Fetal lung fibroblasts selectively down-regulate proteoglycan synthesis in response to elevated oxygen.

Cell proliferation is in part regulated by extracellular matrix. Therefore, it is possible that elevated O2 may indirectly affect lung fibroblast growth via modulation of extracellular matrix. In the present study, we investigated the effect of elevated O2 on the synthesis of glycosaminoglycans (GAGs) and proteoglycans (PGs) by fetal lung fibroblasts. A 48-h exposure to >/=50% O2 reduced the incorporation of [3H]glucosamine and 35SO4 into GAGs by fetal lung fibroblasts. The relative proportion of the individual GAG molecules was not altered by elevated O2. Fibroblasts exposed to 50% O2 secreted less [35S]proteoglycans into the medium than controls. Specifically, the synthesis of the small soluble PG, biglycan, was decreased by exposure to 50% O2. Fetal lung fibroblasts did not synthesize the small chondroitin/dermatan sulfate PG, decorin. Elevated O2 concentrations also reduced the synthesis of membrane- and matrix-associated PGs. Furthermore, exposure of fetal lung fibroblasts to >/=50% O2 resulted in a decreased mRNA expression for biglycan and versican core protein sequences. In contrast, elevated O2 increased the message for type I collagen and fibronectin without affecting that of beta-actin. The inhibitory effect of elevated O2 on biglycan mRNA and protein expression was overcome by incubating the cells in 3% O2 after the 48-h exposure to 50% O2. The latter treatment also reversed the increased mRNA expression of type I collagen associated with elevated O2 but not that of fibronectin. These results demonstrate that fetal lung fibroblasts, in response to elevated oxygen concentrations, selectively down-regulate their GAG and PG synthesis and that this O2 effect is reversible.

Disruption of cell-matrix interactions during lung development may induce cellular responses, which can result in scar formation. Administration of high concentrations of oxygen has been shown to induce changes in extracellular matrix (1). Glycosaminoglycans (GAGs) 1 are matrix components, which may mediate matrix-dependent events during development, including branching, tissue remodeling, and cell differentiation (2,3). GAGs are long polyanionic carbohydrate chains composed of repeating disaccharide units (uronic acid and hexosamine) carrying a negative charge, which results in their binding to other matrix components and cell adhesion molecules (4,5). Most GAGs are covalently linked to a core protein as proteoglycans (PGs). PGs show a marked variation in the size of the core protein and the number and size of GAG side chains attached. Some proteoglycans are cell-associated, and others are present in interstitial matrices (4,5). Increasing evidence suggests that GAGs and PGs play an important role as modulators of growth factor activities (6). Several isoforms of fibroblast growth factor bind to high affinity receptors and are active when bound to specific cell surface heparan sulfate PGs (7). Similarly, the membrane proteoglycan betaglycan increases the binding of TGF-␤ 1,2,3 to the signaling TGF-␤ receptor (8). The small interstitial proteoglycans biglycan, decorin, and fibromodulin have been shown to bind TGF-␤ via their core proteins (9 -11). Recent studies have demonstrated that GAGs also directly affect cell proliferation. Heparan sulfate molecules are potent inhibitors of smooth muscle cell growth (12). Additionally, hyaluronic acid inhibits fetal skin fibroblast proliferation but stimulates collagen and noncollagen protein synthesis (13). These properties suggest that GAGs and PGs may mediate, at least in part, cell adhesion, structural organization, and cell proliferation during fetal lung development. Thus, alterations in GAG and PG content in the interstitial matrix following elevated oxygen may disrupt normal fibroblast adhesion and affect fibroblast proliferation. It has been shown that the distribution and density of PGs in alveolar basement membranes is decreased in lung injury resulting from short term oxygen exposure (14). In contrast, exposure of newborn rats to 85% oxygen for up to 6 weeks has been reported to result in an increase in biglycan mRNA expression and immunoreactivity, specifically in alveolar cells in areas of increased cellularity (15). The effect of oxygen on PG metabolism may be due to increased intracellular generation of partially reduced oxygen species. Reactive oxygen species have been demonstrated to affect GAG and PG synthesis in cultured chondrocytes (16). Physiological concentrations of reactive oxygen species stimulate PG formation, while higher concentrations decrease PG synthesis (16). Additionally, high reactive oxygen species concentrations damage existing cartilage PGs (16). Whether elevated oxygen concentrations directly affect GAG and PG synthesis or destroy existing PGs has yet not been determined. In the present study, we examined the effect of hyperoxia on GAG and PG synthesis by fetal lung fibroblasts. The data suggest that in vitro fetal lung fibroblasts selectively down-regulate their GAG and PG synthesis as an early response to elevated oxygen concentration.
