Proteoglycans decorin and biglycan differentially modulate TGF-β-mediated fibrotic responses in the lung

Martin Kolb, Peter J. Margetts, Patricia J. Sime, Jack Gauldie


Transforming growth factor (TGF)-β is a key cytokine in the pathogenesis of pulmonary fibrosis, and pharmacological interference with TGF-β can ameliorate the fibrotic tissue response. The small proteoglycans decorin and biglycan are able to bind and inhibit TGF-β activity in vitro. Although decorin has anti-TGF-β properties in vivo, little is known about the physiological role of biglycan in vivo. Adenoviral gene transfer was used to overexpress active TGF-β, decorin, and biglycan in cell culture and in murine lungs. Both proteoglycans were able to interfere with TGF-β bioactivity in vitro in a dose-dependant manner. In vivo, overexpression of TGF-β resulted in marked lung fibrosis, which was significantly reduced by concomitant overexpression of decorin. Biglycan, however, had no significant effect on lung fibrosis induced by TGF-β. The data suggest that differences in tissue distribution are responsible for the different effects on TGF-β bioactivity in vivo, indicating that decorin, but not biglycan, has potential therapeutic value in fibrotic disorders of the lung.

  • pulmonary fibrosis
  • extracellular matrix
  • treatment

fibrosis is a main feature of various human organ disorders, from lung and kidney fibrosis to liver cirrhosis and atherosclerosis. It is commonly present in chronic and end stages of disease processes. Morphologically, fibrosis is characterized by a disproportionate increase and disordered deposition of extracellular matrix (ECM), resulting in distortion and irreversible loss of organ function. Once present, there is currently no curative treatment for fibrosis (19).

Cytokines have a crucial role in the pathophysiology of fibrosis independent of the system involved. Fibrotic diseases are likely the product of an overwhelming repair process after tissue injury (5,28). The “normal scenario” after injury shows initial expression of proinflammatory cytokines such as interleukin-1, interleukin-6, and tumor necrosis factor-α, which help the immune system to eliminate the injurious agent. Shortly afterward, other cytokines and growth factors, among them transforming growth factor (TGF)-β, connective tissue growth factor, and platelet-derived growth factor, are expressed to limit the inflammation and repair the damage. Usually, all those cytokines are in a subtle balance. During the development of fibrosis, however, profibrotic cytokines are overexpressed and induce overwhelming repair and accumulation of ECM (5, 18, 28).

One of the key profibrotic cytokines is TGF-β, which is chemotactic for fibroblasts, induces the synthesis of matrix proteins and glycoproteins, and inhibits collagen degradation by induction of protease inhibitors and reduction of metalloproteases (16,28). We have previously shown that transient overexpression of TGF-β by adenoviral gene transfer induces a severe fibrotic reaction in the lungs (25). On the other hand, neutralization of TGF-β using antibodies significantly reduced experimental lung and kidney fibrosis (4, 10).

ECM components such as collagen, fibronectin, and proteoglycans also have effects on stromal cell growth and are able to stimulate protein synthesis. These effects can be either pro- or antifibrotic. Collagen and fibronectin are known to act as chemoattractants for fibroblasts and enhance connective tissue synthesis (11). Other components such as decorin have opposite effects on the ECM, acting likely through interference with profibrotic cytokines (22). Decorin belongs to the group of small, leucine-rich proteoglycans and is thought to be a natural inhibitor of TGF-β, capable of binding and neutralizing significant amounts of the cytokine (12, 29). Decorin has been successfully employed to reduce tissue fibrosis in different disease models in kidney, lung, and vasculature (8, 9, 14, 17). Biglycan is another small proteoglycan related to decorin, and its ability to interfere with TGF-β has been demonstrated in vitro (12). To date, biglycan has not been shown to interfere with the activity of TGF-β in vivo.

Small proteoglycans have important structural and functional roles in tissue remodeling and fibrosis (13, 22). We report here about the ability of two different proteoglycans, human decorin and human biglycan, to interfere with TGF-β in vitro and in vivo in a model of TGF-β-mediated fibrogenesis. Proteoglycans and active TGF-β were delivered using adenoviral gene transfer. Decorin was able to block TGF-β in vitro and in vivo as we have previously shown in a model of bleomycin-induced lung fibrosis. In contrast, biglycan was effective in vitro to inhibit TGF-β but failed to reduce the fibrotic tissue response in vivo.


