Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Tumor necrosis factor (TNF)- is a pivotal mediator in the pathogenesis of fibrosing alveolitis. TNF- not only enhances adherence of inflammatory cells to the endothelium, and their migration into alveolar inflammatory sites, but also activates the production of inflammatory mediators by monocytes and neutrophils (1). In addition, TNF- induces fibroblast proliferation and collagen messenger RNA (mRNA) expression, and can inhibit collagen elimination by phagocytic pathways (2). In rat and mouse models of pulmonary fibrosis induced by silica or bleomycin, treatment with anti-TNF- antibodies caused a significant reduction of the fibrotic response as indicated by lung hydroxyproline content (5, 6). Recent studies with transgenic animals have elegantly completed these data: expression of a TNF- transgene in the murine lung caused spontaneous alveolitis, alveolar disruption, and a progressive fibrotic reaction (7). Using viable motheaten mutant mice, which naturally develop progressive pulmonary inflammation and fibrosis, Thrall and colleagues (8) have shown a close association between pulmonary injury and pulmonary TNF- levels. In addition, TNF-/lymphotoxin (LT)- double-deficient mice were found resistant to bleomycin-induced fibrosis (9). Several human data have also shown a clear association between overproduction of TNF- and the occurrence of lung fibrosis. Human interstitial lung diseases induced by inhalation of inorganic dusts such as asbestos and coal mine particles were accompanied by a significant increase in TNF- release by alveolar macrophages (AM) or blood monocytes (3, 10, 11). In patients with idiopathic pulmonary fibrosis, high TNF- levels were also found in bronchoalveolar lavage fluid (BALF) and in supernatants from cultured AM (3, 12).

Besides this wealth of evidence supporting a profibrotic role for TNF-, at least two experimental studies have shown that the lung fibrotic process may in some models be accompanied by a reduction of TNF- activity. First, Ouellet and coworkers (13), using Wistar rats (a strain considered as relatively resistant to the fibrotic effect of silica), have shown a reduction in TNF- production by bronchoalveolar lavage (BAL) cells from silica-treated rats at all time points considered (up to 28 d after administration). Second, the lung fibrotic response induced by silica in C3H/Hej mice (also considered a relatively resistant strain) was associated with a reduction of TNF- production by BAL cells (14). However, both the mechanism and the possible pathophysiologic significance of such a downregulation of TNF synthesis remain unclear.

We observed a downregulation of TNF- accompanying the response to silica in NMRI mice. Our study examined the mechanisms involved in this apparently aberrant response in NMRI mice. The possible relationship of this downregulation of TNF- with the lung fibrotic process was investigated in a comparative mouse model described previously (15): intratracheal instillation of crystalline silica or manganese dioxide particles was used to produce a fibrosing or a resolving alveolitis; tungsten carbide (WC) particles served as innocuous control particles (noninflammatory response).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Animals

Female NMRI mice (Swiss mice that were initially bred at the Naval Medical Research Institute) and Sprague-Dawley rats were purchased from Iffa Credo (Brussels, Belgium). The animals were housed in air-conditioned positive- pressure units (25°C, 50% relative humidity) on a 12-h light/ dark cycle. Interleukin (IL)-10-deficient mice (C57BL/6) (16) were obtained from B&K Universal Limited (Hull, UK). Homozygous mutants were generated by intercrossing of mice heterozygous for IL-10 (17).

Instillation Method

To allow sterilization and inactivation of any trace of endotoxin, particles were heated at 200°C for 4 h before use. Suspensions of WC (d₅₀ = 1 µm; Johnson Matthey, Royston, UK), MnO₂ (d₅₀ = 3.7 µm; Sedema, Division of Sadacam SA, Tertre, Belgium), or crystalline silica particles (DQ12; d₅₀ = 2.2 µm; Dr. L. Armbruster, Essen, Germany), or sterile 0.9% saline (controls) were injected directly into the lungs by intratracheal instillation (100 µl/mouse or 400 µl/rat). All instillations were performed on anesthetized animals (sodium pentobarbital, 2 mg/mouse or 6 mg/rat) after surgical opening of the neck.

