Role of transforming growth factor-β1 in down-regulating TNF production by alveolar macrophages during asbestos-induced pulmonary fibrosis

Activation of alveolar macrophages (AM) for tumour necrosis factor production is suppressed initially during the inflammatory response to fibrogenic dusts. We investigated the mechanisms involved in TNF suppression, notably the role of other AM-derived mediators including prostaglandin E2 (PGE2), transforming growth factor-β1 (TGF-β1), and interleukin 6 (IL-6). The action of PGE2 and TGF-β1, on AM was different. At physiologically relevant doses (25–300 pg/ml), PGE2 did not cause significant inhibition of Hpopolysaccharide (Lps)-induced TNF release by AM in vitro but stimulated IL-6 (up to six fold), an inhibitor of AM-derived TNT. In contrast, TGF-β1 (0.5–50 ng/ml) inhibited both LPS-induced TNT and IL-6 release by 50% but had no effect on PGE2 production by AM. To determine the respective contribution of these different inhibitors in TNF suppression, AM from rats exposed to fibrogenic asbestos for weeks were treated with neutralizing antibody against TGF-β1 or indomethacin, an inhibitor of PGE2 synthesis. Treatment of rat AM with anti-TGF-β1 but not indomethacin, abrogated the observed TNT suppression. These results suggest that an autocrine, TGF-β1-dependent mechanism is involved in the down-regulation of TNF production by rat AM from animals with lung fibrosis.


Introduction
Alveolar macrophages (AM) are an important source of numerous cytokines and are a primary cell type in chronic inflammatory reactions induced by fibrogenic dusts. 2 It is likely that the contribution of AM-derived cytokines plays an important role in the maintenance and progression of pulmonary inflammation. Among these, tumour necrosis factor-or (TNF) has attracted much attention owing to a wide range of activities including fibroblast rowth-promoting and growth-inhibiting activities. Together with other cytokines, TNF is frequently found at sites of inflammatory reactions and higher levels of TNF messenger RNA were identified in lung tissue of mice treated with silica 5 and bleomycin. 6 With regard to basal and LPS-stimulated TNF production by AM, variable responses including stimulation as well as inhibition or no change have been reported in experimental silicosis. 7-I0 Less information is available concerning the contribution of AM-derived TNF in asbestos-induced fibrosis. In a previous study, we reported bidirectional changes in LPS-stimulated TNF production by rat AM exposed to fibrogenic asbestos. 9 These changes were characterized initially by a 50% suppression of TNF and were associated with the development of lung fibrosis but were absent in resolving lung granuloma. TNF inhibition was observed at a time when the bronchoalveolar cell population was composed exclusively of macrophages, raising the possibility of a macrophagerelated inhibitory mechanism(s). Therefore, the present work was undertaken to study the role of macrophage-derived mediators, notably PGE2 and TGF-, in the observed TNF suppression.
We report here that down-regulation of AMderived TNF production was completely abrogated by TGF-I3 neutralization whereas indomethacin treatment had no effect. These results suggest that TNF response in AM from animals with lung fibrosis is inhibited in an autocrine fashion by TGF-.