Recombinant adenovirus.

The construction of adenoviral vectors is described in detail elsewhere (2, 17, 25). For the current study, full-length human decorin and human biglycan cDNAs (gift of L. W. Fisher, National Institutes of Health, Bethesda, MD) were cloned into shuttle vectors with a human CMV promoter and cotransfected with a virus-rescuing vector. The resulting replication-deficient virus (AdDec and AdBig) was amplified and purified by CsCl gradient centrifugation and PD-10 Sephadex chromatography, and finally plaque titered on 293 cells. AdTGFβ223/225 (a mutant TGF-β1 which was translated into spontaneously bioactive TGF-β) and control vectors (AdDL) with no insert in the E1 region were produced in the same way.

TGF-β bioassay.

Murine lung fibroblasts (ATCC CCL-206) were plated in a 100-cm2 flask and left in MEM supplemented with 1%l-glutamine, 1% penicillin-streptomycin, 0.4% amphotericin, and 10% newborn calf serum until confluent. The cells were infected with AdDec, AdBig, AdTGFβ223/225, or AdDL at a multiplicity of infection of 20 pfu/cell. After 14 h, supernatants were removed and cells were washed six times with phosphate-buffered saline (PBS). Medium containing 1% serum was added, and supernatants generated over 48 h were saved for further analysis. After the supernatant was removed, cells were washed twice with PBS and lysed in Trizol for RNA extraction.

Bioactive TGF-β was detected using an established bioassay (1). Mink lung epithelial cells (MLEC clone 32, provided by D. Rifkin) with a stable transfection of an 800-bp fragment of the 5′-end of the human plasminogen activator inhibitor-1 (PAI-1) gene fused to the firefly luciferase reporter gene were cultured in six-well dishes in DMEM containing 1% penicillin-streptomycin, 1%l-glutamine, 10% FCS, and 200 μg/ml Geneticin (Sigma Chemicals, Oakville, ON, Canada) until they were confluent. After three washes with PBS, MLEC were exposed to conditioned medium of infected CCL-206 cells: AdTGFβ223/225 supernatants in different dilutions in medium, AdTGFβ223/225 supernatants (1:4 diluted) combined with AdDec or AdBig supernatants (pure, 1:1, and 1:4 diluted), AdDL (pure), or monoclonal antibody against mouse TGF-β1–3 (50 μg/ml; R&D Systems, Minneapolis, MN). As well, different doses of recombinant human TGF-β (R&D Systems) were used to generate standards and estimate TGF-β concentration in AdTGFβ223/225-derived supernatants. The cells were then incubated for 16 h and washed again three times with PBS. A total of 250 μl of lysis buffer (25 mM Tris phosphate, 2 mM dithiothreitol, 2 mM 1,2-diaminocyclohexane-N,N,N′,N′-tetraacetic acid, 10% glycerol, and 1% Triton X-100, pH 7.8) was added, and cells were scraped off. Lysates were pelleted and supernatants assayed for luciferase activity using (D)-luciferin (Luciferase Assay System, Promega, Madison, WI) and multiplate luminometer (Tropix TR717, Applied Biosystems, Foster City, CA). Data are presented in relative light units.

Animal treatment.

The 6-wk-old female C57BL/6 mice were obtained from Charles River Laboratories (Montreal, PQ, Canada) and housed under specific pathogen-free conditions. Rodent laboratory food and water were provided ad libitum. The animals were treated in accordance with the guidelines of the Canadian Council of Animal Care. All animal procedures were performed with inhalation anesthesia with isoflurane (MTC Pharmaceuticals, Cambridge, ON, Canada). Then 4 × 108 pfu of AdTGFβ223/225 were given together with 4 × 108 pfu of AdDec, AdBig, or AdDL by intranasal injection, suspended with PBS in a total volume of 25 μl. Different control groups of animals received AdDec plus AdDL, AdBig plus AdDL, or AdDL only (8 × 108 pfu total dose). Mice were killed by abdominal aortic bleeding at days 3,7, and 21 after injection of adenovectors.

Bronchoalveolar lavage.

