INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Granulocyte-macrophage colony-stimulating factor (GM-CSF) was initially identified as a growth factor for promyelotic cells. Recombinant GM-CSF (Leucomax) is clinically used either to shorten neutropenia after chemotherapy in order to reduce the risk of infection, or in cases of bone marrow or stem cell transplantation in leukemic patients. Unexpectedly, GM-CSF-deficient mice showed no overt alterations in their leukocyte formation. However, a prominent defect in these mice was their development of an alveolar proteinosis (1). Consequently, it was shown that aerosolized GM-CSF ameliorates alveolar proteinosis in GM-CSF-deficient mice (2). In line with these animal data, after GM-CSF treatment arterial oxygenation and exercise capacity were increased in a single patient with alveolar proteinosis (3). Taken together, these findings provide strong evidence of an important role for GM-CSF in alveolar homeostasis.

In addition to these well-documented growth factor and regulatory properties for alveolar homeostasis, GM-CSF also has potent proinflammatory activity, which was mainly attributed to its ability to enhance the functional capacity of phagocytes. GM-CSF primes neutrophils and monocytes for enhanced production and release of cytokines, oxygen radicals, platelet-activating factor, and eicosanoids in response to endotoxin or other inflammatory stimuli (4). We showed previously in vivo in mice, that otherwise subtoxic doses of endotoxin become lethal after pretreatment with GM-CSF and that neutralization of endogenous GM-CSF protects mice from lethal endotoxic shock (5). These findings demonstrate that exogenous as well as endogenous GM-CSF functions as a potent proinflammatory mediator in the cytokine network response.

However, in these studies, the influence of GM-CSF on endotoxin-induced lung injury was not investigated. Previously, we have shown in blood-free perfused rat lungs that high doses of endotoxin cause bronchoconstriction that is dependent on induction of cyclooxygenase 2 and thromboxane formation (6). In view of the clinical use of GM-CSF and its lung-associated origin and functions, comparatively little is known about the effects of GM-CSF on pulmonary pathophysiology. The primary objective of this study was to examine the effects of GM-CSF on the pulmonary alterations induced by endotoxin exposure. To eliminate the effect of endotoxin on circulating leukocytes, the pulmonary responses to endotoxin were studied in isolated perfused rat lungs prepared from GM-CSF-pretreated rats.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Materials

Female Wistar rats (220-250 g body weight, Harlan-Winckelmann, Borchen, Germany) were used in all experiments. Vinblastine sulfate, cetyltrimethylammonium chloride, 3,3',5,5'-tetramethyl-benzidine liquid substrate solution (TMB), lipopolysaccharide (LPS) from Salmonella minnesota, and resorcinol were from Sigma-Aldrich Chemie (Steinheim, Germany); bovine serum albumin fraction V (BSA) was from Serva (Heidelberg, Germany); HEPES and tetramethylbenzidine were from Boehringer Mannheim (Mannheim, Germany); human recombinant G-CSF (Neupogen 48) was from Amgen (Munich, Germany); and SQ29.548 (5-heptenoic acid, 7-[3-[[2-[(phenylamino)carbonyl]hydrazino] methyl]-7-oxabicyclo[2.2.1]hept-2-yl]-,[1S-[1,2(Z),3,4]]- ) and NS-398 (methanesulfonamide, N-[2-[cyclohexyloxy]-4-nitrophenyl]) were from Cayman (Ann Arbor, MI). Pentobarbital sodium (Narcoren) was supplied by the Wirtschaftsgenossenschaft Deutscher Tierärzte (Hannover, Germany); streptavidin-peroxidase was from Jackson ImmunoResearch (West Grove, PA). Recombinant murine GM-CSF was a generous gift from G. R. Adolf (Bender Company, Vienna, Austria). Polyclonal sheep anti-mouse tumor necrosis factor (TNF) Ab was from our own laboratory (7).

Isolated Perfused Rat Lung Preparation

The rat lungs were prepared and perfused essentially as described previously (8). Briefly, female Wistar rats were anesthetized by intraperitoneal injection of pentobarbital sodium (16 mg/kg). Lungs were perfused at constant hydrostatic pressure (12 cm H₂O) through the pulmonary artery, which resulted in a flow rate of about 30 ml/min. For lung perfusion, we used Krebs-Henseleit buffer (37° C) that contains 2% albumin, 0.1% glucose, and 0.3% HEPES. The total recirculating amount of buffer was 100 ml. The lungs were suspended by the trachea and were ventilated by negative pressure ventilation with 80 breaths/min and a tidal volume between 1.8 and 2.2 ml in an artificial thorax chamber at 38° C. A hyperinflation (16 cm H₂O) was initiated every 5 min. Artificial thorax chamber pressure was measured with a differential pressure transducer (Validyne DP 45-24; Hugo Sachs Elektronik, March, Germany), and airflow velocity was measured with a pneumotachograph tube (Hugo Sachs Electronik) connected to a differential pressure transducer (Validyne DP 45-14; Hugo Sachs Elektronik). The lungs respired room air. The perfusate flow () (Narcomatic RT 500; Narco Biosystems, Houston, TX), the arterial (Pa) and venous pressure (Pv) (Isotec; Hugo Sachs Electronik), and the pH (MX3000; WTW, Weilheim, Germany) were continuously monitored. The pH of the perfusate before entering the lung was kept at 7.35 by automatic bubbling of CO₂ into the buffer when the pH exceeded this value (Uni-Werkstätten, Konstanz, Germany). Lung weight was continuously monitored by a specially constructed weight transducer (Hugo Sachs Elektronik). All data were transmitted to a computer (PCD-4RA; Siemens, Munich, Germany) via an A/D converter (Metrabyte 16; MBC, Taunton, MA) or an RS232 serial interface and analyzed by special software (programming language Asyst 3.1; S. Uhlig, Borstel, Germany). Chamber pressure, tidal volume (by electronic integration), perfusate flow, pulmonary artery pressure, pulmonary vein pressure, and lung weight were recorded simultaneously on a linear recorder (Graphtec WR 3310; Hugo Sachs Electronik). For calculation of lung mechanics, the data were analyzed by applying the following formula: P = (1/C)VT + RL(dVT/dt), where P is the chamber pressure, C is pulmonary compliance, VT is tidal volume, RL is pulmonary resistance, and t is time. Vascular resistance Rva was calculated as (Pa  Pv)/.

