Brilliant SYBR Green QRT-PCR master mix kit, 1 step (Stratagene), as per the manufacturer’s instructions. cDNA conversion, amplification, and data analysis were performed using a Mx3000P real-time PCR system computerized cycler from Stratagene, as previously described (29). We used the following primers, designed using software available at http://labtools.stratagene.com (Stratagene), and synthesized and HPLC purified by Invitrogen Life Technologies: GM-CSF (forward primer AGATATTCGAGCAGGGTCTAC; reverse primer GGGATATCAGTCAGAAAGGTT) and GAPDH (forward primer TATGTCGTGGAGTCTACTGGT; reverse primer GAGTTGTCATATTTCTCGTGG). Primers were used at 75 nM each in 25-μl reactions. Cycle parameters were as follows: 40 min at 55°C (reverse transcription step); 10 min at 95°C (denaturation step), and 40 cycles composed of 30 s at 95°C, 1 min at 58°C, and 30 s at 72°C. Control wells containing no template were used to exclude the presence of contaminating template molecules and to identify potential primer-dimer products from the dissociation curve analysis.
Total RNA from each sample was analyzed for GM-CSF and GAPDH mRNA in separate triplicate wells. The fluorescence values of the threshold cycles were collected at the end of the annealing step from each reaction. The threshold cycle values obtained from GAPDH amplification were used to normalize specific mRNA quantification (28). Data are expressed as relative increase of specific mRNA in the treated samples compared with an untreated control sample, used as calibrator. To correct for possible volume differences, transparency of the caps, or other well-to-well differences, the passive reference dye 5(6)-carboxy-X-rhodamine-C5-maleimide (also known as ROX; Stratagene) was used in all reactions. A dissociation curve analysis was included in the experiments to verify the absence of nonspecific products such as primer-dimers.
Processing of lung lavage fluid.
At specified times, bronchoalveolar lavage (BAL) was performed in mice from hyperoxia and control groups using previously described methods (1). Lung lavage was performed using 1-ml aliquots of PBS, which were pooled for each animal. Total numbers of viable cells were determined using a hemocytometer with trypan blue exclusion. Cytospins were prepared from BAL cells, 5 × 105 cells/mouse, stained with a modified Wright-Giemsa stain (Diff-Quik; Baxter, McGaw Park, IL) for differential counts. From these stained cytospin preparations, the percentages of mononuclear cells and neutrophils were determined by microscopic counting of 200 cells/slide. Pooled BAL fluid was centrifuged at 800 g for 10 min at 4°C. Cell-free supernatants were stored at −70°C for subsequent analysis.
Isolation and culture of murine AM and AEC.
AM were adherence purified from BAL fluid as described previously (1). BAL cells were resuspended in DMEM (GIBCO, Grand Island, NY), supplemented with penicillin-streptomycin (GIBCO), and then plated on 96-well plastic plates (Corning, Corning, NY) at 1 × 105 cells/well in DMEM supplemented with penicillin-streptomycin. AM were allowed to adhere for 60 min at 37°C and then gently washed with warm PBS to remove nonadherent cells. AM were cultured in DMEM supplemented with penicillin-streptomycin and 10% FBS (Sigma) (complete media).