After the chest cavity was opened, the lungs were removed and rinsed with PBS. Bronchoalveolar lavage (BAL) was performed as described earlier (17). A total of 0.5 ml of PBS was injected intratracheally and retrieved. The fluid was centrifuged at 1,500 rpm for 10 min, and the supernatant was taken and frozen at −70°C for determination of TGF-β1. BAL cells were counted with a hemocytometer, centrifuged in a cytospin, and stained for differential cytology (Hema3 solution, Biochemical Sciences, Swedesboro, NJ). A total of 300 cells per sample were counted for differentials.

The right main bronchus was tied, and the lung was removed, rinsed in PBS again, and frozen immediately in liquid nitrogen. Tissue samples were stored at −70°C until further processing for mRNA extraction or hydroxyproline determination. The left lung was fixed for histological examination.

RNA extraction and mRNA analysis.

Frozen lung samples were homogenized in Trizol with a tissue homogenizer. Chloroform was added, and the samples were centrifuged at 3,000 rpm for 30 min. The aqueous layer was aspirated, and RNA was precipitated with isopropanol. After centrifugation at 9,000 rpm for 10 min and washing the pellet with 75% ethanol, total RNA was dissolved in RNase-free water and the concentration was determined with a spectrophotometer.

Northern blot technique was used to detect mRNA specific for human decorin and human biglycan in treated cells and lungs. Then 15 μg of total RNA extracts from CCL-206 cells or total lung homogenate were separated on a 1% formaldehyde gel and transferred to a nylon membrane (ICN Pharmaceuticals, Montreal, Canada). Blots were hybridized with a 1.6-kb cDNA probe for human decorin or a 1.6-kb cDNA probe for human biglycan, both EcoRI fragments of the original cDNA used for the construction of the adenovectors. Blots were stringently washed and exposed to film for 1–3 days (Kodak XAR, Rochester, NY). Equal loading was confirmed by hybridization with 300-bp cDNA probe for GAPDH.

Determination of TGF-β levels in BAL fluid.

Total TGF-β1 in BAL fluid was determined after acid activation using ELISA (R&D Systems). Level of active TGF-β1 was measured using the assay without acid activation. The sensitivity of the assay is 7 pg/ml.

Histology and immunohistochemistry.

After fixation in 10% buffered formalin for 24 h, a longitudinal section of the lung was paraffin embedded, sectioned, and stained with hematoxylin and eosin and Masson-Trichrome.

Immunohistochemistry was performed to stain cells and structures positive for α-smooth muscle actin (α-SMA). On the day 7and 21 samples, 3-μm sections were cut on aptex-coated slides, and immunohistochemistry was carried out with antibodies against α-SMA (Dako, Carpenteria, CA). Negative control sections were run in parallel with nonimmune mouse or rabbit immunoglobulin diluted to the same concentration as the primary antibodies. All sections were deparaffinized in xylene followed by 100% ethanol and then placed in a freshly prepared methanol-H2O2 solution for 30 min to block endogenous peroxidase activity. After hydration to water with graded alcohol, the sections were placed in 0.05 M Tris-buffered saline, pH 7.6. The sections were blocked with 1% normal swine serum in Tris-buffered saline for 20 min, followed by incubation in 1:100 mouse anti-human α-SMA in 1% normal swine serum overnight. Sections were then incubated with a 1:300 biotinylated rabbit anti-mouse (Dako) for 1 h followed by 45 min of incubation with a 1:600 streptavidin-peroxidase conjugate (Dako). All sections were rinsed in 0.05 M acetate buffer, pH 5.0, before development in an AEC chromogen substrate for 15 min. All sections were counterstained with Mayer's hematoxylin for 2 min before being mounted with glycerin gelatin.

Hydroxyproline assay.

Frozen lung samples were homogenized in 5 ml of deionized water. The homogenate (1 ml) was hydrolyzed in 2 ml of 6 N HCl for 16 h at 110°C. Hydroxyproline content was determined by a colorimetric assay described earlier (27). Briefly, the reaction was started by adding 1 ml of chloramine T solution to 400 μl of sample (diluted with 1.6 ml of water after the pH was adjusted to 7.0). The reaction was stopped with 1 ml of 70% perchloric acid, and 1 ml of dimethylbenzaldehyde solution was added. After an incubation period of 20 min at 60°C, optical density was determined within 30 min at a wavelength of 557 nm. The results were calculated as micrograms of hydroxyproline per milligram of wet lung weight with hydroxyproline standards (Sigma Chemicals).

Statistical analysis.

Data are shown as means ± SE unless otherwise mentioned. For evaluation of group differences, we used the Student t-test assuming unequal variances. P < 0.05 was considered significant.


