INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

THE CELLULAR AND MOLECULAR EVENTS leading to fibrogenesis in the lung are believed to be intimately related to alveolitis and alveolar epithelial repair processes (11). Investigations of the damage caused by hyperoxic gas on murine whole lung tissue (2) or explants derived from such tissue (3) were the first to suggest that repair of alveolar epithelial damage was critical for healing without fibrosis. Subsequent studies (17, 18) of the ability of oxygen to potentiate lung damage induced by butylated hydroxytoluene supported the hypothesis that the severity of fibrogenesis was inversely related to the efficiency of alveolar epithelial repair. Consistent with that theory is the observation that foci of accumulating fibroblasts in human lung tissue are found in close proximity to unrepaired or abnormal alveolar epithelium (22).

In the fibrotic human lung and in animal models of pulmonary fibrosis, the alveolar epithelium can be absent or metaplastic and often is primarily cuboidal rather than the normal mixture of squamous type I cells interspersed with cuboidal type II cells (11, 33). Electron microscopy of human tissue suggests that the cuboidal cells are of both alveolar and bronchiolar origin (21), but the relative importance of alveolar versus bronchial cells in the genesis of new alveolar epithelium has not been explored rigorously. For many years, it has been believed that the cuboidal epithelium is composed of rapidly proliferating cells that are attempting to repair the denuded epithelium (11, 21, 26). It has been suggested that the high ratio of cuboidal to squamous cells in the abnormal epithelium might be due to an inability of dividing type II cells to differentiate into type I cells, possibly as a result of altered composition or integrity of the underlying extracellular matrix (26).

Earlier work from this laboratory (38) demonstrated that fibroblasts isolated from fibrotic human or rat lung produce factor(s) capable of inducing apoptosis and necrosis of alveolar epithelial cells in vitro. These observations suggested that apoptosis might be an important mechanism of epithelial cell death in the fibrotic lung. Consistent with that theory, apoptotic alveolar epithelial cells have been recently observed in sections of fibrotic lung of both human and mouse origin (15, 24). We hypothesized that the abnormal fibroblast phenotypes that emerge in fibroblastic foci might constitute an important source of factor(s) capable of inducing epithelial cell apoptosis. In the lung, these phenotypes are known to include the myofibroblast subtype termed "VA" that expresses the intermediate filaments vimentin and -smooth muscle actin (-SMA) (23). Accordingly, we further theorized that the cuboidal epithelium might contain apoptotic and/or necrotic cells in addition to proliferating cells, particularly in proximity to the VA fibroblast subtype. To test this hypothesis, we examined the distribution of dying cells within the cuboidal epithelium relative to underlying fibrotic foci in the same biopsy samples that we used earlier for fibroblast isolation (38). We report that the cuboidal epithelium of fibrotic human lung contains apoptotic and necrotic cells in close proximity to foci of -actin-positive fibroblasts and collagen accumulation.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Materials. Monoclonal antibodies to -SMA (clone 1A4) and biotinylated dUTP were obtained from Sigma Immunochemicals (St. Louis, MO). DNA polymerase I was obtained from Promega (Madison, WI). Solutions A and B of the Vectastain ABC Elite formulation were purchased from Vector Laboratories (Burlingame, CA). All other materials were of reagent grade.

Tissue samples and handling. Human lung tissue was obtained by open lung biopsies performed at the National Institute of Respiratory Diseases (Tlalpan, Mexico). Lung tissue was obtained from six patients with idiopathic pulmonary fibrosis (IPF). All patients had clinical, functional, and radiological features that fulfill the diagnostic criteria for an interstitial lung disease (29, 32). Briefly, all had progressive dyspnea, bilateral reticulonodular images on chest roentgenogram, restrictive lung functional impairment with decreased lung volumes and compliance, and hypoxemia at rest that worsened with exercise. Patients with IPF had neither antecedents of any occupational or environmental exposure nor other known cause of interstitial lung disease; samples from these patients were the same as those designated human IPF (HIPF) in the earlier study by Uhal et al. (38) of primary fibroblasts isolated from fibrotic human and rat lung tissue. Normal lung tissue was obtained from three individuals having a lobectomy for removal of a primary lung tumor. No histopathological evidence of disease was found in those tissue samples. All tissue was fixed in 10% neutral-buffered Formalin for 16 h and embedded in paraffin. Sections were cut at 6.0-µm thickness and mounted on glass coverslips.