Cell Culture-Rats were sacrificed by diethylether excess on day 19 of gestation (term ϭ 22 days). The fetuses were aseptically removed from the mothers, and the fetal lungs were dissected out in cold Hanks' balanced salt solution without calcium or magnesium (HBSS (Ϫ)) and cleared of major airways and vessels. The lungs were washed twice in HBSS (Ϫ), minced, and suspended in HBSS (Ϫ). Fibroblasts were isolated from the fetal lungs as described in detail previously (17) except that cells were grown in a gas phase of 3% O 2 , 5% CO 2 , 92% N 2 to mimic the normal fetal arterial oxygen concentration of ϳ20 mm Hg observed in vivo. Viability and purity of the fibroblast cultures were comparable with previously published data (17).
Effect of O 2 on GAG Synthesis-To investigate the effect of O 2 on GAG synthesis, cells grown in 75-cm 2 tissue culture flasks were trypsinized and diluted in Eagle's minimal essential medium (MEM) with 10% (v/v) fetal bovine serum to a seeding density of 1 ϫ 10 5 cells/cm 2 . Cells were seeded in 24-well plates and grown to confluence in MEM with 10% (v/v) fetal bovine serum in 3% O 2 and were then maintained in serum-free medium in a gas phase containing 3, 50, or 95% O 2 with 5% CO 2 for 24 h prior to a 24-h incubation in the same O 2 concentration in serum-free MEM containing either 5 Ci/ml [6-3 H]glucosamine or 10 Ci/ml Na 2 35 SO 4 . Radioactive glucosamine is a precursor for both sulfated (heparan sulfate, chondroitin sulfate, and dermatan sulfate) and nonsulfated (hyaluronan) GAGs, whereas 35 S-labeled sulfate is only incorporated into sulfated GAG molecules. Total GAGs were isolated and characterized as described by Castor et al. (18). After incubation with [ 3 H]glucosamine, both medium (extracellular compartment) and cell layer fractions (cellular compartment), were adjusted to 0.2 M NaOH and left overnight to hydrolyze the core protein-glycosaminoglycan linkage. The solutions were then neutralized (pH 6 -8) with 2 M HCl. An aliquot of each solution was taken for estimation of radiolabeled total GAG synthesis using a modification of the cetylpyridinium chloride method (18,19). The remainder of the cell layer and medium fractions was subjected to anion exchange HPLC to separate individual GAG molecules, as described previously (20). Briefly, hyaluronan and chondroitin sulfate C (5 g each) were added as carrier material to the remaining cell-matrix and medium fractions. Fractions were then boiled, and GAG molecules were collected by ethanol precipitation at Ϫ20°C. The samples were dissolved in 150 mM Tris, pH 8.3, 10 mM CaCl 2 , and 100 mM glucosamine and treated with Pronase to digest all residual protein. GAGs were then precipitated by ethanol containing 1.3% (w/v) sodium acetate. The precipitated GAGs were collected by centrifugation, and the GAG pellet was dissolved in H 2 O. Hyaluronan, heparan sulfate, and chondroitin/dermatan sulfate were separated on a DEAE-5-PW column (75 ϫ 7.5 mm) with a sodium chloride gradient (0 -1 M in 10% methanol) and a flow rate of 0.6 ml/min. Identification of GAG molecules was performed by their destruction using Streptomyces hyaluronidase (8 units/ml), heparitinase (0.06 units/ml), and chondroitinase ABC (0.25 units/ml) as described by Copp and Bernfield (21).
To determine the effect of O 2 on cell number, parallel culture wells were incubated with serum-free MEM. After the 48-h incubation period, the cells were trypsinized and counted electronically using a Coulter particle counter (Coulter Electronics, Hialeah, FL).