Gene expression of human decorin and human biglycan in cell culture and lungs.

Murine lung fibroblasts infected with AdDec and AdBig and incubated overnight showed a strong positive signal for human decorin and biglycan mRNAs by Northern gel analysis (Fig.1). In cells infected with control virus AdDL, no mRNA for human decorin or biglycan was detected. In animals treated with AdDec or AdBig, mRNA signals for human decorin or biglycan were found in total lung homogenates 3 days after infection (Fig. 1). Again, no mRNA for human decorin or biglycan was detectable in lungs treated with control virus AdDL.

Fig. 1.

Northern gel analysis of mRNA extracted from murine lung fibroblasts (CCL-206) treated with AdDec, AdBig, or AdDL and lungs treated with AdTGFβ223/225 plus AdDec, AdBig, or AdDL. Cells were harvested after 14 h, and lungs were removed 3 days after administration of adenovectors. Human decorin (1.9 kb) and biglycan (2.5 kb) mRNAs were detectable in cells and lungs treated with AdDec and AdBig, respectively, but not in control animals treated with AdDL. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Bioactivity of TGF-β in vitro and abrogation by adenoviral-derived decorin and biglycan.

To demonstrate the inhibitory effect of transgene-derived proteoglycans on TGF-β, we used an established bioassay for active TGF-β using MLEC transfected with the luciferase gene under control of a PAI-1 promoter. With infection of murine lung fibroblasts with AdTGFβ223/225, we generated supernatants containing bioactive TGF-β (Fig.2 A). This supernatant diluted 1:1 and 1:4 in medium showed the same bioactivity as 1.0 and 0.5 ng/ml recombinant human TGF-β, respectively. Supernatants generated by infecting cells with AdDec, AdBig, and AdDL resulted in luciferase activities similar to supernatants without virus treatment (negative control). When supernatants of AdTGFβ223/225-infected cells (1:4 diluted) were combined with supernatants of AdDec- or AdBig-infected cells, a significant and dose-dependent reduction of TGF-β-induced luciferase activity was observed (Fig. 2 B). However, the concentration of transgene-derived proteoglycan in the supernatant was not sufficient to completely block TGF-β activity. With anti-TGF-β antibodies, complete abrogation of TGF-β activity was achievable. Supernatants of AdDL-infected or noninfected cells did not alter TGF-β bioactivity in this assay.

Fig. 2.

Luciferase induction by active transforming growth factor (TGF)-β in a bioassay as described in methods and materials. A: TGF-β activity in supernatants of murine lung fibroblasts infected with AdTGFβ223/225. Activity in 1:1 diluted supernatant is equivalent to 1 ng/ml recombinant human TGF-β. B: significant reduction of TGF-β activity in presence of supernatants of cells infected with AdDec and AdBig but not with AdDL (*P < 0.02, **P < 0.0001). RLU, relative light units; rh, recombinant human.

Effect of AdDec and AdBig on TGF-β in BAL fluid of AdTGFβ223/225-injected mice.

In AdDL/AdTGFβ223/225- and AdBig/AdTGFβ223/225-treated animals, total and active TGF-β in BAL fluid (measured by ELISA) was significantly increased byday 3 and further increased by day 7 compared with AdDL/AdDL-treated mice (total 4,752 ± 1,211 and 3,571 ± 1,222 vs. 119 ± 32 pg/ml, active 480 ± 174 and 294 ± 106 vs. <10 pg/ml, P < 0.0001, Fig.3). AdDec/AdTGFβ223/225-treated mice showed a similar increase of TGF-β by day 3. However, by day 7, these animals had significantly lower TGF-β concentration in BAL fluid compared with AdDL/AdTGFβ223/225-treated animals (total 1,170 ± 565 pg/ml, active 61 ± 43 pg/ml,P < 0.05). AdBig/AdDL and AdDec/AdDL groups showed no significant changes of TGF-β in BAL fluid. By day 21, TGF-β was not found elevated in any group compared with control.

Fig. 3.

Concentration of total (A) and active (B) TGF-β in bronchoalveolar lavage (BAL) fluid of animals treated with AdTGFβ223/225 plus AdDL, AdDec, or AdBig (measured by ELISA). There was a significant increase of total and active TGF-β in all mice exposed to AdTGFβ223/225compared with mice treated with AdDL only 3 and 7 days after infection (P < 0.01). By day 7, animals treated with AdDec/AdTGFβ223/225 had significantly lower concentrations of both total and active TGF-β in BAL fluid compared with AdBig or AdDL plus AdTGFβ223/225 (*P< 0.05).