In situ end labeling and immunohistochemistry. Tissue sections were deparaffinized by being passed through xylene, 1:1 xylene-alcohol, 100% alcohol, and 70% alcohol for 10 min each. In situ end labeling (ISEL) of fragmented DNA was conducted by a modification of the method of Mundle et al. (27). Briefly, ethanol was removed by rinsing in distilled water for at least 10 min. The coverslips were then placed in 0.23% periodic acid (Sigma) for 30 min at 20°C. Samples were rinsed once in water and three times in 0.15 M phosphate-buffered saline (PBS) for 4 min each and were then incubated in saline-sodium citrate solution (0.3 M NaCl and 30 mM sodium citrate in water, pH 7.0) at 80°C for 20 min. After four rinses in PBS and four rinses in buffer A (50 mM Tris · HCl, 5 mM MgCl₂, 10 mM -mercaptoethanol, and 0.005% BSA in water, pH 7.5), the coverslips were incubated at 18°C for 2 h with ISEL solution (0.001 mM biotinylated dUTP, 20 U/ml of DNA polymerase I, and 0.01 mM each dATP, dCTP, and dGTP in buffer A). Afterward, the sections were rinsed thoroughly five times with buffer A and three additional times in 0.5 M PBS. For detections based on diaminobenzidine, the tissue was then incubated at 20°C with a solution consisting of 80 µl each of solutions A (avidin solution) and B (biotin-peroxidase solution) of the Vectastain Elite Kit in 3.84 ml of buffer B (1% BSA and 0.5% Tween 20 in 0.5 M PBS). After 30 min, the sections were washed four times in PBS and were then immersed for 10 min in a solution of 0.25 mg/ml of diaminobenzidine in 0.05 M Tris · HCl, pH 7.5, containing 0.01% hydrogen peroxide. Alternatively, detection of incorporated dUTP was achieved with a fast blue chromogen system. The tissues were rinsed in distilled water three times and mounted under Fluoromount solution (Southern Biotechnology Associates, Birmingham, AL).

For double labeling with ISEL and antibodies to -SMA, fluorescein (FITC)-conjugated monoclonal antibodies to -actin were diluted 1:100 in ISEL solution and incubated with the tissue as described above for single ISEL. Control samples, treated as described above but without DNA polymerase I, gave no positive signal (data not shown).

Identification of collagen. Collagens were localized through staining of lung sections by the picrosirius red technique (30). Briefly, sections deparaffinized with xylene were rehydrated through a descending series of ethanols to distilled water. The hydrated sections were immersed in 0.2% aqueous phosphomolybdic acid for 2 min, rinsed in distilled water, and then stained for 110 min with 0.1% sirius red F3BA (Pfaltz and Bauer, Stamford, CT) in saturated aqueous picric acid, pH 2.0. The sections were washed for 2 min in 0.01 N HCl, rinsed for 45 s in 70% ethanol, dehydrated in three changes of absolute alcohol, cleared in xylene, and mounted in Permount. Fibrotic foci were detected by polarization microscopy, under which collagen fibers appear yellow-red (30) or white in gray-scale images.

Microscopy and image analysis. The prepared sections were photographed under transmitted, epifluorescent, or polarized light on an Olympus BH2 epifluorescence microscope fitted with automatic photographic equipment. For electron microscopy, portions of some biopsies were fixed in 1% gluteraldehyde in cacodylate buffer. The tissues were osmicated and stained with uranyl acetate, dehydrated with a graded series of alcohol solutions, and embedded in Epon/Araldite. Thin sections were examined and photographed on a Philips EM 300 instrument in the Electron Microscopy Facility, Rush Medical College (Chicago, IL).

For image analysis of ISEL and -SMA labeling, transmitted light and fluorescence digital images were obtained by scanning color photographic slides made with tungsten film. Scanned images were first converted to 256-level gray scale, then to pseudocolor on the basis of intensity thresholds derived by dividing each 256-level gray scale into three equal levels of intensity (low, moderate, and high). Once established, the same intensity thresholds were applied to all images processed for colocalization studies (see Fig. 5).

For analyses of the relationship between epithelial cell death and immunoreactivity to -SMA (Table 1), ISEL or -SMA intensity thresholds were defined as described above. The property of epithelial integrity was defined as the presence of morphologically intact epithelium within 50 µm of any border of the -SMA intensity threshold under evaluation. In the compilation of data from a series of lung sections (Table 1), the scale for scoring epithelial integrity was from zero to five marks (from least to highest, respectively). A score of zero marks indicates complete absence of epithelial cells overlying the stroma within 50 µm of the indicated -SMA threshold. A score of five marks reflects morphologically intact epithelium within the entire scoring region.