Effect of O 2 on 35 SO 4 Incorporation into Proteoglycans-In order to investigate the effect of O 2 on sulfated PG synthesis, fibroblasts were grown to confluence in 75-cm 2 tissue culture flasks in MEM with 10% (v/v) fetal bovine serum in 3% O 2 and were then maintained in serumfree medium in a gas phase containing 3 or 50% O 2 for 24 h prior to a 24-h incubation in the same O 2 concentration in serum-free MEM containing 50 Ci/ml Na 2 35 SO 4 . After the incubation with radioactive sulfate, media were collected and dialyzed extensively against distilled water (M r Ͻ 3,500) in the presence of 1 mM phenylmethylsulfonyl fluoride. The cell layer was washed several times with large volumes of PBS, and the cell membrane fraction was extracted on ice for 10 min with 2% (v/v) Triton X-100 in PBS containing 1 mM phenylmethylsulfonyl fluoride. The remaining extracellular matrix fraction was rinsed with PBS and treated with either chondroitinase ABC or heparitinase or left untreated prior to extraction by scraping with a rubber police- Aliquots (100 l) of dialyzed media samples, cell membrane fraction, and extracellular matrix fraction were then incubated in PBS with or without chondroitinase ABC (100 milliunits/ml) or heparitinase (20 milliunits/ml) at 37°C. After a 24-h incubation, samples were boiled in SDS-sample buffer and fractionated on 5% (w/v) SDS-polyacrylamide gel. Gels were fixed in 10% (v/v) acetic acid, prepared for fluorography by soaking in En 3 Hance (DuPont NEN), dried, and exposed to Kodak XAR-5 film using Dupont Cronex intensifying screens. The films were quantified using an Ultroscan XL laser densitometer (Pharmacia Biotech Inc.).
Northern Analyses-Total cellular RNA was isolated from the fibroblast cultures by lysing the cells in 4 M guanidinium thiocyanate followed by centrifugation on a 5.7 M cesium chloride cushion to pellet RNA. This total RNA (15 g) was size-fractionated on a 1% (v/v) agarose gel containing 3% (v/v) formaldehyde, transferred to Hybond N membranes, and immobilized by UV cross-linking. The cDNA probes were labeled with [␣-32 P]dCTP using random primers. Prehybridization and hybridization were performed in 50% (v/v) formamide, 5 ϫ SSPE, 0.5% (w/v) SDS, 5 ϫ Denhardt's solution, and 100 g/ml denatured salmon sperm DNA at 42°C. Following hybridization, the blots were washed with 0.5 ϫ SSC containing 0.2% (w/v) SDS at 65°C and autoradiographed with Kodak XAR-5 film overnight at Ϫ70°C. Blots were then stripped and, for normalization, rehybridized with radiolabeled rat ␤-actin cDNA. The autoradiographs were quantified using an Ultroscan XL laser densitometer (Pharmacia).
Statistical Analysis-Statistical significance (p Ͻ 0.05) was determined by analysis of variance followed by assessment of differences using Dunnet's two-sided test (22) or Duncan's multiple range test (23).

RESULTS
Effect of O 2 on GAG Synthesis-Exposure of fetal lung fibroblasts to 50% O 2 for 48 h reduced the incorporation of [ 3 H]glucosamine into total GAGs by ϳ40% without changing the number of cells adhered to the plastic (Fig. 1). A 95% O 2 exposure reduced both cell number and GAG synthesis. However, GAG synthesis of cells exposed to 95% O 2 was not further reduced from that of fibroblasts exposed to 50% O 2 . The relative distribution of GAGs between medium and cell-matrix layer was not  Effect of O 2 on GAG Composition-Fetal lung fibroblast, grown in 3% O 2 , synthesized and deposited hyaluronan, heparan sulfate, and chondroitin/dermatan sulfate in the cell-matrix compartment. Heparan sulfate was the main GAG component secreted into the medium. The effect of elevated O 2 on the individual GAG molecules produced by the fibroblasts is shown in Fig. 2. Although total GAG synthesis decreased, the composition of GAG molecules in the medium and in the cell-matrix layer was not significantly changed by exposing the cells to either 50 or 95% O 2 .