Murine type II AEC were isolated as described previously (27). After the pulmonary vasculature was perfused free of blood with PBS, type II AEC were freed from the lung by enzymatic digestion with Dispase (Worthington Biochemical, Lakewood, NJ) infused via the trachea. The collected cells were filtered successively through 100-, 40-, and 25-μm nylon mesh filters to create a single cell suspension. Contaminating leukocytes were bound with biotinylated anti-CD32 (FcγR) (0.65 μg/million cells, BD Pharmingen) and anti-CD45 (common leukocyte antigen) (1.5 μg/million cells, BD Pharmingen) and removed using a magnetic cell separator (Magnasphere, Promega) following addition of streptavidin-coated magnetic particles. Cells not bound with magnetic particles were recovered and suspended in culture media. Viability was >97% by trypan blue exclusion. The cells were plated overnight in 60-mm culture plates. The nonadherent cells, including type II AEC, were recovered and counted. Viability was >97% by trypan blue exclusion. Fibroblasts, endothelial cells, and bone marrow-derived cells (including macrophages) express the intermediate filament protein, vimentin. These cells were >95% vimentin negative, supporting their epithelial origin. This result was confirmed by staining for prosurfactant protein C, a specific type II cell marker. AEC were then cultured in fibronectin-coated 96-well tissue culture plates (Corning) at 1.5 × 105 cells/well in complete media. AEC were cultured in triplicate wells for determination of cytokine secretion.
For measurement of ex vivo production of cytokines, AM or AEC were harvested as above and incubated in complete medium with or without appropriate stimuli. AM were treated with LPS (Escherichia coli-derived; Sigma, St. Louis, MO) at doses of 1, 10, and 100 ng/ml or with TNF (R&D Systems, Minneapolis, MN) at 10 ng/ml. AEC were treated with TNF (10 ng/ml, R&D Systems) or IL-1α (1 ng/ml, Sigma). After 24 h in culture, cell-free supernatants were collected, centrifuged to remove cellular debris, and assayed for cytokine levels by ELISA. For in vivo measurement of cytokines and chemokines, cell-free BAL fluid and lung homogenates were collected from infected and uninfected mice from hyperoxia and control. In selected instances, AM were lysed in lysis buffer (Roche) for determination of cell-associated IL-1β. We measured the levels of GM-CSF, monocyte chemoattractant protein-1 (MCP-1), TNF, IL-6, IL-1β, and macrophage inflammatory protein-2 (MIP-2) in duplicate using murine Quantikine kits (R&D Systems) as directed by the manufacturer.
Flow cytometry of AM.
AM were harvested by BAL as above and resuspended at 1 × 105 cells/100 μl in cold fluorescent antibody diluting (FA) buffer (Difco, Detroit, MI) containing 1% BSA and 0.1% sodium azide. AM were incubated for 30 min with phycoerythrin (PE)-labeled anti-murine-TLR4/MD-2 antibody (eBioScience, www.eBioscience.com) or a nonspecific PE-labeled IgG2b (BD Pharmingen), both diluted to a final concentration of 0.4 μg/100 μl. AM were also stained with FITC-conjugated Bandeiraea simplicifolia lectin (Sigma), a marker of mature AM phenotype (25). AM were washed with FA buffer to remove unbound antibodies or lectin and fixed with FA buffer containing 1% formalin. Flow cytometry was performed using a FACScan flow cytometer (Becton Dickinson, Mountain View, CA), and analysis was done using the CELLQuest software package (Becton Dickinson). Thresholds for positive staining were determined from the isotype-matched control samples.
Preparation and inoculation of K. pneumoniae.
K. pneumoniae strain 43816, serotype 2 (American Type Culture Collection, Manassas, VA) was used for these studies. This strain is known to produce clinically significant pneumonia in mice (15). Bacteria were grown in LB broth (Difco) for 18 h at 37°C overnight. The morning of each experiment, the culture was diluted to a low optical density (OD), and the OD value (600 nm) of the culture was monitored to insure log phase growth. The concentration of bacteria in broth was calculated by measuring absorbance at 600 nm. Standard values based on known colony-forming units (CFU) previously established in our lab were used to calculate inoculum concentration. Final concentration was confirmed by quantitative culture of the inoculum in serial 10-fold dilutions on soy-base blood-agar.