Materials and Methods
Induction of lung inflammation and fibrosis: Male Wistar rats weighing 225-250g were pur-(C) 1996 Rapid Science Publishers chased from Charles River Canada, Inc. (St Con-lected by centrifugation and frozen at -80C stant, Quebec). These animals were derived from until assayed. a pathogen-free colony, shipped behind filter barriers, and housed in isolated temperature-con-Assays of TNF, IL-6 and TGF-flz: TNF activity was trolled quarters in an animal isolator unit (Johns determined as described 9 using the L929 routine Scientific Inc., Toronto). Lung inflammation and fibroblast lysis assay. Serial dilutions of test fibrosis were induced by intratracheal instillation supernatant were added to 4.5 x 104 L929 cells of a preparation of chrysotile asbestos fibres as in the presence of actinomycin D (1 l.tg/ml) described previously. 9 UICC Canadian chrysotile (Boehringer Mannheim) and incubated for 18h B asbestos fibres, (21% > 10gm) ) were in microtitre plates. The supematants were disobtained from the National Research Institute for carded; the remaining adherent viable cells were Occupational Diseases, Johannesburg, South stained with crystal violet (0.5% in 2% methanol) Africa. Asbestos fibres were autoclaved for 45 and the absorbance of each well was read at min and suspended in sterile phosphate-buffered 540 nm using an automated Bio-Tek microplate saline (PBS, pH 7.4) with a Dounce glass homoreader (Mandel Scientific). Each assay was stangenizer (Fisher Scientific, Ottawa) before instilla-dardized with murine recombinant TNF tion into the animals. Under anaesthesia, the (4 x 107U/mg, Genzyme) and TNF units were trachea was exposed surgically and saline or calculated by probit analysis. TNF bioassay was chwsotile B fibres were briskly injected through validated by using a rabbit antimurine TNF antian 18-gauge needle. Two groups of seven rats body (Genzyme Corporation, Boston, MA) which each, received respectively, a single intratracheal completely neutralizes the cytotoxicity of AM injection of saline (control) or chrysotile B conditioned media.
(5 mg). The rats from each group were sacrificed IL-6 activity was measured with the standard 3 weeks after treatment and were analysed by B9 cells proliferation assay as described. B9 bronchoalveolar lavage (BAL), for BAL cell popu-cells (0.5 x 104) were incubated in 0.2ml of lations and AM-derived TNF production. Histolo-Iscove's Modified Dulbecco's Medium (IMDM) gical examination of lung sections stained with (Grand Island Biological Co.)supplemented with haematoxylin-eosin or Masson's trichrome 5% FBS and 5 x 10-5M 2-mercaptoethanol in revealed the presence of fibrosis in the lungs of the presence of AM supematants at various dilurats exposed to UICC chrysotile fibres. 9 The tions. Tritiated thymidine (1 btCi/well) was added lesions consisted of granulation tissue with fibro-after 66 h of incubation and cultures were harblastic proliferation and collagen deposition and vested at 72h with a Skatron filtration device. were localized predominantly in and around The sample dilution curve was related to a stanterminal bronchioles, dard curve generated with recombinant murine IL-6 (Genzyme, 108U/mg) and IL-6 units were Alveolar macropbage culture: Mveolar macro-calculated by probit analysis. phages were obtained by bronchoalveolar lavage TGF-[31 activity was determined using the selecas described previously. 9 Cells were counted in a tively sensitive MvlLu, mink lung connective tissue haemocytometer chamber and viability (98cell line (American Type Culture Collection, Rock-100%) was determined by trypan blue exclusion, ville, MD: no. CCL-64) as described. 14 Cells Differential analysis of lavage cells made from (2.5 x 104/well) were cultured for 4h in 96-well cytocentrifuge smears (Shandon, 2.5 x 104 cells) flat-bottomed microtitre plates (Costar, Camand stained with Wright-Giemsa indicated that bridge, MA) in MEM supplemented with 10% the BAL cell population was essentially corn-foetal bovine serum (FBS) (Grand Island Biologiposed of macrophages (99% AM) in normal rats. cal Co., Grand Island, NY). AM culture super-Similarly, in saline and asbestos-treated rats, the natants were then added in triplicate and BAL cell population was 99% AM, the major difincubation was performed for 18 h. Cells were ference between these two groups being a sig-pulsed for 6h with [H]thymidine (1.SgCi/well) nificant increase in AM number in asbestos-(Dupont NEM). After washing with PBS, cells treated animals (11 x 10 6 and 19 x 10 6 respecwere treated with trypsin-EDTA (0.5-0.03%) for tively, for saline and asbestos groups). For deter-40 min at room temperature and collected with a mination of TNF, IL-6 and TGF-[3 bioactivities, Skatron harvester. Results were calculated based AM (0.2 x 106) from normal rats and when indion the decrease of [H]thymidine incorporation cated, from saline and asbestos-treated rats, were compared with natural human TGF-]3 standard incubated in 0.2ml of DMEM supplemented with (Upstate Biotechnology Inc., NY) and data are 0.5% FBS for 18h at 37C in the presence and expressed as ng/ml. The specificity of the assay absence of LPS (1 I.tg/ml) (Sigma Chemical Co., was demonstrated by using chicken anti-human St Louis, MO). Culture supematants were col-TGF-[3 (R&D Systems)which inhibits the activity. Prostaglandin E2 determination: Prostaglandin E2 (PGE2) was determined from AM supematants using an ELISA kit (Cayman Chemical, Ann Arbor, MI). The ELISA was a competitive acetylcholinesterase-linked immunoassay and was performed according to the manufacturer's instruction.
Neutralization and blocking experiments: AM obtained from animals exposed to saline or UICC chrysotile B asbestos for 3 weeks were incubated as follows: AM (10*/ml) from each group were incubated for 18h at 37C with LPS (1 l.tg/ml) alone or in the presence of a purified turkey IgG antibody (101g) to human transforming growth factor-]3 (anti-TGF-J3)which displays neutralizing activity against rat TGF-[31 (Collaborative Research, Bedford MA), turkey serum as a control for anti-TGF-J3 or indomethacin (10-5 M) (Sigma, St Louis, MO). AM culture supematants were collected and frozen at -80C for TNF measurement as described above.
Statistical analysis: Results were expressed as mean values _ _ _ S.E.M. Statistical significance of differences between treated and control groups were determined using a one-way analysis of variance and Bonferrani test (Instat) (p < 0.05).