Effect of O 2 on Soluble PG Synthesis-As exposure to 95% O 2 affected cell number (Fig. 1) and cell viability (41), we carried out all further experiments at 50% O 2 . Confluent fibroblast cultures were incubated with 35 SO 4 to label the PGs, and the conditioned medium was analyzed by SDS-polyacrylamide gel electrophoresis to identify the soluble PGs. The results demonstrate that fetal lung fibroblasts synthesized [ 35 S]sulfate-labeled macromolecules with high M r values that did not enter into the gel and remained on top of the separating gel (Fig. 3). In addition, the fibroblasts synthesized [ 35 S]sulfate-labeled macromolecules with relatively small molecular weights (Fig.  3). The approximate M r ranges from 190,000 to 250,000, which is in the same range as that of biglycan (24 -26 (24 -26), was synthesized by the fibroblasts. Exposure of fetal lung fibroblasts to 50% O 2 decreased the incorporation of 35 SO 4 into both biglycan and larger PGs by 3.4 Ϯ 0.6-and 3.3 Ϯ 0.7-fold (mean Ϯ S.E., n ϭ 3 separate experiments), respectively (Fig. 3). To evaluate reversibility, fetal lung fibroblasts were first exposed for 48 h to 50% O 2 and then for 48 h to 3% O 2 . As can been seen in Fig. 4, the incorporation of [ 35 SO 4 ] into biglycan in these cells was similar to that of cells exposed to 3% O 2 , suggesting that the effect of elevated O 2 on PG synthesis is reversible.
Effect of O 2 on PGs in Matrix and Cell Membrane-To examine the effect of O 2 on PGs associated with the cell membrane or extracellular matrix, fibroblasts were incubated in 3 or 50% O 2 and labeled with 35 SO 4 . The membrane fraction of fetal lung fibroblasts was extracted by 2% (v/v) Triton X-100, and the remaining fraction was the extracellular matrix fraction. The cell-associated (membrane) fraction also contained [ 35 S]biglycan and [ 35 S]sulfate-labeled large PGs (Fig. 5). A major portion of the large macromolecules was resistant to chondroitinase ABC but not to heparitinase digestion, suggesting that cellassociated (membrane) PGs are mainly heparan sulfate PGs. Exposure of fibroblasts to 50% O 2 decreased the synthesis of cell-associated (membrane) PGs. Besides [ 35 S]biglycan, the cell-matrix fraction was also found to contain PGs with high M r values (Fig. 5). The large PGs were partly degraded by chondroitinase ABC but completely by heparitinase, suggesting that extracellular matrix PGs consist of large chondroitin sulfate as well as heparan sulfate PGs. Again, fibroblasts exposed to 50% O 2 incorporated less 35 SO 4 into PGs associated with the matrix fraction. The O 2 effect on membrane and extracellular matrix PGs was also reversed by incubating the cells in 3% O 2 after the 50% O 2 exposure (not shown).
Effect of O 2 on PG Gene Expression-To determine whether the alterations noted in [ 35 S]PG synthesis by elevated O 2 concentrations might be accompanied by similar changes in mRNA expression for PGs, total mRNA was extracted from fetal lung fibroblasts and hybridized with cDNAs coding for three different PG core proteins (Fig. 6). The biglycan cDNA detected a single mRNA transcript of 2.5 kb, consistent with biglycan mRNA transcript sizes reported for other cell types (26,27). Expression of biglycan did not change with time in culture. Exposure of fibroblasts to 50% O 2 for 24 h resulted in a 4-fold decrease in biglycan mRNA levels (Fig. 7). Longer exposures to 50% O 2 (48 -72 h) or exposures to higher concentrations of O 2 (90%) did not result in a further decrease in biglycan mRNA expression. In contrast to biglycan probe, a cDNA for decorin did not hybridize to mRNA of fetal lung fibroblasts (Fig. 6). The decorin cDNA detected, however, a transcript of 1.9 kb in fetal skin fibroblasts. 2 This finding and the observation that fetal lung fibroblasts did not incorporate detectable amounts of 35 SO 4 into chondroitinase ABC-sensitive macromolecules with M r of 120,000 -160,000 suggest that fetal lung fibroblasts do not synthesize decorin. The effect of O 2 on the expression of the large fibroblast proteoglycan, versican, core protein gene was also examined in the same mRNA preparations by Northern hybridizations. The versican cDNA hybridized to a 10-kb mRNA transcript (Fig. 6), a size identical to that of versican mRNA from several other cell types previously described (28). The versican mRNA expression in fetal lung fibroblasts did not change with time in culture. However, a 2-fold decrease in versican mRNA levels was noted in fibroblasts cultured for 24 h in a gas phase