C57BL/6 mice were exposed to hyperoxia or room air as described above. After 3 days in hyperoxia, mice were inoculated intratracheally with K. pneumoniae (5 × 102 CFU) and then returned to hyperoxia for one additional day (1). Mice kept under similar housing conditions in room air were used as controls. For intratracheal inoculation, mice were anesthetized with pentobarbital sodium (50 mg/kg ip; Abbott Laboratories, North Chicago, IL). The trachea was exposed, and K. pneumoniae, in a final volume of 50 μl in sterile normal saline solution, was inoculated using a 26-gauge needle. Uninfected control mice were inoculated with the same volume of sterile normal saline solution. The skin incision was closed with 4–0 surgical prolene with simple stitches. Mice were allowed to recover from the procedure and were returned to appropriate housing. Mice in the hyperoxia group were returned to hyperoxia to complete 4 days of total exposure.
Determination of lung CFU and systemic dissemination of K. pneumoniae.
Twenty-four hours after intratracheal inoculation with K. pneumoniae, mice were anesthetized with intraperitoneal Euthasol (Diamond Animal Health) and exsanguinated by sectioning of the abdominal aorta as above. The pulmonary vascular bed was perfused with 10 ml of PBS via the right ventricle. Lungs and spleens were removed aseptically, placed in sterile PBS (1 ml), and ground with a tissue homogenizer (Biospec Products, Bartlesville, OK) under aseptic conditions. Serial 10-fold dilutions were made to 10−8 (lungs) or 10−4 (spleens); 10 μl of each dilution was plated on soy-base blood-agar plates (Difco) and incubated for 24 h at 37°C. Colony counts were then determined for individual mice. A priori, we defined positive spleen cultures as those containing >10 CFU of K. pneumoniae (1).
Data are expressed as means ± SE. Groups were compared using a two-tailed Student’s t-test or ANOVA, if more than two groups were compared, with the InStat software package from GraphPad Software (San Diego, CA). The Bonferroni multiple comparisons test was used to compare CFU in different groups of mice in ⇓⇓⇓⇓⇓⇓Fig. 7. Differences were considered statistically significant if P values were <0.05.
Effect of sublethal hyperoxia on GM-CSF expression by alveolar wall cells.
We first determined the effect of hyperoxic stress in vivo on expression of GM-CSF in whole lung. Relative GM-CSF mRNA expression was evaluated by real-time PCR in lung homogenates from mice exposed to hyperoxia for increasing numbers of days. Whole lung GM-CSF mRNA was significantly reduced after 3 days in hyperoxia (Fig. 1). When mice were returned to room air following 4 days in hyperoxia, lung GM-CSF mRNA expression returned toward baseline after 48 h of normoxia. Thus exposure to transient hyperoxia resulted in reduced lung GM-CSF expression.
GM-CSF is a tightly regulated growth factor that acts locally and is normally undetectable in BAL fluid. Although a number of different cells in the lung can produce GM-CSF, AEC clearly are an important source of this growth factor in the alveolar space. To investigate the effect of hyperoxia on AEC expression of GM-CSF in vivo, experimental mice were placed in hyperoxia for 4 days, whereas control mice remained in ambient air. The lungs then were harvested, frozen, and sectioned. Alveolar wall cells (200 cells/section) were captured using laser capture microscopy, and expression of GM-CSF and GAPDH was determined using real-time PCR. Expression of GM-CSF was normalized to expression of the housekeeping gene in each mouse. As shown in Fig. 2, exposure to hyperoxia resulted in a significant decrease in alveolar GM-CSF mRNA expression.
Effect of sublethal hyperoxia on AEC secretion of GM-CSF ex vivo.
