Increased Cell Surface Fas Expression Is Necessary and Sufficient To Sensitize Lung Fibroblasts to Fas Ligation-Induced Apoptosis: Implications for Fibroblast Accumulation in Idiopathic Pulmonary Fibrosis

Progressive pulmonary fibrosis, especially idiopathic pulmonary fibrosis (IPF), is thought to arise following injury to, and abnormal repair of, the distal alveolar-capillary units (1). Although little is known about how the alveolar epithelium is injured, the ensuing fibrotic response is associated with unrestrained accumulation of fibroblasts and myofibroblasts that synthesize and deposit collagen fibrils within fibroblast foci located in the pulmonary parenchyma (2, 3). Recent studies have suggested that pulmonary fibroblasts arise by several routes including increased migration and proliferation of resident pulmonary fibroblasts, mesenchymal transition of the alveolar epithelium, and recruitment of bone marrow-derived progenitor cells (4–7). In the presence of TGF-β and other agonists, resident fibroblast subsets transdifferentiate into α-smooth muscle actin-positive myofibroblasts (8, 9), produce increased amounts of collagen, and, through their contractile activities, distort the parenchymal lung architecture (9, 10). Furthermore, in contrast to normal resolution of the repair process, in which fibroblasts and myofibroblasts are eliminated by apoptosis (11), in situ studies with fibrotic lung tissues from IPF patients and bleomycin-induced pulmonary fibrosis in mice have shown that fibroblasts and myofibroblasts are resistant to apoptosis and accumulate in the lung parenchyma (12–14). Remarkably, little is known about the physiologic and pathologic mechanisms that contribute to fibroblast survival or to their susceptibility to apoptosis.

Fibroblast apoptosis can be induced by multiple agonists, cognate receptors, and signaling pathways that converge to promote caspase activation. Conversely, resistance to apoptosis occurs following exposure of fibroblasts to an array of prosurvival factors including TGF-β (15, 16). Recent studies have shown that Fas, a cell-surface death receptor of the TNFR superfamily, paradoxically initiates both survival/differentiation and apoptosis in a context-specific fashion. Ligation of Fas alone activates ERK and NF-κB signaling pathways, promotes neuronal and epithelial cell survival (17–19), and induces minimal apoptosis in lung fibroblasts (20). In contrast, prior exposure to TNF-α and IFN-γ renders fibroblasts and myofibroblasts exquisitely sensitive to Fas ligation-induced apoptosis and allows them to overcome the prosurvival effect of TGF-β (20). Two recent studies have shown a similar sensitizing effect of PGE₂ on Fas-induced fibroblast apoptosis (14, 21). Although the source(s) of Fas ligand (FasL) are not completely known, recent studies have suggested that myofibroblasts themselves may be an important source in the fibrotic lung (22, 23). Taken together, these findings suggest that fibroblast and myofibroblast accumulation in the lungs of IPF patients could occur as a result of impaired sensitization to Fas-induced apoptosis. They are also consistent with the developing notion that fibrosis can progress in the absence of a robust inflammatory response associated with reduced levels of proinflammatory sensitizing molecules such as TNF-α, IFN-γ, and PGE₂ (1, 24–26).

Seeking to understand how TNF-α and IFN-γ sensitize fibroblasts to Fas-induced apoptosis, the goal of this study was to investigate the mechanisms that couple cytokine-induced sensitization to the acquisition of susceptibility to Fas-induced apoptosis. Based on an initial analysis of the transcriptomes of unstimulated and TNF-α– and IFN-γ–stimulated lung fibroblasts, we addressed the functional necessity of increased Fas expression in cytokine-induced sensitization of fibroblasts to Fas ligation-induced apoptosis. In this study, we show that exposure to TNF-α and IFN-γ increases cell surface Fas expression in an NF-κB– and p38^mapk-dependent fashion. Furthermore, through the use of small interfering RNA (siRNA)-mediated Fas knockdown and adenovirus-mediated cell surface Fas expression, we show that among 603 genes for which expression was altered in response to TNF-α and IFN-γ, increased cell-surface expression of Fas is necessary and sufficient to promote sensitization to Fas-induced apoptosis. Furthermore, immunohistochemical staining of lung sections from IPF patients revealed minimal expression of Fas by fibroblasts and myofibroblasts in fibroblastic foci. Collectively, these findings emphasize the importance of proinflammatory cytokines in the initiation of Fas-induced fibroblast apoptosis and suggest that the underlying mechanism of their action involves raising Fas cell-surface expression above a basal threshold.

