An Essential Role for Fibronectin Extra Type III Domain A in Pulmonary Fibrosis

Primary Lung Fibroblast Isolation

Primary human lung fibroblasts and lung fibroblasts from EDA^−/− and wild-type (WT) mice were isolated as we have previously described (14). Written, informed consent was obtained from all subjects in accordance with the University of Michigan Institutional Review Board. Cells were maintained in Dulbecco's modified Eagle medium with 10% fetal calf serum, antibiotics, glutamine, and N-2-hydroxyethylpiperazone-N′-ethane sulfonic acid.

FN Purification

For experiments using purified EDA-containing FN, IMR-90 (Institute for Medical Research, Camden, NJ) human lung fibroblasts were cultured in growth media containing TGF-β1 (2 ng/ml) for 5 days after achieving confluence. Cell surface–associated FN was purified using affinity chromatography as described elsewhere (15). Purity was confirmed by Coomassie staining and by Western blot for EDA FN (not shown).

EDA-Transgenic Mice, Intratracheal Bleomycin Administration, and Bronchoalveolar Lavage

EDA^−/− mice have been described previously (7). WT littermates were used as controls. Animals were bred and housed at the International Centre for Genetic Engineering and Biotechnology (ICGEB) (Trieste, Italy) and the University of Michigan (Ann Arbor, MI). Experimental procedures were approved by the Animal Care and Use Committees of each institution. Bleomycin experiments were performed as we have previously described (16). Bronchoalveolar lavage (BAL) was performed by tracheal cannulation, and lungs were lavaged twice with 750 μl phosphate-buffered saline (PBS)/ethylenediaminetetraacetic acid. Cytospin preparations of equal numbers of cells were stained for differential cell count determination.

Immunohistochemistry

Immunohistochemistry on paraffin-embedded sections was performed as previously described (16) with Masson's trichrome, or with anti–α-smooth muscle actin (anti–α-SMA) antibody (clone 1A4; Dako Corporation, Carpinteria, CA) or anti–phospho-Smad2 antibody (catalog no. 3101; Cell Signaling Technology, Beverly, MA) followed by color development with DAB. A Nikon Eclipse 50i microscope (Nikon Instruments, Melville, NY) was used to visualize and photograph sections. Immunohistochemistry for β6-integrins was performed as previously described (17).

Hydroxyproline Assay

Collagen content was measured in lungs via a conventional hydroxyproline assay (18). Experimental results were quantitated by comparison to a standard curve of known hydroxyproline concentrations.

Western Blot

Western blot analysis of cellular lysates was performed as previously described (16). Antibody against the EDA of FN (clone 3E2) was from Sigma (St. Louis, MO).

Sircol Assay

Sircol assay (Biocolor Ltd., Carrickfergus, UK), measuring soluble collagens, was performed following the manufacturer's instructions as previously described (16).

Semiquantitative Reverse Transcriptase–Polymerase Chain Reaction

Semiquantitative reverse transcriptase–polymerase chain reaction (RT-PCR) was performed as previously described (16). Primer and probe sequences are reported in the online supplement (Table E1).

TGF-β Activation Assay

TGF-β activation was assessed using the method of Munger and colleagues (17) with minor modifications. Briefly, 2 × 10⁵ EDA^−/− or WT lung fibroblasts were cocultured overnight with equal numbers of mink lung epithelial cells (MLECs) stably expressing luciferase coupled to the plasminogen activator inhibitor (PAI)-1 promoter. Cells were assessed for luminescence using a standard reporter assay (Promega, Madison, WI). Luminescence from cocultures was normalized to untreated MLECs alone. MLECs treated with escalating concentrations of active TGF-β1 were used to generate a luminescence standard curve against which raw data were plotted. Results were expressed as relative TGF-β activity.

Lung Fibroblasts from Patients with IPF Express More EDA cFN and α-SMA Than Those from Control Patients

EDA cFN is expressed in the fibroblastic foci of IPF (Reference 13 and Figure E1); thus, we postulated that EDA cFN might be an important contributor to lung fibrosis. To assess this possibility, we determined EDA cFN expression by Western blot in primary human IPF and normal lung fibroblasts. We found that IPF lung fibroblasts expressed markedly more EDA cFN than control lung fibroblasts (Figure 1A). When the membrane was stripped and reprobed for α-SMA, a marker of fibroblast activation, we observed that increased α-SMA expression in IPF fibroblasts accompanied the increase in EDA cFN (Figure 1A). To ensure that the increase in EDA cFN expression was not due solely to increased total FN production, we performed quantitative RT-PCR for both EDA cFN and total FN on RNA harvested from IPF lung fibroblasts (n = 8) and normal lung fibroblasts (n = 8). Although we observed an increase in total FN expression from IPF lung fibroblasts (not shown), the proportion of transcripts including the EDA domain had also increased to approximately 1.5 times that of normal lung fibroblasts (Figure 1B).

