In Vitro Tumor Necrosis Factor-a Secretion by Monocytes from Patients with Cystic Fibrosis J. Stuart Elborn, Debbie Norman, Finola M. Delamere, and Dennis J. Shale Respiratory Medicine Unit, University of Nottingham, City Hospital, Nottingham, United Kingdom
Immunoreactive tumor necrosis factor-a (TNF-a) concentration is increased in plasma from patients with cystic fibrosis and chronic Pseudomonas aeruginosa pulmonary infection. To determine if circulating monocytes could be the source of plasma TNF-a, we determined in vitro basal and endotoxin-stimulated TNF-a secretion by monocytes. In 10 adult patients studied at the time of a symptomatic respiratory exacerbation, basal secretion of TNF-a was significantly less than that for 10 matched healthy controls (median 265 pgl/-tg DNA, nonparametric 95 % confidence interval 193 to 463 pg//-tg DNA versus 575, 298 to 923 pgl/-tg DNA; P < 0.006), although both groups responded equally effectively to added Escherichia coli endotoxin at ~ 25 ng/mI. In six patients and six matched controls, monocyte culture was repeated after completion of2 wk anti-pseudomonal antibiotic treatment in the patients. The reduced basal TNF-a secretion in the patients had reversed and was not significantly different to that of controls. This effect mirrored a significant reduction in plasma immunoreactive TNF-a in these patients (mean ± SD, 258 ± 59.3 pg/mI pretreatment versus 133 ± 47.8 pg/rnl post-treatment; P < 0.05). These findings suggest that a reversible downregulation of TNF-a secretion occurred at the time of a symptomatic respiratory deterioration in the presence of chronic P. aeruginosa infection. This may represent a physiologic regulatory mechanism to maintain a local inflammatory response to chronic pulmonary infection in cystic fibrosis.
The main factor determining survival in cystic fibrosis (CF) is progressive lung destruction secondary to chronic Pseudomonas aeruginosa infection. Infection with P. aeruginosa is associated with a local inflammatory response that is responsible for lung injury (1). There is evidence of a systemic component to the host response with increased circulating neutrophil granule products, increased acute phase proteins, and increased immunoreactive tumor necrosis factor-o/cachectin (TNF-a) (2, 3). The plasma concentration of immunoreactive TNF-a is increased in patients with CF and chronic P. aeruginosa lung infection (2, 3), although some centers have been unable to confirm this (4). Moreover, there is a reduction in plasma immunoreactive TNF-a levels after specific antibiotic treatment (3). TNF-a produced by cells of the monocyte-macrophage lineage is an important regulator of the inflammatory response to bacterial infection and endotoxemia (5, 6). The wide range of effects of TNF-a on a variety of cells influences the magnitude of the inflammatory response and (Received for publication May 2. 1991) Address correspondence to: D. 1. Shale, M.D., M.R.C.P., Respiratory Medicine Unit, University of Nottingham, City Hospital, Hucknall Road, Nottingham NG5 IPB, United Kingdom. Abbreviations: cystic fibrosis, CF; fetal calf serum, FCS; lipopolysaccharide, LPS; phosphate-buffered saline, PBS; tumor necrosis factor-a, TNF-a. Am. J. Respir. Cell Mol. BioI. Vol. 6. pp. 207-211, 1992
its duration (7). The plasma concentration of bioactive TNF-a is increased in acute infective disorders, such as meningococcal septicemia and malaria. In these acute septicemia states, the magnitude of the initial plasma TNF-a concentration and those of interleukin (IL)-1 and IL-6 are predictive of a poor outcome (8-10). The pathophysiologic role of circulating TNF-a in localized infection, such as in CF, is unknown. There are a variety of potential sources of circulating TNF-a in CF. The most obvious is the lung, although the precise origin from within the lung is unknown it is likely to be from macrophages in the alveolar, interstitial, or intravascular compartments (11, 12). Investigation of the source is hampered because pulmonary macrophages are largely inaccessible and the alveolar cell profile in such patients is dominated by the neutrophil (13). The presence of circulating Pseudomonas antigen-antibody complexes, which includes complexed endotoxin, in patients with chronic P. aeruginosa infection raises the possibility that the monocyte or other macrophage cells, such as the Kupffer cell, may be contributing to or be the source of circulating TNF-a (14, 15), whereas downregulation ofIL-l secretion by monocytes from patients with CF and chronic P. aeruginosa infection suggests suppression of secretion by inflammatory cells not in the infected pulmonary compartment (16). We studied the circulating monocyte, an accessible precursor of pulmonary macrophages, to determine the state of TNF-a secretory activity in CF patients chronically infected with P. aeruginosa. The in vitro secretion of TNF-a, in the presence and absence
208
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 6 1992
TABLE 1
Details of lung function, Shwachman score, and treatment in 10 patients with cystic fibrosis at the time of a symptomatic deterioration of respiratory function Patient No.