View this table: [in this window] [in a new window]   Table 1.   Relationship of epithelial integrity and ISEL to -actin-positive interstitium

The scale for scoring ISEL intensity also was from zero to five marks (from least to highest, respectively). In the compiled intensity data, a score of zero marks indicates that all epithelial cells within the 50-µm limit defined above display no ISEL. A score of five marks indicates that all epithelial cells within the 50-µm limit display the highest ISEL intensity. Regions in which both ISEL and -SMA labels exactly colocalized were scored at the highest ISEL intensity. Processing of scanned images was performed with MOCHA image-analysis software (Jandel Scientific, San Raphael, CA). The scoring of samples from IPF patients was performed in a blinded fashion.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

The histology of normal and fibrotic lung specimens is shown in Fig. 1, A-C. Tissue from patients with IPF showed patchy alveolar septal fibrosis and interstitial inflammation consisting mostly of mononuclear cells but also of neutrophils and eosinophils. A variable macrophage accumulation was observed in the air spaces as was cuboidalization of the alveolar epithelium. Biopsies lacked granulomas, vasculitis, microorganisms, and inorganic material by polarized light microscopy (4).

View larger version (175K): [in this window] [in a new window]   Fig. 1.   Histology and in situ end labeling (ISEL) of biopsies of normal and fibrotic lung tissue. A: hematoxylin and eosin staining of normal lung tissue. B and C: hematoxylin and eosin preparations of biopsies from 2 patients with idiopathic pulmonary fibrosis (IPF). D: ISEL of fragmented DNA in normal lung tissue. E: ISEL of biopsy from same IPF patient as in B; positive reaction is intense black (actually blue generated by fast blue detection system). Bars, 50 µm; B and D are the same as A.

ISEL of fragmented DNA in sections of normal human lung tissue revealed very few labeled cells (Fig. 1D). In contrast, sections prepared from lung tissue of patients with IPF (Fig. 1E) contained many labeled cells within the interstitium, air spaces, and epithelium. Electron microscopy of the alveolar epithelium revealed cells with morphological characteristics of both apoptosis and necrosis (Fig. 2, B and C); dying epithelial cells were often observed in locations adjacent to fibrotic foci containing numerous collagen bundles (Fig. 2B, CB) interspersed with fibroblast-like interstitial cells (Fig. 2C, F).

View larger version (142K): [in this window] [in a new window]   Fig. 2.   Electron microscopy of alveolar epithelial cells in fibrotic human lung. A: alveolar epithelial cells (AECs) distal to fibrotic foci have uniform cytoplasmic density and abundant microvillae and mitochondria, which appear normal. B: significant number of AECs near fibrotic foci contain swollen cytosol characteristic of necrosis. Note collapsed microvillae and collagen bundles (CB) immediately underlying epithelial cell. C: many AECs adjacent to fibrotic foci display characteristics of both necrosis and apoptosis, such as basolateral swelling together with condensed and marginated chromatin (arrow) and apical shrinkage (9, 36). Abundant collagen bundles underlying these cells were interspersed with fibroblast-like interstitial cells (F). All samples were from a patient with IPF. Magnification, ×5,000.

Staining of lung sections by the picrosirius red technique (Fig. 3) confirmed the presence of collagen in these foci and suggested a relationship between the locations of fibrotic foci and epithelial cell death. The condensed nuclei of apoptotic cells were found within the epithelium near regions of moderate accumulation of collagen (Fig. 3C); the epithelium adjacent to regions of very dense collagen accumulation was clearly injured and often absent (Fig. 3E).

View larger version (164K): [in this window] [in a new window]   Fig. 3.   Identification of collagen in normal and fibrotic human lung. Deparaffinized biopsy sections were stained with picrosirius red technique (30). B, D, and F: polarization microscopy of the same fields in A, C, and E, respectively, under transmitted light. Under polarized light, collagen appears white. A and B: normal human lung. Small amounts of collagen are seen only in thickest parts of interstitium but not in alveolar septa. Bar, 50 µm. C and D: area of moderate accumulation of collagen in thickened septum in biopsy from a patient with IPF. Note condensed nucleus surrounded by halo in adjacent epithelium (C, arrowhead). Bar, 20 µm. E and F: area of severe accumulation of collagen within septum of a separate IPF patient. Note damaged or absent epithelium adjacent to most abundant collagen (E, arrows). Bar, 20 µm.

