Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

There is considerable evidence that active oxygen species (AOS) are important mediators of asbestos toxicity (1). Asbestos fibers themselves catalyze the formation of AOS in aqueous media: Fe(II) on the surface of the fiber reduces molecular oxygen to superoxide anion, which then dissociates to hydrogen peroxide, and hydrogen peroxide reacts with Fe(II) to yield the highly reactive hydroxyl radical. Whether the catalytic iron is that actually on the fiber surface or is iron that is mobilized from asbestos into the medium is as yet not established, but it is clear that iron is crucial to the process because chelation of fiber iron with agents such as deferoxamine (DFX) that render it redox inactive prevents the generation of AOS, and conversely, chelation with agents such as ethylenediaminetetraacetic acid (EDTA) that mobilize iron from the surface but leave it redox active increases AOS generation (3).

Asbestos-induced AOS are believed to produce a variety of acute abnormalities, including lipid peroxidation, protein oxidation, damage to DNA, activation of cell signaling cascades, particularly nuclear factor (NF)-B, NF-interleukin (IL)-6, and activator protein-1, and release of acute inflammatory mediators such as IL-1, IL-8 (or the murine macrophage inflammatory peptide-2), and tumor necrosis factor (TNF)- (1). AOS also increase adhesion of fibers to the surface of epithelial cells (6) and increase fiber uptake by epithelial cells (7), events that presumably lead to increases in the generation of the mediators just listed. Many of these effects can be prevented with DFX, again implicating fiber surface iron as a fundamental mediator. AOS may also be responsible for asbestos-induced autophosphorylation of the epidermal growth factor (EGF) receptor and subsequent activation of the extracellular signal-regulated protein kinase (ERK) pathway (2, 8).

The role of AOS in chronic forms of asbestos toxicity is much less well defined and is somewhat controversial. Mossman and coworkers (9) showed that administration of polyethylene glycol-conjugated catalase decreased inflammation and fibrosis in a rodent inhalation model of early asbestosis, implying that rapid removal of hydrogen peroxide, and/or prevention of formation of hydroxyl radical, was protective. Kamp and colleagues (10) reported that the iron chelator phytic acid had a similar effect in an instillation model, again implicating iron and AOS in the development of fibrosis. However, the mechanisms by which AOS might produce fibrosis are poorly defined, and other data suggest that fibrogenic mediators are more important than AOS in fibrogenesis. Platelet-derived growth factor (PDGF), transforming growth factor (TGF)-, and TGF- are all profibrotic peptides that cause fibroblast as well as epithelial proliferation and increased matrix production. Liu and associates (11, 12) and Perdue and Brody (13) have shown, using a high concentration-brief exposure inhalation model, that chrysotile asbestos exposure upregulates macrophage and/or epithelial production of PDGF, TGF-, and TGF- as well as TNF- and that increased TNF- production is crucial to the induction of both mediators and fibrosis because mice with both TNF- receptors knocked out are protected against the fibrogenic effects of asbestos (14). However, the relationship of fiber-induced AOS generation to the induction of fibrogenic mediators is as yet uncertain, and it is also uncertain which mediator(s) are most important in fibrogenesis.

Examination of fibrogenic events in whole animal models is complicated by the presence of both inflammatory and tissue reactions, and their interactions. We have developed a tracheal explant model that is free of exogenous inflammatory cells and thus allows a simpler investigation of fibrogenic processes in tissue. Using this model, we have previously shown that exposure of explants to asbestos results in increases in gene expression for PDGF-A, TGF-, and procollagen, as well increases in actual collagen content over a period of 7 d (15). In this study we use the explant model to examine the effects of iron and AOS on the production of fibrogenic mediators and matrix.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Dusts and Iron Loading

The asbestos used was the International Union Against Cancer amosite standard reference sample. The geometric mean fiber size ± standard deviation, as determined by counting by electron microscopy in our laboratory, was 3.8 ± 2.7 µm in length and 0.26 ± 1.9 µm in width. To load iron onto the fiber surface, the fibers were treated overnight with various concentrations of a freshly prepared mixture of equimolar Fe(II)/Fe(III) chloride. The fibers were then washed three times with saline to remove unbound iron and resuspended in culture medium. We have previously shown that this procedure results in reproducible increases in measurable surface iron (6).