Experimental Protocol

All lungs were finally studied in the model of the isolated perfused rat lung described above. Prior to perfusion, some donor animals were pretreated with GM-CSF, G-CSF, vinblastine, or TNF Ab by intravenous injection into the tail vein. A dose of 50 of murine GM-CSF or human G-CSF per kilogram (both diluted in phosphate-buffered saline-0.1% human serum albumin [PBS-0.1% HSA]) was injected intravenously 60 min before starting lung perfusion. Vinblastine (0.75 mg/ kg) was injected intravenously 4 d before the perfusion experiments to cause severe neutropenia (9). Polyclonal sheep anti-mouse TNF Ab (500 µl) was injected intraperitoneally 90 min before starting lung perfusion; in these experiments TNF antibody was also added to the perfusion buffer (5%, vol/vol).

The start of perfusion was set as time t = 0 min. The substances SQ29.548 and NS-398 were diluted in ethanol and added to the perfusion buffer at t = 10 min or t = 30 min, respectively, to yield a final concentration of 10 µM. LPS from S. minnesota was dissolved in 500 µl of PBS-0.005% hydroxylamine and injected into the tube leading to the pulmonary artery. LPS was administered in two different concentrations at t = 40 min: LPS low (2 µg/ml) or LPS high (50 µg/ml). All perfusion experiments were carried out for 150 min, except those experiments in which GM-CSF (500 ng/ml) was added directly to the perfusate. The GM-CSF pretreatment period in these experiments lasted 100 min, leading to a total perfusion time of 210 min. An overview of all perfusion experiments that were done is given in Table 1. The LPS concentration of 2 µg/ml was chosen on the basis of preliminary experiments, in which we found that this concentration caused no changes in lung mechanics.

Measurement of Thromboxane

Samples taken from the perfusate (times indicated) were stored at 20° C. Thromboxane A₂ (TX) (t_1/2 = 30 s) was assessed as the stable by-product thromboxane B₂ (TXB₂) by enzyme immunoassay (EIA) (Cayman Chemicals) according to the manufacturer instructions. The cross-reactivity of the detecting antibody was TXB₂, 100%; 2,3-dinor-TXB₂, 8.2%; prostaglandins (e.g., PGD₂, PGE₂, 6-keto-PGF_1a), < 0.5%. Recovery of thromboxane in the perfusate was > 95%.

Measurement of Tumor Necrosis Factor

For TNF measurements, samples from the perfusate were stored at 70° C and quantified by sandwich enzyme-linked immunosorbent assay (ELISA). Antibody pairs for TNF were purchased from Endogen (Munich, Germany). Recombinant TNF (Bender) was used as standard. ELISA plates (Greiner, Nürtingen, Germany) were coated overnight at 4° C with coat Ab (50 µl/well) in 0.1 M NaHCO₃, pH 8.2. After blocking with PBS (200 µl/well) supplemented with 3% bovine serum albumin (BSA), pH 7.0, for 2 h at room temperature, the plates were washed twice with PBS-0.05% Tween 20. Sample (50 µl/well) and tracer antibody (50 µl/well) in PBS-3% BSA were added and incubated for 2 h at room temperature. After six wash cycles, plates were incubated for 30 min with streptavidin-peroxidase (1 µg/ml in PBS-3% BSA, 100 µl/well; Dianova, Hamburg, Germany). After eight washes, TMB liquid substrate solution (100 µl/well) was added and incubated at room temperature for 5 to 30 min. After addition of 50 µl of stop solution (1 M H₂SO₄) to each well, absorption was measured at 450 nm, using a reference wavelength of 690 nm, in an ELISA reader. The detection limit of the assay was 10 pg/ml.