Age (yr)
Weight (kg)
Shwachman Score
No. of Years Colonized with P. aeruginosa
1 2 3 4 5 6 7 8 9 10
16 19 24 21 26 20 19 23 19 25
38.2 61.1 48.3 34.7 70.0 53.7 41.7 44.2 43.3 49.8
45 70 52 43 75 55 40 62 45 60
6 4.5 8 9.5 5 2 8 2 9 2
FEV,
FVC
PEFR
(% predicted; liter)
(% predicted; liter)
(% predicted; literls)
0.7 (26) 1.76 (42) 2.09 (57) 0.68 (17) 3.55 (70) 2.43 (46) 1.49 (34) 2.14 (68) 0.62 (15) 1.78 (51)
186 (52) 360 (75) 312 (74) 114(18) 306 (70) 348 (49) 150 (30) 252 (65) 168 (28) 252 (61)
0.56 1.36 1.27 0.42 1.80 1.47 1.03 1.03 0.43 1.30
(22) (35) (40) (12) (42) (33) (27) (38) (15) (43)
Intravenous Treatment'
A A A
C C C C C C A
• A = aztreonam; C = ceftazidime.
of added endotoxin, was determined both before and after antibiotic treatment of a respiratory exacerbation.
Materials and Methods Patients Ten consecutive patients (mean age, 21 yr; range, 16 to 26 yr) with a symptomatic respiratory exacerbation of their CF (five males) were studied. All had a positive sweat test (Na", Cl- > 60 mmol/liter) and significant respiratory disease with P. aeruginosa chronically cultured from sputum (>3 times/yr). At the time of the study, all had increased respiratory symptoms and a reduction in pulmonary function defined as a fall in FEV 1 > 15% of the best value recorded in the preceding 6 mo (Table 1). Before the first intravenous dose of antibiotics, blood was drawn at 9:00 A.M. for monocyte studies and plasma TNF-a determination. In six patients, a further sample was obtained after 2 wk of intravenously administered antibiotics, when pulmonary function, body weight, and 24-h sputum weight were repeated (Table 1). Ten sex- and age-matched to within 5 yr, apparently healthy, free of infection, non-CF subjects were also studied (mean age, 24 yr; range, 20 to 28 yr). A patient and a matched control subject were studied simultaneously to control for within-day variations in cell separation and culture conditions. The six control subjects matched with six of the patients with CF, in whom post-antibiotic studies were available, also had coincidental monocyte studies repeated to assess the reproducibility of the TNF-a secretory response. This study had local Ethics Committee approval. Separation and Culture of Peripheral Blood Monocytes Blood (30 ml) was drawn by venepuncture and anticoagulated with disodium EDTA (170 mM, 600 ttl). Anticoagulated blood was mixed with 6 % dextran 500 in phosphatebuffered saline (PBS) (pH 7.4) in a proportion of 10 parts of blood to one part of dextran. After approximately 20 min, when the red blood cells had sedimented, buffy coat was removed and layered over Nycodenz monocyte separation medium (Nycomed Ltd). After centrifugation, the monocyterich band was removed and the cells were washed twice in PBS containing 0.13% EDTA and 1% fetal calf serum (FCS). The resulting cell pellet was suspended in RPMI-1640 (l ml)
supplemented with 5 % FCS (Biological Industries Ltd., Glasgow, UK), 2 mM L-glutamine, penicillin 100 IU/ml, and streptomycin 100 ttg/ml (culture medium). Purity ranged from 45 to 75 %, with a viability of> 95 % (trypan blue exclusion). The cells were counted and adjusted to 5 X 105/ml. A l-ml aliquot of this suspension was centrifuged, and the cells were lysed in 1 ml of distilled water for DNA determination. Monocytes in 1-ml aliquots were cultured in 12-well culture plates (Costar, Cambridge, MA) and incubated for 2 h at 37 0 C in air with 5 % CO 2 to allow adherence to occur. Dilutions of Escherichia coli (serotype 055: 85) lipopolysaccharide (LPS) (Sigma Chemical Co., St. Louis, MO) were made in culture medium from a stock solution of 1 mg/ml in sterile water. Stock LPS was vortexed vigorously for 3 min, and each subsequent dilution was vortexed for 1 min to ensure