Effect of TGF-β overexpression on inflammation, tissue fibrosis, and hydroxyproline content in the lung in the presence of AdDec and AdBig.

Administration of AdTGFβ223/225 induced a transient inflammatory reaction in the lung present as early as 3 days after injection and most pronounced after 7 days. Total cells in BAL were significantly elevated in AdDL/AdTGFβ223/225-, AdDec/AdTGFβ223/225-, and AdBig/AdTGFβ223/225-treated animals compared with untreated control animals: approximately threefold at day 3 and fivefold at day 7 (Table1). Injection of AdDL/AdDL resulted in a twofold increase of total cells, likely an acute inflammatory reaction against adenovectors. Throughout the experiment, no significant differences were observed in total cell counts between AdDec/AdTGFβ223/225- and AdBig/AdTGFβ223/225-treated animals. Cell differentials showed mainly alveolar macrophages and an increased percentage of neutrophil granulocytes. No difference in cell differentials between treatment groups was observed. AdDec/AdDL and AdBig/AdDL had no further effect on BAL cells compared with AdDL/AdDL (data not shown).

View this table:
Table 1.

Cell content of BAL fluids

Morphologically, animals treated with AdTGFβ223/225 in either combination showed an inflammatory reaction consisting of macrophages, perivascular and peribronchial lymphocytes, and some neutrophil granulocytes in the lungs 7 days after injection. At this early time point, no obvious difference was present among AdDL/AdTGFβ223/225, AdDec/AdTGFβ223/225, and AdBig/AdTGFβ223/225 in hematoxylin and eosin-stained sections. However, by day 7, striking differences were seen between AdDL/AdTGFβ223/225 and AdDec/AdTGFβ223/225 in immunohistochemistry. In the lungs of AdDL/AdTGFβ223/225-treated mice, significantly more α-SMA-positive cells (presumably myofibroblasts) appeared throughout the tissue, predominantly in areas of inflammatory reactions (Fig.4 A). AdDec/AdTGFβ223/225-treated mice showed reduced numbers of these cells, whereas AdBig/AdTGFβ223/225-treated mice were not different from AdDL/AdTGFβ223/225-treated animals (Fig. 4, B and C). Consistent with this observation, collagen accumulation in lungs 21 days after injection of AdTGF-β223/225 plus AdDL or AdBig was much more diffuse and collagen bundles denser compared with AdDec/AdTGFβ223/225 (Fig. 4, DF). It has to be noted that the fibrotic changes induced by overexpression of TGF-β were not completely abrogated in AdDec/AdTGFβ223/225-treated lungs. Lung morphology in animals exposed to AdDL/AdDL, AdDec/AdDL, or AdBig/AdDL did not differ from normal untreated mouse lungs (not shown).

Fig. 4.

A–C: immunohistochemistry for α-smooth muscle actin (α-SMA) in lung tissues 7 days after administration of adenovectors (representative sections, magnification ×160). There was intense staining of interstitial cells in lungs treated with AdDL/AdTGFβ223/225 (A), whereas treatment with AdDec/AdTGFβ223/225 did not induce the presence of α-SMA-positive cells in areas other than bronchi or vessels (B). The amount of α-SMA-positive cells in AdBig/AdTGFβ223/225 was not different from the first group (C). D–F: histology of lungs 21 days after administration of adenovectors (Masson trichrome, representative sections, magnification ×40). There was severe fibrotic tissue reaction with accumulation of collagen (green staining) in animals treated with AdTGFβ223/225 plus AdDL (D) or AdBig (F). Treatment with AdDec/AdTGFβ223/225resulted in a significant reduction of lung fibrosis (E); collagen deposition was only present in patchy peribronchial areas.

On day 21, AdDL/AdTGFβ223/225-treated animals showed an almost twofold increase of hydroxyproline concentration in the lung compared with AdDL/AdDL mice (2.40 ± 0.11 vs. 1.37 ± 0.11 μg/mg lung, P < 0.0001, Fig.5). This TGF-β-induced fibrosis was unchanged in AdBig/AdTGFβ223/225-treated animals (2.22 ± 0.09 μg/mg but was reduced by more than 50% in AdDec/AdTGFβ223/225-treated mice (1.84 ± 0.14 μg/mg, P < 0.01 vs. AdDL/AdTGFβ223/225).