Myeloperoxidase Extraction from Rat Lungs

Myeloperoxidase (MPO) activity was assayed according to a method of Schneider and Issekutz (10). Briefly, lungs were removed, washed in isotonic NaCl solution, dried on a paper towel, and weighed (0.14-0.23 g). The lungs were pooled and stored at 80° C until they were freeze-dried. For enzyme extraction, lyophilized samples were crushed with a stainless steel stick and homogenized in 2 ml of HEPES (50 mM, pH 8.0) with a 2-ml pestle homogenizer in glass tubes (size S; Braun Melsungen, Melsungen, Germany) with 12 passages at 3,000 rpm. Samples were spun at 10,000 × g for 30 min at 4° C, and the supernatant was discarded. The pellet was resuspended in 2 ml of 0.5% cetyltrimethylammonium chloride in distilled water, rehomogenized (three passages), and spun again as before. This resulted in a pellet with a clear supernatant and a thin lipid layer on the top. An aliquot of the clear supernatant was taken for MPO analysis and stored at 20° C.

MPO Detection in Lung Extracts

Lung extracts were diluted 1:5 in 0.5% cetyltrimethylammonium chloride. Aliquots of 75 µl of each sample were pipetted into four wells of a 96-well tissue culture plate (Greiner). Cold stop solution (4 N H₂SO₄) was added to two wells (150 µl/well) to stop the reaction at t = 0 min for measurement of background optical density (OD). The MPO-substrate solution was a TMB ready-to-use liquid substrate system containing also 120 µM resorcinol. Substrate solution (75 µl) was added to each well, the reaction was stopped after 2 min with 150 µl of cold stop solution, and the OD_450nm was determined in an ELISA reader.

Calculation of Total Number of Neutrophils in Rat Lungs

To obtain a standard curve for calculating the neutrophil content in the lungs based on the MPO enzyme activity, neutrophils were isolated from rat blood. Blood (7-8 ml) was withdrawn after heparin injection (500 IU) during preparation of the lungs for perfusion. One part of the heparinized blood was layered over one part Lympholyte-M medium (Cedar Lane, Hornby, ON, Canada). After centrifugation (1,750 × g, 20° C, 20 min) the supernatant was discarded. The pellet was washed twice with 40 ml of PBS-EDTA (1 mM) (370 × g, 4° C, 10 min). The erythrocytes were removed from the pellet by triple water lysis and centrifugation (270 × g, 4° C, 5 min): twice with 30 ml of ultrapure water, and 10 ml of 0.6 M KCl and once with 3 ml of ultrapure water and 1 ml of 0.6 M KCl. The resulting PMN pellet was resuspended in RPMI 1640. Viability was tested by the trypan blue exclusion method. We obtained an average neutrophil viability of 97%. Purity was higher than 93%, as judged by myeloperoxidase staining of cytocentrifuge preparations according to Kaplow (11). These isolated neutrophils were injected at different concentrations, into pieces (0.2 g) of noninflamed control lung, using a 0.5-ml syringe with a 22-gauge needle. Lung pieces were then lyophilized and extracted as described above for lung tissue. The different extracts were analyzed for MPO activity and plotted against number of neutrophils injected. This MPO standard curve was used to estimate the neutrophil content of the lung samples.

Reverse Transcription-Polymerase Chain Reaction Analysis

At the end of experiments, that is, after 150 min of perfusion, rat lungs were cut into pieces and stored in RNA-clean isolation solvent (AGS, Heidelberg, Germany) at 80° C. Total RNA was extracted from rat lung by using an Ultraturrax device (Jahnke & Kunkel, Staufen, Germany) and the acid guanidinium thiocyanate-phenol-chloroform method (12). After quantification, 1 µg of total RNA was reverse transcribed into cDNA with sequence-specific primers and the polymerase chain reaction (PCR) was performed as described (12). The primer 5'-A(G/C)AGCTCAGT(G/T)GA(A/G)CG(C/T)CT-3', complementary to the 3'-part of cyclooxygenase 2 (COX-2), was used for reverse transcription of the mRNA of this enzyme, and for -actin the primer 5'-CTAGAAGCATTTGCGGTGGAC-3' was used. After cDNA synthesis, excess primers were removed and PCR amplification was performed with the cDNA template and the following nested primer pairs: for COX-2 5'-ATCTAGTCTGGAGTGGGAGG-3' and 5'-AATGAGTACCGCAAACGCTT-3' were used, and for -actin 5'-CTAGAAGCATTTGCGGTGGAC-3' and 5'-CATCACCATTGGCAATGAGCG-3' were used. The reactions were cycled 32 times with a cycle profile of 30 s at 94° C, 30 s at 56° C, and 30 s at 72° C after a 5-min denaturing step at 95° C. Amplification products were analyzed by 1.8% agarose gel electrophoresis and ethidium bromide staining. No amplification products were found when the reverse transcription was performed without specific primer or when the PCR was carried out without template. The identity of the fragments was evaluated by their molecular mass or sequencing. Furthermore, in pilot experiments, samples were assayed in various dilutions to ensure proportionality in the yield of PCR products. For -actin detection, the template was diluted 2,500-fold and for COX-2 it was diluted 500-fold. The proportion of PCR products was estimated relative to the control fragment of -actin by measuring the intensity of ethidium bromide luminescence with a CCD image sensor in combination with the BIOPROFIL program (LTF, Wasserburg, Germany), and data are presented as arbitrary units.