Laser capture microscopy allowed us to evaluate expression of GM-CSF mRNA by cells of the alveolar wall. We next performed experiments to specifically determine the effects of in vivo hyperoxia on GM-CSF protein expression by type II AEC. C57BL/6 mice were placed in hyperoxia or room air for 4 days. Type II AEC were then isolated and placed in culture for 24 h in the presence or absence of inflammatory cytokines (TNF-α or IL-1α). Following exposure to hyperoxia in vivo, type II AEC production of GM-CSF was significantly reduced, both at baseline and following cytokine stimulation (Fig. 3A). This decreased growth factor production did not appear to be a consequence of cellular injury, based on exclusion of trypan blue. However, to determine whether reduced GM-CSF expression was a consequence of generalized cellular dysfunction, we also measured release of MCP-1 in these AEC cultures. Interestingly, secretion of MCP-1 at baseline was not influenced by sublethal hyperoxia, whereas chemokine release in response to cytokine stimulation was enhanced in AEC isolated from mice undergoing sublethal hyperoxia compared with controls in room air (Fig. 3B). These data indicate that hyperoxic stress results in a shift in behavior of AEC away from production of GM-CSF and toward expression of MCP-1.
Effect of systemic treatment with GM-CSF on inflammatory cells in whole lung lavage.
Having found that sublethal hyperoxia resulted in diminished GM-CSF expression by AEC, experiments were performed to determine whether systemic treatment with recombinant murine GM-CSF would prevent the adverse consequences of hyperoxia for pulmonary innate immunity. We first determined the effect of treatment with GM-CSF on the composition of cells within the alveolar space. In uninfected mice, exposure to hyperoxia for 4 days had little effect on BAL cell counts or differentials compared with normoxic control mice. Similarly, treatment with GM-CSF during hyperoxia did not cause a significant change in alveolar inflammatory cells. In each instance, AM were the predominant cell type, with very few polymorphonuclear leukocytes (PMN) (data not shown). Both room air and hyperoxia-exposed mice developed significant PMN accumulation in the alveolar space following inoculation with K. pneumoniae. The total number of leukocytes was increased slightly in hyperoxia mice compared with controls. Following treatment of hyperoxia-exposed mice with GM-CSF, there was a slight decrease in total lung leukocytes recovered by lavage compared with mice exposed to hyperoxia alone before infection (Fig. 4A). The proportion of PMN recruited to the lungs did not differ significantly between room air, hyperoxia, and hyperoxia plus GM-CSF groups (Fig. 4B). Thus inflammatory cell numbers in the lungs of K. pneumoniae-infected mice were little changed by treatment of hyperoxia-exposed mice with GM-CSF.
Effect of systemic treatment with GM-CSF during sublethal hyperoxia on AM cytokine production.
Previously, we had found that impaired pulmonary host defense in the setting of sublethal hyperoxia was associated with diminished AM production of early response cytokines. Having determined that GM-CSF production by AEC was reduced during hyperoxia, we next examined the effect of systemic treatment with GM-CSF during the hyperoxic stress on AM production of TNF and IL-6 following stimulation with LPS ex vivo. As anticipated, AM harvested from mice exposed to sublethal hyperoxia released significantly less TNF and IL-6 in response to LPS compared with AM from normoxic controls. However, AM cytokine production in response to LPS was largely restored by treatment of hyperoxic mice with GM-CSF (Fig. 5, A and B). Similarly, IL-1β in cell lysates of AM stimulated with LPS was decreased in AM from mice exposed to hyperoxia compared with normoxic controls (50 ± 2 vs. 406 ± 112 pg/106 cells). AM harvested from mice treated with GM-CSF demonstrated significantly increased LPS-stimulated IL-1β in cell lysates (2,081 ± 135 pg/106 cells).
To determine whether the alteration in AM cytokine production following sublethal hyperoxia was limited to LPS signaling, AM from hyperoxic and normoxic mice were exposed to complete medium or treated with TNF ex vivo, and release of the neutrophil chemoattractant, MIP-2, was measured in the culture supernatants. Previously, we had found that MIP-2 expression in lung homogenates from hyperoxia-exposed mice was reduced compared with room air controls (1). In isolated AM, we now find that MIP-2 production was significantly reduced in AM from mice exposed to sublethal hyperoxia (Fig. 5C). Interestingly, in contrast to LPS-stimulated cytokine production, TNF-induced MIP-2 production was not preserved by treatment of mice with GM-CSF in vivo during hyperoxia. These data confirm that AM responses to multiple stimuli are significantly impaired by in vivo hyperoxia and demonstrate that short-term systemic treatment with GM-CSF protects against some, but not all, of these changes.