BAL and Cell Culture

At selected time intervals after treatment, a first group of animals was killed with sodium pentobarbital (20 mg/ mouse or 60 mg/rat, intraperitoneally) and BAL was performed by cannulating the trachea and infusing the lungs six times with a volume of 1.5 ml (mouse) or 8 ml (rat) of sterile saline. The BALF was centrifuged (1,000 × g, 10 min, 4°C ) and the cell-free supernatant of this first lavage fraction was used for cytokine measurements. BAL was done with three volumes (1.5 or 8 ml) of sterile saline. The lavage fluids were centrifuged and the cell pellets from all animals within each treatment group were pooled. Aliquots of the cell suspensions were then used to determine cell number and viability (trypan blue exclusion method). Cell differentials were performed on cytocentrifuge preparations fixed in methanol and stained with Diff-Quik (Dade A. G., Düdingen, Switzerland; 200 cells counted). The alveolar cell suspensions were adjusted to a concentration of 0.5 × 10⁶ viable BAL cells/ml of RPMI 1640 medium (GIBCO BRL, Merelbeke, Belgium) containing lactalbumin hydrolysate (0.1%) and antibiotics (1%). Aliquots of 1 ml of the cell suspensions were seeded into 24-well culture plates. BAL cell cultures were stimulated or not with 1 µg/ml of lipopolysaccharide (LPS) (Sigma; Escherichia coli 055:B5) for 3 h. After this period, the culture supernatants were harvested, centrifuged (1,000 × g, 10 min, 4°C), and kept at 80°C until analysis.

Preparation of BAL Cells and Lung Homogenates

In a second group of mice, BAL cells were obtained from each animal individually through four successive lavages with sterile saline. After centrifugation, cells obtained from each mouse were resuspended in 1 ml of cold phosphate-buffered saline (PBS) (GIBCO BRL) and homogenized for 30 s with a Polytron PT1200 homogenizer (Kinematica AG, Lucerne, Switzerland). A third group of mice was used for measurements of cytokines in the lung; their whole lungs were excised and placed into Falcon tubes (Becton Dickinson, Meylan, France) chilled on ice, and 3 ml of cold PBS was added. The content of each tube was then homogenized as described earlier. The tubes were centrifuged at 4°C, 2,000 rpm for 10 min, and supernatants were kept at 80°C until used. The same procedure was used with rats, except that cells and lungs were homogenized in 3 and 10 ml of PBS, respectively.

Semiquantitative Polymerase Chain Reaction Analysis

RNA from whole lung and BAL cells obtained from saline- or dust-treated animals (fourth group) were isolated with the Trizol method (GIBCO BRL). Reverse transcription (RT) was performed on total RNA (2.5 µg) with an oligodeoxythymidine (oligo[dT]) primer, and amplification was done for 25-40 cycles by polymerase chain reaction (PCR) with specific primers as follows: for TNF-, sense: 5'-GGCCCA GACCCTCACACTCA-3; antisense: 5'-TGGGTGAGGAGCACGTAGTC-3'. For TNFRI (p55), sense: 5'- CAGGTGGAGATCTCTCCTTG-3'; antisense: 5'-CGAT CTCGTGCTCGCTCAGC-3'. For TNFRII (p75), sense: 5'-GCAAGCACAGATGCAGT CTG-3'; antisense: 5'- GGTCAGAGCTGCTACAGACG-3'. For interleukin (IL)-10, sense: 5'-GAGACTTGCTCTTGCACTAC-3'; antisense: 5'-CCTGGAGTCCAGCAGACTCA-3'. For -actin, sense: 5'-ATGGATGACGATATCGCTGC-3'; antisense: 5'-GCTGGAAGGTGGACAGTGAG-3'.

An aliquot of the PCR reaction was electrophoresed in a 1% agarose gel and stained with ethidium bromide. Gel analysis was performed densitometrically through use of the MCID software system (Imaging Research, St. Catherines, ON, Canada). Results were expressed as arbitrary units, which were calculated by integration of the intensity of each pixel over the spot area and normalized for the intensity of the actin signal (cytokine-to--actin ratio).