Recombinant human and mouse TNF-α and IFN-γ were purchased from R&D Systems (Minneapolis, MN). Agonistic anti-human Fas (CH-11) and anti-mouse-Fas (Jo-2) Abs were from Upstate Biotechnology (Lake Placid, NY) and BD Pharmingen (San Diego, CA), respectively. Fas probe and primer sets (Hs00531110_m1) for quantitative real-time PCR were from Applied Biosystems (Foster City, CA). Human Genome U133 Plus 2.0 microarray chips were purchased from Affymetrix (Foster City, CA). Fas (sc715-R) and IκB-α (sc371R) Western Abs were from Santa Cruz Biotechnology (Santa Cruz, CA), the actin (mab1501) Western Ab was from Millipore (Temecula, CA) and the phospho-specific HSP27 (PA1-005) Ab was from Affinity Bioreagents (Golden, CO). Conjugated Abs for flow cytometry (allophycocyanin-labeled Fas Ab [558814] and control Ab [555751]) were from BD Pharmingen. The inhibitors SB 203580 (p38^mapk) and Bay 11-7082 (NF-κB, PD98059 [MEK1] and LY294002 [PI3K]) were purchased from Calbiochem (San Diego, CA).

Human primary lung fibroblasts from an IPF patient (FS087) and nondisease control subject (N78), the human fetal lung fibroblast cell lung MRC-5 (American Tissue Culture Collection, Manassas, VA), and mouse primary lung fibroblasts were maintained in modified Eagle’s MEM, supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin, and 10% (v/v) heat-inactivated FBS under a 5% CO₂ atmosphere as previously described (20). Primary cultures of human lung fibroblasts were derived from nondiseased human lungs obtained from the International Institute for the Advancement of Medicine (Edison, NJ), in accordance with an approved Institutional Review Board protocol (HS-1539). All donors suffered brain death and were evaluated for organ donation before research consent. All lung samples failed regional lung selection criteria for transplantation. Individuals had no evidence of current systemic or pulmonary infection, a clear chest radiograph, and partial pressure for oxygen/fraction of inspired oxygen ratio >250 mm Hg O₂. Patients were excluded if they had any history of lung disease or a history of systemic disease that commonly affects the lungs (e.g., rheumatoid arthritis or systemic lupus erythematosus). Fibrotic lung fibroblasts from IPF patients were derived from surgical lung biopsy specimens obtained as part of a prospective, Institutional Review Board-approved, longitudinal study of the pathobiology of fibrotic lung disease within the Interstitial Lung Disease (ILD) Program at National Jewish Health (HS1603). Pathologic diagnoses were determined via review by an expert pulmonary pathologist with the ILD Program, and a pathologic diagnosis of usual interstitial pneumonia was required for a clinical diagnosis of IPF. All clinical diagnoses were determined by an ILD Program physician who was caring for the patient, adhering to established pathologic and clinical criteria for the diagnosis of idiopathic pulmonary fibrosis (27, 28). Freshly harvested explants, surgical lung biopsies, or murine lungs were minced into 1 to 2 mm³ sections and cultured in DMEM containing 10% heat-inactivated FCS media on scored tissue-culture dishes. After 3 to 4 d, fibroblasts derived from the tissue were trypsinized and maintained in tissue culture. All experiments were performed on early passage (2–4) cell cultures.

Total RNA was isolated using TRIzol (Life Technologies, Grand Island, NY). Northern blot analysis was performed as previously described (29). Prior to quantitative RT-PCR (qPCR) analysis, total RNA was cleaned up using the Qiagen RNeasy mini kit (Qiagen, Valencia, CA). Reverse transcription was performed on 1 μg total RNA with random hexamers in a 50-μl reaction using TaqMan RT reagents (Applied Biosystems). The single-plex qPCR reactions for Fas or GAPDH were performed using the probe/primer sets Hs00531110_m1 and 4326317E, respectively, with 40 ng cDNA, 250 nM fluorogenic probe, and 900 nM each primer for Fas and 150 nM each primer for GAPDH. The specificity of PCR was verified by the lack of signal in the no-template and no-reverse transcriptase controls. The threshold cycle was recorded for each sample. The relative Fas mRNA expression levels were determined using the comparative threshold cycle method, as previously described (30). Microarray analysis was conducted with 5 μg total RNA, which was labeled and hybridized to Affymetrix U133 Plus2.0 arrays (Affymetrix) using standard methods. The complete array data set and experimental details are available from the Gene Expression Omnibus repository under accession number GSE26594 (http://www.ncbi.nlm.nih.gov/geo/).

Fas cell-surface expression was determined by flow cytometry. Lung fibroblasts were harvested by trypsinization and stained for 4 h on ice using either the APC-conjugated anti-Fas Ab or APC-conjugated nonimmune IgG. After washing, the cells were analyzed on an FACScalibur flow cytometer (BD Biosciences, San Jose, CA). For total Fas expression (cell surface and intracellular), the cells were first fixed and permeabilized with 1.5% (w/v) paraformaldehyde and 0.1% (v/v) saponin, for 30 min, washed, stained, and analyzed as above but with the inclusion of saponin in the stain and subsequent washes. The data were analyzed with FlowJo (Tree Star, Ashland, OR) and Cell Quest software (BD Biosciences).

