Alveolar epithelial cells express mesenchymal proteins in patients with idiopathic pulmonary fibrosis

Cecilia Marmai, Rachel E. Sutherland, Kevin K. Kim, Gregory M. Dolganov, Xiaohui Fang, Sophia S. Kim, Shuwei Jiang, Jeffery A. Golden, Charles W. Hoopes, Michael A. Matthay, Harold A. Chapman, Paul J. Wolters


Prior work has shown that transforming growth factor-β (TGF-β) can mediate transition of alveolar type II cells into mesenchymal cells in mice. Evidence this occurs in humans is limited to immunohistochemical studies colocalizing epithelial and mesenchymal proteins in sections of fibrotic lungs. To acquire further evidence that epithelial-to-mesenchymal transition occurs in the lungs of patients with idiopathic pulmonary fibrosis (IPF), we studied alveolar type II cells isolated from fibrotic and normal human lung. Unlike normal type II cells, type II cells isolated from the lungs of patients with IPF express higher levels of mRNA for the mesenchymal proteins type I collagen, α-smooth muscle actin (α-SMA), and calponin. When cultured on Matrigel/collagen, human alveolar type II cells maintain a cellular morphology consistent with epithelial cells and expression of surfactant protein C (SPC) and E-cadherin. In contrast, when cultured on fibronectin, the human type II cells flatten, spread, lose expression of pro- SPC, and increase expression of vimentin, N-cadherin, and α-SMA; markers of mesenchymal cells. Addition of a TGF-β receptor kinase inhibitor (SB431542) to cells cultured on fibronectin inhibited vimentin expression and maintained pro-SPC expression, indicating persistence of an epithelial phenotype. These data suggest that alveolar type II cells can acquire features of mesenchymal cells in IPF lungs and that TGF-β can mediate this process.

  • fibronectin
  • usual interstitial pneumonia
  • flow cytometry
  • epithelial-to-mesenchymal transition

idiopathic pulmonary fibrosis (IPF) is a lung disorder that is believed to be the consequence of abnormal healing of the lung in response to injury (30). It is histopathologically heterogeneous with regions of normal lung alternating with regions of abnormal lung, which are characterized by type II cell hyperplasia, fibroblastic foci, excessive collagen deposition, and destruction of normal lung architecture (14). Because of the destruction of normal lung architecture, patients suffer from progressive loss of lung function and die, on average, 3 yr after diagnosis (3, 16). Recent epidemiological studies report that IPF is more prevalent in older individuals with a mean age at diagnosis of 65 yr and that its incidence is increasing (27, 28). Therefore, because of the increasing disease burden and lack of effective therapies, it is imperative more is learned of the pathogenesis of IPF.

The pathogenesis of the fibrotic process in IPF lungs is unknown, but increasingly it is thought the epithelium plays a direct role in the disease process (31). Recognition that the epithelium is abnormal in IPF lungs began with histopathological studies showing prominent type II cell hyperplasia overlying a denuded basement membrane in IPF lungs (14, 31). These type II cells have increased levels of apoptosis (17, 34) and an altered phenotype supported by studies showing that they synthesize cytokeratins not commonly expressed in normal type II cells (10) and have genomic mutations of the transforming growth factor-β (TGF-β) receptor type II (23). Type II cells may promote lung fibrosis by secreting the profibrotic growth factors PDGF or endothelin-1 (6, 8). In addition, recent data in mouse models of lung fibrosis suggest the epithelial cell can directly promote lung fibrosis by acquiring a mesenchymal phenotype through a process termed epithelial-to-mesenchymal transition (EMT) (15, 39).

EMT is a multistage process wherein epithelial cells acquire mesenchymal features following a molecular reprogramming mediated by growth factors, of which TGF-β is the prototype (12). EMT is characterized genetically by decreased expression of epithelial cell-associated genes (e.g., E-cadherin) and increased expression of mesenchymal cell-associated genes [e.g., type I collagen (Col1) or α-smooth muscle actin (α-SMA)] (12, 38). Phenotypically, EMT is characterized by a loss of cellular polarity and increased mobility (12, 20). Tissue homeostasis does not appear to require EMT. In contrast, EMT occurs during vertebrate embryo development, the progression of epithelial cell cancers, and remodeling of injured tissue (12, 20, 38).