Cytokine and Inflammatory Mediator Measurements

TNF- and IL-10 levels in the rat model were measured with commercial enzyme-linked immunosorbent assay (ELISA) kits (Biosource International, Camarillo, CA; detection limit: 4 pg/ml and 5 pg/ml, respectively). The concentrations of mouse TNF- and IL-10 in BALF and in BAL cell and lung homogenates were measured with cytokine-specific ELISAs obtained from Biosource International; the detection limits of these ELISAs were 3 pg/ml and 0.178 pg/ml, respectively. Mouse soluble TNF receptors (sTNF-R) (p75 and p55) were measured with ELISAs obtained from Genzyme (Cambridge, MA), with detection limits of 12 pg/ml and 15 pg/ml, respectively. Cytokines were expressed per milliliter of recovered lavage fluid. Levels of TNF- in culture supernatants were measured both with ELISA (pg/ml) and through bioassay (arbitrary units/ml). TNF- activity was estimated as described previously (18).

Prostaglandin (PG) E₂ was measured with an enzyme immunoassay (EIA) from Amersham Life Science (Buckinghamshire, England). The biologic activity of transforming growth factor (TGF)- was estimated with a bioassay (19), and was expressed as percent of the maximal activity obtained with a serially diluted TGF- standard (R&D Systems, Minneapolis, MN).

In Vitro Neutralization of sTNF-R (p55 and p75)

Neutralization of mouse sTNF-R p55 and p75 was done with soluble antibodies (20) obtained from Genzyme. The mix of antibodies (anti-p55 and anti-p75, 10 µg/ml each) was added to BAL cell cultures 2 h before addition of LPS.

Fluorescence-Activated Cell Sorter Purification of Neutrophils

BAL cells obtained from silica-treated mice (24 h after treatment) were stained with a fluorescein isothiocyanate-conjugated rat monoclonal antimouse Ly-6G (Gr-1) antibody (PharMingen, San Diego, CA) in Hanks' balanced salt solution (GIBCO BRL) and purified by sorting the positive cells (ATC 300; Bruker Spectrospin, Milton, Canada). The purity and viability of purified cells were both > 98%.

Statistics

Treatment-related differences were evaluated through one-way analysis of variance (ANOVA), followed by pairwise comparison with the Student-Newman-Keuls test. Statistical significance was considered at P < 0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Levels of TNF- Protein and mRNA in the Fibrotic, Resolving Inflammatory and Noninflammatory Models in NMRI Mice

The expression of TNF- protein and mRNA levels was examined in three mouse models of pulmonary response to mineral particles. Levels of TNF- protein in BALF from fibrotic, inflammatory, and noninflammatory models were significantly lower than in controls at 3 d after treatment. Although this downregulation of TNF- was maintained for up to 120 d in the fibrotic model, TNF- levels returned to control values after 15 and 30 d in the noninflammatory and resolving inflammatory models, respectively (Figure 1, upper panel).

View larger version (29K): [in this window] [in a new window]   Figure 1.   Levels of TNF- protein in BALF (upper panel ) and TNF- mRNA expression in the lung (lower panel) in the control, noninflammatory (NI), resolving alveolitis (RA), and fibrosing alveolitis (FA) models. Bars represent the means ± SD for at least four animals (* P < 0.05 versus control group, Student- Newman-Keuls multiple comparison test).

The pattern of TNF- mRNA expression in the lung was opposite to that of the protein. In the resolving and fibrotic inflammatory models, mRNA levels were upregulated as compared with those of controls. This effect was present for up to 30 d in the resolving alveolitis model, and at all time points examined in the fibrotic model. No effect on TNF- mRNA levels was observed in the noninflammatory model (Figure 1, lower panel).