Statistics

Data in the figures are given as means ± SEM, while data in the tables and in the text are given as means ± SD. Data were analyzed by one-way analysis of variance (ANOVA), and in case of differences among the groups, the Dunnett multiple comparison test was performed to test which group was different from the respective control group (Prism 3.0; GraphPad Software, San Diego, CA). p < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

GM-CSF- and LPS-induced Bronchoconstriction

The isolated and perfused rat lung (IPL) was used to study whether and how colony stimulation factors affect LPS-induced lung damage. Lungs from rats treated with GM-CSF alone (50 µg/kg) showed no differences compared with lungs from sham-injected animals with respect to pulmonary resistance, pulmonary compliance, tidal volume, vascular resistance, or lung weight over a period of 150 min. Perfusion of rat lungs with LPS alone at 50 µg/ml increased pulmonary resistance as described previously (13), while perfusion with LPS at 2 µg/ml was the highest LPS concentration that caused almost no change in this parameter compared with lungs perfused with solvent alone (Figure 1, Table 2). In lungs from rats that were pretreated with GM-CSF (50 µg/kg) in vivo 60 min before starting the perfusion, an impairment of lung function was observed already with the low and otherwise ineffective LPS concentration of 2 µg/ml (Figure 1). In these experiments (GM-CSF/LPS), RL began to rise 30 min after LPS infusion and reached a plateau after 130 min. Similarly airway compliance and tidal volume decreased at this time (Table 2). On the other hand, vascular resistance and lung weight (W) did not change significantly under any condition (Table 2). These experiments show that GM-CSF pretreatment leads to an enhanced susceptibility of rat lungs to LPS-induced toxicity.

View larger version (22K): [in this window] [in a new window]   Figure 1.   Time course of airway resistance in isolated and ventilated perfused rat lungs after granulocyte-macrophage colony-stimulating factor (GM-CSF) and lipopolysaccharide (LPS) administration. Rats were pretreated with GM-CSF (50 µg/kg, intravenous) 60 min prior to perfusion and stimulated at t = 40 min with LPS (2 µg/ml) by addition to the perfusion buffer (GM-CSF/LPS) (n = 11). Controls were either left untreated (n = 13), pretreated with GM-CSF alone (n = 8), or perfused only with LPS (2 µg/ml) (n = 10). RL/RL40min: airway resistance was normalized to time t = 40 min and was continuously monitored. Data are expressed as means ± SEM, with the number of individual experiments (n) in parentheses.

To examine whether the exacerbation of the LPS-induced bronchoconstriction by GM-CSF was due to a direct interaction of this factor with lung cells, GM-CSF was perfused at the high concentration of 500 ng/ml 100 min prior to LPS. Under these conditions, RL was not enhanced after a total perfusion time of 210 min, and also C and VT decreased only slightly, probably because of the long time of perfusion (Table 2). This finding indicates that GM-CSF is unlikely to act directly on lung tissue but rather initiates extrapulmonary processes sensitive to LPS.

G-CSF- and LPS-induced Bronchoconstriction

G-CSF is a growth factor specific for neutrophils and, in contrast to GM-CSF, does not prime cells from the monocytic lineage (4). Like GM-CSF, G-CSF (recombinant human G-CSF [rhu G-CSF], 50 µg/kg, intravenous) pretreatment alone did not affect the lung functions investigated, as shown in Table 2. However, when lungs obtained from G-CSF-pretreated rats were subsequently perfused with LPS at 2 µg/ml, RL increased (Table 2). Like GM-CSF/LPS, G-CSF/LPS also decreased airway compliance and tidal volume (Table 2). These data show that G-CSF sensitizes the lung for LPS-induced bronchoconstriction.

RT-PCR Analysis of Lung Tissue for COX-2 mRNA

An increased expression of COX-2 mRNA was found in isolated perfused rat lungs after infusion of LPS at 50 µg/ml. Subsequent TX release resulted in bronchoconstriction (6). To investigate whether GM-CSF/LPS-induced bronchoconstriction also depends on COX-2, mRNA expression in lung tissue was measured with RT-PCR at the end of the perfusion experiments (Figure 2). As seen in controls, LPS infusion at the subtoxic dose of 2 µg/ml alone elicited a modest but not significant increase in COX-2 mRNA expression. However, COX-2 mRNA expression was significantly elevated in GM-CSF-pretreated rat lungs (p < 0.01). Treatment with GM-CSF and LPS caused a further increase in COX-2 mRNA expression (p < 0.001 versus control). This indicates that GM-CSF stimulates COX-2 expression, which can be further enhanced by an otherwise nearly ineffective dose of LPS.

View larger version (41K): [in this window] [in a new window]   Figure 2.   Analysis of COX-2 mRNA expression in rat lung tissue by RT-PCR after various treatments. (A) Rats were pretreated with GM-CSF (50 µg/ kg, intravenous) 60 min prior to perfusion and stimulated at t = 40 min with LPS (2 µg/ml) by addition to the perfusion buffer (GM-CSF/LPS) (n = 5). Controls were either left untreated (n = 3), pretreated with GM-CSF alone (n = 5), or perfused only with LPS (2 µg/ml) (n = 4). Shown is the ratio between COX-2 and -actin message. The cDNA dilution was as follows: COX-2, 1:500; -actin, 1:2,500. Data are expressed as means ± SEM, with the number of individual experiments (n) in parentheses. *p < 0.01, **p < 0.001 versus control. (B) Representative image of COX-2 and -actin amplification products. Amplification products were analyzed by 1.8% agarose gel electrophoresis and ethidium bromide staining.