Cell surface staining of AM.
Diminished cytokine expression in response to LPS by AM from hyperoxia-exposed mice was associated with decreased cell surface expression of TLR4, the receptor by which mammalian cells recognize LPS. Treatment of mice with GM-CSF during sublethal hyperoxia resulted in AM expression of TLR4 resembling that in cells from normoxic controls (Fig. 6A).
In normal mice, mature AM stain positively with the lectin BS-1 (32). In contrast, AM from GM-CSF-deficient mice demonstrate significantly reduced staining for BS-1 compared with wild-type controls (25). Similarly, the fraction of AM that stained positively for BS-1 was decreased following sublethal hyperoxia (normoxia = 93.5%; hyperoxia = 70.5%). However, treatment of mice with GM-CSF resulted in BS-1 staining similar to that in normoxic controls (GM-CSF = 91.5%). As an additional indication of AM maturational state, cell surface staining for CD11b was determined by flow cytometry (Fig. 6B). In the setting of hyperoxic stress, staining for CD11b was increased compared with room air controls, consistent with a less mature AM phenotype. Treatment with GM-CSF during hyperoxia resulted in return of AM CD11b expression to that observed in room air controls. Thus treatment with GM-CSF during hyperoxic stress led to maintenance of cell surface characteristics similar to those found in normoxic control mice.
Lung bacterial burden in the setting of K. pneumonia.
Having demonstrated that GM-CSF treatment of hyperoxic mice restored AM innate immune function measured ex vivo, experiments were performed to determine whether this effect was translated into reduced progression of bacterial infection 24 h after intratracheal inoculation with K. pneumoniae (500 CFU/mouse). Treatment of mice in normoxia with GM-CSF led to a trend toward decreased bacterial CFU in the lung compared with placebo controls, although the difference did not achieve statistical significance. However, in mice exposed to hyperoxia, treatment with GM-CSF led to a significant (3 log) reduction in bacterial burden in the lung compared with placebo-treated controls (Fig. 7). Thus these data show that GM-CSF treatment during oxygen exposure substantially reverses the defect in pulmonary host response induced by sublethal hyperoxia.
Our previous studies have demonstrated that exposing mice to a sublethal period of hyperoxia resulted in significantly impaired pulmonary innate immune response, as a consequence of disruption of normal AM activity. The present study provides new information that helps define the mechanism by which pulmonary host defense is impaired in this setting. First, transient exposure of mice to hyperoxia leads to diminished expression of GM-CSF by AEC, both in vivo and ex vivo. Second, pharmacological treatment of mice with GM-CSF during hyperoxic exposure prevents impaired alveolar macrophage function. This preservation of AM function, in turn, is associated with restoration of the normal pulmonary host defense against K. pneumoniae, as reflected by decreased growth of organisms in the lung. Thus these studies indicate that GM-CSF plays a pivotal role in the impaired innate immune response in this model of lung injury.
High concentrations of inspired oxygen can be life saving in individuals with respiratory failure. However, exposure to very high concentrations of oxygen for extended periods may also result in lung injury and death. The features of oxygen-induced lung injury resemble those of other types of ALI (14). AEC and pulmonary capillary endothelial cells are injured, and cell death may result from either necrosis or apoptosis (3). The barrier function of the alveolar wall is compromised, and fluid leaks into the alveolar space, causing impaired gas exchange. Although brief periods of hyperoxia are tolerated, with return to normal lung function, more prolonged exposure to hyperoxia leads to respiratory failure and death. Because the features of hyperoxic lung injury are predictable and are quite similar to those found in other forms of ALI, hyperoxia has become a well-recognized model of ALI in mice. We have used limited exposure to hyperoxia as a clinically relevant model of oxidant stress and sublethal lung injury (1, 4). When C57BL/6 mice are placed in an atmosphere of >95% oxygen for 4 days, they demonstrate mild tachypnea but typically recover completely. However, this sublethal hyperoxia results in significant impairment of AM function, leading to increased susceptibility to lethal infection with K. pneumoniae. The extent to which the defect in pulmonary innate immunity in the setting of lung injury due to exposure to >95% oxygen might be a common feature of any diffuse lung injury, even in the absence of this degree of hyperoxia, has not yet been determined.