Fig. 5.

Hydroxyproline concentration was increased in lungs treated with AdTGFβ223/225 plus AdDL, AdDec, or AdBig compared with control animals 21 days after adenovector administration. Animals who received AdDec/AdTGFβ223/225 showed a significant reduction of hydroxyproline compared with AdDL/AdTGFβ223/225 (*P < 0.02, **P < 0.0001).


Fibrotic diseases are characterized by a disproportionate accumulation of ECM after tissue injury. The major component of the matrix is collagen, predominantly types I and III (20). Others are proteoglycans and glycoproteins. In the context of tissue injury and loss of organ-specific cells, ECM proteins are used to restructure the defect (21). Other, less-recognized functions of ECM molecules are to participate actively in intercellular communication (13, 22). This can happen either by direct chemotactic stimulation or by interference with cytokines and thus affect cell traffic and function.

Decorin, biglycan, fibromodulin, and lumican are small proteoglycans with leucine-rich repeat structures (13, 22). Decorin carries a single glycosaminoglycan chain and is widely distributed in mesenchymal tissues, associated and bound to collagen, which gains stability through this interaction (3, 13). One of the key features of decorin knockout mice is fragile skin, probably due to irregularly shaped collagen (6). Biglycan has two glycosaminoglycan chains and is localized closely around cells (3, 13). The precise physiological role of biglycan is still under discussion; various interactions with collagen and glycoproteins have been suggested (23). The distribution of fibromodulin and lumican is somewhat more restricted, with fibromodulin present mainly in cartilage and tendons and lumican in the cornea (12). However, recent reports about the composition of pulmonary proteoglycan matrix showed significant quantities of lumican in normal lung and fibromodulin in bleomycin-injured lung, which brought these molecules more into the focus of pulmonary matrix research (7, 26). It has been demonstrated that decorin, biglycan, and fibromodulin are able to interact in vitro with TGF-β, which is a profibrotic key mediator in tissue fibrosis (12,28). The affinity to TGF-β is similar for all, and it has been suggested that these proteoglycans may be able to sequester an overwhelming amount of TGF-β into the matrix and thus control its biological effects (12, 22). For decorin, the in vitro data have been confirmed in different disease models in animals, all of them TGF-β mediated. Decorin was successful in reduction of experimental pulmonary fibrosis induced with bleomycin given either repeatedly as proteoglycan or once as gene using adenoviral gene transfer (9, 17). Furthermore, decorin was employed to reduce fibrotic kidney disease and neointimal proliferation in arteries after balloon angioplasty (8, 14). Current data imply a positive role of biglycan in the course of fibrosis and potential therapeutic value; however, the application of biglycan in an animal model of fibrotic disease has not been reported.

In this study, we used adenoviral transient gene transfer to generate prolonged production and presence of active TGF-β, decorin, and biglycan in a model of pulmonary fibrosis. AdTGFβ223/225induces the synthesis of a mutated TGF-β molecule, which is spontaneously active (25). Murine lung fibroblasts infected with AdTGFβ223/225 produce a high amount of active TGF-β as shown in a bioassay using MLEC transfected with the luciferase gene under control of the PAI-1 promoter. When injected intratracheally into rat lungs, this adenovector leads to severe interstitial fibrosis (25). AdDec and AdBig encode the gene for human decorin and biglycan, respectively. We infected murine lung fibroblasts with these vectors, which transcribed the foreign cDNA into mRNA in vitro. Supernatants of cells infected with either AdDec and AdBig, but not with control vector AdDL, were able to inhibit the activity of AdTGFβ223/225 supernatants in a dose-dependent manner. The ability of both proteoglycans to interfere with TGF-β was comparable, which is consistent with earlier reports.