TNF- Levels in Different Pulmonary Compartments 24 h after Silica Treatment in NMRI Mice

In subsequent experiments, attention was focused on the early response to silica particles (24 h), in order to investigate the possible determinants of the downregulation of TNF- protein observed earlier. Since TNF- can be produced both extracellularly in BALF and in a membrane-bound form (21), the amounts of TNF- were assessed in both lung tissue and in BAL cell homogenates after silica treatment. Twenty-four hours after silica administration, levels of TNF- in both lung and BAL cells were significantly decreased as compared with control values, ruling out a possible redistribution mechanism (in lung: 1,824.9 ± 349.9 pg/ml with saline and 1,152.6 ± 104.7 pg/ml with silica; P < 0.05; in BAL cells: 353.3 ± 93 pg/ml with saline and 171.4 ± 29.6 pg/ml with silica; P < 0.05). Again mRNA levels after 24 h did not reflect the TNF- protein levels. Following silica administration, TNF- mRNA expression was increased as compared with that in the control group both in BAL cells and in whole lung (data not shown). A similar pattern was observed in cultures of BAL cells incubated with LPS for 3 h. Although the release of TNF- by BAL cells obtained from animals treated with silica was significantly reduced both in the supernatant (saline: 15,906.4 ± 2,435.7 pg/ml; and silica: 4,694.1 ± 2,164.1 pg/ml; P < 0.05) and in adherent cell homogenates (saline: 2,509.8 ± 29.1 pg/ml; and silica: 711.4 ± 55.8 pg/ml; P < 0.05), TNF- mRNA was upregulated in BAL cells obtained from silica-exposed animals (data not shown). These ex vivo results were confirmed in the absence of LPS stimulation and for different durations of culture, for up to 24 h. Measurements of the biologic activity of TNF- also showed that supernatants from BAL cell cultures obtained from mice treated with silica contained significantly lower levels of TNF- activity than in cultures from controls (data not shown). This reduction of TNF- protein and activity was not prevented by the addition of an antiprotease cocktail (phenylmethylsulfonyl fluoride, 2 mM; leupeptin, 1 µg/ml; pepstatin A, 1 µg/ml; and ₁- antitrypsin, 0.5 mg/ml) to the culture medium (TNF- biologic activity: saline without antiproteases: 319.6 ± 100.9 U/ml; silica without antiproteases: 56.6 ± 11.6 U/ml; saline with antiproteases: 408.8 ± 124 U/ml; silica with antiproteases: 58.6 ± 10 U/ml). In addition, the activity of recombinant TNF- (50, 100, and 500 U/ml) was not degraded by incubation with BAL cell cultures, culture supernatants, or BALF from silica-treated animals (data not shown). TNF- production by cultured BAL cells obtained from the comparative models at different time points was qualitatively comparable to the in vivo patterns shown in the upper panel of Figure 1 (data not shown).

To investigate the mechanisms responsible for the downregulation of TNF- in NMRI mice, we examined the production of mediators known to limit the activity or synthesis of TNF-. When compared with those of control cultures, levels of TGF- were significantly reduced in supernatants from BAL cell cultures obtained 24 h after silica treatment (saline: 51.7 ± 11.1%; and silica: 29.3 ± 3.8% of maximal TGF- biologic activity; P < 0.05). In contrast, PGE₂ was upregulated after silica treatment in both BAL cell cultures (saline: 10.5 ± 4.8 pg/ml; and silica: 24.7 ± 5 pg/ml; P < 0.05) and in BALF (saline: < 1 pg/ml; and silica: 26.7 ± 2.3 pg/ml). However, blocking the production of PGE₂ by the addition of indomethacin (1 µg/ml) did not restore the ex vivo production of TNF- after silica treatment (data not shown). Twenty-four hours after instillation, silica treatment stimulated the expression of mRNAs for the p75 and p55 soluble TNF receptors in BAL cell homogenates (data not shown). This was accompanied by increased levels of both of these soluble TNF receptor proteins in BALF (p55, saline: 6.9 ± 5.3 pg/ml; and silica: 121.7 ± 36.1 pg/ml; p75, saline: 59.8 ± 8.0 pg/ml; and silica: 760.2 ± 0.9 pg/ml; P < 0.05) and in BAL cell homogenates (p55, saline: not detected; and silica: 13.5 ± 6.5 pg/ml; p75, saline: 45.7 ± 1.3 pg/ml; and silica: 71.9 ± 6.1 pg/ml; P < 0.05). Treatment with silica induced a concurrent and pronounced upregulation of IL-10 mRNA expression (data not shown) as well as increased IL-10 protein levels in BAL cell homogenates (saline: 1.3 ± 2 pg/ml; and silica: 67.0 ± 5.6 pg/ml; P < 0.05).