Results
PGE2 inhibited LPS-induced TNF release from AM by 50% in a dose-dependent manner with half-maximal effect observed at 25 Ig/ml (Fig. 1A). This is much higher than the levels of PGE2 produced by LPS activated AM (805 pg/ml) ( Table 1) indicating that physiological concentrao tions of PGE2 are not sufficient to inhibit TNF production by rat AM. In contrast, relatively small concentrations of PGE2 (25-300 pg/ml) had a direct stimulatory effect on spontaneous IL-6 release (7 U/ml), an inhibitor of TNF production by rat AM. Up to six-fold stimulation (46 U/ml) was obtained with 300pg/ml PGE2 (Fig. 1B). In addition, LPS-stimulated IL-6 release (75U/ml) was further increased two-to five-fold by PGE2 within the same concentration range.
On the other hand, TGF-[3 at concentrations ranging between 0.5 and 50ng/ml caused significan inhibition of LPS-induced TNF ( Fig. 2A) and It-6 ( Fig. 2B) release by rat AM with maximal inhibition (50%) obtained with 2.5-5 ng/ml. In addition, TGF-]3 had a potent inhibitory effect (54%) on LPS-induced PGE2 production by AM (Table 1) To investigate which pathway(s) may be involved in the down-regulation of AM-derived TNF following asbestos exposure, AM obtained from rats exposed to UICC chrysotile asbestos L Lemaire and S. Ouellet for 3 weeks were incubated in the presence and absence of anti-TGF-l known to neutralize rat TGF-[ ,14 turkey serum as a control, and indomethacin, an inhibitor of PGE2 synthesis. Treatment with anti-TGF-but not with turkey serum or indomethacin abrogated the 50% suppression of AM-derived TNF seen 3 weeks after asbestos exposure (Fig. 3). At this time, AM culture media from asbestos exposed rats contained levels of TGF-(1.2ng/ml) sufficient for inhibition of TNF release (Table 2).

Discussion
We have investigated the mechanisms involved in TNF suppression during the development of asbestos-induced fibrosis, 9 at the macrophage level. A variety of mediators are known to inhibit 5 TNF production from cells including IL-4, IL-  IL-4 is produced by activated T-lymphocytes, 2 a cell type which is not present in the bronchoalveolar compartment of our inflammatory lung model. 2' Because the alveolar macrophage (AM) is a prominent feature of lung reactions to asbestos, 2 our study focused on mediators produced by AM, notably PGE2, IL-6 and TGF-[. As reported in other systems, m TGF-[ exerts significan inhibitory effects on TNF release by activated rat AM, conditions that are consistent with inflammatory states. TGF-I inhibits TNF release within the concentration range produced by AM suggesting that it could act in an autocrine fashion to modulate TNF production in the pulmonary microenvironment. In addition, TGF-[ inhibits LPS-induced PGE2 indicating that it suppresses TNF production through a PGE2-independent pathway. This is consistent with previous findings in peripheral blood mononuclear cells 19 and murine macrophage cell lines. 2 Furthermore, TGF-] also inhibits IL-6 release from activated AM, ruling out the contribution of IL-616 in mediating its suppressive effect on TNF. Although TGF-fl has been reported to stimulate IL-6 production in fibroblasts, 22 chondrocytes and peripheral blood lymphocytes, 24 it has been shown to suppress IL-6 in bone marrow cells. 25  inhibition of TNF whereas indomethacin treatment has no effect. Furthermore, in contrast to controls, AM from animals exposed to asbestos produce amounts of TGF-I3 sufficient to cause inhibition. The contribution of other inhibitors, notably IL-6 and IL-10, cannot be completely ruled out. However, our observations that PGE2 at physiological concentrations stimulate IL-6, do not suggest a role for IL-6. In support of this, previous work from this laboratory demonstrated a lack of correlation between AM-derived IL-6 and TNF production following asbestos exposure. Similarly, since PGE2 deactivation of macrophages has recently been shown to be mediated by IL-10, 29 our results are not consistent with a role for IL-10. Further experiments would be necessary to address these issues. Collectively, our findings support a role for