Alveolar macrophages harvested from mice exposed to sublethal hyperoxia demonstrate profound abnormalities in host defense function (1). These cells fail to recognize pathogens, due to loss of cell surface expression of TLR4, and fail to secrete critical early response cytokines (TNF and IL-6) following exposure to LPS. In addition to this defect in signaling to other cells, AM from hyperoxia-exposed mice demonstrate significantly impaired bacterial phagocytosis and killing. The functional abnormalities in AM following in vivo hyperoxia are quite similar to those found in cells from mice genetically deficient in GM-CSF. GM-CSF is a potent growth factor with wide-ranging effects on mononuclear phagocytes. It is a mitogen and survival factor for tissue macrophages. Studies with GM-CSF-deficient (GM-CSF−/−) mice have demonstrated a critical role for GM-CSF in regulation of AM maturation in the lung. In the absence of GM-CSF, AM metabolism of pulmonary surfactant is severely impaired, resulting in alveolar proteinosis (13, 18). AM from GM-CSF−/− mice demonstrate reduced phagocytic activity for microbes (25, 26, 31), impaired production of reactive oxygen intermediates in the respiratory burst (leading to defective bacterial killing) (22), and reduced TNF expression (25, 26, 31). Replacement of GM-CSF in the alveolar space by nebulized administration of recombinant protein (30), adenoviral gene transfer (38), or transgenic expression of GM-CSF under control of the surfactant protein C promoter (16, 26) restores AM function and reverses alveolar proteinosis and susceptibility to infection in GM-CSF−/− mice. These data confirm that exposure of AM to GM-CSF within the alveolar milieu is crucial for normal AM function.
We found that GM-CSF expression is significantly diminished in the setting of sublethal hyperoxic stress in vivo. GM-CSF normally is expressed in a tightly controlled fashion in the lung and is not detectable in normal BAL fluid. Therefore, to assess the impact of sublethal hyperoxia on alveolar GM-CSF expression, we used laser capture microdissection followed by real-time PCR to determine relative GM-CSF mRNA expression by cells of the alveolar wall. We found that hyperoxia significantly reduced baseline GM-CSF mRNA expression compared with normoxic controls. Although a number of different pulmonary cell types, including AM, may contribute GM-CSF to the alveolar space, AEC are a particularly prominent source of this growth factor. To specifically assess the impact of in vivo hyperoxia on AEC GM-CSF protein secretion, we measured GM-CSF protein in the culture supernatants of cytokine-stimulated AEC that had been isolated from hyperoxic or normoxic mice. The observation that ex vivo GM-CSF secretion was reduced in cells harvested from hyperoxic mice indicates that the mRNA data from laser capture microdissection indeed reflect changes in gene expression in AEC. The effect of hyperoxia on both GM-CSF expression and many aspects of AM function are reversible upon return of mice to normoxia (Fig. 1 and data not shown). Given the time these cells spend in culture after removal from the experimental and control mice, it is likely this approach may underestimate the true impact of hyperoxia on AEC GM-CSF protein expression in vivo.
AEC harvested from hyperoxic mice produce increased MCP-1 compared with cells from normoxic mice. This finding suggests that diminished AEC GM-CSF expression in the setting of sublethal hyperoxia is unlikely to be a manifestation of global cellular dysfunction, but instead is a consequence of hyperoxia-induced alterations in the cellular program of response to inflammatory cytokines. This circumstance is reminiscent of that found in GM-CSF mutant mice, which have significantly increased MCP-1 protein in BAL fluid, in association with increased numbers of AM recovered by lung lavage (25). Conversely, transgenic mice lacking CC-chemokine receptor 2, the receptor for MCP-1, have significantly increased GM-CSF in the lung (24). When activity of either of these AEC products is impaired, increased expression of the other molecule may be associated with preservation or increase in AM numbers. However, AM phenotype in the setting of reduced GM-CSF activity is clearly altered, as demonstrated by decreased numbers of cells staining positively for TLR4 and BS-1, and increased numbers of cells staining positively for CD11b. This pattern is consistent with a less well-differentiated state, similar to that found in AM from GM-CSF-deficient mice (25, 31).
Our studies suggest that continuous GM-CSF exposure may be necessary for maintenance of normal AM function. Even a relatively limited period of reduced AEC GM-CSF production, especially in the setting of hyperoxic stress, may be sufficient to impair macrophage function. Support for this conclusion is provided by the work of Bozinovski et al. (5, 6). These investigators found that a single intratracheal inoculation with neutralizing anti-GM-CSF antibody was sufficient to greatly reduce pulmonary inflammation in response to LPS and to decrease lung expression of inflammatory cytokines, including TNF (5, 6). Inhibition of lung inflammation measured 24 h after LPS administration was obtained when the neutralizing antibody was given before or up to 6 h after LPS (5). The consequences of neutralization of GM-CSF in the lung included decreased AM expression of TLR4 (5) and reduced activation of the transcription factors Akt, Erk1/2, NF-κB, and AP-1 in response to LPS (6). Together, these studies support the hypothesis that hyperoxia leads to impaired pulmonary host defense as a consequence of diminished AEC GM-CSF production, resulting in loss of vital support for maintenance of normal AM function.
In these studies, we have focused on the effects of GM-CSF treatment on AM function in the setting of in vivo hyperoxia. Although GM-CSF is a survival factor for PMNs, its major effects on functional maturation are in macrophages. Recent studies of infection due to another gram-negative pathogen, Pseudomonas aeruginosa, found increased susceptibility to pneumonia in GM-CSF−/− mice compared with wild-type controls, associated with major defects in AM function and limited alterations in PMN number or functional characteristics (2). Our initial studies of impaired pulmonary host defense following in vivo hyperoxia found only modest effects on PMN function. In the present work, we found no significant difference in PMN numbers in the lung in hyperoxia plus GM-CSF compared with hyperoxia alone at 12 h (data not shown) or at 24 h (Fig. 4). We found that hyperoxic stress had a profound effect on AM TNF expression and that treatment with GM-CSF resulted in restoration of TNF production by AM to that of cells from normoxic mice. Pulmonary expression of TNF is essential for normal innate immune response to K. pneumoniae (20, 21). Thus it is likely that AM expression of TNF plays a pivotal role in both the response to hyperoxia and its reversal with GM-CSF.
It is striking that short-term systemic treatment with GM-CSF is sufficient to preserve elements of normal AM function in the face of hyperoxia. In the setting of life-long, total deficiency of GM-CSF, local delivery of GM-CSF to the lung, by gene transfer, transgenic expression exclusively in the lung, or aerosol delivery of protein, is required to reverse the profound macrophage abnormalities (16, 30). However, in the acute setting, in which GM-CSF activity in the lung must be maintained over only a few days, systemic treatment is sufficient. In human studies of pulmonary alveolar proteinosis, a condition of relative GM-CSF deficiency due to endogenous neutralizing antibodies (34), subcutaneous administration of GM-CSF has resulted in restoration of normal AM behavior (19). We have used the same dose of GM-CSF (9 μg/kg, given subcutaneously) as was used in that human study. We have found previously that this approach to treatment with GM-CSF can influence AEC apoptosis in the setting of hyperoxia (27). Thus short-term systemic treatment with subcutaneous recombinant GM-CSF can affect the behavior of lung cells in biologically important ways. It is not yet clear whether the administered GM-CSF is simply affecting AM directly or is also acting through effects on AEC.