Protein expression in cell lysates was determined by Western blot analysis. Four hundred microliters lysis buffer (50 mM Tris-HCl [pH 7.4] containing 150 mM NaCl, 1 mM Na₂EDTA, 1% [v/v] Nonidet P-40, 0.1% [w/v] SDS, 1% [w/v] sodium deoxycholate, 20 mM β-glycerophosphate, 1 mM Na₃VO₄, 1 μg/ml leupeptin, 2 μg/ml aprotinin, and 1 mM PMSF) were added to each well of a six-well plate. The cells were scraped off the plate, and 10 μg protein from the postnuclear lysate was analyzed as previously described (31).

Fas knockdown was achieved in MRC5 cells using a pool of four Fas-specific Accell SMARTpool siRNAs (E-003776-00-0010 human Fas, NM_152876; Dharmacon). Fas siRNAs or a nontargeting Accell control siRNA (D-001910-10-20) were prepared according to the manufacturer’s instructions (Dharmacon) and incubated with MRC5 cells in six-well plates in delivery medium at a final concentration of 1 μM. Additional plates of control cells were grown in parallel. After 36 h, TNF-α (10 ng/ml) and IFN-γ (50 U/ml) were added, and the incubations continued for a further 36 h. The cells were then either stained for cell-surface Fas expression or incubated with agonistic anti-Fas Ab (CH-11; 250 ng/ml) for 4 h and assessed for caspase-8 activation, as described above.

Enforced cell-surface expression of Fas was accomplished using an adenovirus expressing human Fas under the control of the chicken actin promoter. The adenoviral construct was created by ligating EcoRI and ClaI-digested full-length human Fas into EcoRI and ClaI pAd5 CMV GFP vector to create a bistronic construct in which Fas and GFP were separated by an internal ribosome entry site. Next, the majority of the CMV enhancer was excised by digestion with NdeI and ClaI, and NdeI/ClaI-digested chicken β-actin promoter was ligated in place of the CMV promoter to yield the final construct in which Fas-GFP is transcribed under control of the chicken β-actin promoter. The fidelity of the construct was confirmed by: 1) diagnostic restriction enzyme digestion; 2) nucleotide sequencing; and 3) expression of Fas and GPF in HEK293 cells as determined by Western blotting and microscopy to visualize GFP-labeled cells. Adenoviruses expressing the Fas-GFP bistronic sequence were prepared by the Gene Transfer Vector Core at the University of Iowa (http://www.uiowa.edu/∼gene/). MRC cells plated in six-well plates were transduced with Fas and empty vector adenoviruses suspended in 1 ml Optimem (Invitrogen) for 1 h at 37°C. Preliminary experiments were conducted to determine the level of cell surface Fas expression achieved by transduction with the Fas-adenovirus at multiplicities of infection (MOI) of 10, 30, 100, and 300. Fas expression was found to be slightly higher than the level seen following coincubation with TNF-α (10 ng/ml) and IFN-γ (50 U/ml) following transduction with the Fas adenovirus at an MOI of 300, and so this MOI was used in all experiments. Following transduction, an additional 2 ml Eagle’s modified MEM containing 10% (v/v) heat-inactivated FCS were added, and 24 h later, the cells were placed in medium containing 0.1% (v/v) heat-inactivated FCS and incubated for a further 36 h prior to analysis.

Lung tissues were fixed with 10% (v/v) formalin and embedded in paraffin. Each block was sequentially cut into 4-μm-thick serial sections on Superfrost/Plus slides (Fisher Scientific, Pittsburgh, PA) and prepared for immunohistochemical staining as previously described (13). Sections were stained with: 1) anti-Fas (Santa Cruz Biotechnology; Ab SC 715) used at a 1:100 dilution; or 2) anti-human α-smooth muscle actin Ab (Sigma-Aldrich, St. Louis, MO; A2547) used at a 1:200 dilution. Negative controls consisted of nonimmune mouse or rabbit IgG and were used at the same concentration as the primary Abs. Immunohistochemistry was performed either manually or using an automated immunohistochemical stainer (DakoCytomation, Carpinteria, CA) in which slides were incubated sequentially with each primary or control Ab (30 min, 25°C) followed by secondary Ab conjugated to polymerized HRP enzymatic domains (DAKO Envision FLEX system; DakoCytomation), and chromogenic enzyme substrate (4 min, 25°C) according to the manufacturer’s protocol. The slides were counterstained with hematoxylin (4 min, 25°C), rinsed in water, dehydrated with graded alcohols/xylene, and coverslipped with permanent mounting solution.