of 50% O 2 (Fig. 7). Rehybridization of the filters with ␤-actin cDNA did not reveal any significant differ-ences in ␤-actin mRNA levels (Figs. 6 and 7). To examine whether the decrease of PG mRNA by elevated O 2 was a general metabolic effect, total RNA from fibroblasts exposed for 48 h to 3 or 50% O 2 was also hybridized with cDNAs coding for procollagen type I and fibronectin. Again, exposure to 50% O 2 reduced biglycan mRNA expression (Fig. 8). In contrast, procollagen type I and fibronectin mRNA expression were upregulated by elevated O 2 (Fig. 8). As mentioned previously, reversibility was tested by exposing fetal lung fibroblasts first to 50% O 2 and then to 3% O 2 . In these cells, message levels for biglycan and type I collagen returned to that of control. However, fibronectin gene expression increased even further (Fig.  8). ␤-Actin gene expression was not affected by these treat- The cells were exposed to various concentrations of O 2 for 0 -72 h. Total cellular RNA (10 g) was analyzed by successive Northern hybridizations of the same filter (ethidium bromide-stained gel) using cDNA probes indicated. A, a representative autoradiogram is shown. B, autoradiograms of Northern blots were quantified by laser scanning densitometry. Open bar, cells exposed to 3% O 2 ; dark gray bar, cells exposed to 50% O 2 . Similar results were obtained in a second experiment. ments (Fig. 8). These findings suggest that elevated O 2 selectively down-regulates PG mRNA expression in fetal lung fibroblasts and that the O 2 effect on PG synthesis is reversible. DISCUSSION Increasing evidence suggests that cell-matrix interactions play an important role in lung morphogenesis. Several studies suggest that matrix molecules can alter the growth of cells, with some components increasing and others decreasing proliferation (29). Thus, changes in matrix molecules following an insult may modulate cell interaction with these matrix molecules, thereby disrupting normal proliferation. This is of particular importance in the preterm human infant, exposed to elevated oxygen concentrations for weeks or months, since a major portion of lung growth and development occurs over these first weeks or months of life. Chronic exposure to elevated concentrations of O 2 is known to reduce lung growth and to cause pulmonary fibrosis in these infants (30,31).
In the present study, we report that elevated O 2 concentrations decrease the synthesis of GAGs and PGs by fetal lung fibroblasts. Using metabolic labeling with Na 2 35 SO 4 , we showed that biglycan was the most abundant proteoglycan secreted into the medium by fetal lung fibroblasts. In contrast to several different types of cells, including human skin and gingival lung fibroblasts (32), fetal rat lung fibroblasts did not synthesize the small soluble proteoglycan, decorin. Cultured bovine aortic endothelial cells (27), human umbical vein endothelial cells (27), and rat pleural mesothelial cells (33) have also been found to express biglycan but not decorin. The lack of decorin synthesis may be due to the use of primary cell cultures, because the human embryonic lung fibroblast cell line, HFL-1, has been shown to synthesize decorin (34). Although cell culture findings should be interpreted with caution, it is likely that the absence of decorin synthesis has a functional role in late fetal lung development. It should be noted that under similar experimental conditions as used for fetal rat lung fibroblasts, fetal rat skin fibroblasts expressed decorin. Thus, it is unlikely that the decorin gene is silenced by methylation of the control regions in isolated fetal lung fibroblasts. The exact role of soluble PGs is unknown, but they have been implicated in cell adhesion. Soluble PGs may inhibit cell adhesion to fibronectin and collagen by binding to the GAG binding site of these matrix molecules, thus making it inaccessible to the cell surface PGs (4, 5). Soluble as well as membrane-and matrix-bound small PGs may also play a role in the control of cell proliferation. Increasing evidence suggests that small PGs control TGF-␤ bioactivity by sequestering TGF-␤ in the extracellular matrix (9,11). The larger membrane-and matrix-associated PGs have also been implicated in binding and regulating the bioactivity of several growth factors (7,35,36). It has been suggested that the binding of growth factors to extracellular matrices is of general significance and may explain the growthpromoting or inhibiting activities of extracellular matrices (5).
We have previously reported growth inhibition of cultured fetal lung fibroblasts in response to elevated oxygen exposure (37). It is possible that altered PG production may contribute to this growth inhibition. We found that elevated O 2 mainly decreased biglycan synthesis of fetal lung fibroblasts. However, cell-matrix-associated PGs were also affected by elevated O 2 concentrations. The results obtained at the protein level were confirmed by mRNA observations. Whether similar changes in proteoglycan production by fetal lung fibroblasts in response to elevated concentrations of O 2 occur in vivo remains to be elucidated. Short term exposure of newborn rats to 95% O 2 has been reported to decrease the density of PGs in alveolar basement membranes (14). In contrast, an increase in biglycan synthesis has been reported in a newborn rat model of chronic hyperoxia-induced lung injury (15). Increases in hyaluronan and PGs have also been described in animal models of pulmonary fibrosis using N-nitroso-N-methylurethane (38) or bleomycin (39,40) and in humans with fibroproliferative lung diseases (41,42). These animal models of pulmonary fibrosis showed that changes in hyaluronan and PGs occurred prior to collagen accumulation, suggesting that hyaluronan and PG may modulate the later fibrotic response (43). Exposure of animals to elevated O 2 concentrations has also been shown to increase the amount of lung collagen as well as elastin and fibronectin (44 -46). To our knowledge, no studies have reported a temporal relationship of PG and collagen in an O 2 toxicity model of pulmonary fibrosis. In the present study, we found that fetal lung fibroblasts increased type I collagen and fibronectin mRNA expression in response to elevated concentrations of O 2 , suggesting that fetal lung fibroblasts selectively up-regulate and down-regulate gene expression of individual extracellular matrix molecules. The observation that the effect of elevated O 2 concentrations on PG and type I collagen gene expression was reversible but that on fibronectin mRNA expression was not strongly supports such selective regulation of extracellular matrix expression by fetal lung fibroblasts in response to O 2 .
The exact mechanism by which oxygen may modulate GAG and PG synthesis of fetal lung fibroblasts is not known. The inhibition of GAG and PG synthesis by 50% O 2 is not due to the loss of cell viability, and the cell number remained constant. Furthermore, we have previously demonstrated that fetal lung fibroblasts developed tolerance to 50% O 2 , as measured by LDH release (37). Similarly, an alteration of 35 SO 4 incorporation induced by elevated oxygen is not due to the changes in sulfotransferase activities because [ 3 H]glucosamine incorporation into GAGs was also altered. The toxic effects of oxygen are believed to be initiated through increased intracellular generation of partially reduced oxygen species. Superoxide, generated by the action of xanthine oxidase on hypoxanthine, has been reported to decrease the synthesis of PGs in cultured bovine chondrocytes (16), bovine articular cartilage explants (47), and isolated perfused rat kidneys (48). Superoxide also damaged PG molecules at the level of the core protein, while GAG side chains were resistant to free radical attack (16). In the present study, we did not investigate whether elevated concentrations of O 2 damaged intact PGs synthesized by fetal FIG. 8. The effect of elevated O 2 concentrations on biglycan, type I collagen, fibronectin, and ␤-actin mRNA levels in fetal rat lung fibroblasts. The cells were exposed for 48 h to either 3 or 50% O 2 . To examine reversibility, cells exposed to 50% O 2 were incubated for 48 h in 3% O 2 . Total cellular RNA (10 g) was analyzed by successive Northern hybridizations of the same filter using cDNA probes indicated. A representative autoradiogram is shown. Autoradiograms of Northern blots were quantified by laser scanning densitometry. Filled bars, cells exposed to 3% O 2 ; gray bars, cells exposed to 50% O 2 ; black bars, cells exposed to 50% O 2 followed by 3% O 2 . Similar results were obtained in a second experiment. lung fibroblasts. However, the parallel decreases in PG mRNA and 35 SO 4 incorporation into PGs make it unlikely that the observed decrease in PGs is due to such an action by reactive oxygen species. Consistent with our findings, superoxide has been shown to increase collagen synthesis in the human fetal fibroblast cell line, IMR-90 (49).