We found that treatment with GM-CSF had only a modest effect on clearance of K. pneumoniae from the lungs of normoxic mice, which did not achieve statistical significance. We evaluated a bacterial inoculum size that results in limited bacterial growth in normal C57BL/6 mice. It is unknown whether experiments using larger inocula, different pathogens, or other treatment regimens would unmask beneficial effects of GM-CSF in normal mice. Conversely, a number of studies have demonstrated that GM-CSF could augment host defense function in pathological states in which the function of macrophages from sites outside the lung is impaired. Short-term treatment with GM-CSF has been shown to counteract impaired innate immune function due to a variety of insults. Ex vivo exposure to GM-CSF has restored function in monocytes harvested from humans with septic shock (36) or from immunosuppressed liver transplant recipients (37). In a rat model of hemorrhagic shock, GM-CSF treatment in vivo has corrected tissue macrophage dysfunction (17). Our data are of particular interest because we have shown both that reduced local expression of GM-CSF is likely to account for AM dysfunction following sublethal hyperoxia and that treatment with GM-CSF largely corrects this defect.
The decrease in TNF-stimulated MIP-2 expression by AM following hyperoxic stress demonstrates that impaired macrophage signaling in AM from mice exposed to hyperoxia is not limited to responses initiated by TLR4. Direct responses to stimulation with TNF are also impaired. The mechanism by which this response to TNF is impaired in the setting of hyperoxia has not yet been defined. It particular, it has not been determined whether TNF receptors have been shed from the AM surface. However, this observation suggests that there is a global defect in innate immunity following hyperoxic stress that is not limited to a single category of pathogens recognized by specific pattern recognition receptors. It is of interest that treatment with GM-CSF restores the response to LPS, but does not restore the response to TNF. The consequences of continued impaired MIP-2 production for infection with organisms other than K. pneumoniae and whether this continued defect is a reflection of inadequate dose or duration of GM-CSF treatment have not yet been determined.
Limited data are available concerning GM-CSF expression in the lungs of humans with acute respiratory failure. In studies focused on lung neutrophil apoptosis in the setting of ALI, Matute-Bello and colleagues (23) evaluated growth factor expression in patients with ARDS. On days 1 and 3 after onset of ARDS, they found higher concentrations of GM-CSF in BAL fluid of patients who ultimately survived compared with patients who subsequently died. Thus these investigators found that increased pulmonary expression of GM-CSF was associated with a better outcome.
In conclusion, we have found that impaired pulmonary innate immune response due to sublethal hyperoxia is associated with diminished GM-CSF expression by AEC. Short-term treatment with recombinant GM-CSF during hyperoxia is sufficient to preserve normal AM function and decrease bacterial growth in the lungs. These studies point to a critical role for continuous GM-CSF activity in the lung in maintenance of the normal AM phenotype and demonstrate that effects of stress on AEC can alter the pulmonary microenvironment, with important consequences for AM function. Finally, they suggest that therapy with GM-CSF may have a role in the prevention of ventilator-associated pneumonia in humans with ALI or ARDS.
This work was supported by grants from the Department of Veterans Affairs (Merit Review Grants to R. Paine III and P. J. Christensen; and Research Enhancement Award Program) and from the National Heart Lung and Blood Institute (HL-64558 and SCCOR Grant HL-074024 to R. Paine III).
We thank Drs. Theodore Standiford and Bethany Moore for useful discussions and review of this manuscript. We also acknowledge helpful discussions of this work with members of the Pulmonary Research Enhancement Award Program at the Ann Arbor Veterans Affairs Medical Center, supported by the Department of Veterans Affairs.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society