To determine whether alveolar epithelial cell death was related to the proximity of the affected epithelium to abnormal interstitial cells within the fibrotic foci, double labeling of fibrotic lung specimens was performed with ISEL and simultaneous immunolabeling of -SMA. The specificity of the antibodies to -SMA was demonstrated first with sections of normal human lung, in which immunoreactivity was found only within the smooth muscle underlying vessels or bronchi (Fig. 4). In contrast, sections of fibrotic human lung were immunoreactive to anti--SMA in numerous fibroblastic foci scattered throughout the parenchyma; one of these foci is displayed in Fig. 5. The -SMA labeling (Fig. 5B) was extensive in the interstitium immediately underlying "sheets" of in situ end-labeled cuboidal epithelium (Fig. 5A, arrowheads). In isolated areas corresponding to the most intense -SMA immunoreactivity (Fig. 5C, green areas), the adjacent alveolar epithelium was absent or partially detached from the underlying stroma (arrows).

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Specificity of antibodies to -smooth muscle actin (-SMA). A: phase-contrast micrograph of deparaffinized section of normal human lung. Note large airway at left. B: fluorescence micrograph of same field shown in A. Section was immunolabeled with FITC-conjugated monoclonal antibodies to -SMA. Note immunoreactivity of smooth muscle underlying airway wall. Magnification, ×100.

View larger version (77K): [in this window] [in a new window]   Fig. 5.   Double labeling of fibrotic human lung with ISEL and antibodies to -SMA. A: in situ end-labeled lung section from a patient with IPF. Positive labeling is brown color of diaminobenzidine detection system. Note "sheets" of ISEL-positive cuboidal epithelium (arrowheads). B: fluorescence image of the same field shown in A. Section also was immunostained with FITC-conjugated monoclonal antibodies to -SMA during ISEL procedure as described in METHODS. C: composite pseudocolor image of ISEL and -SMA labeling superimposed. Inset: pseudocolor thresholds denote light, moderate, and heavy intensity of each label, increasing from left to right. Pseudocolor black (far right) denotes regions in which both labels colocalized at highest intensity. Note ISEL of moderate intensity (red) or, less frequently, of high intensity (blue) adjacent to interstitial cells reacting moderately with -SMA (yellow). More often, epithelium is absent (arrows) in areas adjacent to most intense -SMA labeling (green). Bar, 50 µm. See Table 1 for evaluation of relationship between ISEL and -SMA.

The pseudocolor image (Fig. 5C) displays three intensity threshold levels for both ISEL (white, red, and blue) and -SMA labeling (white, yellow, and green), constructed from images of each label analyzed individually. These thresholds were used to obtain a semiquantitative evaluation of the relationship between the two labels in a series of fibrotic lung samples. Table 1 displays the results of this analysis. Scoring of epithelial integrity, defined as the presence of intact epithelium within 50 µm of -actin immunoreactivity, revealed that the most damaged or absent epithelia were those in close proximity to interstitia that reacted most intensely with -SMA antibodies (-SMA intensity level 3, green). The remaining epithelial cells were then scored on the basis of ISEL intensity as defined in the pseudocolor image; the most intense ISEL of epithelia was found adjacent to foci of the most intense -SMA labeling. The epithelial layer adjacent to interstitia of light or absent -SMA immunoreactivity (intensity level 1, white) was of higher integrity and relatively free of ISEL.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

It is generally accepted that alveolitis and damage to the alveolar epithelium are important antecedents of fibrogenesis in the lung. As discussed by Crystal et al. (11), data supporting this concept come from both animal models and studies of familial cases of IPF in which the children of afflicted parents exhibit alveolitis well before clinical evidence of the disease (7). In studies of intact animal models, the experimental inhibition of alveolar repair in mice previously injured with butylated hydroxytoluene led to an acceleration of fibrogenesis (17, 18). Similarly, mouse lung explants with severe epithelial damage induced by prior hyperoxic lung injury exhibited marked interstitial cell proliferation and collagen deposition in culture, but explants of less severely injured tissue did not (3). As detailed by Witschi (39), these observations suggested that the critical determinant in the pathway to fibrogenesis was not the infiltrating inflammatory cells but rather the epithelial damage and its repair. Alveolar epithelial repair normally occurs through the proliferation and differentiation of type II alveolar epithelial cells, a process that is believed to be regulated by factors produced by lung fibroblasts (25, 28). The many important "antifibrotic" functions of the intact alveolar epithelium have been discussed in detail by Simon (35).

Cuboidalization of the alveolar epithelium is a common feature of the fibrotic lung regardless of the animal species or the initiating stimuli (11, 33, 35). Numerous investigations (1, 13, 19, 20) of animal models of lung injury have documented a transient cuboidalization as type II pneumocytes proliferate to replace damaged or lost type I cells. Largely on the basis of these observations, it has been presumed that epithelial cuboidalization in the fibrotic human lung is indicative of sustained epithelial proliferation in the absence of appropriate factors to signal the cessation of cell division and the initiation of subsequent differentiation (26, 35). This interpretation, however, fails to consider other possible mechanisms underlying cuboidalization, such as reversion of type I cells into type II cells by transdifferentiation (12, 34) or rapid loss of type I cells by apoptosis (37, 38).