To further investigate, if decorin and biglycan are able to interfere with TGF-β in vivo, we used a mouse model of pulmonary fibrosis. Intranasal injection of AdTGFβ223/225 resulted in transient overexpression of TGF-β in the lung as measured by elevated cytokine concentration in BAL fluid of treated mice after 3 and 7 days. When AdDec was administered simultaneously with AdTGFβ223/225, the increase of TGF-β in BAL fluid was reduced substantially compared with the combination AdDL/AdTGFβ223/225; although total and active TGF-β were still elevated above control, the concentration was four and eight times lower than in AdDL/AdTGFβ223/225. This observation is in agreement with an earlier report, which has shown that decorin mainly binds the active form of TGF-β (12). In contrast, AdBig given together with AdTGFβ223/225 did not affect cytokine concentration in BAL fluid. After 21 days, considerable interstitial fibrosis was present in mice treated with AdDL/AdTGFβ223/225. The fibrotic response was significantly reduced when AdDec was administered together with AdTGFβ223/225, whereas AdBig had no positive effect. Lung fibrosis was determined by histology and hydroxyproline concentrations in lung homogenates and was preceded by accumulation of myofibroblasts in the tissue 1 wk after administration. Although we did not measure proteoglycans in BAL fluid or lung tissue, we attribute the effects of AdDec and AdBig on TGF-β concentration in BAL fluid and on fibrotic tissue responses to the presence of transgene product in the lungs. We demonstrated strong mRNA signals in the tissue of treated mice 3 days after infection. In previous experiments, we have shown that mRNA of decorin persists at least for 7 days in mouse lungs after injection (17).

These data show that the previously documented inhibitory effect of decorin on TGF-β in vitro can be successfully transferred to an in vivo model of disease in which tissue fibrosis and matrix accumulation are induced by transient overexpression of active TGF-β. The data also support earlier reports that biglycan is able to interfere with TGF-β function in vitro (12). However, biglycan did not affect TGF-β activity in the animal model. The failure of biglycan to induce an antifibrotic tissue effect in this model could be explained by differential binding of the TGF-β molecule through either lowered affinity or altered site specificity. However, earlier reports and the data presented here showing in vitro anti-TGF-β properties of both proteoglycans suggest that biglycan and decorin have similar binding properties to TGF-β (12). Therefore, we speculate that the tissue localization of the two proteoglycans might account for differences in their biological effect in vivo. Decorin is bound to collagen and is probably able to bind TGF-β and prohibit its interaction with cellular receptors, thus controlling its biological effects on matrix-producing cells. Biglycan in contrast is more closely associated to the pericellular space and cell surface (3,22). In this localization, it could bind TGF-β but still allow interaction with the receptor. This hypothesis is strengthened by the observation that AdDec/AdTGFβ223/225-treated animals had a significantly lower concentration of TGF-β in BAL fluid compared with AdBig/AdTGFβ223/225-treated animals. It suggests that the decorin-bound transgene TGF-β is fixed to collagen, whereas biglycan-bound TGF-β remains in the pericellular space where it is more easily liberated by BAL procedure and released into BAL fluid. In previous experiments with rats, we observed that biglycan can induce transient accumulation of myofibroblasts and ECM in the lung after intratracheal injection without resulting in persistent tissue fibrosis (24). It is possible that under certain circumstances biglycan could even act as a profibrotic agent by binding TGF-β and presenting it to the receptor similar to the related molecule betaglycan. Betaglycan is the TGF-β receptor III, which can bind and store TGF-β and eventually presents it to the signal transducing receptors I and II (15).

In summary, the current study confirms the antifibrotic properties of the proteoglycan decorin in an animal model of pulmonary fibrosis in which fibrosis was induced by transient overexpression of active TGF-β using adenoviral gene transfer. The data demonstrate that decorin binds and inhibits TGF-β in vitro and in vivo. Biglycan, another proteoglycan with the capability of binding TGF-β in vitro, failed to inhibit the fibrogenic effects of the cytokine in vivo. We conclude that decorin but not biglycan has a potential role in the future treatment of fibrotic disorders.


We thank Duncan Chong, Xueya Feng, and Mary Jo Smith for outstanding technical help, and Tom Galt, Zhou Xing, and Michael Schmidt for helpful discussions and advice during the course of the experiments.


  • The work was supported by the Medical Research Council (MRC) of Canada. P. J. Sime is supported by James P. Wilmot Foundation and P. J. Margetts by MRC and Kidney Foundation of Canada.

  • Address for reprint requests and other correspondence: J. Gauldie, Dept. of Pathology and Molecular Medicine and Centre for Gene Therapeutics, McMaster Univ., 1200 Main St., West Hamilton, Ontario, Canada L8N 3Z5 (E-mail: gauldie{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract