Regulation of human lung fibroblast alpha 1 ( I ) procollagen gene expression by tumor necrosis factor alpha , interleukin-1 beta , and prostaglandin E 2

We investigated the participation of prostaglandin (PG) E2 in the regulation of the alpha 1(I) procollagen gene expression by tumor necrosis factor alpha (TNF alpha), and interleukin-1 beta (IL-1 beta) in normal adult human lung fibroblasts. TNF alpha (100 units/ml) and IL-1 beta (100 units/ml) stimulated the production of PGE2 and caused a dose-dependent inhibition of up to 54 and 66%, respectively, of the production of type I procollagen. Preincubation of cultures with indomethacin partially reversed the inhibition of procollagen production induced by the cytokines. Cytokine-stimulated endogenous fibroblast PG accounted for 35 and 68% of the inhibition induced by TNF alpha and IL-1 beta, respectively. Steady-state mRNA levels for alpha 1(I) procollagen paralleled the changes in collagen production. The transcription rate of the alpha 1(I) procollagen gene was reduced by 58% by TNF alpha and by 43% by IL-1 beta. Cytokine-stimulated endogenous PG production accounted for half of these effects. These results indicate that TNF alpha and IL-1 beta inhibit the expression of the alpha 1(I) procollagen gene in human lung fibroblasts at the transcriptional level by a PGE2-independent effect as well as through the effect of endogenous fibroblast PGE2 released under the stimulus of the cytokines.

Fibrillar collagens are the most abundant proteins in the lung interstitium and constitute about 15% of the dry weight of the human lung (1). Because of the high turnover of the connective tissue of adult lung (2), the balance between synthesis and degradation must be accurately controlled in order to insure the preservation of normal structure and function. Exaggerated tissue deposition of extracellular matrix proteins is the final outcome of several diseases in which an inflammatory process triggered by various stimuli is the earliest event (3). Tissues undergoing a chronic inflammatory process are often infiltrated by macrophages and lymphocytes. These are the main cell lineages responsible for the production of various cytokines that have been implicated in the initiation, progression, and eventual modulation of a variety of inflam- matory and immunologic responses. It has been shown that in addition to their participation in inflammation and the immune response, several cytokines can exert profound effects on fibroblast production of extracellular matrix proteins (4). Thus, it is very likely that they would participate in tissue remodeling and, perhaps, in the development of fibrosis (2).
Tumor necrosis factor a (TNFa)' and interleukin-lfl (ILlfl) are cytokines produced mainly by activated cells of the monocyte/macrophage lineage. These cytokines have many overlapping activities and play a central role in inflammation, T cell activation, and cytotoxicity (5, 6). It is thought that TNFa and IL-1fl participate in tissue remodeling because of their ability to promote fibroblast growth and angiogenesis and to stimulate the production of collagenase. Multiple studies have examined the effects of TNFa and/or IL-1 on the production of extracellular matrix proteins by various mesenchymal cells . In adult and fetal dermal fibroblasts, TNFa inhibited the production of type I and I11 procollagens and decreased the levels of their corresponding mRNAs (17)(18)(19)26). However, other reports showed that TNFa stimulated collagen production in dermal and lung fibroblasts (20,21,25). The effects of IL-1 on fibroblast collagen production are also controversial. IL-1 has been shown both to increase and inhibit collagen production. These effects were accompanied by parallel changes in the steady-state levels of the corresponding mRNAs in most studies (20,(22)(23)(24)(25), although in other studies an inhibition of collagen production with a paradoxical increase in collagen mRNA levels was found (26,27).