To date, the evidence that EMT occurs during the remodeling of lungs in IPF is limited to immunohistochemical studies colocalizing epithelial and mesenchymal proteins within fibrotic lungs (15, 39). To gather further evidence that EMT occurs in IPF lung, this study applies novel gene expression profiling, immunohistochemical, and type II cell purification and culture techniques to further support the hypothesis that during specific pathological conditions human alveolar epithelial cells acquire mesenchymal features, that TGF-β may mediate the transition, and that epithelial cell-derived mesenchymal cells may contribute to the development of idiopathic pulmonary fibrosis.


Lung tissue processing for laser capture.

Subjects enrolled in this study underwent a history, physical examination, high-resolution computed tomography, pulmonary function tests, and diagnostic lung biopsy. In all cases, the pathological diagnosis was usual interstitial pneumonia and the consensus clinical diagnosis IPF. To acquire tissue, at the time of lung biopsy, a portion of excised lung was immediately inflated with sterile PBS, embedded in Tissue-Tek OCT compound (Miles, Elkhart, IN) and snap frozen in liquid nitrogen. Written, informed consent was obtained from all subjects, and the study was approved by the University of California, San Francisco Committee on Human Research.

Laser capture.

The P.A.L.M. Laser Microbeam system (P.A.L.M. Microlaser Technologies, Bernried, Germany) was used to acquire isolates of epithelial cells from IPF tissues. Briefly, 5-μm cryosections of IPF tissue were mounted on polyethylene naphthol membrane-coated slides (P.A.L.M. Microlaser Technologies), fixed for 30 s with 70% ethanol, stained with Mayer's hematoxylin (Sigma, St. Louis, MO) and eosin Y (Sigma), and dried with successive ethanol rinses and forced air. These sections were then visualized via an inverted microscope, and the alveolar epithelial cells were identified by histological appearance and recovered by laser capture microdissection (see Supplemental Fig. S1; the online version of this article contains supplemental data). To maintain RNA quality, time from sectioning to completion of laser capture did not exceed 15 min. Captured epithelial cells were lysed immediately in RLT RNA isolation buffer (Qiagen, Valencia, CA) containing 10 μl/ml mercaptoethanol and 20 μg/ml linear acrylamide as carrier (Ambion, Austin, TX), and RNA was isolated by use of the RNeasy Mini Kit (Qiagen). RNA samples eluted from RNeasy columns were treated with RQ1 RNase-free DNase (Promega, Madison, WI) and purified a second time on RNeasy columns. RNA quality and quantity were analyzed via the Agilent 2100 Bioanalyzer and RNA 6000 Pico LabChip kit (Agilent Technologies, Palo Alto, CA). Typically, 200–300 epithelial cells were captured per patient sample. To ensure reproducibility, each sample was captured at least three times on separate days.

TaqMan two-step RT-PCR.

Our two-step real-time RT-PCR transcript quantification method uses two sets of separate nested gene-specific primers to produce amplicons smaller than 250 bp (4). Briefly, in the first step, RNA was converted into single-stranded cDNA by use of PowerScript (BD Biosciences, Palo Alto, CA) and random hexamers (Biosearch Technologies, Novato, CA) followed by PCR for ∼25 cycles by using a cocktail of the outer gene-specific primers and Advantage 2 polymerase Mix (BD Biosciences). The second step used real-time PCR in a 384-well format for quantifying individual transcripts via TaqMan in PCR products generated in step one. Aliquots of PCR product from step one were used in duplicate for individual TaqMan quantification. Nested gene-specific forward and reverse primers (internal to those used in step one) and a TaqMan probe were used along with ABI Prism 7700/7900 Sequence Detection Systems (Applied Biosystems, Foster City, CA). Threshold cycle values were converted to relative transcript copy numbers as described (4) and normalized to geomean of two most stable housekeeping genes of the specimens via GeNorm (35). Normalized relative gene copy numbers were then compared between IPF epithelial cells and those measured in normal type II cells, and the ratio of gene expression in IPF epithelial cells vs. normal type II cells was calculated.


The 5-μm cryosections of tissue from IPF (obtained from diagnostic lung biopsies) or normal lung and cultured cells were fixed and then blocked with PBS containing 5% nonimmune goat serum and 1% BSA. Rabbit polyclonal anti-human pro-surfactant C (Chemicon International), mouse anti-human calponin 1 (Sigma), mouse anti-human α-SMA (Sigma), mouse anti-human N-cadherin (Chemicon International), and mouse anti-human vimentin (Sigma) were used as primary antibodies. Dilutions of primary antibody were applied to the fixed tissue sections overnight at 4°C. Bound primary antibodies were detected with either fluorescein- or Texas red-conjugated secondary antibodies. All sections were counterstained with 1 mM 4,6-diamidino-2-phenylindole (DAPI) (Molecular Probes) prior to being mounted in Prolong (Molecular Probes). Immunostained sections were visualized via a Nikon fluorescent microscope and images captured with a SPOT 2.3.1 camera (Diagnostic Instruments) and analyzed with SPOT 4.0.9 software (Diagnostic Instruments).

Type II cell isolation.

Human alveolar epithelial type II cells were isolated as previously described (5, 36) from explanted IPF lungs or human lungs not used by the Northern California Transplant Donor Network; our studies indicate that these lungs are physiologically and pathologically normal (37). Cells were isolated from lungs that had been preserved for no longer than 24 h at 4°C. Briefly, the pulmonary artery was perfused with PBS and distal air spaces were lavaged several times with PBS. Then HBSS containing elastase was instilled into distal air spaces and the lung was incubated at 37°C for 60 min. The elastase-digested lung was minced in the presence of bovine serum and DNase and the cell-rich fraction was sequentially filtered through nylon meshes. Filtered cells were separated by using a discontinuous Percoll density gradient centrifuged at 400 g for 20 min. The band containing type II cells was collected, washed, then resuspended in PBS containing FCS and incubated with anti-CD-14 antibody-coated magnet beads for 40 min. Macrophages adhering to the beads were depleted with a Dynal magnet, and the remaining cell suspension was incubated on human IgG-coated tissue culture-treated petri dishes for 90 min. Unattached type II cells were collected and counted. Purity of human alveolar type II cells, assessed by pro-surfactant protein C (SPC) staining on cytospin, was ≥95% for normal lung.

Type II cell culture.

Human primary type II cells were cultured on tissue culture plates coated with either Matrigel (BD Biosciences) supplemented with 5% collagen type I (Sigma) or fibronectin (Roche) 100 μg/ml. Cells were maintained in StemPro-34 SFM (serum-free medium) Complete Medium (GIMCO Invitrogen Cell Culture) containing 10 ng/ml keratinocyte growth factor (Peprotech) and antibiotics in a 37°C, 5% CO2 incubator. The medium was replaced every 48 h.

Flow cytometry and FACS sorting.

Type II cells isolated by the method outlined above were suspended in DMEM H-21 containing 10% fetal bovine serum and incubated for 30 min on ice with primary antibodies: rat anti-human E-cadherin (Zymed Laboratories) and mouse anti-human CD-45 (Dako Cytomation). After being washed with PBS, the resuspended cells were incubated for 20 min on ice with appropriate FITC- and phycoerythrin-conjugated secondary antibodies. Samples were then analyzed via a LRS II flow cytometer (Becton Dickinson) and sorted by a Moflo High Performance Cell Sorter (Dako Cymation).


Cell lysates were subjected to SDS-PAGE under reducing conditions and transferred to nitrocellulose membrane (Life Sciences Products, Boston, MA). The membrane was washed with 50 mM Tris·HCl containing 0.5 M NaCl, 0.01% Tween-20 (TBS; pH 7.5) and incubated overnight in 5% milk containing a 1:1,000 dilution of primary antibody. The membrane was then washed with TBS, incubated in TBS for 30 min containing a 1:2,000 dilution of horseradish peroxidase (HRP)-conjugated secondary antibody (New England Biolabs, Beverly, MA) and washed again. Immunoreactivity was detected by using the Phototope-HRP detection kit (New England Biolabs).

Statistical analysis.

Normalized gene expression data were analyzed using the Significance Analysis of Microarrays (SAM) statistical package available at∼tibs/SAM/index.html. SAM identifies genes with statistically significant changes in expression by assimilating a set of gene-specific t-tests. Each gene is assigned a score on the basis of its change in gene expression relative to the standard deviation of repeated measurements for that gene. Genes with scores greater than a threshold (defined as the delta) are considered significant. This delta can be adjusted to identify smaller or larger sets of genes, and a false discovery rate (FDR) is calculated for each set. The FDR is the percentage of such genes identified by chance. For example, an FDR of 1% indicates that, on average, 1% of genes called significantly regulated are not actually regulated. In these studies, two sets of analysis were performed. First, SAM was used with the one class response to analyze variation of gene expression in individual isolates of normal type II cells. Second, a two-class response was used to analyze variation of gene expression between captured IPF epithelial cells and normal type II cells. For this analysis, the delta value was acquired by fixing the median FDR to 0.69%.