Thromboxane Release

Because TX is known to be responsible for bronchoconstriction induced with LPS at 50 µg/ml (6), TX concentrations in the perfusion buffer obtained from the experiments described above were measured by EIA. The time course of the TX concentrations in the perfusate for the experiments with GM-CSF is shown in Figure 3 and those with G-CSF in Figure 4. TX concentrations in control experiments remained almost constant over 150 min (26 ± 15 pg/ml). A modest increase in TX release was observed after 150 min in rat lungs perfused with LPS alone at 2 µg/ml (134 ± 72 pg/ml) and in experiments with GM-CSF pretreatment alone (151 ± 150 pg/ml). In GM-CSF/LPS experiments, TX concentrations started to rise 40 min after LPS administration and were significantly elevated from 50 min after LPS administration until the end of the experiment at t = 150 min (293 ± 155 pg/ml; p < 0.01 versus control). In the G-CSF/LPS experiments perfusate TX concentrations were significantly elevated (284 ± 245 pg/ml, p < 0.05 versus control), but only modestly in lungs perfused with G-CSF alone (98 ± 67 pg/ml). These results indicate that the effects of G(M)-CSF pretreatment and LPS on TX liberation into the perfusate are additive.

View larger version (21K): [in this window] [in a new window]   Figure 3.   Time course of GM- CSF/LPS-induced release of TX into the perfusate of isolated perfused and ventilated rat lungs. Rats were pretreated with GM-CSF (50 µg/kg, intravenous) 60 min prior to perfusion and stimulated at t = 40 min with LPS (2 µg/ml) by addition to the perfusion buffer (GM-CSF/LPS) (n = 12). Controls were either left untreated (n = 8), pretreated with GM-CSF alone (n = 9), or perfused only with LPS (2 µg/ml) (n = 11). Data are expressed as means ± SEM, with the number of individual experiments (n) in parentheses. *p < 0.01 versus control.

View larger version (19K): [in this window] [in a new window]   Figure 4.   Time course of G-CSF/LPS-induced release of TX into the perfusate of isolated perfused and ventilated rat lungs. Rats were pretreated with GM-CSF (50 µg/kg, intravenous) 60 min prior to perfusion and stimulated at t = 40 min with LPS (2 µg/ml) by addition to the perfusion buffer (GM-CSF/LPS) (n = 4). Controls were either left untreated (n = 8), pretreated with GM-CSF alone (n = 4), or perfused only with LPS (2 µg/ml) (n = 11). Data are expressed as means ± SEM, with the number of individual experiments (n) in parentheses. *p < 0.05 versus control.

TNF Release

TNF is considered to be responsible for the priming effects of GM-CSF in vivo, thus amplifying LPS toxicity in mice (5). Therefore, the TNF concentrations in the perfusate were measured under the various experimental conditions. In GM-CSF/ LPS experiments (Figure 5), TNF levels in the perfusate buffer were elevated already 20 min (t = 60 min) after LPS infusion (p < 0.05 versus control). At t = 150 min, the TNF concentrations in the perfusate reached 3,500 ± 1,000 pg/ml (p < 0.001 versus control, 400 ± 320 pg/ml). Although both GM-CSF (1,670 ± 1,180 pg/ml) and LPS alone (1,470 ± 1,350 pg/ ml) also appeared to increase TNF perfusate levels, this effect was not significant. Almost no increase in TNF levels was seen in the G-CSF (680 ± 510 pg/ml) and G-CSF/LPS (890 ± 550 pg/ ml) experiments (Figure 6). These data indicate that only GM-CSF pretreatment significantly enhances LPS-triggered TNF release into the perfusate.

View larger version (21K): [in this window] [in a new window]   Figure 5.   Time course of GM-CSF/LPS-induced release of tumor necrosis factor (TNF) into the perfusate of isolated perfused and ventilated rat lungs. Rats were pretreated with GM-CSF (50 µg/kg, intravenous) 60 min prior to perfusion and stimulated at t = 40 min with LPS (2 µg/ml) by addition to the perfusion buffer (GM-CSF/LPS) (n = 12). Controls were either left untreated (n = 8), pretreated with GM-CSF alone (n = 9), or perfused only with LPS (2 µg/ml) (n = 11). Data are expressed as means ± SEM, with the number of individual experiments (n) in parentheses. *p < 0.05, **p < 0.01 versus control.

View larger version (18K): [in this window] [in a new window]   Figure 6.   Time course of G-CSF/LPS-induced release of TNF into the perfusate of isolated perfused and ventilated rat lungs. Rats were pretreated with G-CSF (50 µg/kg, intravenous) 60 min prior to perfusion and stimulated at t = 40 min with LPS (2 µg/ml) by addition to the perfusion buffer (G-CSF/ LPS) (n = 4). Controls were either left untreated (n = 8), pretreated with G-CSF alone (n = 4), or perfused only with LPS (2 µg/ml) (n = 11). Data are expressed as means ± SEM, with the number of individual experiments (n) in parentheses.