The products of the cyclooxygenase pathway, such as PGE2, also participate in inflammatory and immune responses. TNFa and IL-lp stimulate PGE2 production in several cell lines including macrophages (28), synovial cells, and fibroblasts (29,30) and increase the steady-state levels of cyclooxygenase mRNA (31). There is evidence that PGE2 inhibits collagen production (32) by several mechanisms, including a reduced uptake of proline (33) and an increase in the intracellular degradation of the protein (34). Furthermore, decreased steady-state mRNA levels for al(1) procollagen have been shown in PGEn-treated fibroblasts, indicating that PGE2 also acts at pretranslational levels (35). Several studies have examined the contribution of increased endogenous PG production by TNFa and IL-1 on the modulation of fibroblast collagen gene expression (18,19,(22)(23)(24)(25)(26)(27). In only two of these studies, it was shown that PGs play a modulatory role (24,27). Because of our interest on the regulation of fibroblast * The abbreviations used are: TNFa, tumor necrosis factor a; IL-16, interleukin-lp; PG, prostaglandin; PGEZ, prostaglandin Ez; IFNy, y-interferon. collagen gene expression by cytokines and cytokine-stimulated endogenous PG, we conducted the studies presented here to clarify the conflicting results described above. We postulated that stimulation of fibroblast PGE2 production by TNFa and IL-lp must play a role in the net effect of these cytokines on type I collagen production in a manner similar to that shown with TGFB (36). We present evidence that TNFa and IL-lP inhibit lung fibroblast type I collagen production and decrease the corresponding steady-state mRNA levels and that these effects are mediated by PG-dependent and -independent mechanisms. Furthermore, we found that TNFa, IL-lP, and PGEz modulation of the steady-state mRNA levels for al(1) procollagen is largely exerted at the transcriptional level. Analysis of Labeled Proteins-Aliquots of media and cell layers were dialyzed extensively to remove unincorporated radioactive precursors. Total incorporation of ~-[U-'~C]proline, ~-[~'S]-methionine, and D-[6-3H]glucosamine hydrochloride into newly synthesized macromolecules was measured by scintillation spectroscopy. The L-[U-"Clproline labeled proteins in the media, and cell layers were analyzed by polyacrylamide gel electrophoresis under reducing conditions. After electrophoresis the gels were processed for fluorography. The fluorographs were scanned with a laser densitometer (UltroScan XL, Pharmacia LKB Biotechnology Inc.). The amount of radioactive collagen synthesized by the fibroblasts was determined by a bacterial collagenase assay (37) and the relative proportions of collagen calculated following the formula of Breul et al. (38).

Human
Determination of PGE,-PGE, was measured in undialyzed samples of culture media by a radioimmunoassay as described previously (39).
Isolation (1 pg/ml), or PGE, (100 ng/ml) for 24 h. At the end of the incubations, the cell layers were washed in Hanks' solution and harvested immediately in 4 M guanidinium isothiocyanate. Total RNA was isolated in a CsC1, discontinuous gradient as described previously (40). For Northern blot hybridizations, aliquots containing equal amounts of total RNA were denatured in formaldehyde, electrophoresed in 0.8% agarose/formaldehyde gels, and then transferred in a vacuum blotting system (LKB 2016 VacuGene) to nitrocellulose filters (Optibind S & S) and UV-cross-linked (UV Stratalinker 2400, Stratagene). The human cDNA clone Hf677 specific for al(1) procollagen (41) and a mouse cDNA specific for glyceraldehyde-3-phosphate dehydrogenase (42) were nick-translated with [a-32P]dCTP to specific activities > 1.0 X lo8 cpm/pg of DNA. The filters were prehybridized overnight and hybridized at 42 "C for 24 h in 50% formamide, 2 X SSC, 2 X Denhart's solution, and 0.1% SDS. For quantitative analysis of the mRNA levels, the filters were submitted to autoradiography, and autoradiographs were scanned in a laser densitometer.