All experiments were conducted a minimum of three times. Statistical analyses using GraphPad InStat version 3.0b for Macintosh (GraphPad, San Diego CA) were performed by one-way ANOVA, and comparisons among groups were performed with a Newman-Keul multiple comparison test. A p value <0.05 was considered to be significant. All microarray data met the quality control criteria established by the Tumor Analysis Best Practices Working Group (32). Microarray data were analyzed in Partek Genomics Suite (Partek, St. Louis, MO), using GC robust multiarray average probe set summarization. Transcripts that were differentially expressed between cytokine-stimulated and control fibroblasts were identified using a Student t test, with a significance threshold defined by the 5% false discovery rate (FDR). Biological themes differentially affected by cytokine stimulation were identified among the 603 differentially expressed transcripts using a modified Fischer’s exact test as implemented in Ingenuity Pathway Analysis (http://www.ingenuity.com).

To investigate the mechanism by which TNF-α and IFN-γ reprogram fibroblasts from resistance to sensitivity to Fas ligation-induced apoptosis, we exposed representative normal and fibrotic human lung fibroblasts and the human lung fibroblast cell line MRC5 to the combination of TNF-α (10 ng/ml) and IFN-γ (50 U/ml) for 36 h and analyzed changes in their transcriptomes using Affymetrix microarrays. We observed that the transcriptional responses to cytokine stimulation were large and consistent within the set of fibroblastic cells examined. Approximately 50% of the transcripts with 2-fold or greater expression change upon stimulation within a single fibroblast line were likewise changed in all three fibroblast groups (Fig. 1B). This consistent pattern of gene expression change is visible in an unsupervised principal components analysis (Fig. 1A). In this plot, each experimental sample is represented as a solid shape for which the position in space is determined by genome-wide transcript levels. The consistent pattern of expression changes upon stimulation appears as a vector of similar orientation and length connecting the control and stimulated samples from each line. The most robustly induced family of transcripts was chemokine receptor ligands (e.g., CCL20, CCL5, CX3CL1, CXCL1, CXCL10, CXCL11, CXCL3, and CXCL9), which exhibited nominal increases from 18.2- to 5930-fold.

A statistically defined set of 603 differentially expressed transcripts was identified using a Student t test, with a significance threshold determined by the 5% FDR (Supplemental Table I). Biological themes underlying this expression pattern were identified based on the overrepresentation of predefined groups of transcripts within the statistically defined set. As expected, the most overrepresented canonical pathways (Fig. 1D) include pathways associated with IFN signaling, the inflammatory response, and death receptor signaling (Fig. 1E). Among the latter, Trail (APO2L) mRNA expression was increased 1051-fold, whereas expression of the antiapoptotic genes cIAP and c-FLIP was increased 132-fold and 9.1-fold, respectively. Notably, within the Fas signaling pathway, Fas expression was increased 6.1-fold, whereas transcripts encoding other members of the Fas-signaling cascade were not affected by cytokine stimulation. Together, these findings suggest that the level of Fas expression might play a dominant role in the sensitization to Fas ligation-induced apoptosis.

Based on the findings from the microarray experiments, we confirmed the effect of TNF-α and IFN-γ on Fas mRNA levels by stimulating human lung fibroblastic MRC5 cells with TNF-α (10 ng/ml) and IFN-γ (50 U/ml) for up to 48 h and measured Fas mRNA levels by qPCR and Northern analysis. qPCR analysis showed that Fas mRNA expression peaked at ∼6-fold above basal levels by 24 h and declined to ∼4-fold above basal levels at 48 h (Fig. 2A). Similarly, Northern analysis indicated that Fas mRNA was increased ∼3.5 fold following 36 h of stimulation with TNF-α and IFN-γ (Fig. 2A).

We next measured the effect of TNF-α and IFN-γ on expression of Fas protein. MRC5 cells were incubated in either medium alone or with TNF-α (10 ng/ml) and IFN-γ (50 U/ml) for up to 36 h. The cells were then lysed and evaluated for Fas expression by Western blotting (Fig. 2B) or harvested and subjected to flow cytometry analysis following staining with allophycocyanin-conjugated anti-Fas Ab (Fig. 2C). Both approaches indicated that stimulation with TNF-α and IFN-γ for 36 h resulted in an ∼6-fold increase in total Fas protein levels and cell-surface Fas expression. In addition, Fig. 2C (upper panel) shows that the increase in Fas expression in response to TNF-α and IFN-γ was seen in a single population of cells.

Previous studies have suggested that Fas is localized both at the cell surface and within intracellular organelles, including the Golgi complex (33, 34). Although the data shown in Fig. 2A–C suggest that the increase in cell-surface Fas expression is most likely due to a coordinated increase in Fas transcription, translation, and translocation to the cell surface, they do not completely rule out the possibility that translocation of a pre-existing intracellular pool contributes to the TNF-α– and IFN-γ–dependent increase in cell-surface Fas. To address this possibility, we compared total cellular Fas levels (intracellular and cell-surface) and cell-surface Fas levels by flow cytometry in detergent-permeabilized and nonpermeabilized cells. Fig. 2D (left and right panels) shows that total cellular Fas staining was the same as cell-surface staining in unstimulated cells. In addition, the TNF-α– and IFN-γ–induced increase in total Fas and cell-surface Fas staining was identical (Fig. 2D, center and right panels), suggesting that Fas is expressed at the cell surface and not associated with an intracellular pool. Together, these findings indicate that TNF-α and IFN-γ increase Fas transcription and translation and lead to increased cell-surface expression of Fas.

TNF-α–initiated signaling involves two receptors (TNF-R1, TNF-R2) and multiple signaling pathways (35). To investigate the role of TNF-R1 and TNF-R2 in TNF-α– and IFN-γ–induced fibroblast sensitization to Fas-induced apoptosis, primary cultures of lung fibroblasts from TNF-R1^−/−, TNF-R2^−/−, and C57BL/6 mice were stimulated with TNF-α (10 ng/ml), IFN-γ (50 U/ml), or both cytokines for 48 h and analyzed for cell-surface Fas expression by flow cytometry. Fig. 3A shows that stimulation with TNF-α alone and TNF-α plus IFN-γ significantly increased Fas expression on fibroblasts from TNF-R2^−/− mice (p < 0.01) and C57BL/6 mice (p < 0.01), but failed to increase in Fas expression on lung fibroblasts from TNF-R1^−/− mice (p > 0.05). Similarly, Fig. 3B shows that lung fibroblasts from TNF-R1^−/− mice were not sensitized to Fas-induced apoptosis (p > 0.05). Interestingly, compared with fibroblasts from wild-type C57BL/6 mice, lung fibroblasts from TNF-R2^−/− mice exhibited a significant increase in caspase-8 activation following stimulation with TNF-α alone (p < 0.05) and trended toward a significant increase in response to TNF-α plus IFN-γ (p = 0.08). Together, these data indicate that signaling via TNF-R1 is required for cytokine-induced upregulation of Fas and sensitization to Fas-induced apoptosis. In addition, signaling via TNF-R2, although not being required for Fas expression, delivers an antiapoptotic signal that reduces the level of apoptosis induced by ligation of TNF-R1.

Next, we investigated the role of NF-κB, Akt, ERK, and p38^mapk signaling in the TNF-α– and IFN-γ–induced increase in Fas expression and Fas-induced apoptosis using pharmacologic inhibitors. Incubation of MRC5 cells with TNF-α and IFN-γ in the presence of the PI3K and MEK1 inhibitors LY294002 (20 μM) and PD98059 (20 μM) failed to block the cytokine-induced increase Fas expression (data not shown). In contrast, pharmacologic inhibition of p38^mapk and NF-κB with SB203580 (20 μM, Fig. 3C) and Bay 11-7082 (3 μM, Fig. 3D), respectively, significantly reduced the TNF-α– and IFN-γ–induced increase in Fas expression (SB203580, p < 0.001; Bay11-7082, p < 0.001), suggesting that both p38^mapk and NF-κB are required for the TNF-α– and IFN-γ–induced increase in Fas expression. Similarly, pharmacologic inhibition of p38^mapk and NF-κB significantly, though incompletely, inhibited cytokine-induced sensitization to Fas-induced apoptosis (p < 0.01 and p < 0.01, respectively; Fig. 3E, 3F). In control experiments, the concentrations of SB203580 and Bay11-7082 were found to completely block p38^mapk and NF-κB activation as reflected by inhibition of the phosphorylation of IκBα and Hsp-27, respectively (Supplemental Fig. 1). Collectively, these findings suggest that the TNF-α– and IFN-γ–induced increases in Fas expression and sensitization to Fas ligation-induced apoptosis require occupation of TNF-R1 and signaling via p38^mapk and NF-κB.

To determine if the increase in Fas expression was necessary for cytokine-induced sensitization to Fas-induced apoptosis, we used siRNA knockdown to block the TNF-α and IFN-γ increase in cell-surface Fas expression. We then determined the effect of Fas knockdown on cell-surface Fas expression and Fas ligation-induced caspase-8 activation. MRC5 cells were incubated with Fas or control siRNAs (1 μM) or medium alone for 36 h. Next, TNF-α (10 ng/ml) and IFN-γ (50 U/ml), or medium alone, were added to the cells and the incubations continued for an additional 36 h. The cells were then evaluated for cell-surface Fas expression. Fig. 4A shows that neither Fas siRNA nor control siRNA significantly affected basal cell-surface Fas expression (p > 0.05 and p > 0.05, respectively). However, whereas the control siRNA had no effect on cytokine-induced Fas expression (p > 0.05), incubation with Fas siRNA almost completely blocked the increase seen in response to TNF-α and IFN-γ (p < 0.01) (Fig. 4A). Next, we investigated the effect of Fas knockdown on Fas ligation-induced caspase-8 activation. Fig. 4B shows that compared with cells incubated in the absence of siRNAs or incubated with control siRNA, specific Fas knockdown resulted in a complete inhibition of Fas-induced caspase-8 activation in TNF-α– and IFN-γ–exposed cells. These data indicate that the cytokine-induced increase in cell-surface Fas expression is required for Fas ligation-induced caspase-8 activation and apoptosis in fibroblasts.

To determine if increased Fas expression was sufficient to sensitize fibroblasts to Fas ligation-induced apoptosis, we constructed an adenovirus expressing human Fas under the control of the chicken β-actin promoter and transduced MRC5 cells with Fas-expressing or empty vector adenoviruses at an MOI of 300 for 36 h. Fig. 5A shows that transduction with the Fas adenovirus led to an ∼8-fold increase in cell-surface Fas expression, whereas transduction with the empty vector adenovirus had no effect on Fas expression. We next determined the effect of adenovirus-mediated increased Fas expression on Fas ligation-induced apoptosis. In these experiments, apoptosis was quantified by measuring exposure of phosphatidylserine using allophycocyanin-conjugated Annexin V because the intrinsic fluorescence of adenovirus-encoded GFP was found to interfere with the caspase-8 activation assay. Fig. 5B shows that following transduction with the Fas adenovirus, MRC5 cells exhibited increased Fas ligation-induced Annexin V binding compared with cells transduced with empty vector adenovirus (p < 0.01). Of interest, we noticed that whereas the level of cell-surface Fas expression achieved by transduction with Fas-expressing adenovirus was greater than that induced by TNF-α and IFN-γ (Fig. 5A), the level of Fas ligation-induced apoptosis was lower in cells transduced with the Fas adenovirus compared with fibroblasts exposed to TNF-α and IFN-γ (Fig. 5B). These findings suggest that although increased cell-surface Fas expression is sufficient to render the cells susceptible to Fas-induced apoptosis, additional mechanisms induced by TNF-α and IFN-γ may contribute to the efficiency of the sensitization. To explore this notion further, we investigated the relationship between the level of adenoviral-transduced cell-surface Fas expression and sensitivity to Fas ligation-induced apoptosis by quantifying the sensitivity to Fas ligation-induced apoptosis at incremental quartiles of Fas expression. Fig. 5C shows that the induction of apoptosis following Fas ligation increased linearly with increasing cell-surface Fas expression, and the slopes of the lines obtained in the presence and absence of Fas ligation were significantly different (p < 0.001). However, significant increases in Fas ligation-induced apoptosis were only detected in the third and fourth quartiles of Fas expression (p < 0.05 and p < 0.01, respectively). Taken together, these data suggest that Fas expression must exceed a threshold level before the cells can undergo Fas ligation-induced apoptosis. In addition, because the level of Fas expression seen following transduction with the Fas adenovirus was higher than that induced by cytokine stimulation, the data provide further support to the notion that exposure to TNF-α and IFN-γ also reduces the threshold for Fas ligation-induced apoptosis.

Our overall rationale for investigating the mechanisms of sensitization of fibroblasts to Fas ligation-induced apoptosis was to learn more about how this process is impaired in the fibroblastic foci of IPF patients. Given our data showing that increased Fas expression overcomes the basal resistance of lung fibroblasts to Fas ligation-induced apoptosis, we next investigated Fas expression in the fibroblastic foci of lung tissues from IPF patients by immunostaining paraffin sections of lung tissues from IPF patients with an anti-Fas Ab. As shown in Fig. 6A, 6C, and 6D, we observed low to minimal Fas staining by fibroblasts and myofibroblasts in the fibroblastic foci of IPF patients, with only occasional positive Fas staining by interstitial macrophages. In contrast, Fas was abundantly detected in type I and type II cells overlying the fibroblastic foci, as well as in alveolar macrophages (Fig. 6A, 6C) and lymphocytes clustered in lymphoid aggregates (Fig. 6A). Fas was detected primarily in alveolar macrophages in the lung parenchyma of nondiseased control subjects (Fig. 6E).

Building on earlier work showing that the proinflammatory Th1 cytokines TNF-α and IFN-γ reprogram lung fibroblasts and myofibroblasts from resistance to sensitivity to Fas-induced apoptosis, the goal of the current study was to investigate the mechanisms underlying this phenotypic change. In this study, we show for the first time, to our knowledge, that among 603 transcripts for which expression was reprogrammed by TNF-α and IFN-γ, increased cell-surface expression of Fas was necessary and sufficient to overcome the basal resistance of lung fibroblastic cells to Fas ligation-induced apoptosis. In addition, we show that the apoptosis-resistant phenotype of fibroblastic cells in the fibroblast foci of IPF/usual interstitial pneumonia patients is associated with low Fas expression. Together, these findings suggest that therapeutic interventions aimed at increasing Fas expression by fibrotic lung fibroblasts may reverse their state of basal resistance and may potentially provide a new avenue to reduce fibroblast accumulation in the lungs of IPF patients.

The dramatic and durable phenotypic changes in fibroblasts exposed to TNF-α and IFN-γ are reflected in the large number of transcriptional changes this exposure elicits. These expression changes are similar in direction and magnitude in each of the three cell lines for a majority of transcripts, even though the cell lines originate from vastly different lungs, suggesting that reprogramming by TNF-α and IFN-γ is a fundamental pathway open to lung fibroblasts in multiple contexts. The biological themes that these expression changes represent include those that are likely to be the direct result of cytokine stimulation (i.e., the activation of IFN signaling and inflammatory responses). In contrast, changes associated with genes involved in death receptor signaling appear to reflect a change in the poise of the cell with respect to signaling by these receptors rather than activation of this pathway. Interestingly, expression of transcripts encoding the antiapoptotic molecules cIAP and c-FLIP was found to increase in response to TNF-α and IFN-γ and may afford protection against the induction of apoptosis by cytokines during the sensitization process. However, their expression is clearly not sufficient to protect the cells from apoptosis following Fas ligation. Taken together, the microarray data suggest that fibroblast programming by TNF-α and IFN-γ does not result in autocrine activation of death receptor signaling, but reflects changes that may underlie their increased sensitivity to Fas ligation.

Among the genes for which expression was increased by TNF-α and IFN-γ, Fas mRNA and cell-surface protein expression increased ∼5- to 6-fold. Both the increase in Fas expression and sensitivity to Fas ligation-induced apoptosis were found to be dependent on ligation of the TNFR TNF-R1 and activation of p38^mapk and NF-κB. Although little is known about transcriptional regulation of Fas expression in lung fibroblasts, previous studies in other cell types have suggested that activation of NF-κB, c-Jun, and p53 (36–38) contribute to increased Fas expression. In addition, p38^mapk has been implicated in arachidonic acid-induced Fas expression in U937 cells (39). However, although activation of NF-κB and p38^mapk are key steps in increased Fas expression and sensitization to apoptosis induced by TNF-α and IFN-γ, additional signaling mechanisms are likely to be involved because IL-1β and IL-6 both activate NF-κB and p38^mapk yet fail to induce sensitization to Fas ligation-dependent apoptosis (S.K. Frankel and D.W.H. Riches, unpublished observations). Interestingly, although our studies with TNFR-deficient lung fibroblasts revealed that ligation of TNF-R1 was required for the increase in Fas expression and susceptibility to Fas ligation-induced apoptosis, they also suggest a role for TNF-R2 in protecting the cells from apoptosis as previously reported (40). Furthermore, because ligation of TNF-R2 has also been shown to activate NF-κB and p38^mapk signaling (41), it is plausible that the incomplete inhibition of caspase-8 activation seen in the presence of Bay11-7082 and SB203580 may be related to loss of antiapoptotic signaling initiated by ligation of TNF-R2. Collectively, our findings suggest that although p38^mapk and NF-κB activation are necessary in TNF-α– and IFN-γ–induced Fas expression and sensitization to apoptosis, their activation is unlikely to be sufficient to induce these responses. Thus, additional, and as yet unknown, mechanisms likely contribute to the control of fibroblast apoptosis.

Based on in vivo studies in mice (42, 43), we used an siRNA approach to show that increased Fas expression was necessary for TNF-α– and IFN-γ–induced sensitization to Fas ligation-induced apoptosis. In addition, we showed that adenovirus-mediated overexpression of Fas was sufficient to render fibroblasts sensitive to Fas ligation-induced apoptosis. However, although the level of Fas expression achieved following transduction with the Fas-expressing adenovirus was higher than that induced by TNF-α and IFN-γ, the level of apoptosis detected in cells expressing the Fas adenovirus was lower than that induced by cytokines. These findings suggest that although increased Fas expression is necessary and sufficient to enable fibroblasts to undergo Fas ligation-induced apoptosis, other genes for which expression are differentially regulated by TNF-α and IFN-γ may also contribute to reprogramming fibroblast sensitivity to Fas ligation. In particular, as has been reported for several receptors including T cell and B cell Ag receptors (44, 45), our findings suggest that until Fas expression exceeds a certain threshold, its ability to signal apoptosis remains blocked. As illustrated in Fig. 7A, an intriguing possibility is that one of these putative additional gene(s) may contribute to the control of this threshold by preventing death-inducing signaling complex (DISC) assembly at low levels of Fas expression.

Because apoptotic fibroblastic cells are rarely seen in the fibroblast foci of IPF patients (12, 13), and lung fibroblasts are largely resistant to Fas-induced apoptosis when isolated and studied in vitro (14, 20, 46), we investigated the level of Fas expression in lung tissue from patients with IPF. Consistent with previous reports, Fas was strongly expressed by alveolar epithelial cells and alveolar macrophages (47). We also noted abundant Fas expression by cells located in lymphoid aggregates in IPF patients. In contrast, Fas was rarely detected in fibroblastic cells located in fibroblast foci. Recently, we also showed that unlike the overlying alveolar epithelium, fibroblastic cells within fibroblast foci exhibit minimal nuclear translocation of NF-κB (13). Taken together with the present finding that increased Fas expression by fibroblasts is dependent on NF-κB activation, it is possible that the failure of fibroblastic cells to undergo apoptosis in fibroblast foci is due in part to reduced NF-κB activation. In turn, reduced NF-κB activation may contribute to the low level of Fas expression, thereby preventing the ability of these cells to undergo apoptosis. Although many mechanisms could contribute to reduced NF-κB activation, the microarray data presented in this study revealed similar patterns of cytokine-induced increases in NF-κB–dependent gene expression in primary cultures of normal and fibrotic lung fibroblasts. Furthermore, we previously showed that normal and fibrotic lung fibroblasts respond to TNF-α and IFN-γ in vitro with similar increases in Fas expression (20). In addition, we have shown that TGF-β has no effect on the TNF-α and IFN-γ induction of Fas expression (20), suggesting that the TGF-β–rich environment of the lung parenchyma in IPF has little or no impact on the ability of fibroblasts to increase Fas expression upon exposure to the appropriate stimuli. Thus, it is plausible that the failure to activate NF-κB and increase Fas expression in fibroblast foci is due to the absence or reduced presence of the appropriate NF-κB–activating or other sensitizing stimuli within the fibroblastic foci. Consistent with this notion, previous studies have suggested relative deficiencies of key sensitizing molecules including TNF-α, IFN-γ, and PGE₂ in lung tissues and bronchoalveolar lavage specimens from IPF patients (24, 25, 48, 49).

Taken together, our findings suggest a previously unrecognized role for increased cell-surface Fas expression in promoting the sensitization of lung fibroblasts to apoptosis. This finding obviously raises the question of the relevant source(s) of FasL. Constitutive cell-surface FasL expression has been detected in a variety of airway epithelial cells including Clara cells (50, 51). Soluble FasL has also been detected in the bronchoalveolar lavage fluid of patients with a variety of interstitial lung diseases including IPF (52, 53). Previous studies have shown FasL to be expressed by monocytes, macrophages, and NK cells (54, 55), and recent studies have suggested that myofibroblasts themselves are an important source of cell-surface FasL (22, 23). Myofibroblast expression of FasL has been implicated in epithelial injury in mice following intratracheal instillation with bleomycin (22, 56), and epithelial cell apoptosis in IPF patients is often seen in areas overlying fibroblastic foci (56). Thus, it is conceivable that FasL may be constitutively available to engage Fas on fibroblasts and myofibroblasts. When Fas is expressed at comparatively low levels, Fas ligation has been shown to stimulate a variety of nonapoptotic responses including proliferation, survival, tumor growth, differentiation, and chemokine production (57–62). In contrast, when Fas is expressed at comparatively high levels (e.g., in hepatocytes and TNF-α– and IFN-γ–stimulated lung fibroblasts) (20, 42), its ligation results in DISC assembly and induction of apoptosis. Thus, as illustrated schematically in Fig. 7B, we speculate that resistance or sensitivity of lung fibroblasts to Fas ligation-induced apoptosis is controlled by the level of cell-surface Fas expression, even if FasL is ubiquitously present. It is also possible that expression of FasL on the surface of myofibroblasts may allow the cells to interact with Fas on adjacent fibroblasts and other cell types in trans, suggesting that cell–cell contact and the three-dimensional structure of the fibroblastic foci may play an important role in the regulation of Fas signaling and function. Clearly, understanding the events that control fibroblast and myofibroblast sensitivity to apoptosis should provide insights into how these cells persist in the lungs of IPF patients and, desirably, how they may be therapeutically eliminated. Approaches aimed at increasing fibroblast Fas expression might therefore be therapeutically relevant.

We thank Linda Remigio and Jane Parr for outstanding technical assistance. We also thank Dr. Doug Everett-Curran (Division of Biostatistics and Bioinformatics, National Jewish Health) for advice on the statistical analysis of the data.

The authors have no financial conflicts of interest.