Role of COX and TNF

To clarify the role of TX in GM-CSF/LPS-induced bronchoconstriction, the TX receptor antagonist SQ29.548 and the selective cyclooxygenase 2 inhibitor NS-398 were used as pharmacological tools. NS-398 (10 µM) administered 10 min prior to LPS completely blocked GM-CSF/LPS-induced bronchoconstriction (Figure 7A) and TX release (Figure 7B), but not TNF release (Figure 7C). SQ29.548 (10 µM), administered 10 min after starting the perfusion, blocked bronchoconstriction (Figure 7A) without affecting TX (Figure 7B) and TNF concentrations (Figure 7C). These findings indicate that GM-CSF/ LPS-induced bronchoconstriction is COX-2 dependent and TX receptor mediated.

View larger version (21K): [in this window] [in a new window]   View larger version (16K): [in this window] [in a new window]   View larger version (20K): [in this window] [in a new window]   Figure 7.   Involvement of TX, COX-2, and TNF in GM-CSF/LPS-induced bronchoconstriction. All groups, except the control group (n = 8), were challenged with GM-CSF and LPS. GM-CSF/LPS: GM-CSF was given intravenously at a dose of 50 µg/kg 60 min prior to perfusion. LPS (2 µg/ml) was infused 40 min after starting perfusion (n = 12). NS-398: The COX-2 inhibitor NS-398 (10 µM) was given 10 min prior to LPS via perfusion buffer (n = 4). SQ29.548: The thromboxane receptor antagonist SQ29.548 (10 µM) was given 30 min prior to LPS infusion (n = 3). Anti-TNF Ab: Neutralizing anti-TNF Ab (500 µl) was injected intraperitoneally in vivo 90 min before starting perfusion and, in addition, neutralizing anti-TNF Ab was dissolved at 5% (vol/vol) in the perfusion buffer (n = 3). Airway resistance (RL) (A) was monitored continuously. Here end points (110 min after LPS infusion) are shown. TX (B) and TNF (C) concentrations were measured by ELISA (110 min after LPS infusion). Data are expressed as means ± SEM, with the number of individual experiments (n) in parentheses. **p < 0.01, *p < 0.05 versus control. RL/RL40min: Airway resistance normalized to time t = 40 min.

To test whether TNF is involved in GM-CSF/LPS-induced lung failure, endogenously formed TNF was neutralized by an anti-TNF Ab. Anti-TNF Ab (500 µl) was injected intraperitoneally 90 min before starting the perfusion and anti-TNF Ab was also added to the perfusion buffer at 5% (vol/vol). Although TNF release was largely inhibited (760 ± 130 pg/ml) compared with GM-CSF/LPS experiments (3,500 ± 1,000 pg/ ml) (Figure 7C), this had no effect on either elevated TX concentrations (354 ± 23 pg/ml) (Figure 7B) or bronchoconstriction (RL = 0.46 ± 0.03 cm H₂O · s/ml) (Figure 7A). We conclude from these data that elevated TNF release does not contribute to GM-CSF/LPS-induced bronchoconstriction.

Involvement of Neutrophils

Our observation that both GM-CSF and G-CSF primed rat lungs for LPS-induced bronchoconstriction suggests that activation of neutrophils might be responsible for this effect. To quantitatively assess neutrophil invasion into lung tissue after pretreatment with GM-CSF or G-CSF in vivo, a neutrophil myeloperoxidase (MPO) assay was used. A massive invasion of neutrophils into the lung tissue occurred after GM-CSF or G-CSF pretreatment (Figure 8A). Subsequent perfusion with LPS had no influence on the number of infiltrated neutrophils in the lungs: GM-CSF (n = 4), 9.8 (± 4.9) × 10⁶ neutrophils/g lung wet weight (g LW); GM-CSF/LPS (n = 4) 9.5 (± 2.6) × 10⁶ neutrophils/g LW; G-CSF (n = 6), 7.9 (± 1.9) × 10⁶ neutrophils/g LW; and G-CSF/LPS (n = 4), 8.1 (± 1.0) × 10⁶ neutrophils/g LW. These findings indicate that neutrophils are invading lung tissue as a consequence of G- or GM-CSF pretreatment, and remain attached to the lung vasculature throughout our experiments.

View larger version (26K): [in this window] [in a new window]   View larger version (27K): [in this window] [in a new window]   Figure 8.   Neutrophil invasion and bronchoconstriction in isolated perfused and ventilated rat lungs after GM-CSF and G-CSF pretreatment. Rats were pretreated with GM-CSF or G-CSF (50 µg/kg, intravenous) 60 min prior to perfusion and stimulated at t = 40 min with LPS (2 µg/ml) by addition to the perfusion buffer (GM-CSF/LPS, G-CSF/LPS). Controls were either left untreated, pretreated with GM-CSF or G-CSF alone, or perfused only with LPS. Vinblastine was injected intravenously 4 d before the perfusion experiments at a dose of 0.75 mg/kg to cause severe neutropenia. Vin./GM-CSF/LPS: Vinblastine pretreatment was combined with GM-CSF/LPS. Vin./LPS 5: Vinblastine pretreatment was combined with LPS (50 µg/ml) infusion solely, t = 40 min. (A) Lungs were processed at the end of the perfusion (t = 150 min) to estimate neutrophil content in lung tissue by the myeloperoxidase assay. Data are expressed as means ± SEM; number of individual experiments n = 4 except for G-CSF (n = 6). *p < 0.05, **p < 0.01 versus control. (B) Airway resistance (RL) was continuously monitored. End points at t = 150 min are presented. RL/RL40min: Airway resistance normalized to time t = 40 min. Data are expressed as means ± SEM, with the number of individual experiments (n) in parentheses. GM-CSF/LPS (n = 11), G-CSF/LPS (n = 4), controls (n = 13), GM-CSF (n = 8), G-CSF (n = 4), LPS (n = 10), Vin./GM-CSF/LPS (n = 4), Vin./LPS 5 (n = 4). *p < 0.05, **p < 0.01 versus control.