In Vitro Nuclear Transcription Assay-The transcription rate was measured by an in vitro nuclear run-off assay as described previously (43). Lung fibroblasts were cultured in T-175 flasks for 24 h in the presence or absence of TNFa (100 units/ml), IL-1P (50 units/ml), indomethacin (1 pg/ml), PGE, (100 ng/ml), or vehicle alone (10 pl/ ml 10% ethanol) as described above. At the end of the incubation period, cell layers were trypsinized, and nuclei were prepared as described previously (44) and were stored at -70 "C until used. The transcription reactions were carried out in volumes of 100 pl (Experiment 1) or 300 pl (Experiment 2) in 10 mM Tris, pH 8.0, 90 mM KCl,, 3 mM MgCl,, 2 mM dithiothreitol, 1 unitlpl RNasin, 0.4 mM each of ATP, UTP, and GTP, and 0.5 mCi of [a-"PICTP (Du Pont-New England Nuclear, 800 Ci/mmol). Incubations were for 25 min at 25 "C, and incorporation of [c~-~'P]CTP was followed by trichloroacetic acid precipitation of 1-p1 aliquots. Transcription was terminated by the addition of 900 pl of buffer containing 100 mM NaCl, 10 mM Tris-HC1, pH 7.5, 2 mM KCl, 1 mM EDTA, and 0.5% SDS. To each sample 100 pg of yeast tRNA was added, and the samples were digested with 100 pg/ml proteinase K for 60 min at 42 "C, extracted with phenol/chloroform, and precipitated in 10% trichloroacetic acid and 10% saturated sodium pyrophosphate. The pellets were washed with 70% EtOH, dried, and dissolved in 100 pl of 10 mM Tris-HCI, pH 7.4, containing 1 mM EDTA and 0.1% SDS. An additional 100 pg of yeast tRNA was added, and nucleic acids were ethanol-precipitated in 2.5 M ammonium acetate. The pellets were dissolved in 100 pl of buffer that contained 20 mM Tris-HC1, pH 7.5, 10 mM MgCl,, and 2 mM CaC1, and incubated for 30 min at 37 "C with 100 pg/ml RNasefree DNase and 1 unit/pl RNasin. Samples were extracted with phenol/chloroform and ethanol-precipitated in 0.3 M sodium acetate. Labeled transcripts were resuspended in prehybridization buffer, and duplicate 5-pl aliquots were trichloroacetic acid-precipitated and their radioactivity determined by scintillation counting. Aliquots of each sample containing equal counts/min were adjusted to 400-pl volume by addition of the same buffer and were hybridized to filters containing dot-blotted and immobilized purified al(1) procollagen and glyceraldehyde-3-phosphate dehydrogenase cDNAs in pBR-322 plasmid or pBR-322 alone. The dots were previously cut out from the filters and prehybridized in 50% formamide, 5 X Denhart's solution, 4 X SSC, 0.1% SDS, 0.1 mg/ml salmon sperm DNA. Hybridizations were performed with continuous shaking at 42 "C for 72 h. After hybridizations, the filters were washed for 15 min in 2 X SSC at room temperature and then for 15 min in 0.2 X SSC at 65 "C and treated with RNase A (10 pg/ml in 2 X SSC) for 15 min at 37 "C. The filters were then washed in 2 X SSC, 0.1% SDS for 15 min at room temperature and dried. Autoradiographs were obtained and scanned in a laser densitometer. The amount of 32P hybridized to each dot blot was determined by scintillation counting. TNFa and IL-10 was related to stimulation of endogenous fibroblast PG synthesis by the cytokines, parallel cultures were preincubated with indomethacin (1 pg/ml), a concentration shown previously to produce complete inhibition of cyclooxygenase in these cells. The preincubation of cultures with indomethacin reversed only partially the inhibitory effects of TNFa (Fig. 1B) and IL-10 ( Fig. 2B) on type I procollagen production. Densitometric scanning of fluorographs showed a maximal inhibition of newly synthesized type I procollagen of 54% at a concentration of 100 units/ml of TNFa. The concomitant treatment of cells with indomethacin resulted in an inhibition of only 35%. Therefore, endogenous PG accounted for 35% of the total inhibitory effect of TNFa (Fig. IC). Similar analysis showed that 100 units/ml of IL-l@ caused a maximal inhibition of 66% on type I procollagen production.

Effects of
The concomitant treatment of cells with indomethacin resulted in an inhibition of only 21%. Therefore, endogenous PG accounted for 68% of the total inhibitory effect of IL-1@ (Fig. 2C). However, at lower concentrations of IL-lP, indomethacin completely abolished IL-10 inhibition of collagen production, suggesting that at these concentrations, the collagen inhibitory effects of the cytokine are entirely PG-dependent. Next, we examined the effect of the combination of both cytokines on type I procollagen production. Incubation of lung fibroblasts in the presence of constant concentrations of TNFa and increasing concentrations of IL-10 showed that the inhibition of procollagen production was more pronounced (72%) than when the cultures were incubated with a single cytokine (Fig. 3A). Preincubation of cultures with indomethacin showed only a partial reversal of the inhibitory effect of the combination of both cytokines to 55% (Fig. 3B). Thus, the contribution of endogenous PG accounted for only 23% of the total inhibitory effect of that of the combination of both cytokines (Fig. 3C). These results were confirmed by a specific collagenase assay in two separate experiments (Table   I).
To exclude the possibility that the observed effects were the result of a global cytotoxic effect of TNFa, cells were labeled with [35S]methionine or with [3H]glucosamine. As shown in Table 11, incubation of cells with TNFa (100 units/ ml) did not affect the incorporation ~f[~~S]methionine into total proteins. Furthermore, the incorporation of [ 3 H ] g l u~~samine into glycoproteins and glycosaminoglycans was increased in a dose-dependent manner by TNFa, reaching a maximal stimulation of 39% at 100 units/ml. In addition, trypan blue exclusion showed a viability greater than 90% in cells cultured under either control conditions or treated with 50 or 100 units/ml TNFa.
Effects of TNFa and IL-10 on PGE2 Production-As shown in Table 111, incubation of lung fibroblasts with TNFa (100 units/ml) caused an increase in PGE, production from 1.06 to 13.74 ng/ml, and incubation with IL-10 (50 units/ml) caused an increase from 2.16 to 30.56 ng/ml. The incubation with both cytokines resulted in higher PGE, production than A.

C.
Collagen Regulation by TNFa, IL-1 P, and PGE:! Effects of TNFa and IL-10 on al(I) Procollagen mRNA Levels-In order to investigate the mechanisms of the inhibition of fibroblast type I procollagen production by TNFa and IL-18, the steady-state mRNA levels for al(1) procollagen were examined by Northern blot hybridizations with a specific human cDNA. TNFa produced a dose-and length of incubation-dependent reduction of the steady-state mRNA levels for al(1) procollagen with maximal reduction of 76% at a concentration of 100 units/ml (Fig. 4A) and of 67% after 16 h of incubation with the cytokine (Fig. 4B).
Northern hybridization analysis of total RNA from lung fibroblasts treated with increasing concentrations of IL-18 showed a dose-dependent reduction of al(1) procollagen steady-state mRNA levels that reached a 68% a t 100 units/ ml (Fig. 5A). IL-18 induced a length of incubation-dependent reduction of al(1) procollagen steady-state mRNA levels that reached 91% a t 24 h (Fig. 5 B ) .
Participation of Endogenous PG on the Reduction of al(I) Procollagen mRNA Levels by TNFa and IL-18"To investigate the participation of PG on the effects of TNFa on the al(1) procollagen steady-state mRNA levels, control and TNFa-treated cells were incubated with or without indometh- 24  Northern hybridizations (Fig. 6). In agreement with a previous report (35), the treatment of control cultures with PGE2 (100 ng/ml) resulted in a marked decrease (up to 60%) of the al (1) procollagen steady-state mRNA levels.   Methods." Samples containing 10 pg of total RNA were denatured, electrophoresed on a 0.8% agarose gel, and transferred to a nitrocellulose filter. The filter was hybridized to a radiolabeled human cDNA specific for al(1) procollagen and processed by autoradiography.

TABLE I Effect of indomethacin on TNFa-and IL-10-induced inhibition of total protein and collagen production by cultured human lung fibroblasts Confluent human lung fibroblasts (cell line ID) were incubated in triplicate for
indomethacin. Therefore, endogenous PG accounted for 43% of the effect of TNFa. The addition of exogenous PGE, (100 ng/ml) to cultures treated with T N F a (100 units/ml) plus indomethacin (1 pg/ml) resulted in an additive decrease of 70% on the al (1) procollagen steady-state mRNA levels. Incubation of lung fibroblasts with IL-lD decreased the al(1) procollagen steady-state mRNA levels by 61%. However, when IL-lp-treated cells were preincubated with indomethacin the al(1) procollagen mRNA steady-state levels decreased by only 35% (Fig 7). Incubation of cultures with T N F a plus IL-lP caused a 72% reduction of the steady-state mRNA levels for al (1) procollagen. This inhibition was not reversed by preincubation of the cultures with indomethacin, indicating that when the cells were exposed to a combination of TNFa plus IL-lP at these concentrations, the main mechanisms affecting the al(1) procollagen steady-state mRNA levels were PG-independent.
Effect of TNFa, IL-ID, and PGE2 on the Transcription Rate of the al(I) Procollagen Gene-In order to investigate whether A.

RNA was extracted as described under "Material and
Methods." Samples containing 10 pg of total RNA were denatured, electrophoresed on a 0.8% agarose gel, and transferred to a nitrocellulose filter. The filter was hybridized to a radiolabeled human cDNA specific for al(1) procollagen and processed by autoradiography. kb, kilobase. the decrease on the a1(I) procollagen steady-state mRNA levels induced by T N F a is mediated by transcriptional mechanisms, lung fibroblasts were incubated for 24 h with TNFa with or without indomethacin or with PGEz plus indomethacin, and the transcription rates of the al(1) procollagen gene were measured by an in vitro nuclear transcription assay.
Control cultures were incubated with indomethacin to eliminate any influence of endogenous PG and allow maximal expression of the al(1) procollagen gene. Treatment of cells with indomethacin plus PGE, (100 ng/ml) resulted in a 37% inhibition of the transcription rate of the al(1) procollagen gene (Fig. 8). Treatment of cells with TNFa (100 units/ml) resulted in greater inhibition (58%) of the transcription rate of the gene. This effect was partially reversed by the preincubation of cells with indomethacin (inhibition of only 25%). Therefore, endogenous PG accounted for 57% of the inhibitory effect of TNFa. In a separate experiment lung fibroblasts were incubated with vehicle alone (EtOH), indomethacin alone, IL-l@ alone, or IL-lD plus indomethacin (Fig. 8). Cells treated with vehicle alone showed a modest inhibition of the transcription rate of the al(1) procollagen gene (20%) as compared with indomethacin-treated cells. This level of inhibition, therefore, reflects the level of inhibition induced by endogenous PG under basal conditions. Treatment with ILlp alone caused a 43% inhibition of the transcription rate of the al(1) procollagen gene, and preincubation of IL-ID-treated cultures with indomethacin partially reversed the inhibitory effect of IL-lP to only 18%. Therefore, endogenous PG accounted for 58% of the inhibitory effect of IL-1fl. These observations indicate that the reduction in the transcription rate of the gene by TNFa and IL-lP results from a combination of direct inhibitory effects of the cytokines plus the inhibitory effects of endogenous PGE,.

DISCUSSION
Here we present evidence that TNFa and IL-10 downregulate the production of type I procollagen in normal human lung fibroblasts by modulating the steady-state mRNA levels PGE2 (100 ng/ml). Total RNA was extracted as described under "Material and Methods." Samples containing 8 pg of total RNA were denatured, electrophoresed on a 0.8% agarose gel, and transferred to a nitrocellulose filter. The same filter was hybridized to a radiolabeled human cDNA specific for al (1) procollagen and to a murine cDNA specific for glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Filters were processed by autoradiography. B, autoradiographs were scanned in a laser densitometer. The integrated areas are expressed in arbitrary densitometric units (ADU) as a percentage relative to values from samples cultured with medium alone. The numbers under the burs correspond to the conditions described in A.
for the protein and that these effects occur largely through inhibition of transcription. Blockade of endogenous fibroblast PG production reversed only partially these effects. Addition of exogenous PGE, reduced the steady-state mRNA levels and the transcription rate of the al(1) procollagen gene. These results demonstrate that the inhibitory effect of TNFa and IL-1P on lung fibroblast collagen production is partially due to the effects of newly synthesized fibroblast PG in response to TNFa and IL-1P stimulation. The inhibitory effects of T N F a were selective for collagen and were not related to toxicity as demonstrated by the absence of changes in the incorporation of [35S]methionine, the increased synthesis of glycosaminoglycans and glycoproteins (20,45), and the high level of cell viability as measured by trypan blue exclusion. Although the results presented have clearly demonstrated an effect of IL-1 and TNFa on collagen synthesis, an additional mechanism for the inhibition on the production of collagen by TNFa and IL-1@ must be that of intracellular and extracellular degradation induced by the increased production of PG and increased production of collagenase, respectively (6).