adequate mixing. Adherent cells were washed with culture medium before the addition of 1 ml of 0, 25, 50, or 100 ng/ml of LPS in culture medium. All determinations were in duplicate, and cells were incubated for a further 3 h as above. At the end of the incubation period, supernatants were aspirated, centrifuged to remove any cells, and stored at -70 0 C before the measurement of TNF-a. Adherent cells were lysed with 1 ml of distilled water/well, and the plates vortexed to complete lysis. Plates were stored at -20 0 C before DNA assay by a modified specific flurometric method (17). At all stages of cell culture, every effort was made to minimize exogenous endotoxin contamination. Endotoxin content of solutions used was assayed using the limu1us amoebocyte lysate assay (Kabi Vitrum Ltd), dextran 500/PBS 25 pg/ml and FCS 120 pg/ml. Manufacturers' quoted endotoxin content for Nycodenz and RPMI was < 10 and < 50 pg/ml, respectively. Enzyme-linked Immunosorbent Assay for Human TNF-a Immulon IV, irradiated, 96-well microtiter plates (Dynatech Laboratories, Alexandria, VA) were coated with a monoclonal antibody to recombinant human TNF-a (1:1,000; National Institute of Biological Standards, London, UK) in bicarbonate buffer (pH 9.6) at 4 0 C overnight. After draining, wells were blocked with 50 ttl/well of 2 % bovine serum albumin in PBS (pH 7.4) for 1 h at 37 0 C. Assay plates were washed with PBS containing 0.1 % Tween 20. Three rapid washes were followed with three 3-min soaks with the same
Elborn, Norman, Delamere et al.: TNF-a Secretion by Monocytes in Cystic Fibrosis
buffer. Recombinant human TNF-a (interim standard; National Institute for Biological Standards) or sample (50 Ill/well), diluted in PBS containing 0.1% gelatin and 0.05 % Tween 20, was added, incubated at 4°C overnight, and washed as before. Then 50 Ill/well of rabbit anti-TNF-a (1:500 in sample buffer; Genzyme, Boston, MA) was allowed to act for 2 h at 37° C. After washing as before, 50 Ill/well of biotinylated anti-rabbit antibody (1:1,000 in sample buffer; Sigma) was added for 1 h at 37° C. Plates were washed as previously, and avidin-peroxidase complex 50 Ill/well (1:1,000 in borate saline buffer [pH 8.6]) was added and incubated at room temperature for 1 h. After four washes with borate-saline buffer with a 4-min soak during each wash period, peroxidase substrate 100 Ill/well was allowed to act for 30 min at room temperature. This reaction was stopped with 3 M NaOH, 50 Ill/well. Absorbance was read at 455.5 nm. Each assay included eight buffer (zero TNF-a) wells per plate with a TNF-a standard curve from 62.5 to 2,000 pg/ml, with each concentration in duplicate. Plasma samples were assayed at a 1:2 dilution, and conditioned medium at a 1:2 or greater dilution. Quality controls were included in each assay. The TNF-a concentration of specimens was calculated from a regression line derived from the standard curve. Sensitivity, defined as the mean + 2 SD of the absorbance of the eight buffer-containing background wells, was 28.3 pg/ml based on 20 assays. The bioactivity ofTNF-a detected in the enzyme-linked immunosorbent assay used in this study was determined in the WEHI 164 cell line bioassay (18) on six conditioned media samples from control subjects and patients with CF with an immunoreactive TNF-a content ranging from 0 to 8,683 pg/ml (r = +0.97, P < 0.01). Statistical Methods The TNF-a data was skewed; therefore, the significance of differences was determined using the Mann-Whitney U test for unpaired data and the Wilcoxon test for paired data. These results are expressed as medians and nonparametric confidence intervals (19). The DNA data were normally distributed and were analyzed by Student's t test and expressed as mean ± SD.