Epithelial cells.

To develop evidence that epithelial cells in IPF patients express mesenchymal genes, we studied the atypical collections of cuboidal-shaped epithelial cells overlying regions of excess matrix deposition in IPF lungs (Fig. 1A). Because their histological appearance suggested these cells are alveolar type II cells, we immunostained normal and IPF lungs for the type II cell-specific protein SPC and found that, in contrast to the distribution of SPC-immunoreactive cells in normal lungs (Fig. 1B), all the epithelial cells in regions of matrix remodeling were immunoreactive for SPC (Fig. 1C). These findings confirm the cuboidal-shaped epithelial cells are type II cells and that their spatial distribution is abnormal in regions of matrix remodeling in IPF lungs.

Fig. 1.

Epithelial cell characterization in idiopathic pulmonary fibrosis (IPF) lung. A: scout slide of lung tissue stained with hematoxylin and eosin. Note the atypical collection of cuboidal-shaped epithelial cells (arrowheads) overlying regions of matrix deposition (*). B and C: 5-μm cryosections were immunostained sequentially with rabbit anti-human surfactant protein C (SPC) and then Texas red-conjugated anti-rabbit IgG. Note the scattered distribution of SPC-immunoreactive cells in normal lungs (arrowheads, B) compared with prominent immunostaining of layers of cuboidal-shaped epithelial cells for SPC (arrowheads) in IPF lungs harvested at the time of diagnosis (C). Images were obtained via a ×60 immersion objective.

Our plan to use laser capture microdissection and quantitative real-time PCR to profile genes expressed in the IPF epithelial cells prompted us to identify appropriate epithelial cells to use as controls. Because normal type II cells exist as single cells and are relatively few in number, it was technically not feasible to use laser capture to isolate them from normal controls. Instead, type II cells purified from unused normal lungs of transplant donors were used because this was the richest source of a pure population of normal type II cells available. To ensure these cells were representative of in situ normal alveolar type II cells, they were immediately lysed after purification for total RNA isolation and real-time PCR for comparison to expression profiles of in situ captured cells.

Gene expression profiling in IPF epithelial cells isolated by laser capture microdissection.

Because of the prominence of SPC-immunoreactive cells in IPF lungs, we hypothesized that they were potentially molecularly reprogrammed compared with normal type II cells. To address this hypothesis, relative expression of 205 genes (see Supplemental Table S2) that could potentially play a role in fibrogenesis, matrix remodeling, or EMT were compared in captured IPF epithelial cells (3 donors) to normal type II cells (2 donors). From this data set, we selected a subset of 76 reliably amplified genes, differentially expressed in IPF epithelial cells, for confirmation in a second analysis of samples obtained from six IPF patients (clinical data are summarized in Supplemental Table S1) and six normal controls. We then used a two-class analysis to compare levels of gene expression in captured IPF epithelial cells to normal type II cells and found that 27 genes were differentially expressed in IPF epithelial cells by using parameters set at a median FDR of 0.69% and a cutoff of at least a twofold difference in expression (Table 1). Of these 27 genes, 12 were expressed at higher levels and 14 at lower levels in IPF epithelial cells. Interestingly, several genes expressed at higher levels in IPF epithelial cells include the mesenchyme-associated genes calponin 1, α-SMA, and Col1. However, not all the mesenchyme- or epithelium-associated genes we tested were differentially expressed in IPF epithelial cells. For example, the mesenchyme-associated gene S100A4 (see Supplemental Table S2), epithelium-associated surfactant and mucin genes were not deregulated in captured IPF epithelial cells (Table 1). The selective pattern of increased mesenchyme-associated gene expression indicates that the captured IPF epithelial cells express genes commonly associated with a mesenchymal cell phenotype.

View this table:
Table 1.

Expression profile of IPF alveolar type II cells obtained by laser capture

Immunohistochemical evidence for EMT in IPF lungs.

Next, tissue sections of diagnostic lung biopsies obtained from IPF patients were immunostained to confirm that IPF epithelial cells express mesenchyme-associated genes (Table 1) as proteins. Tissue sections immunostained for both the alveolar epithelial cell-specific protein SPC and the smooth muscle cell-associated protein calponin 1 showed the luminal layer of alveolar cells to be immunoreactive for SPC and calponin 1. In addition, bundles of subepithelial cells typical for myofibroblasts were also strongly immunoreactive for calponin 1 (Fig. 2, AD). Next, when IPF tissue sections were immunostained for both α-SMA and SPC, select cells immunoreactive for both α-SMA and SPC were found in isolated cells in regions of matrix remodeling (Fig. 2, EH). SPC and calponin or α-SMA did not colocalize to individual cells in normal lung tissues (Fig. 2, J and K). These findings in IPF lungs of luminal type II cells coexpressing both epithelium- and mesenchyme-specific genes, and the identification of cells immunoreactive for SPC within remodeling matrixes support the possibility that some type II cells are undergoing EMT in IPF lung tissue.

Fig. 2.

Immunofluorescent colocalization of SPC and calponin 1 (CNN1) or α-smooth muscle actin (α-SMA) in IPF lungs. AD: cryosections of IPF lung tissue harvested at the time of a diagnostic lung biopsy immunostained with both rabbit anti-human SPC and mouse anti-human calponin 1 antibodies. Note the calponin 1 immunostaining in the alveolar epithelium (arrowheads) and in cells underneath the epithelium (arrows) on the merged image (D). DAPI, 4,6-diamidino-2-phenylindole. E and F: cryosections of IPF lung tissue harvested at the time of a diagnostic lung biopsy immunostained with both rabbit anti-human SPC and mouse anti-human α-SMA antibodies. Note the submucosal (arrowheads) cells immunoreactive for both SPC and α-SMA on the merged image (H). I: cryosections of IPF lung immunostained with rabbit and mouse nonimmune IgG and the same secondary antibodies used in AH. J: cryosections of normal lung tissue immunostained with both rabbit anti-human SPC and mouse anti-human calponin 1 antibodies. K: cryosections of normal lung tissue immunostained with both rabbit anti-human SPC and mouse anti-human α-SMA antibodies. SPC immunoreactivity was imaged by using Texas red-conjugated anti-rabbit IgG and calponin 1 and α-SMA with FITC-conjugated anti-mouse IgG. Images are representative of findings in 3 different IPF lungs and were obtained via ×60 (AI) or ×20 (J and K) objectives.

Validation of gene expression profiling in IPF epithelial cells isolated by FACS sorting.

The gene profiling of type II cells isolated by laser capture ensured analysis of type II cells in diseased regions of IPF lungs. However, laser capture purification of type II cells allows the possibility that fragments of mesenchymal cells are present in the captured cells. To overcome these potential experimental limitations and to validate differences in gene expression defined in laser capture type II cells (Table 1), we next compared levels of gene expression in alveolar type II cells freshly isolated from single cell preparations of IPF and normal lungs. In preliminary experiments, we note that our preparations of type II cells from fibrotic lungs were contaminated with mesenchymal cells (Fig. 3A). Although the number of contaminating mesenchymal cells were few (typically <20%), to eliminate the possibility that relative differences in mesenchymal cell numbers could confound our analysis, we fluorescence-activated cell sorting (FACS) sorted type II cells both from normal and IPF lungs for E-cadherin and CD-45 and isolated the ECad+/CD45 population (Fig. 3C), which were reliably 100% type II cells (Fig. 3B).

Fig. 3.

Fluorescence-activated cell sorting (FACS) purifies type II cells to homogeneity. Type II cells isolated from IPF lungs harvested at the time of a lung transplant by either a standard protocol (No FACS; A), or FACS sorting (B) for E-cadherin-positive cells from normal lung cultured on Matrigel for 24 h, then coimmunostained for pro-SPC (red) and vimentin (Vim; green). Note the vimentin-immunoreactive mesenchymal cells (arrowheads) that contaminate type II cell preparations isolated by a standard protocol. C: representative FACS plot of cells isolated from disaggregated lung tissues, stained for E-cadherin (Ecad) and CD45. The stained cells were FACS sorted to isolate the Ecad+/CD45 population (outlined in the box, R4).

Genes selected for validation included the most highly differentially expressed gene in the laser capture studies (FoxG1), a gene previously reported to have increased expression in IPF type II cells (MMP7) (29, 40), a transcription factor that mediates EMT in some experimental systems (Slug) (21), and mesenchymal cell-associated genes, two of which were shown to be expressed type II cells (Col1, α-SMA, and calponin 1, Fig. 2). Similar to the findings in laser-captured epithelial cells, the levels of expression for five genes were increased, albeit at relatively lower levels, in type II cells isolated by FACS sorting from lungs of IPF patients (Table 2). One explanation for the relatively lower differences in gene expression of FACS sorted type II cells is that FACS sorting for E-cadherin selects for a more “normal” population of type II cell present in IPF lungs. Nonetheless, these results confirm that IPF type II cells are transcriptionally different than normal type II cells and that they express increased levels of select mesenchymal genes.

View this table:
Table 2.

Relative expression of select genes in IPF type II cells

Fibronectin induces expression of mesenchymal genes in human type II cells.

The gene profiling and immunohistochemical studies enable spatial colocalization of epithelial and mesenchymal proteins within IPF tissues. However, they are limited because they do not provide temporal information defining whether these cells are derived from epithelial cells via EMT. To obtain temporal information that human type II cells can undergo EMT, we studied whether, similar to mouse type II cells (15), human type II cells acquire features of mesenchymal cells when cultured on fibronectin. When cultured on Matrigel/collagen I, the type II cells isolated from normal lungs maintain their rounded epithelial shape and express pro-SPC but not mesenchymal proteins after 5 days in culture (Fig. 4). In contrast, after 5 days of culture on fibronectin, the type II cells lose their rounded epithelial shape, flatten, and spread into a shape more characteristic of mesenchymal cells (Fig. 4), and significantly increased mRNA expression of slug, vimentin, α-SMA, Col1, and N-cadherin (Fig. 5). Additionally, type II cells cultured on fibronectin lose protein expression of pro-SPC, decrease expression of E-cadherin, and increase expression of vimentin, α-SMA, and N-cadherin (Fig. 4). Similar findings were observed for type II cells isolated from fibrotic lungs (see Supplemental Fig. S2). Collectively, these data show that the culture of type II cells on fibronectin induces cell biological changes consistent with a more mesenchymal phenotype.

Fig. 4.

Fibronectin promotes epithelial-to-mesenchymal transition (EMT) of human type II cells. A: FACS purified normal type II cells cultured for 5 days on Matrigel then immunostained for pro-SPC (red) and vimentin (green). Note the cells maintain the morphology of type II cells are strongly immunoreactive for SPC and lack expression of vimentin. B and C: in contrast, when cultured on fibronectin, human type II cells develop a flattened morphology, decrease expression of SPC, and increase expression of mesenchymal cell-associated proteins vimentin (B) or N-cadherin (C). D and E: to evaluate EMT in a nonbiased way, FACS-purified normal type II cells cultured for 5 days (5 d) on Matrigel (MG) or fibronectin (FN) were lysed, and relative levels of pro-SPC, E-cadherin (E-Cad), α-SMA, or β-actin (Actin) were quantified by immunoblot. Note that type II cells lose expression of SPC and E-cadherin, and increase expression of mesenchymal cell-associated α-SMA when cultured on fibronectin compared with Matrigel. Immunohistochemistry images were obtained via a ×60 immersion objective, except the inset of B, where a ×100 objective was used. Data are from type II cells isolated from 5 individuals.

Fig. 5.

TGF-β inhibition blocks EMT of human type II cells. A: immunoblot of FACS-purified type II cells lysed after 3 days of culture on fibronectin in the absence (−) or presence (+) of the kinase inhibitor SB431542 (10 μM) confirms that SB431542 blocked phosphorylation of the TGF-β signaling molecule smad-2 and vimentin expression. B: FACS-purified normal type II cells were cultured for 3 days on fibronectin in the absence (no inhibitor) or presence of the kinase inhibitor SB431542 (10 μM) were coimmunostained for pro-SPC (red) and vimentin (green). Note that type II cells cultured in the presence of SB431542 maintain a shape consistent with an epithelial phenotype, maintain expression of SPC, and have less increase in vimentin expression. Images are representative of findings of 5 lungs. Immunohistochemistry images were obtained via a ×60 immersion objective. C: normalized mRNA levels of slug, α-SMA, type I collagen (Col1), vimentin, or N-cadherin (N-cad) in type II cells cultured 5 days on Matrigel or fibronectin in the absence or presence (MG+, FN+) of SB431542 (n = 3/group, *P < 0.02 vs. MG, **P < 0.05 vs. FN).

TGF-β mediates expression of mesenchymal genes in human type II cells.

Prior work has shown that activation of latent TGF-β via integrin αvβ6 mediates the EMT of mouse type II cells cultured on fibronectin (15, 24). To examine whether a similar mechanism could mediate EMT in human type II cells, normal human type II cells were cultured on fibronectin in the presence and absence of the TGF-β receptor kinase inhibitor (SB431542). Type II cells cultured on Matrigel had no detectable phospho-Smad2 (not shown). In contrast, type II cells cultured on fibronectin had a significant increase in phospho-Smad2, suggesting they are activated by TGF-β. Addition of SB431542 largely blocked the formation of phospho-Smad2, the morphological responses of cells to fibronectin, and expression of vimentin and maintained the expression of pro-SPC (Fig. 5). SB431542 also limited the increase of slug, vimentin, α-SMA, Col1 and N-cadherin mRNA in type II cells cultured on fibronectin (Fig. 5). These observations indicate that TGF-β receptor kinase inhibition maintained the epithelial phenotype of type II cells cultured on fibronectin. Similar findings were observed for type II cells isolated from IPF lungs (see Supplemental Fig. S2).


This work provides gene expression and immunohistochemical evidence that a subset of epithelial cells in fibrotic lungs coexpress both epithelial and mesenchymal proteins and that some of these cells can be identified in regions of matrix remodeling. In addition, data are provided showing human type II cells cultured on the provisional matrix protein fibronectin temporally acquire a morphological appearance more typical of mesenchymal cells, downregulate expression of epithelial cell-associated proteins, and increase expression of proteins associated with mesenchymal cells. Blockade of these mesenchymal changes using the tyrosine kinase inhibitor SB431542 suggests that they are mediated by TGF-β. These data provide additional evidence that human lung epithelial cells can transition into cells with a mesenchymal phenotype and that this biological phenomenon may occur in the lungs of patients with IPF.

Fibrotic lung diseases consist of a heterogeneous group of disorders, the pathogenesis of which remains undefined. One recurring theme from animal models and translational studies is that TGF-β is a driving force behind the fibrotic process (2, 7). In recent years, attention has been directed toward understanding where fibroblasts originate from in scarring lung diseases. These studies have defined three potential sources: resident lung fibroblasts, circulating fibrocytes (1, 18, 22), or epithelial cells via EMT (15, 39), all of which likely contribute to the development of lung fibrosis. Defining the unique contribution that each of these sources play in the fibrotic process could lead to new therapies selectively modulating the contribution from each cellular source.

Currently, the data supporting the hypothesis that EMT contributes to the pathogenesis of lung fibrosis is limited to rodent studies, where lineage-tracing studies showed that a fraction of mesenchymal cells are derived from epithelial cells in a models of lung fibrosis (15, 33) and that TGF-β and endothelin mediate transition of cultured primary lung epithelial cells obtained from rodents (11, 15, 39). Evidence that EMT occurs in humans is limited to immunohistochemical studies on fibrotic lungs (15, 39), a report that fibroblasts isolated from IPF lungs express increased levels of epithelial intermediate filament keratin 18 (19), and studies showing that TGF-β can increase mesenchymal gene expression in transformed lung epithelial cells lines (13). This report is important because it provides new data showing that alveolar type II cells isolated by laser capture from fibrotic human lungs have increased expression of genes encoding mesenchymal proteins and potential regulators of EMT. The most notable candidate is FoxG1, a transcription factor that purportedly plays a role in regulating TGF-β signaling (32). Whether FoxG1 plays a specific role in the development of lung fibrosis will require further investigation.

Surprisingly, IPF type II cells isolated by laser capture expressed lower levels of mRNA for vimentin and plasminogen activator inhibitor type 1 (PAI-1). The changes in vimentin levels are small and of unclear biological significance. The reduced PAI-1 levels were unexpected because PAI-1 is classically described as a TGF-β-responsive gene (25). Nevertheless, other mediators can both positively or negatively regulate its expression (25). Our data suggest that in aggregate its expression is downregulated in IPF type II cells.

A limitation of the laser capture results is the analysis used to compare type II cells isolated by different methodologies. Therefore, it is possible that measured differences may be due to the dissimilar methods of type II cell isolation. To address this limitation, differential expression of a subset of genes reported in Table 1 were confirmed in normal and IPF type II cells isolated by identical methods (Table 2). Although every effort was made to handle the normal and IPF lungs in similar manner, it is possible that unrecognized differences in processing prior to our handling the lungs could have confounded these results. Further validation of the differences in gene expression in Table 1 and identification of unrecognized differences will require future studies using more comprehensive methods to compare mRNA expression in the FACS-sorted cells.

This study is the first to show dynamic changes that primary human type II cells undergo when transitioning into cells with a more mesenchymal phenotype. When cultured on fibronectin, primary human type II cells undergo shape changes consistent with mesenchymal cells, downregulate expression of epithelial cell-associated proteins (SPC, E-cadherin), and increase expression of mesenchymal cell-associated proteins (vimentin, N-cadherin, and α-SMA). This work overcomes several limitations of prior studies reporting EMT in human type II cells. First, by studying primary cells, any influence that the transformation of type II cells could have on signaling within these cells that may predispose them to undergo EMT is removed. In addition, the unique method of FACS sorting type II cells provides a uniformly pure population of type II cells and eliminates the possibility that overgrowth of contaminating mesenchymal cells contribute to the phenotypic differences detected in fibronectin cultured type II cells. Thus primary human type II cells can be reprogrammed to undergo a mesenchymal change and fibronectin can mediate that change.

EMT associated with lung fibrosis can be promoted by a number of mediators, of which TGF-β is the prototype. In murine type II cells, engagement of the integrin αvβ6 to fibronectin triggers activation of TGF-β by type II cells (15). This integrin-activated TGF-β then reprograms the type II cells to develop a more mesenchymal phenotype. The experiments showing that human type II cultured on fibronectin develop morphological features and protein and mRNA expression characteristic of mesenchymal cells and that development of these changes can be blocked by the TGF-β inhibitor SB431542 suggest that αvβ6 activation of TGF-β also mediates mesenchymal transition of human type II cells. Although the source of TGF-β in cell culture system was not formally characterized, because the type II cells were cultured in serum-free media and express TGF-β mRNA, the alveolar type II cells are the likely source of activatable TGF-β in the cell culture. Fibronectin mediated EMT in type II cells isolated from either fibrotic or normal lungs, indicating that fibronectin-mediated TGF-β activation provides a strong signal that promotes cellular changes consistent with EMT in both normal and IPF type II cells. Finally, recent reports demonstrating increased expression of the integrin β6 in fibrotic lungs adjacent to regions of matrix remodeling indicate that αvβ6 may active TGF-β in fibrotic lungs, thereby providing a potential source to promote EMT of epithelial cells (9).

The evidence for mesenchymal gene expression by IPF type II cells was heterogeneous, both in the immunohistochemical and gene profiling studies using en bloc tissues and in the studies using isolated cells. For example, in fibrotic lungs, not all epithelial cells coexpressed epithelial and mesenchymal proteins. During the gene profiling studies, we noted that although the pattern of difference in gene expression was remarkably consistent (e.g., Col1 was always elevated in IPF epithelial cells), the magnitude of difference varied (7- to 50-fold) between experiments. When purified type II cells were cultured on fibronectin, the increase in mesenchymal protein expression was not uniform among the cultured cells. These observations do not diminish the probability that human type II cells undergo EMT in fibrotic lungs, but rather prompt the question of why heterogeneity exists.

One explanation for why only a subpopulation of type II cells may undergo EMT in fibrotic lungs is that the majority of epithelial cells are maintained as epithelial cells to maintain the “normal” integrity of the epithelium. If all epithelial cells transitioned to mesenchymal cells, there would be no epithelial cells remaining that could perform necessary duties of the lung epithelium. A second explanation is that the biological signals that reprogram epithelial cells such that they transition to a mesenchymal cell occur only in discrete microenvironments of the lung. This concept is consistent with the temporally heterogeneous pattern of fibrosis seen within IPF lungs. Finally, the observation that not all type II cells undergo EMT when cultured on fibronectin suggests that there may be select subpopulations of type II cells within the lung that either are predisposed to undergo EMT or may be at a less differentiated stage of development that is more biased to undergo EMT. Whether these possibilities are true will require further investigation.

In summary, these results stress the importance of alveolar epithelial cells in the pathogenesis of IPF and suggest they directly contribute to the fibrotic process by acquiring features of mesenchymal cells. Whether these changes are the cause or consequence of fibrotic lung disease will require further study. Nevertheless, the data strengthen the evidence that EMT is a feature of pathological lung fibrosis and suggest that therapies directed toward inhibiting pathways that promote EMT should be considered for development as targets for the treatment of IPF.


This work was supported in part by grants from the Resource Allocation Program at University of California, San Francisco (to P. J. Wolters); the Parker B. Francis Foundation (to K. K. Kim); and National Heart, Lung, and Blood Institute Grants HL104422 (to P. J. Wolters), HL085290 (to K. K. Kim), and HL44712 (to H. A. Chapman).


No conflicts of interest, financial or otherwise are declared by the author(s).


We thank the patients who generously donated lung tissue for this study, Kirk Jones for helpful discussions, and Alexis Brumwell for technical support.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
View Abstract