EDA^−/− Mice Are Protected from the Development of Experimentally Induced Fibrosis

We have previously shown that EDA^−/− mice are incapable of adequately healing cutaneous wounds (7) due to ulceration of the newly formed epidermis, abnormal reepithelialization, and ongoing proliferation of infiltrating inflammatory cells. To explore the active role of EDA cFN in lung repair and fibrosis, EDA^−/− and WT littermate control mice received a standard dose of intratracheal bleomycin (0.025 U/mouse) to induce lung injury. We observed that WT mice developed a significant degree of fibrosis, as evidenced by elevated levels of lung collagen 21 days after injury (Figure 2A). However, EDA^−/− mice failed to develop significant increases in lung collagen (Figure 2A). Histologic evaluation of trichrome-stained lung tissue revealed significant patchy fibrosis after bleomycin in WT animals, which is typical of this model. Conversely, trichrome-stained lung sections from EDA^−/− mice demonstrated variable thickening of alveolar walls with a conspicuous lack of collagen deposition but with continued interstitial inflammation (Figure 2B). WT and EDA^−/− mice receiving diluent alone (PBS) demonstrated no inflammation or fibrosis 21 days after challenge, and no differences in lung architecture were observed (not shown). We next assessed lung tissue sections for the presence of α-SMA–expressing myofibroblasts. As expected, fibrotic regions of lung in WT mice expressed clusters of α-SMA–expressing myofibroblasts. However, thickened alveolar septae and perivascular thickened areas lacking collagen in EDA^−/− mice appeared devoid of such cells (Figure 2B), although airway smooth muscle cells stained positive for α-SMA, providing evidence for the in vivo necessity of EDA cFN for fibroblast α-SMA expression but not for smooth muscle α-SMA expression. To ensure that differences in proliferative potential of the fibroblasts did not account for the relative lack of α-SMA–expressing cells in EDA^−/− lungs, we assessed [³H]thymidine incorporation rates in WT and EDA^−/− fibroblasts. We found no difference between the two genotypes with respect to thymidine incorporation (Figure E2).

The bleomycin-induced inflammatory pattern in EDA^−/− mice is similar to that observed in bleomycin-treated β6-integrin–null mice (17). β6-integrin is an epithelium-restricted integrin that binds TGF-β latency-associated peptide (LAP) complexes and, through conformational changes, activates TGF-β1. Thus, to determine whether differences in β6-integrin expression might account for differences in development of lung fibrosis, histologic sections of PBS- and bleomycin-treated WT and EDA^−/− mice were assessed for β6-integrin by immunohistochemistry. We did not detect substantial differences in β6-integrin expression between strains in either condition (Figure E3).

Kinetic Analysis of Inflammatory Cell Recruitment after Intratracheal Bleomycin Reveals No Differences between EDA^−/− and WT Animals

Intratracheal bleomycin instillation typically induces a brisk inflammatory response in the lung beginning within 3 days of injury (19). Examination of the inflammatory response in BAL fluid revealed no difference in total cell counts between EDA^−/− and WT mice 3, 7, and 14 days after bleomycin-induced lung injury (Figure 3A). However, by Day 21 after bleomycin instillation, WT mice showed marked resolution of the inflammatory response as assessed by decreasing BAL fluid cell count. In contrast, BAL fluid cell counts from EDA^−/− mice remained elevated and significantly higher than WT mice, indicating a persistent inflammatory response. Determination of inflammatory cell subtypes by differential counting revealed equivalent recruitment of macrophages, lymphocytes, and polymorphonuclear leukocytes (Figure 3B). These data suggest that lack of fibrosis after bleomycin in EDA^−/− mice may be due to a failure to transition from an inflammatory response to a fibrotic one, a role typically ascribed to the effects of TGF-β1. Thus, we next sought to determine the ability of EDA^−/− cells to produce and activate latent TGF-β. Activation of TGF-β requires conformational changes that allow TGF-β receptors to access the mature TGF-β, which can occur under acid conditions, via αvβ6 or other integrins (17) or via plasmin (20). To assess TGF-β production, lung fibroblasts from untreated EDA^−/− and WT mice were cultured for 24 hours in serum-free media. Conditioned media from these cells were harvested and TGF-β was measured using a commercially available ELISA assay (Assay Designs, Ann Arbor, MI). This kit requires acid treatment of conditioned media and, as such, measures both latent and active forms of TGF-β1. We observed no significant difference in production of TGF-β1 between the two genotypes (Figure 3C). Next, EDA^−/− or WT lung fibroblasts were cocultured in FN-depleted media overnight with MLECs stably expressing a TGF-β–responsive portion of the PAI-1 promoter, followed by lysis for assessment of luciferase activity as previously described (17). Results were calculated based on a standard curve of luciferase activity in reporter cells after stimulation with known concentrations of active TGF-β1. We noted that EDA^−/− fibroblasts were significantly less able to activate TGF-β1 than WT fibroblasts in vitro (Figure 3D).