We found that the inhibitory effects of the cytokines on the steady-state mRNA levels were variable within the same cell line (60-90% for IL-1P in PM) or for the same cytokine in different cell lines (TNFa, -70% in P M and 40% in ID). Total RNA was extracted as described under "Material and Methods." Samples containing 10 pg of total RNA were denatured, electrophoresed in a 0.8% agarose gel, and transferred to a nitrocellulose filter. The same filter was hybridized to a radiolabeled human cDNA specific for al(1) procollagen and to a murine cDNA specific for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and processed by autoradiography. Autoradiographs were scanned in a laser densitometer and the integrated areas expressed in arbitrary densitometric units (ADU) as a percentage relative to values from samples incubated in media without the cytokines. Samples without

(C) or with indomethacin (D).
Despite this variability, the effects of TNFa and IL-1 on the two cell lines described here and in an additional cell line (CCL-210; not shown) were consistently inhibitory on the al(1) procollagen gene expression, at the three levels examined, i.e. rates of gene transcription, steady-state mRNA levels, and protein production. Furthermore the participation of endogenous prostaglandins on this inhibitory effect was found also at the three levels of protein biosynthetic pathway examined. This consistency makes it very unlikely that a clonal selection of a particular cell could be responsible for the results we obtained (46). The discrepancies with previous reports that examined the influence of PG on TNFa and IL-1P effects on collagen production could be due to intrinsic differences in the ability of different cell types to produce or to respond to endogenous PG. It is also possible that under particular experimental conditions such as serum-free or low serum conditions, the endogenous PG production could be too low to cause detectable effects on collagen production. On the other hand, very high concentrations of TNFa or IL-lP or their combination could inhibit the expression of the procollagen gene by PG-independent mechanisms that cannot be reversed by inhibition of endogenous PG production. The observations described here when cultures were exposed to a A. indomethacin (1 pglml). Nuclei were isolated and in oitro transcription assays performed as described under "Materials and Methods." Labeled transcripts from each sample (3 X lo6 cpm in Experiment 1 and 18 X lo6 cpm in Experiment 2) were hybridized to the filterbound cDNAs for al(1) procollagen and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and to the plasmid pBR-322. After washing and digestion with RNase A, the filters were processed by autoradiography ( A ) ; autoradiographs were scanned in a laser densitometer, the pBR-322 background was subtracted, and the integrated areas corresponding to hybridized d ( 1 ) procollagen transcripts were normalized with the areas corresponding to hybridized glyceraldehyde-3-phosphate dehydrogenase transcripts. The resulting values are expressed in arbitrary densitometric units ( A D U ) as a percentage relative to values from samples incubated with indomethacin alone ( B ) .
combination of T N F a plus IL-1P support this notion.
It has been shown previously that TNFa and IL-lD exert a proliferative response in fibroblasts and that human alveolar macrophages produced higher amounts of biologically active TNFa and expressed higher steady-state T N F a mRNA levels than autologous peripheral blood monocytes in response to lipopolysaccharide stimulation (47). Furthermore, TNFa has been implicated in the development of fibrosis in murine models (48-50). However, our findings support the notion that in the human lung these cytokines may reduce the deposition of extracellular matrix by decreasing collagen gene expression. Therefore, it is likely that a more complex network of cell to cell and cytokine interactions is required for the increased deposition of extracellular matrix in lung fibrosis. For example, T N F a in combination with IL-1 and interferony (IFNy) increases the adherence of T-lymphocytes to human lung fibroblasts (51), presumably through the induction of the intercellular adhesion molecule-1 (ICAM-1). T N F a in conjunction with IFNy induces or amplifies the expression of HLA class I1 antigens in monocytes (52) and T cells (53), providing an additional mechanism for cell adherence, and more importantly, enhancing their antigen presenting capability. TNFa and IL-lP also increase the expression of high affinity IL-2 receptors in T cells (53), enhancing their proliferative response. The macrophage and lymphocyte activation caused by TNFa and IL-1P would stimulate the production of powerful fibrogenic factor(s) such as TGFP by these cells. These cytokine interactions turn more complex if their interdependence with the products of arachidonic acid metabolism are considered. In conclusion, in human lung fibroblasts TNFa and IL-lP inhibit the production of al(1) procollagen largely at the transcriptional level by PGE2-dependent andindependent mechanisms. Although these cytokines appear to be important mediators in the early inflammatory stages of lung fibrosis, the chronic deposition of extracellular matrix proteins leading to fibrosis must be the result of more complex cellular responses.