Results Monocyte Adherence Rates There were no significant differences in the adherence rates for monocytes, as indicated by DNA determination, from patients with CF 1.79 ± 0.94 Ilg/ml and healthy control subjects 1.34 ± 0.86 Ilg/ml, or between the samples from patients with CF before and after antibiotic treatment. There was no significant difference in the DNA per well expressed as a percentage of total cells aliquoted: patients, 43 % (nonparametric 95% confidence interval, 16 to 79%); normal subjects, 33% (17 to 79%). TNF-a Secretion
In the patients with CF at the time of an exacerbation, there was a significantly lower basal secretion rate for TNF-a compared with healthy control subjects (P = 0.006) (table 2). However, the response to added LPS was retained and the release ofTNF-a was not significantly different from that of the monocytes of healthy control subjects (Table 2)_
209
Effects of Antibiotic Therapy In the six patients with CF studied after 2 wk of antibiotic teratment, significant improvements were observed for FEV l (1.14 ± 0.54 to 1.55 ± 0.7 liters [mean ± SD]; P < 0.05), FVC (1.86 ± 1.09 to 2.67 ± 1.16 liters; P < 0.05), PEFR (257 ± 135 to 331 ± 106 liters/min; P < 0.05), and body weight (50.9 ± 13.4 to 52.5 ± 12.7 kg; P < 0.05). Daily sputum production was reduced from 51.3 ± 8.7 to 25 ± 10.5 g (P < 0.05). Circulating immunoreactive TNF-a was also reduced from 258 ± 58.3 to 133 ± 47.8 pg/ml (P < 0.05) (range for 32 healthy control subjects, 20 to 65 pg/ml). Basal (0 LPS) secretion of TNF-a by the monocytes from the patients with CF increased significantly over pretreatment values after 2 wk of antibiotic treatment (P < 0.05). The basal TNF-a secretion after 2 wk of antibiotic therapy was not significantly different from healthy control subjects (Table 3). No effect of antibiotics was demonstrated on the monocyte response to added LPS in patients with CF. The six healthy control subjects studied on two occasions, simultaneously with six patients with CF, demonstrated no significant difference in basal or endotoxin-stimulated TNF-a secretion.
Discussion Apparent attenuation of spontaneous TNF-a secretion occurred at the time of a symptomatic deterioration of respiratory function in patients with chronic pulmonary P. aeruginosa infection. This coincided with significantly increased concentrations of circulating immunoreactive TNF-a at a time of increased circulating concentrations of neutrophil granule products, C-reactive protein, and other inflammatory mediators (2, 3, 20, 21). This reduction in spontaneous secretion suggests that monocytes from the patients with CF were reacting to regulatory stimuli that limit systemic production of TNF-a in the presence of localized sepsis and complements the report that IL-l secretion by monocytes from patients with CF and chronic P. aeruginosa infection are downregulated (16). The mechanism underlying the reduction ofTNF-a secretion is unknown, but it seems likely that it is related to chronic infection and the intensity of the inflammatory response and might be mediated by a variety of inflammatory products. The relationship to the intensity of the inflammatory response was demonstrated by the return of basal TNF-a release by monocytes from patients, after 2 wk of specific antibiotic treatment, to levels similar to control subjects. At this time, circulating TNF-a levels were significantly reduced in the patient. In addition, for the control group studied on two occasions 14 days apart, there was no significant difference in spontaneous or stimulated monocyte secretion of TNF-a. This suggests that the difference in TNF-a secretion after antibiotic treatment was due to real differences in monocyte function in the setting of an active localized inflammatory response. A direct effect of antibiotic treatment on TNF-a secretion by monocytes cannot be excluded, though two unrelated antibiotics were used. It is almost impossible to completely exclude endotoxin from cell separation and culture systems (22). For all the experiments described in this study, contamination was kept to a minimum by using infusion-quality solutions whenever
210
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY YOLo 6 1992
TABLE 2
Monocyte secretion of tumor necrosis factor-a (TNF-a) in response to endotoxin (LPS) in 10 control subjects and 10 patients with cystic fibrosisrt LPS Concentration (ng/ml)
o
25
50
100
Control subjects
583 (298-993)
1,028§ (468-2,340)
778 (375-1,588)
945 (233-1,945)
Cystic fibrosis patients
265* (193-463)
1,180§ (598-2,048)
1,173 (680-2,825)
1,098 (345-2,108)
• Data are given as median and nonparametric 95% confidence intervals. t TNF-a production is expressed as pg/ug DNA. P 0.006 versus control.
*P
Immunoreactive tumor necrosis factor-a (TNF-a) concentration is increased in plasma from patients with cystic fibrosis and chronic Pseudomonas aeruginosa pulmonary infection. To determine if circulating monocytes could be the source of plasma TNF-a, we determined in vitro basal and endotoxin-stimulated TNF-a secretion by monocytes. In 10 adult patients studied at the time of a symptomatic respiratory exacerbation, basal secretion of TNF-a was significantly less than that for 10 matched healthy controls (median 265 pgl/-tg DNA, nonparametric 95 % confidence interval 193 to 463 pg//-tg DNA versus 575, 298 to 923 pgl/-tg DNA; P < 0.006), although both groups responded equally effectively to added Escherichia coli endotoxin at ~ 25 ng/mI. In six patients and six matched controls, monocyte culture was repeated after completion of2 wk anti-pseudomonal antibiotic treatment in the patients. The reduced basal TNF-a secretion in the patients had reversed and was not significantly different to that of controls. This effect mirrored a significant reduction in plasma immunoreactive TNF-a in these patients (mean ± SD, 258 ± 59.3 pg/mI pretreatment versus 133 ± 47.8 pg/rnl post-treatment; P < 0.05). These findings suggest that a reversible downregulation of TNF-a secretion occurred at the time of a symptomatic respiratory deterioration in the presence of chronic P. aeruginosa infection. This may represent a physiologic regulatory mechanism to maintain a local inflammatory response to chronic pulmonary infection in cystic fibrosis.
The main factor determining survival in cystic fibrosis (CF) is progressive lung destruction secondary to chronic Pseudomonas aeruginosa infection. Infection with P. aeruginosa is associated with a local inflammatory response that is responsible for lung injury (1). There is evidence of a systemic component to the host response with increased circulating neutrophil granule products, increased acute phase proteins, and increased immunoreactive tumor necrosis factor-o/cachectin (TNF-a) (2, 3). The plasma concentration of immunoreactive TNF-a is increased in patients with CF and chronic P. aeruginosa lung infection (2, 3), although some centers have been unable to confirm this (4). Moreover, there is a reduction in plasma immunoreactive TNF-a levels after specific antibiotic treatment (3). TNF-a produced by cells of the monocyte-macrophage lineage is an important regulator of the inflammatory response to bacterial infection and endotoxemia (5, 6). The wide range of effects of TNF-a on a variety of cells influences the magnitude of the inflammatory response and (Received for publication May 2. 1991) Address correspondence to: D. 1. Shale, M.D., M.R.C.P., Respiratory Medicine Unit, University of Nottingham, City Hospital, Hucknall Road, Nottingham NG5 IPB, United Kingdom. Abbreviations: cystic fibrosis, CF; fetal calf serum, FCS; lipopolysaccharide, LPS; phosphate-buffered saline, PBS; tumor necrosis factor-a, TNF-a. Am. J. Respir. Cell Mol. BioI. Vol. 6. pp. 207-211, 1992
its duration (7). The plasma concentration of bioactive TNF-a is increased in acute infective disorders, such as meningococcal septicemia and malaria. In these acute septicemia states, the magnitude of the initial plasma TNF-a concentration and those of interleukin (IL)-1 and IL-6 are predictive of a poor outcome (8-10). The pathophysiologic role of circulating TNF-a in localized infection, such as in CF, is unknown. There are a variety of potential sources of circulating TNF-a in CF. The most obvious is the lung, although the precise origin from within the lung is unknown it is likely to be from macrophages in the alveolar, interstitial, or intravascular compartments (11, 12). Investigation of the source is hampered because pulmonary macrophages are largely inaccessible and the alveolar cell profile in such patients is dominated by the neutrophil (13). The presence of circulating Pseudomonas antigen-antibody complexes, which includes complexed endotoxin, in patients with chronic P. aeruginosa infection raises the possibility that the monocyte or other macrophage cells, such as the Kupffer cell, may be contributing to or be the source of circulating TNF-a (14, 15), whereas downregulation ofIL-l secretion by monocytes from patients with CF and chronic P. aeruginosa infection suggests suppression of secretion by inflammatory cells not in the infected pulmonary compartment (16). We studied the circulating monocyte, an accessible precursor of pulmonary macrophages, to determine the state of TNF-a secretory activity in CF patients chronically infected with P. aeruginosa. The in vitro secretion of TNF-a, in the presence and absence
208
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 6 1992
TABLE 1
Details of lung function, Shwachman score, and treatment in 10 patients with cystic fibrosis at the time of a symptomatic deterioration of respiratory function Patient No.
Age (yr)
Weight (kg)
Shwachman Score
No. of Years Colonized with P. aeruginosa
1 2 3 4 5 6 7 8 9 10
16 19 24 21 26 20 19 23 19 25
38.2 61.1 48.3 34.7 70.0 53.7 41.7 44.2 43.3 49.8
45 70 52 43 75 55 40 62 45 60
6 4.5 8 9.5 5 2 8 2 9 2
FEV,
FVC
PEFR
(% predicted; liter)
(% predicted; liter)
(% predicted; literls)
0.7 (26) 1.76 (42) 2.09 (57) 0.68 (17) 3.55 (70) 2.43 (46) 1.49 (34) 2.14 (68) 0.62 (15) 1.78 (51)
186 (52) 360 (75) 312 (74) 114(18) 306 (70) 348 (49) 150 (30) 252 (65) 168 (28) 252 (61)
0.56 1.36 1.27 0.42 1.80 1.47 1.03 1.03 0.43 1.30
(22) (35) (40) (12) (42) (33) (27) (38) (15) (43)
Intravenous Treatment'
A A A
C C C C C C A
• A = aztreonam; C = ceftazidime.
of added endotoxin, was determined both before and after antibiotic treatment of a respiratory exacerbation.
Materials and Methods Patients Ten consecutive patients (mean age, 21 yr; range, 16 to 26 yr) with a symptomatic respiratory exacerbation of their CF (five males) were studied. All had a positive sweat test (Na", Cl- > 60 mmol/liter) and significant respiratory disease with P. aeruginosa chronically cultured from sputum (>3 times/yr). At the time of the study, all had increased respiratory symptoms and a reduction in pulmonary function defined as a fall in FEV 1 > 15% of the best value recorded in the preceding 6 mo (Table 1). Before the first intravenous dose of antibiotics, blood was drawn at 9:00 A.M. for monocyte studies and plasma TNF-a determination. In six patients, a further sample was obtained after 2 wk of intravenously administered antibiotics, when pulmonary function, body weight, and 24-h sputum weight were repeated (Table 1). Ten sex- and age-matched to within 5 yr, apparently healthy, free of infection, non-CF subjects were also studied (mean age, 24 yr; range, 20 to 28 yr). A patient and a matched control subject were studied simultaneously to control for within-day variations in cell separation and culture conditions. The six control subjects matched with six of the patients with CF, in whom post-antibiotic studies were available, also had coincidental monocyte studies repeated to assess the reproducibility of the TNF-a secretory response. This study had local Ethics Committee approval. Separation and Culture of Peripheral Blood Monocytes Blood (30 ml) was drawn by venepuncture and anticoagulated with disodium EDTA (170 mM, 600 ttl). Anticoagulated blood was mixed with 6 % dextran 500 in phosphatebuffered saline (PBS) (pH 7.4) in a proportion of 10 parts of blood to one part of dextran. After approximately 20 min, when the red blood cells had sedimented, buffy coat was removed and layered over Nycodenz monocyte separation medium (Nycomed Ltd). After centrifugation, the monocyterich band was removed and the cells were washed twice in PBS containing 0.13% EDTA and 1% fetal calf serum (FCS). The resulting cell pellet was suspended in RPMI-1640 (l ml)
supplemented with 5 % FCS (Biological Industries Ltd., Glasgow, UK), 2 mM L-glutamine, penicillin 100 IU/ml, and streptomycin 100 ttg/ml (culture medium). Purity ranged from 45 to 75 %, with a viability of> 95 % (trypan blue exclusion). The cells were counted and adjusted to 5 X 105/ml. A l-ml aliquot of this suspension was centrifuged, and the cells were lysed in 1 ml of distilled water for DNA determination. Monocytes in 1-ml aliquots were cultured in 12-well culture plates (Costar, Cambridge, MA) and incubated for 2 h at 37 0 C in air with 5 % CO 2 to allow adherence to occur. Dilutions of Escherichia coli (serotype 055: 85) lipopolysaccharide (LPS) (Sigma Chemical Co., St. Louis, MO) were made in culture medium from a stock solution of 1 mg/ml in sterile water. Stock LPS was vortexed vigorously for 3 min, and each subsequent dilution was vortexed for 1 min to ensure adequate mixing. Adherent cells were washed with culture medium before the addition of 1 ml of 0, 25, 50, or 100 ng/ml of LPS in culture medium. All determinations were in duplicate, and cells were incubated for a further 3 h as above. At the end of the incubation period, supernatants were aspirated, centrifuged to remove any cells, and stored at -70 0 C before the measurement of TNF-a. Adherent cells were lysed with 1 ml of distilled water/well, and the plates vortexed to complete lysis. Plates were stored at -20 0 C before DNA assay by a modified specific flurometric method (17). At all stages of cell culture, every effort was made to minimize exogenous endotoxin contamination. Endotoxin content of solutions used was assayed using the limu1us amoebocyte lysate assay (Kabi Vitrum Ltd), dextran 500/PBS 25 pg/ml and FCS 120 pg/ml. Manufacturers' quoted endotoxin content for Nycodenz and RPMI was < 10 and < 50 pg/ml, respectively. Enzyme-linked Immunosorbent Assay for Human TNF-a Immulon IV, irradiated, 96-well microtiter plates (Dynatech Laboratories, Alexandria, VA) were coated with a monoclonal antibody to recombinant human TNF-a (1:1,000; National Institute of Biological Standards, London, UK) in bicarbonate buffer (pH 9.6) at 4 0 C overnight. After draining, wells were blocked with 50 ttl/well of 2 % bovine serum albumin in PBS (pH 7.4) for 1 h at 37 0 C. Assay plates were washed with PBS containing 0.1 % Tween 20. Three rapid washes were followed with three 3-min soaks with the same
Elborn, Norman, Delamere et al.: TNF-a Secretion by Monocytes in Cystic Fibrosis
buffer. Recombinant human TNF-a (interim standard; National Institute for Biological Standards) or sample (50 Ill/well), diluted in PBS containing 0.1% gelatin and 0.05 % Tween 20, was added, incubated at 4°C overnight, and washed as before. Then 50 Ill/well of rabbit anti-TNF-a (1:500 in sample buffer; Genzyme, Boston, MA) was allowed to act for 2 h at 37° C. After washing as before, 50 Ill/well of biotinylated anti-rabbit antibody (1:1,000 in sample buffer; Sigma) was added for 1 h at 37° C. Plates were washed as previously, and avidin-peroxidase complex 50 Ill/well (1:1,000 in borate saline buffer [pH 8.6]) was added and incubated at room temperature for 1 h. After four washes with borate-saline buffer with a 4-min soak during each wash period, peroxidase substrate 100 Ill/well was allowed to act for 30 min at room temperature. This reaction was stopped with 3 M NaOH, 50 Ill/well. Absorbance was read at 455.5 nm. Each assay included eight buffer (zero TNF-a) wells per plate with a TNF-a standard curve from 62.5 to 2,000 pg/ml, with each concentration in duplicate. Plasma samples were assayed at a 1:2 dilution, and conditioned medium at a 1:2 or greater dilution. Quality controls were included in each assay. The TNF-a concentration of specimens was calculated from a regression line derived from the standard curve. Sensitivity, defined as the mean + 2 SD of the absorbance of the eight buffer-containing background wells, was 28.3 pg/ml based on 20 assays. The bioactivity ofTNF-a detected in the enzyme-linked immunosorbent assay used in this study was determined in the WEHI 164 cell line bioassay (18) on six conditioned media samples from control subjects and patients with CF with an immunoreactive TNF-a content ranging from 0 to 8,683 pg/ml (r = +0.97, P < 0.01). Statistical Methods The TNF-a data was skewed; therefore, the significance of differences was determined using the Mann-Whitney U test for unpaired data and the Wilcoxon test for paired data. These results are expressed as medians and nonparametric confidence intervals (19). The DNA data were normally distributed and were analyzed by Student's t test and expressed as mean ± SD.
Results Monocyte Adherence Rates There were no significant differences in the adherence rates for monocytes, as indicated by DNA determination, from patients with CF 1.79 ± 0.94 Ilg/ml and healthy control subjects 1.34 ± 0.86 Ilg/ml, or between the samples from patients with CF before and after antibiotic treatment. There was no significant difference in the DNA per well expressed as a percentage of total cells aliquoted: patients, 43 % (nonparametric 95% confidence interval, 16 to 79%); normal subjects, 33% (17 to 79%). TNF-a Secretion
In the patients with CF at the time of an exacerbation, there was a significantly lower basal secretion rate for TNF-a compared with healthy control subjects (P = 0.006) (table 2). However, the response to added LPS was retained and the release ofTNF-a was not significantly different from that of the monocytes of healthy control subjects (Table 2)_
209
Effects of Antibiotic Therapy In the six patients with CF studied after 2 wk of antibiotic teratment, significant improvements were observed for FEV l (1.14 ± 0.54 to 1.55 ± 0.7 liters [mean ± SD]; P < 0.05), FVC (1.86 ± 1.09 to 2.67 ± 1.16 liters; P < 0.05), PEFR (257 ± 135 to 331 ± 106 liters/min; P < 0.05), and body weight (50.9 ± 13.4 to 52.5 ± 12.7 kg; P < 0.05). Daily sputum production was reduced from 51.3 ± 8.7 to 25 ± 10.5 g (P < 0.05). Circulating immunoreactive TNF-a was also reduced from 258 ± 58.3 to 133 ± 47.8 pg/ml (P < 0.05) (range for 32 healthy control subjects, 20 to 65 pg/ml). Basal (0 LPS) secretion of TNF-a by the monocytes from the patients with CF increased significantly over pretreatment values after 2 wk of antibiotic treatment (P < 0.05). The basal TNF-a secretion after 2 wk of antibiotic therapy was not significantly different from healthy control subjects (Table 3). No effect of antibiotics was demonstrated on the monocyte response to added LPS in patients with CF. The six healthy control subjects studied on two occasions, simultaneously with six patients with CF, demonstrated no significant difference in basal or endotoxin-stimulated TNF-a secretion.
Discussion Apparent attenuation of spontaneous TNF-a secretion occurred at the time of a symptomatic deterioration of respiratory function in patients with chronic pulmonary P. aeruginosa infection. This coincided with significantly increased concentrations of circulating immunoreactive TNF-a at a time of increased circulating concentrations of neutrophil granule products, C-reactive protein, and other inflammatory mediators (2, 3, 20, 21). This reduction in spontaneous secretion suggests that monocytes from the patients with CF were reacting to regulatory stimuli that limit systemic production of TNF-a in the presence of localized sepsis and complements the report that IL-l secretion by monocytes from patients with CF and chronic P. aeruginosa infection are downregulated (16). The mechanism underlying the reduction ofTNF-a secretion is unknown, but it seems likely that it is related to chronic infection and the intensity of the inflammatory response and might be mediated by a variety of inflammatory products. The relationship to the intensity of the inflammatory response was demonstrated by the return of basal TNF-a release by monocytes from patients, after 2 wk of specific antibiotic treatment, to levels similar to control subjects. At this time, circulating TNF-a levels were significantly reduced in the patient. In addition, for the control group studied on two occasions 14 days apart, there was no significant difference in spontaneous or stimulated monocyte secretion of TNF-a. This suggests that the difference in TNF-a secretion after antibiotic treatment was due to real differences in monocyte function in the setting of an active localized inflammatory response. A direct effect of antibiotic treatment on TNF-a secretion by monocytes cannot be excluded, though two unrelated antibiotics were used. It is almost impossible to completely exclude endotoxin from cell separation and culture systems (22). For all the experiments described in this study, contamination was kept to a minimum by using infusion-quality solutions whenever
210
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY YOLo 6 1992
TABLE 2
Monocyte secretion of tumor necrosis factor-a (TNF-a) in response to endotoxin (LPS) in 10 control subjects and 10 patients with cystic fibrosisrt LPS Concentration (ng/ml)
o
25
50
100
Control subjects
583 (298-993)
1,028§ (468-2,340)
778 (375-1,588)
945 (233-1,945)
Cystic fibrosis patients
265* (193-463)
1,180§ (598-2,048)
1,173 (680-2,825)
1,098 (345-2,108)
• Data are given as median and nonparametric 95% confidence intervals. t TNF-a production is expressed as pg/ug DNA. P 0.006 versus control.
*P