Explant Preparation and Culture

Tracheal explants were prepared from 250 g male Sprague-Dawley mice as previously described (15). Each explant was approximately 2 × 2 mm. Because most of the explant by weight is cartilage and the amount of tissue that actually contributes RNA is extremely small, three explants were used to prepare RNA for each data point for reverse transcriptase polymerase chain reaction (RT-PCR) analysis. We have previously shown that this procedure provides a reliable signal (15). Freshly prepared explants from several different animals were used for each experiment and mixed to ensure that all explants for a given data point did not come from the same animal. Each test group consisted of three data points.

For dust exposure, the explants were submerged, epithelial side up, in a 500-µg/cm² (see DISCUSSION) suspension of dust in Dulbecco's modified Eagle's medium (DMEM) without serum for 1 h. Controls were exposed only to culture medium. At the end of this time, the explants were transferred to petri dishes containing DMEM in agarose supplemented with 1% glutamine, 1% penicillin/streptomycin/fungizone, 1 µg/ml insulin, 0.1 µg/ml hydrocortisone, 1.5× amino acids, and 10% chicken serum. Explants were maintained in air plus 5% CO₂ organ culture with basal feeding in an incubator at 37°C for 7 d because previous investigation has shown that there are few changes in the level of gene expression in this model before 7 d (15). All experiments were repeated. Representative data from single experiments are shown.

Measurement of Hydroxyproline

Hydroxyproline (HP) was measured on individual explants by high performance liquid chromatography using the methodology described previously (15).

Measurement of Phosphorylated ERK Protein Levels

Levels of phosphorylated ERK (phosphoERK) in the explants were very low and six segments were combined to produce each data point. The ERK inhibitor PD98059 was added in the same fashion described in subsequent text to show specificity. To assay phosphoERK levels, tracheal segments were frozen with liquid nitrogen and ground to a fine powder. The pulverized tissue was resuspended in lysing solution (20 mM Tris buffer with 1% Triton X-100, 137 mM sodium chloride, 10% glycerol, 2 mM EDTA, 25 mM glycerophosphate, 2 mM pyrophosphate, 0.5 mM AEBSF, 10 µg/ml leupeptin, 20 µg/ml aprotinin, 1 mM orthovanadate, and 10 mM sodium fluoride; all reagents from Sigma, St. Louis, MO). The lysates were centrifuged at 14,000 rpm for 10 min at 4°C. Phospho-p44/42 mitogen-activated protein kinase was immunoprecipitated by using a specific anti-p44/42 antibody (Cell Signaling, Beverly, MA) and protein A agarose beads (GIBCO-BRL, Grand Island, NY). Proteins were resolved on a 15% polyacrylamide gel under reducing conditions and immobilized onto a nitrocellulose membrane. For immunodetection, anti-p44/42 antibody was used with horseradish peroxidase-labeled secondary antibody and visualized with enhanced chemiluminescence (Amersham, Arlington Heights, IL). As a positive and negative control, phosphorylated and nonphosphorylated ERK proteins were used (Cell Signaling).

Measurement of TGF-₁ Protein Levels

Measurement was carried out using a Quantikine enzyme-linked immunosorbent assay (ELISA) kit from R&D Systems (Minneapolis, MN). Because the explants are maintained in air organ culture, explant tissue itself was used in the analysis. Explants were homogenized under liquid nitrogen, and the powder was resuspended in 500 µl lysing solution on ice for 20 min. The lysing solution contained 10 mM N-2-hydroxyethylpiperazine-N'-ethane sulfonic acid (Hepes), pH 7.5, 0.5% Triton X-100, 150 mM NaCl, 1 mM EDTA, 0.5 mM AEBSF, 1 µg/ml leupeptin, 1 µg/ml aprotinin, 10 µg/ml trypsin-chymotrypsin inhibitor, and 1 µg/ml pepstatin A (all reagents from Sigma). The samples were then centrifuged at 14,000 rpm for 10 min at 4°C, and the supernatants were decanted and kept on ice. TGF- was activated by adding 500 µl 2.5 N acetic acid/10 M urea solution to 500 µl supernatant. This mixture was incubated at room temperature for 10 min and then neutralized with 2.7 N NaOH/1 M Hepes. ELISA was then performed according to the manufacturer's instructions.