To clarify further the relationship between neutrophil invasion and amplification of the LPS-induced bronchoconstriction, rats were made neutropenic by vinblastine treatment (0.75 mg/kg, intravenous, 4 d before onset of the experiments). In neutropenic rats [vinblastine treated: 0.3 (± 0.2) × 10⁶ neutrophils/g LW, n = 4; control: 1.0 (± 0.7) × 10⁶ neutrophils/g LW, n = 4; LPS alone: 0.5 (± 0.1) × 10⁶ neutrophils/g LW, n = 4] pretreated with GM-CSF no neutrophil invasion (Figure 8A) and no bronchoconstriction were seen after infusion of LPS at 2 µg/ml, that is, the low concentration in our setting (Figure 8B). These experiments emphasize that neutrophil invasion is indispensable for GM-CSF/LPS- or G-CSF/LPS-induced bronchoconstriction.

In contrast, if lungs made neutropenic by vinblastine treatment [0.3 (± 0.2) × 10⁶ neutrophils/g LW, n = 4] were perfused with a high concentration of LPS (50 µg/ml), pulmonary resistance increased to the same extent as in lungs from control rats, that is, LPS (50 µg/ml) infusion without vinblastine pretreatment (Figures 8A and 8B). This experiment indicates that neutropenia does definitively not prevent bronchoconstriction induced by high LPS concentrations. This means that invading neutrophils are not necessary to elicit bronchoconstriction at high LPS concentrations but are sufficient to amplify bronchoconstriction induced with a subtoxic dose of LPS. In other words, a neutrophil-triggered mechanism is likely to be the cause for enhanced LPS-induced TX release in G-CSF/ LPS- or GM-CSF/LPS-induced bronchoconstriction.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Previous studies of the effects of GM-CSF in endotoxic or septic lung injury have focused on leukocyte counts, edema formation, and pulmonary hypertension, but not on airway responses. To our knowledge, GM-CSF has not been studied in this context previously. Here, we show that pretreatment with G-CSF or GM-CSF enhances the airway response to endotoxin. Perfusion of rat lungs obtained from G-CSF- or GM-CSF-treated rats with low and otherwise ineffective LPS doses caused bronchoconstriction to an extent similar to that evoked by higher doses of LPS in lungs from control rats. Because the GM-CSF/LPS-induced bronchoconstriction was accompanied by the release of thromboxane and was completely abrogated by the specific TX receptor antagonist SQ29.548 and by the specific COX-2 inhibitor NS-398, we conclude that this effect is COX-2 dependent and TX receptor mediated. Further insight into the mechanism of GM-CSF priming was provided by the finding that GM-CSF caused an accumulation of neutrophils in the lungs and that the priming effect of GM-CSF was absent in neutropenic animals. These data show that intravenously injected GM-CSF causes accumulation of neutrophils in the lungs, where these neutrophils increase the sensitivity of the lungs to LPS-induced pulmonary alterations.

The accumulation of neutrophils in the lungs of GM-CSF-treated animals was accompanied by an increase in COX-2 mRNA and by a minor increase in TX release, but not by bronchoconstriction. We tentatively conclude from this finding that GM-CSF not only recruited neutrophils to the lungs, but also primed them without fully activating these neutrophils. Although we have no direct evidence that the recruited neutrophils express COX-2, this hypothesis is supported by the findings of Pouliot and colleagues that GM-CSF stimulates COX-2 expression in neutrophils (14). However, because LPS can induce COX-2 expression in practically neutrophil-free (13) lungs obtained from control rats (6, 15), it is evident that parenchymal lung cells can also express COX-2 (15). The fact that Ermert and coworkers (15) found increased COX-2 expression at lower LPS concentrations than we did might be explained by the different rat strains (Wistar rats versus Sprague-Dawley rats) and perfusion buffers (2% albumin versus 1.5% plasma) used. Although the extent to which neutrophils and parenchymal lung cells contribute to the final thromboxane release is unknown at present, our data clearly show that COX-2-dependent thromboxane formation can occur dependently as well as independently of neutrophils. The mechanism by which the sequestered neutrophils amplify COX-2 expression, TX generation, and subsequently bronchoconstriction remains elusive. Possible mechanisms might include direct action of PGH₂ (14), derived from GM-CSF-primed neutrophils on TX receptors (16), leading to bronchoconstriction, or the stimulation of lung cells by products from activated neutrophils such as reactive oxygen species (17).