Increased Mortality in EDA^−/− Mice after High-Dose Bleomycin Injury

To test whether EDA^−/− mice would generate lung fibrosis after any inflammatory insult, we used a model in which EDA^−/− and WT mice received a single intratracheal bleomycin injection of 0.075 U/mouse (“high dose,” corresponding to three times the “standard” dose), a dose expected to induce mortality rates of 35–40% (21). Surprisingly, we observed an increased mortality of EDA^−/− mice compared with WT mice (Figure 4). Histologic examination revealed profound and ongoing inflammatory changes in EDA^−/− mice at 12, 19, 23, and 29 days postinjury, whereas WT mice at similar time points showed progressive architectural distortion consistent with lung fibrosis (Figure E4). These data supported our hypothesis that EDA cFN might be necessary for transitioning from an acute inflammatory response to a chronic fibrotic response, and showed a functional significance to our data that EDA^−/− mice were less capable of activating TGF-β1.

Smad Signaling Is Similar between EDA^−/− and WT Fibroblasts

Although our observations suggested a defect in TGF-β1 activation in EDA^−/− animals, we wanted to also explore potential differential TGF-β signaling between WT and EDA^−/− mice; thus, we next evaluated functional TGF-β1 signaling in lung fibroblasts from each genotype. Serum-starved EDA^−/− and WT lung fibroblasts were cultured in the absence or presence of active TGF-β1 (2 ng/ml) for the indicated time points, lysed, and assessed for Smad2 phosphorylation. Blots were stripped and reprobed for total Smad2 as a loading control. Smad2 phosphorylation after TGF-β was similar between the WT and mutant cells, with peak phosphorylation occurring 1 hour after stimulation (Figure 5), suggesting that early TGF-β signaling was intact in both EDA^−/− and WT fibroblasts. To assess this effect in vivo, we next performed immunohistochemistry for phospho-Smad2 in mouse lung samples 21 days after bleomycin administration. Smad2 phosphorylation could be observed in both EDA^−/− and WT mice, although to a much greater degree in WT mice (Figure E5). This finding suggests that, although EDA^−/− cells in vitro are capable of phosphorylating Smad2 equally after active TGF-β1 stimulation, in vivo they do not appear to phosphorylate Smad2 in fibrotic lung injury, consistent with an inability to activate TGF-β in vivo.

Culture of EDA^−/− Fibroblasts on EDA-containing FN Restores TGF-β1–stimulated Increases in α-SMA and Collagen Expression

As confirmation of the critical role of EDA FN in TGF-β1–stimulated α-SMA and collagen induction, we next assessed whether plating EDA^−/− lung fibroblasts on plates coated with EDA cFN (50 μg/ml) was capable of restoring the observed deficiencies. Consistent with previous data (6), TGF-β1 failed to induce α-SMA expression in EDA^−/− cells plated on plastic (Figure 6A). However, if EDA^−/− cells were cultured on EDA-containing cFN, α-SMA induction in response to TGF-β1 stimulation at the gene and protein level was restored (Figure 6A). Similarly, collagen expression in EDA^−/− fibroblasts grown on plastic after TGF-β1 (2 ng/ml) stimulation was not significantly augmented (Figure 6B). Yet again, culture of EDA^−/− fibroblasts on EDA-containing cFN resulted in restoration of TGF-β1 responsiveness as evidenced by increased collagen production at the gene and protein level (Figure 6B). Interestingly, at the gene level, collagen I was markedly induced in EDA^−/− cells plated on EDA cFN, regardless of the presence of TGF-β1 (Figure 6B), suggesting that post-translational regulation of collagen was also influenced by plating cells on EDA cFN. Finally, we observed that EDA^−/− cells plated on EDA cFN significantly enhanced TGF-β1 activation (Figure 6C).

We conclude that EDA^−/− mice are protected from the development of experimental lung fibrosis due to a crucial role of EDA cFN in resolution of inflammation, fibroblast activation, and subsequent tissue fibrosis. Mechanistically, EDA cFN is necessary in vitro for TGF-β activation and myofibroblast differentiation. In aggregate, these data suggest that EDA cFN plays an essential role in fibrogenesis.