Effects of Added Iron Without Fibers

To determine whether the effects observed with iron loading were specifically related to fiber-bound iron, explants without dust were incubated with 1 mM Fe(II)/Fe(III) mixture described previously for 1 h, then transferred to agarose culture dishes with the same iron mixture added to the medium, and assayed after 7 d in culture.

Treatment with Inhibitors and Chelators

Iron chelator DFX. Amosite asbestos was incubated overnight with 10 mM DFX (Desferal; Ciba-Geigy) and excess DFX was removed by washing the fibers in saline before use. Fibers were then resuspended in culture medium and used as described previously.

Proteasome inhibitor MG-132. Explants were submerged in 0.5 µM MG-132 (Peptide Institute Inc., Osaka, Japan) in 0.1% dimethyl sulfoxide/DMEM for 1 h and were then exposed to asbestos fibers as described previously, but with MG-132 in the medium. A total of 0.5 µM MG-132 was also included in the agarose culture medium for the 7-d incubation period.

AOS scavenger tetramethylthiourea. Explants were exposed for 1 h to 10 mM tetramethylthiourea (TMTU) (Sigma) and then to asbestos fibers with TMTU in the medium. A total of 10 mM TMTU was also included in the agarose culture medium for 7 d.

ERK inhibitor PD98059. Explants were exposed for 1 h to 50 µM PD98059 (Calbiochem, La jolla, CA) and then to asbestos fibers with PD98059 in the medium. A total of 50 µM PD98059 was also included in the agarose culture medium for 7 d.

Nonmembrane permeable AOS scavenger glutathione. Explants were exposed for 1 h to 10 mM glutathione (GSH) (Sigma) and then to asbestos with GSH in the medium. GSH was also included in the agarose medium for 7 d.

Gel Shift Assay for NF-B

To confirm that asbestos did activate NF-B in our explant system, additional explants were treated with asbestos or asbestos plus MG-132 as described previously and incubated for 7 d. Explants were snap-frozen and then homogenized in 0.1% Triton X-100, 150 mM NaCl, 10 mM Hepes, pH 7.5, 1 mM EDTA, 0.5 mM AEBSF, 1 mg/ml leupeptin, 1 µg/ml aprotinin, 10 µg/ml soybean trypsin inhibitor, and 1 µg/ml pepstatin A. The homogenate was incubated on ice for 5 min and then centrifuged at 5,000 rpm for 5 min. The pelleted nuclei were resuspended in 100 to 500 µl of a solution of 25% glycerol, 20 mM Hepes, 420 mM NaCl, 1.2 mM MgCl₂, 0.2 mM EDTA, 0.5 mM dithiothreitol (DTT), 0.5 mM AEBSF, 1 µg/ml leupeptin, 1 µg/ml aprotinin, 10 µg/ml soybean trypsin inhibitor, and 1 µg/ml pepstatin A, and left on ice for a 30-min high salt extraction of the nuclear proteins. The lysed nuclei were centrifuged at 5,000 rpm for 15 s and a protein assay was carried on the supernatant. Single stranded NF-B consensus oligonucleotide (5' - AGT TGA GGG GAC TTT CCC AGG C- 3') was random labeled with [³²P]cytidine triphosphate. Binding reactions containing equal amounts of protein (7 µg) and 6.7 pmol of oligonucleotide were performed for 20 min in binding buffer (10 mM Tris HCl, 50 mM NaCl, 1 mM EDTA, 4% glycerol, 67 µg/ml poly(dI-dC)). Reaction products were separated in a 5% polyacrylamide gel in 0.25× TBE buffer and analyzed by autoradiography and densitometry.

Expression of Growth Factors, Fibrogenic Mediators, and Matrix Components by RT-PCR

After 7 d in organ culture, explants were harvested and RNA extracted by the method of Chomczynski and Sacchi (16). First strand complementary DNA (cDNA) was synthesized using superscript RNase H reverse transcriptase (GIBCO-BRL) according to the manufacturer's instruction. Briefly, 5 µg RNA were added to a reaction mixture of 1× first strand buffer, 200 ng oligo(dT)12-18 primer, 0.5 mM each deoxyadenosine triphosphate, deoxythymidine triphosphate, deoxyguanidine triphosphate, and deoxycytidine triphosphate, 0.1 M DTT, plus water to 49 µl. A total of 200 U superscript RT was added and the reaction incubated at 42°C for 1 h.

PCRs contained 1 µM primers, 1.5 mM Mg⁺⁺, 200 µM deoxynucleotide triphosphates, reaction buffer, 2.5 U of Taq DNA polymerase (Perkin-Elmer Cetus Instruments, Norwalk, CT) and 1 or 5 µl of cDNA in a final volume of 20 µl. The PCR temperature profile consisted of 25 or 28 cycles of denaturation at 94°C for 45 s, primer annealing at 60°C for 45 s, and extension at 72°C for 1.25 min, followed by an additional 5 min of final extension at 72°C. The PCR products were size-fractionated on 1.5% agarose gel and quantified from this ethidium bromide-stained gel using a gel documentation system (Bio-Rad Laboratories, Hercules, CA).

Primer design was based on sequences from the Genbank database (Table 1). We optimized the reaction conditions (magnesium concentration, thermocycler temperature, etc.) to produce the greatest amount of a single PCR product. The amount of cDNA used and number of cycles of amplification were adjusted to stay within the linear region of amplification in order to allow quantification. Specificity of the various products was confirmed by restriction digests. Expression of malate dehydrogenase (17) was used as control (housekeeper) gene.

For TNF-, expression was also examined at 2 and 8 h after dust exposure to determine whether the apparent lack of long-term upregulation might be hiding early upregulation. As a positive control, explants were treated for 2 h with 20 ng/ml of recombinant human TNF- (specific activity > 2 × 10⁷ U/mg; Life Technologies, Rockville, MD).

Statistics

Comparisons for gene expression and HP content were made among treatment groups by analysis of variance.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Analysis of type I procollagen gene expression and tissue HP levels were used as the measure of asbestos/iron effects on matrix production. The effect of iron loading on procollagen gene expression is shown in Figure 1. Asbestos by itself roughly doubled procollagen gene expression and there was a further progressive increase with increasing levels of surface iron. Data on tissue HP levels are shown in Figure 2: levels were significantly greater after asbestos treatment compared with control levels, and greater again with loading with 1 mM iron (use of multiple iron loading levels was not attempted because the differences in HP content are small). Figure 2 also shows that incubation of control explants with 1 mM iron as described in MATERIALS AND METHODS, but without dust, did not increase HP content. The effects of the iron chelator DFX and the cell permeable AOS scavenger TMTU, both of which completely abolished the asbestos-induced increases in procollagen gene expression, and the cell membrane impermeable AOS scavenger GSH, which did not prevent increased procollagen gene expression, are shown in Figure 3. The effects of proteasome inhibitor MG-132 (used here as an inhibitor of NF-B activation) and the ERK inhibitor PD98059 are shown in Figure 4; the former completely prevented the asbestos-induced increases, whereas PD98059 did not affect asbestos-mediated increases in procollagen gene expression at all.

View larger version (23K): [in this window] [in a new window]   Figure 1.   Increasing levels of iron loading on amosite asbestos increase procollagen gene expression. Ethidium bromide images show RT-PCR products, with densitometry below. MDH = housekeeper (malate dehydrogenase). A only = asbestos alone; A/[x] mM Fe = asbestos + [concentration] of loaded iron. Values are mean ± SD. *Significantly greater than control. Asbestos/0.1 mM Fe is also significantly greater than asbestos alone, and asbestos/1 mM Fe is greater than asbestos/0.1 mM Fe.

View larger version (10K): [in this window] [in a new window]   Figure 2.   Tissue hydroxyproline content is increased in asbestos-treated explants compared with control explants and further increased by loading with 1 mM iron. These findings indicate that hydroxyproline production is a function of fiber surface iron levels. Addition of iron to the medium ("iron alone") without dust has no effect on hydroxyproline content. Values are mean ± SD. *Significantly greater than control. Asbestos + 1 mM Fe is also significantly greater than asbestos alone.

View larger version (24K): [in this window] [in a new window]   Figure 3.   The iron chelator DFX and the membrane permeable AOS scavenger TMTU protect against asbestos-induced increases in procollagen gene expression, indicating that the iron-catalyzed generation of AOS plays a role in increased procollagen expression. The nonmembrane permeable AOS scavenger GSH is not protective, indicating that the important reactions must be intracellular. A Only = asbestos alone. Values are mean ± SD. *Significantly greater than control.

View larger version (21K): [in this window] [in a new window]   Figure 4.   Proteasome inhibitor MG-132 prevents asbestos-induced increases in procollagen expression, whereas ERK inhibitor PD98059 does not, suggesting that NF-B activation is involved in increased procollagen expression. Values are mean ± SD. A only = asbestos alone. *Significantly greater than control.

To show that asbestos did activate NF-B in our explant system, a gel shift assay was run and is shown in Figure 5. The gel shift assay is carried out on nuclear extracts and indicates the relative amount of NF-B protein in the nucleus (i.e., translocated NF-B). Asbestos increased NF-B activity and this increase was prevented by inclusion of MG-132 in the medium. To show that asbestos fibers did activate the ERK pathway, proteins were extracted and phosphoERK levels were measured on Western blot. As shown in Figure 6, phosphoERK levels were greater in asbestos-exposed explants compared with control explants and greater again in explants treated with iron-loaded asbestos. The ERK inhibitor PD98059 completely abolished asbestos-induced increases in phosphoERK.

View larger version (37K): [in this window] [in a new window]   View larger version (61K): [in this window] [in a new window]   Figure 5.   Gel shift assay showing that asbestos exposure increases NF-B activation in explants after 7 d of culture. Inclusion of MG-132 in the medium prevents this effect. 1 = control; 2 = asbestos; 3 = asbestos + MG-132. Each bar represents data from six explants combined.

View larger version (18K): [in this window] [in a new window]   Figure 6.   Treatment with asbestos increases the level of phosphoERK, and treatment with asbestos + 1 mM iron increases it still further. Addition of the ERK inhibitor PD98059 to the medium prevents asbestos-induced increases in phosphoERK. Data shown are representative Western blots with each signal derived from six explants combined. 1 = control; 2 = asbestos alone; 3 = asbestos + 1 mM Fe iron; 4 = asbestos + PD989059 as described in text; 5 = negative control (nonphosphorylated ERK); 6 = positive control (phosphoERK).

Figures 7-9 show similar data for TGF-₁ gene expression and Figures 10-12 for PDGF-A gene expression. Asbestos by itself increased expression of both mediators, and adding surface iron produced further increases. DFX and TMTU again abolished the asbestos effect, but GSH did not. In contrast to the situation for procollagen, MG-132 did not prevent asbestos-induced increases in PDGF-A or TGF- gene expression, but PD98059 did. Attempts to examine levels of TGF- protein by ELISA were unsuccessful: tissue levels were at the bottom of the ELISA range and showed no differences among treatment groups (data not shown). It is unclear whether there really are no differences in TGF- content (i.e., increased gene expression is not translated into protein) or the protein levels are simply too low to detect; however, in some senses this question is not crucial because it is clear that blocking PDGF or TGF- gene expression with PD98059 does not prevent increases in procollagen gene expression, thus PDGF and TGF- are not driving collagen production in this system.

View larger version (23K): [in this window] [in a new window]   Figure 7.   Increasing levels of iron loading on amosite asbestos increase TGF-1 gene expression. Values are mean ± SD. A only = asbestos alone; A/[x] mM Fe = asbestos + [concentration] of loaded iron. *Significantly greater than control. Asbestos/0.1 mM Fe is also significantly greater than asbestos alone, and asbestos/1 mM Fe is greater than asbestos/0.1 mM Fe.

View larger version (22K): [in this window] [in a new window]   Figure 8.   DFX and TMTU protect against asbestos-mediated increases in TGF- gene expression, whereas GSH does not. Values are mean ± SD. A only = asbestos alone. *Significantly greater than control.

View larger version (21K): [in this window] [in a new window]   Figure 9.   Proteasome inhibitor MG-132 does not prevent asbestos-induced increases in TGF- gene expression, whereas ERK inhibitor PD98059 is protective. Values are mean ± SD. A only = asbestos alone. *Significantly greater than control.

View larger version (22K): [in this window] [in a new window]   Figure 10.   Increasing levels of iron loading on amosite asbestos increase PDGF-A gene expression. Values are mean ± SD. A only = asbestos alone; A/[x] mM Fe = asbestos + [concentration] of loaded iron. *Significantly greater than control. Asbestos/0.1 mM Fe is also significantly greater than asbestos alone, and asbestos/1 mM Fe is greater than asbestos/0.1 mM Fe.

View larger version (23K): [in this window] [in a new window]   Figure 11.   DFX and TMTU protect against asbestos-mediated increases in PDGF-A gene expression, whereas GSH does not. Values are mean ± SD. A only = asbestos alone. *Significantly greater than control.

View larger version (20K): [in this window] [in a new window]   Figure 12.   Proteasome inhibitor MG-132 does not prevent asbestos-induced increases in PDGF-A gene expression, whereas ERK inhibitor PD98059 is protective. Values are mean ± SD. A only = asbestos alone. *Significantly greater than control.

The effects of asbestos and iron loading on TNF- expression are shown in Figure 13: neither asbestos nor iron-loaded asbestos increased expression. A similar lack of effect was seen in explants incubated for only 2 or 8 h (data not shown). Exogenous TNF-, used as a positive control, did upregulate explant TNF- levels within 2 h (Figure 13). Inclusion of AOS scavengers, MG-132, and PD98059 did not affect expression of TNF- (data not shown). Identical results were seen for PDGF-B and TGF- expression (data not shown).

View larger version (22K): [in this window] [in a new window]   Figure 13.   Neither asbestos alone nor iron-loaded asbestos affect gene expression of TNF-. GSH, DFX, and TMTU similarly had no effect, and inclusion of MG-132 or PD98059 in the medium also did not change expression levels (data not shown). Identical results were seen for PDGF-B and TGF-. A only = asbestos alone; A/[x] mM Fe = asbestos + [concentration] of loaded iron. Minus and plus indicate a separate experiment done to show that incubation of explants with 20 ng/ml of TNF- for 2 h (+) leads to upregulation of TNF- compared with untreated explants (). Densitometry is not shown, but increase is about 2.5-fold.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

We have used a tracheal explant system to examine the role of AOS and surface iron on the expression of asbestos-induced fibrogenic mediators and matrix. As noted previously, examining fibrogenic events can be difficult, and the explants simplify this process because they lack inflammatory cells but preserve the important modulating influences of epithelial cells on mesenchymal cells and vice versa (18). However, the explant system also has its peculiarities: dust uptake is very slow and gene expression appears to follow dust uptake, hence 7-d cultures are required to see effects (15). As well, relatively high dust loadings are required to produce effects. However, the dose that the epithelium "sees" is probably very much lower because we have shown in experiments looking at fiber adhesion that the vast majority of fibers on the explant surface are only very loosely adherent (6). Nonetheless, these limitations need to be kept in mind in interpreting our results.

Our data suggest that AOS and surface iron play an important role in the development of asbestos-induced fibrosis and affect expression of both matrix proteins and fibrogenic mediators, but in different ways. The current experiments show that it is a clear relationship between increasing levels of surface iron and increases in the expression of procollagen messenger RNA (mRNA) and HP, and that addition of a cell permeable AOS scavenger (TMTU) or an iron chelator (DFX) that renders surface iron redox inactive prevents asbestos-induced increases in procollagen expression. These observations imply that asbestos-induced increases in collagen production are driven, at least in part, by AOS, presumably through the iron-catalyzed formation of hydroxyl radical. Thus, these observations support the previous findings of Mossman and coworkers (9) and Kamp and colleagues (10) that suggested that asbestos can induce fibrosis through an oxidant mechanism. Because our explant system has no exogenous inflammatory cells, it is clear that AOS derived from the fibers are able, in and of themselves, to cause procollagen expression. However, all dusts evoke an inflammatory response, and it is possible that, in vivo, AOS derived from dust-evoked inflammatory cells augment this process (see subsequent text). Also of interest is the observation that the nonmembrane permeable AOS scavenger GSH was not protective, which implies that fiber-derived AOS must be acting within epithelial and/or interstitial cells rather than on the cell surface, an idea supported by our previous observation that increases in gene expression parallel (over time) increases in dust uptake by the epithelial and interstitial cells (15).

There is evidence in other experimental systems that iron affects collagen production. This effect has been seen most clearly in hepatocytes, where iron loading of the sort found in hemochromatosis has been shown to increase the activity of prolyl hydroxylase, a crucial enzyme for collagen production (19, 20). Iron overload also causes hepatocytes to increase expression of type I procollagen (19, 20). In a model more relevant to this study, Gardi and associates (20) showed that mobilizing iron from crocidolite asbestos with redox active chelators such as citrate increased collagen production by cultured rat lung fibroblasts. Unfortunately, our data do not allow us to determine whether the crucial effects in our explants occur in epithelial or interstitial cells, but it is of interest that adding iron to the culture medium did not affect generation of procollagen, implying either that iron bound to the asbestos fiber surface is crucial to this process or that iron must be mobilized from the fiber surface with the appropriate chelator to be active in fibrogenesis.

NF-B resides in the cytoplasm as an inactive DNA-binding dimer (NF-B itself) and an inhibitory protein, IB. NF-B can be activated by a variety of pathways, most commonly when an external stimulus leads to IB phosphorylation, ubiquitination, and consequent degradation in proteasomes. The NF-B complex then migrates to the nucleus and activates a wide variety of genes involved in the acute inflammatory response (21, 22). MG-132 is a proteasome inhibitor that prevents degradation of IB (23). As noted previously, asbestos fibers have been shown to activate NF-B in monolayer tissue culture systems (1, 2), and we have confirmed with the gel shift assay that NF-B is similarly activated over a relatively long period in our explants. The fact that MG-132 completely prevents the asbestos-induced increases in procollagen expression, not only with the native fibers but also with the iron-loaded fibers (data not shown), suggests strongly that increased procollagen production is mediated in some fashion through AOS-induced activation of NF-B. It is possible that prolyl hydroxylase has an NF-B recognition site in its promoter, but there appears to be no data on this question, and several intermediate steps may also be involved.

Oxidant attack is known to activate NF-B (21, 24, 25), and the present experiments thus support other observations that the generation of AOS from asbestos fibers directly activates NF-B (1, 2). However, what is surprising is that there is no clear evidence for involvement of TNF- in our model. The in vivo studies of Liu and coworkers (11, 12, 14) and Perdue and Brody (13) indicate that TNF- is crucial to chrysotile asbestos-induced fibrogenesis because mice lacking TNF- receptors are protected. TNF- itself activates NF-B (24, 25) and it is possible that, in vivo, the production of TNF- from asbestos-evoked macrophages simply overshadows the direct effects of the dust, particularly with chrysotile asbestos which has relatively little iron compared with the amosite used here.

Explant tracheal epithelial cells certainly can upregulate TNF- expression, as is clear from our positive (TNF- driven) control. In this and our previous work (15), we were unable to find evidence of upregulation of TNF- gene expression from 2 h to 7 d of asbestos exposure, so it is unlikely that we have missed a brief episode of increased transcription followed by prolonged increases in protein production. It is noteworthy that iron loading also had no effect on TNF- expression in our explants. Although it is possible that increased levels of TNF- do occur and are produced purely by increased translation of pre-existing mRNA or by increased release of preformed TNF-, this appears unlikely because in other cell types, asbestos-related stimuli to TNF- production do increase gene expression. For example, Simeonova and Luster (26) have shown that iron loading of asbestos increases TNF- gene expression in cultured alveolar macrophages by greater than 10-fold.

Although we do not have a definite explanation of why TNF- gene expression is not upregulated in our explants, the roles of AOS and TNF- in activating NF-B are complex. It is becoming clear that TNF- and oxidants activate NF-B via quite different mechanisms (21, 24, 27). This applies not only to nuclear translocation but to generation of transcriptionally competent NF-B bound to promotor elements in DNA. Even oxidative stress itself appears to activate separate pathways leading to NF-B DNA binding and NF-B transactivation (21). Whether and to what degree these types of differences occur in tracheal epithelial cells are not known but the important point is that AOS derived from the dust itself and TNF- derived from airspace macrophages that have phagocytized dust may activate different genes. Thus, it is possible that TNF- derived from airspace macrophages is required to upregulate explant TNF- and that AOS are not sufficient. This notion might provide an explanation for the observation that in the model developed by Liu and coworkers (11, 12, 14) and Perdue and Brody (13), chrysotile asbestos upregulates expression of TNF-, PDGF-B, and TGF-, none of which is increased in our explants. Some indirect support for this idea comes from the observation by Gallucci and colleagues (28) that TNF- upregulates expression of TGF- in murine liver cells.

Relatively little information is available about the effects of AOS on induction of TGF- and PDGF. Bellocq and coworkers (29) recently reported that exposure of cultured A549 cells (a model of human type II cells) to AOS resulted in increased TGF-₁ release, largely through increased transcription. AOS are also known to activate latent TGF- (30). The current data support a role for iron-derived AOS in TGF-₁ transcription. It has also been reported that high concentrations of GSH prevent release of PDGF from platelets (31), possibly indicating that PDGF production can be driven by AOS. Our data show that this process differs for different forms of PDGF because in our hands both asbestos and iron-loaded asbestos only upregulate gene expression of PDGF-A, a finding in accord with the observations of Lasky and associates (32, 33) that chrysotile asbestos particularly stimulates production of PDGF-AA and upregulates the PDGF-receptor .

What is clear from our data is that in this model system, AOS drive gene expression of PDGF-A and TGF- in a fashion quite different from that of procollagen because only the ERK inhibitor PD98059, but not the NF-B inhibitor MG-132 prevents PDGF-A and TGF-₁ upregulation and vice versa. As noted previously, Robledo and Mossman (2) and Zanella and coworker (8, 34) have shown that crocidolite asbestos activates the EGF receptor, possibly via AOS, and that this leads, through an ERK-dependent mechanism, to increases in c-fos expression and apoptosis. The same group has recently reported that increased phosphoERK is detectable by immunohistochemistry in the areas of beginning fibrosis in chrysotile-exposed mice (35). Our data also suggest that asbestos-induced ERK pathway activation is mediated, at least in part, by AOS and that this pathway may be involved in asbestos- induced activation not only of genes involved in acute responses but of a variety of genes involved in chronic fibrotic responses. Thus, fibrogenesis, at least in our model, appears to proceed via two separate pathways: one directly affects procollagen production and the other affects expression of fibrogenic mediators.