Levels of sTNF-R and IL-10 in Noninflammatory, Resolving Alveolitis, and Fibrosing Alveolitis Models

To assess the time course of sTNF-R and IL-10 activity and its relation to the in vivo downregulation of TNF- protein (Figure 1, upper panel), the levels of both molecules were measured in BALF and BAL cells in the three comparative models. Lower levels of TNF- observed in the resolving and fibrosing alveolitis models were correlated with an overproduction of sTNF-R p55 and p75 in BALF at up to 30 and 120 d in the resolving and fibrotic alveolitis models, respectively (Figure 2). Similarly, an increased expression of IL-10 mRNA in BAL cells coincided with the reduction of TNF-. Although IL-10 was transiently upregulated in the resolving alveolitis model, persisting upregulation of this cytokine was found in the fibrosing alveolitis model (Figure 3). Measured values for IL-10 protein in lung homogenates were qualitatively comparable with the expression patterns observed in BAL cells (data not shown).

View larger version (24K): [in this window] [in a new window]   Figure 2.   Levels of sTNF-R p55 (upper panel) and p75 (lower panel) in BALF in control, noninflammatory (NI), resolving alveolitis (RA), and fibrosing alveolitis (FA) models. Bars represent the means ± SD for at least four animals (* P < 0.05 versus control group, Student-Newman-Keuls multiple comparison test).

View larger version (34K): [in this window] [in a new window]   Figure 3.   Expression of IL-10 mRNA in BAL cells in control, noninflammatory (NI), resolving alveolitis (RA), and fibrosing alveolitis (FA) models. Shown are photographs of ethidium-bromide-stained gels containing the RT-PCR products for IL-10 and -actin. Relative expression of IL-10 was quantified by densitometric scanning and was normalized for the intensity of the corresponding expression of -actin. The determination was performed on a pool of BAL cells recovered from six mice.

sTNF-R and IL-10 Regulation of TNF- Activity after Silica Administration in NMRI Mice

To assess the functional role of TNF receptor expression in the downregulation of TNF-, BAL cells from silica-treated NMRI mice were first incubated for 2 h with neutralizing anti-sTNF-R p75 and p55 antibodies or with medium alone; LPS was then added to the cultures and incubation was conducted for 3 h. After silica treatment, the levels of TNF- in BAL cell supernatants preincubated with antibodies were higher than in cultures from which antibodies were omitted (Figure 4, upper panel). Similarly, in homozygous IL-10-deficient animals (/), the inhibition of LPS-induced TNF- synthesis ex vivo after silica treatment was found to depend on the ability to produce IL-10 (Figure 4, lower panel).

View larger version (16K): [in this window] [in a new window]   Figure 4.   (Upper panel) Effect of antibodies against sTNF-R p55 and p75 on the ex vivo release of TNF- by BAL cells obtained from NMRI mice at 1 d after silica treatment (5 mg/mouse). Cells were preincubated for 2 h with neutralizing anti-TNFR p75 and p55 antibodies (10 µg/ml) or medium alone; LPS (1 µg/ml) was then added to the cultures and incubation was done for 3 h. (Lower panel) Ex vivo release of TNF- by BAL cells obtained from wild type (+/+, C57BL/6) and homozygous IL-10-deficient (/, C57BL/6) animals at 1 d after silica particle treatment. Cells were cultured for 3 h with LPS (1 µg/ml). Bars represent the means ± SD of at least three separate determinations (*P < 0.05 versus control group, Student-Newman-Keuls multiple comparison test).

Role of Neutrophils in the Modulation of TNF- Production in NMRI Mice

To determine whether a particular type of BAL cell might contribute to the lower levels of TNF- observed after silica administration, naive BAL cells isolated from normal mice (100% of AM) were incubated with BAL cells explanted from silica-treated mice (35% and 65% of AM and neutrophils [PMN], respectively) or with a fluorescence-activated cell sorter (FACS)-purified population of PMN (98% of PMN) also obtained from mice treated with silica particles. Cocultures of naive AM with total BAL cells or with purified PMN clearly reduced the BAL cells' LPS-induced production of TNF- protein (Figure 5). In addition, BAL cells from mice depleted of PMN by cyclophosphamide pretreatment (200 mg/kg, 3 d before culture) and instilled with silica produced more TNF- in culture than did BAL cells obtained from mice not pretreated with cyclophosphamide (saline without cyclophosphamide: 100 ± 34.5%; and silica without cyclophosphamide: 29.0 ± 14.7% of control; and saline with cyclophosphamide = 100 ± 17.8%; and silica with cyclophosphamide = 62.1 ± 17.8% of control).

View larger version (13K): [in this window] [in a new window]   Figure 5.   Effects of BAL cells (AM = 35% and PMN = 65%) or FACS-purified PMN (PMN = 98%) obtained from NMRI mice treated with silica (Day 1, 5 mg/mouse) on LPS-induced TNF- release by naive AM. Cells were cultured for 18 h with LPS (1 µg/ ml), and levels of TNF- were determined in the culture media through bioassay. Bars represent the means ± SD of at least three separate determinations (*P < 0.05 versus control group, Student-Newman-Keuls multiple comparison test).

Comparison of the Response to Silica Particles in Sprague-Dawley Rats and NMRI Mice

Twenty-four hours after instillation of two different doses of silica particles (2 mg/g and 20 mg/g lung weight), TNF- responses in BALF and in BAL cell culture supernatants were compared in rats and in NMRI mice. In NMRI mice, silica induced a dose-dependent downregulation of TNF- activity in both BALF and in BAL cell cultures (Figure 6, upper panel). In contrast, in the rats, ex vivo TNF- production by BAL cells and TNF- levels in BALF were significantly increased (Figure 6, lower panel). The AM:PMN ratio in BALF was typically 30:70 in rats and 35:65 in mice treated with 20 mg silica, respectively.

View larger version (24K): [in this window] [in a new window]   Figure 6.   Comparison of TNF- levels in BALF and in supernatants of BAL cell cultures (LPS stimulation, 1 µg/ml, 3 h) obtained from mice (upper panel) and rats (lower panel) treated with 2 mg silica/g lung and 20 mg silica/g lung (1 d). BALF determination was done on a pool of samples recovered from at least three animals. Values in culture represent the means ± SD of at least three separate determinations (*P < 0.05 versus control group, Student-Newman-Keuls multiple comparison test).

IL-10 protein levels were compared in BAL cell homogenates 24 h after administration of increasing doses of crystalline silica particles to both rats and mice. Although silica induced a dose-dependent upregulation of IL-10 in mice (Figure 7, upper panel), this cytokine was dose-dependently reduced after administration of equivalent doses of silica in rats (Figure 7, lower panel). IL-10 levels were also reduced in BALF but not in whole-lung homogenates from rats treated with silica (not shown).

View larger version (29K): [in this window] [in a new window]   Figure 7.   Comparison of IL-10 levels in BAL cell homogenates from mice (upper panel) and rats (lower panel) treated with 2 mg silica/g lung and 20 mg silica/g lung (1 d). Values represent the means ± SD of at least three separate determinations (*P < 0.05 versus control group, Student-Newman-Keuls multiple comparison test).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In NMRI mouse models of lung alveolitis induced by inorganic particles, inflammation and/or fibrosis were accompanied by a downregulation of TNF- protein. This study examined the mechanisms involved in the control of this apparently aberrant TNF- response, and its possible relationship with the fibrotic process.

What Are the Mechanisms Involved in the Downregulation of TNF- in Mice?

At inflammatory sites, although metalloproteinases have been reported to enhance the release of soluble TNF- activity (22), an enzymatic degradation of TNF-, mainly by elastase, may occur in the vicinity of activated PMN (23). Because addition of antiproteases did not prevent the reduction of TNF- release by BAL cells obtained from silica-treated animals in our study, and because the activity of recombinant TNF- was not abolished when it was incubated with BAL cell cultures, culture supernatants, or BALFs obtained from silica-treated animals, it appears very unlikely that a proteolytic activity could be responsible for the observed inactivation of TNF- in NMRI mice. Furthermore, the upregulation of other cytokines (IL-6, IL-10, p40 IL-12; not shown) at the same time points does not support the involvement of proteolytic degradation in the observed downregulation of TNF- protein in NMRI mice.

TNF- exerts its biologic activity by interacting with two distinct TNF receptors, with molecular masses of 55 kD and 75 kD, respectively (24). Several studies have shown that after appropriate stimulation, both TNF receptors may be shed from the cell surface of various cell types, resulting in the expression of soluble forms of these proteins that retain binding affinities for TNF-. Although the role of sTNF-R is still a subject of speculation, soluble forms of each receptor inhibit TNF- action in vitro and in vivo, and can downregulate the inflammatory response (25, 26). Since addition of a monoclonal anti-TNFR antibody restored to some extent the levels of TNF- produced by BAL cell cultures from silica-treated animals, and because a strong time relationship existed between downregulation of TNF- and BALF levels of sTNF-R, it can be concluded that both the 55 kD and 75 kD sTNF-R contributed to the limitation of TNF- activity observed in the resolving alveolitis and fibrosing alveolitis models. It cannot, however, be formally excluded that overexpression of cell-associated TNF receptors did not also contribute to the limitation of TNF- activity through the binding of TNF- molecules. Moreover, a mechanism based on the neutralization of TNF- proteins by specific receptors is consistent with the observed discordance between TNF mRNA and protein levels. Other investigators have observed that administration of recombinant sTNF-R to normal or silica-exposed mice markedly decreased the collagen content in the lung by limiting the activity of TNF- in collagen synthesis (27, 28). Therefore, it is plausible that the sTNF-R overproduction observed in the resolving and fibrosing alveolitis models in our study, as well as in coal workers' pneumoconiosis (29), represents an attempt of the host to limit the extension of pulmonary inflammation and fibrosis.

It has also been shown that TNF- can be negatively regulated at the posttranscriptional and/or translational levels by several antiinflammatory mediators, including PGE₂, TGF-, IL-4, IL-10, and IL-13 (30), which might also be of particular relevance to understanding the discordance between TNF- protein and mRNA levels observed in the present study. Since critical roles for PGE₂ (36), TGF- (37), and IL-10 (38) in the lung fibrosis induced by silica have recently been demonstrated, their overexpression in our mouse model could also have contributed to the observed downregulation of TNF-. The possible contribution of TGF- to the downregulation of TNF- was not supported by the finding of reduced production of this cytokine after silica treatment. PGE₂ levels were significantly increased after silica administration, but addition of the cyclooxygenase inhibitor indomethacin did not restore TNF- production, suggesting a minor role for PGE₂ in the regulation of TNF-. Using IL-10-deficient animals and assessing IL-10 expression in the comparative models indicated that this cytokine contributed to downregulating TNF- production. Interestingly, several reports have shown associations between sTNF-R and IL-10. Because IL-10 has been reported to regulate the cleavage of TNF receptors from monocytes/macrophages (39) it is possible that, in addition to its effect at the posttranscriptional level, IL-10 could, in the models used in our study, control TNF- activity through an increased production of sTNF-R.

The low levels of TNF- in BALF were associated with the recruitment of PMN in alveolar spaces (resolving and fibrosing alveolitis), and the restoration of TNF- activity was concurrent with the disappearance of PMN from the lung (resolving alveolitis after Day 15) (15). When levels of TNF in BALF measured in all models and at all time points were pooled, an inverse correlation was found with the number of recruited PMN (P < 0.0025, r = 0.6998). In addition, neutrophils stimulated in vivo by silica particles suppressed TNF- release from naive AM. Taken together, these results support a major role of PMN in the restriction of TNF- activity, which is in accord with an earlier report of reduced expression of TNF- by LPS-stimulated monocytes after addition of activated PMN (40). Since PMN can produce significant amounts of sTNF-R and IL-10 both in vitro and in vivo (41, 42), our results are consistent with a downregulation of TNF- in the resolving and fibrosing alveolitis models through the expression of sTNF-R and IL-10 by PMN. In the rat, instillation of silica was also associated with a wide recruitment of PMN, but in contrast to the mouse, TNF- activity was upregulated after treatment with silica (Figure 6). The cause of this different pattern could conceivably be ascribed to the opposite IL-10 response observed in the rat (Figure 7). Further investigation will be necessary to determine the relevance of this observation.

What Is the Relationship between Downregulated TNF- and the Fibrotic Process?

In view of the well-recognized profibrotic activity of TNF-, it remains to be explained how fibrosis can develop in association with a downregulation of this cytokine.

Silica administration in different experimental animals has produced a wide range of pulmonary responses. For example, it has been shown that rats are highly susceptible to silica as compared with mice (43, 44). Differences in susceptibility were also observed in different strains of the same species. To explain these differences in sensitivity, investigators have evoked a relation between the capacity to mount a TNF- response to silica particles and the degree of fibrotic reaction (14, 45). In species known to be relatively resistant to the fibrosing effect of silica, a reduction of TNF- activity may therefore represent an attempt of the host to reduce the excessive activity of the proinflammatory and profibrotic TNF molecule. In other experimental models, the intensity and/or duration of the inflammatory response induced by an overproduction of TNF- is controlled by a secondary overproduction of antiinflammatory cytokines and mediators (46, 47). However, antiinflammatory cytokines involved in this control of TNF activity, including IL-10, which was identified in the present study, also exert profibrotic effects. We have previously shown that IL-10 expression reduces the amplitude of the inflammation in response to silica, but also accentuates the fibrotic reaction (38). Other antiinflammatory mediators in the lung, such as TGF-, IL-11, and IL-13, were also shown to act as profibrotic cytokines (48). Therefore, the results obtained in the comparative models can be interpreted as follows : (1) MnO₂ particles induced a transient inflammatory reaction in the lung (resolving alveolitis) that was accompanied by a transient upregulation of IL-10, contributing to a reduction in TNF- expression and in the intensity of the inflammatory response; (2) crystalline silica particles induced a protracted inflammatory reaction (fibrosing alveolitis) that was accompanied by a persisting overexpression of IL-10, which also contributed to limiting lung inflammation and the fibrogenic activity of TNF-. However, although this reaction of the host can in the first instance be regarded as homeostatic, its persistence (up to 120 d in the fibrosing alveolitis model) promoted the development of lung fibrosis. It is also striking to note that all of these antiinflammatory cytokines described here (IL-10, IL-11, IL-13, TGF-) are representative of a Th2-polarized immune process, which has been reported to be associated with fibrosis (51, 52).

In conclusion, the present study shows that in NMRI mice, the downregulation of TNF- during the course of pulmonary inflammation and fibrosis induced by inorganic particles is associated with and at least partly caused by an overproduction of sTNF-R and IL-10. This downregulation of TNF- is interpreted as an attempt of the host to reduce excessive activity of this proinflammatory and profibrotic cytokine. Although such a response may confer a certain degree of protection against the development of fibrosis, by preventing detrimental effects of TNF-, the protracted synthesis of profibrotic mediators such as IL-10 can in turn elicit and/or exacerbate the pulmonary fibrotic process. This mechanism may help in understanding the apparent contradiction observed in some experimental models of an association between reduced TNF- activity and the development of fibrosis.