GM-CSF is also known to prime cells for increased TNF mRNA production (18) and to amplify LPS-induced toxicity via TNF (5). Thus, the GM-CSF/LPS-induced bronchoconstriction could possibly have been mediated by TNF. However, the lack of any effect of TNF neutralization (5) makes this hypothesis unlikely. The finding that both TX release and bronchoconstriction remained unchanged indicates that TNF release is, rather, a bystander effect, even though TNF has been reported to upregulate COX-2 mRNA in airway smooth muscle cells (19) and in pulmonary artery smooth muscle cells (20). In addition, perfusion with exogenous TNF did not potentiate the LPS response, that is, no bronchoconstriction was seen (data not shown). The fact that G-CSF potentiated the LPS-induced bronchoconstriction without affecting TNF concentrations provides further evidence for the conclusion that TNF is not involved in amplifying LPS-induced lung failure. We conclude from these findings that TNF as such is not responsible for the growth factor-induced sensitization toward LPS-induced bronchoconstriction.

The rationale for the G-CSF experiments was to use a hematopoietic growth factor for which, in contrast to GM-CSF, antiinflammatory properties have been described (21). The novel finding that G-CSF, like GM-CSF, enhanced TX release and amplified LPS-induced bronchoconstriction, suggests a common mechanism in both instances. Because G-CSF, in contrast to GM-CSF, acts predominantly on neutrophils, these cells are likely to be the main effector cells. The failure of GM-CSF perfused through the lungs in high concentrations to sensitize lungs toward LPS (see resistance in Table 2), further suggests that extrapulmonary cells are needed to elicit GM-CSF/LPS-induced lung failure. The decrement in tidal volume seen in these experiments can be explained by the prolonged perfusion time of 210 min, as the normal functional loss in tidal volume in the perfused rat lung model is about 8%/h (13). These results are consistent with the virtual absence of neutrophils from blood-free perfused lungs in our model (Figure 8A [13]).

In vivo, lungs contain a large pool of neutrophils, which are primarily located in the vascular capillary bed. Whereas the average diameter of the pulmonary capillary segments is only 5 µm, that of neutrophils is about 7 µm. This size difference and the stiffness of the neutrophils influence neutrophil sequestration in the lungs, because most neutrophils must deform to pass through the smaller pulmonary capillaries. Yong and coworkers showed that the deformability of neutrophils in the presence of GM-CSF is impaired after 30-120 min, thus leading to neutrophil accumulation in the lung (22). Inano and coworkers demonstrated a similar mechanism for G-CSF, when studying neutrophil sequestration in rabbit lungs (23). In addition, GM-CSF and G-CSF stimulate upregulation of CD11b/ CD18 on neutrophils, thereby enhancing adhesion to pulmonary endothelial cells (22, 23). Because neutrophil counts were elevated in lungs from both GM-CSF- and G-CSF-pretreated rats, this is most likely part of the common priming effect of the growth factors toward LPS-induced bronchoconstriction. Finally, the neutrophil depletion experiments with vinblastine provide the causal connection between neutrophil accumulation in the lung tissue and GM-CSF-amplified lung failure. In addition, these experiments help to clarify the role of neutrophils in LPS-elicited bronchoconstriction in general. Although they showed no sensitization toward low LPS concentrations with G-CSF or GM-CSF, lungs obtained from neutropenic rats still showed a fully expressed bronchoconstriction in response to a high LPS concentration, indicating that neutrophils are not essential for eliciting this type of pulmonary response. Thus, LPS can cause bronchoconstriction dependent on (24) and independent of neutrophils. Because GM-CSF is known to be produced in response to LPS in vivo (5, 25) both mechanism may be relevant in vivo.

Our present findings suggest that by recruiting neutrophils into lungs, GM-CSF and G-CSF may exacerbate acute lung injury. In line with this, it was reported that G-CSF enhanced mortality and edema formation after intrabronchial application of Escherichia coli (26, 27) in rats. However, paradoxically in other instances, G-CSF was reported to have beneficial effects. For instance, also in rats, G-CSF reduced mortality in models of intrabronchial Staphylococcus aureus (26), intravenous Klebsiella pneumoniae (28), or intraabdominal sepsis (29). In addition, G-CSF protected mice from LPS-induced mortality (21), while in contrast GM-CSF aggravated it (5). Furthermore, dependent on the design of the study G-CSF potentiated (30) or ameliorated (31, 32) LPS- or E. coli-induced pulmonary edema formation in rats or guinea pigs. Apparently, growth factors can influence the outcome of sepsis in different ways, depending on the time of intervention and type of infection. The results of this study give rise to reservations regarding the use of GM-CSF under conditions of activation of the immune system by either infectious disease or the presence of bacterial components such as LPS, for example, in the case of antibiotic therapy.

In conclusion, this study demonstrates that GM-CSF as well as G-CSF can cause neutrophil accumulation in lung tissue, which is accompanied by increased COX-2 expression. If the organ is then exposed to LPS, an elevated TX synthesis and release leads to a strong bronchoconstriction. This series of events might have bearing also on humans under clinical conditions. The pulmonary side effects (33) noticed during the clinical use of GM-CSF and G-CSF as recombinant drugs in cancer, or in combination with antibiotics in sepsis patients (36), might in part be explained by mechanisms discussed in this animal study.