Apoptotic cells were observed in the bronchial and alveolar epithelia of mice made fibrotic by intratracheal administration of bleomycin (15); the apoptotic cells appeared 1 day after bleomycin administration and reappeared during the progression of fibrosis at days 7-14 after administration. Although bleomycin is known to induce apoptosis in isolated alveolar macrophages (16), it is not clear whether the agent induces apoptosis directly on alveolar epithelial cells, and thus it is unknown whether epithelial cell apoptosis is a common mechanism underlying fibrogenesis in the mouse lung or an unrelated consequence of direct damage by bleomycin. On the other hand, ligation of Fas antigen with antibodies to Fas induced epithelial apoptosis and pulmonary fibrosis in mice, suggesting involvement of apoptosis in the pathogenesis of pulmonary fibrosis (14).

Our work suggests that the cuboidal epithelium of fibrotic human lung contains dying as well as proliferating cells, at least during the late stage of disease progression at which the tissue biopsies were obtained. Although the study of samples from advanced fibrotic lung may not yield information relevant to early events in the pathogenesis of fibrosis, we contend that the contribution of epithelial cell death to failure of epithelial repair likely deepens and extends the committment to fibrogenesis. As discussed by Simon (35), intact alveolar epithelium produces prostaglandins of the E series, known inhibitors of fibroblast proliferation (6, 10), and protects underlying interstitial cells from exposure to growth factors released by intra-alveolar macrophages and other inflammatory cells. For these reasons, failure of the epithelium to repair may be equally detrimental in advanced fibrosis as well as in the genesis of nascent fibrotic foci. In this study, the finding of apoptotic and necrotic epithelial cells overlying fibroblastic foci in the particular specimens examined is consistent with the earlier study by Uhal et al. (38) of factors produced by fibroblasts isolated from the same biopsy samples.

Our data also confirm the observations of Kuwano et al. (24) that DNA fragmentation and apoptosis in the alveolar epithelium are prominent features of fibrotic human lung tissue with no known prior exposure to bleomycin or other potential xenobiotic inducers of apoptosis. Apoptosis of mesenchymal cells is believed to constitute an important and necessary component of the resolution of granulation tissue in the lung (31); intuitively, apoptotic cells would be expected in a grossly hypercellular tissue that is attempting to restore normal architecture. The notion that excess alveolar epithelial cells are also eliminated by programmed cell death is supported by the recent study of Bardales et al. (5), who found that apoptotic alveolar epithelial cells were relatively abundant in the lungs of patients with resolving acute lung injury but not in tissue samples taken during the proliferative phase of chronic interstitial pneumonia. Those findings are in contrast to the data of Hagimoto and colleagues (14, 15), which suggest a role for apoptosis in the genesis of fibrosis as well as in its resolution.

Abnormal epithelial-mesenchymal interactions in lung injury were first speculated to be important on the basis that unusual cell contacts between fibroblasts and alveolar epithelial cells were observed in butylated hydroxytoluene-injured mouse lung and in lung tissue from patients with IPF (8). In the human lung, the predominant fibroblast subtype that emerges in fibroblastic foci has been designated as the subtype VA, which expresses the intermediate filaments vimentin and -SMA and has been termed the myofibroblast (23). This subtype may be the same as that identified by immunoreactivity to anti-pC1, which recognizes the COOH-terminal polypeptide of human procollagen type I; pC1-positive cells were found adjacent to regions of severe epithelial shedding or cuboidalization in lung tissue from patients with IPF (22). In light of the earlier study (38) of fibroblasts derived from the same specimens studied here, our findings of apoptotic epithelial cells in proximity to -SMA-positive interstitial cells in situ suggests a role for the VA fibroblast subtype in the induction of epithelial cell death in vivo. Characterization of the primary fibroblast strains and the proteins they secrete is currently underway.

In summary, electron microscopy and ISEL of fragmented DNA revealed that the cuboidal epithelium of fibrotic human lung is composed of both proliferating and dying cells. ISEL and simultaneous immunolabeling for -SMA revealed that apoptotic and necrotic epithelial cells were observed primarily in proximity to underlying foci of -actin-positive interstitial cells. These results are consistent with a role for the myofibroblast in the production of inducers of epithelial cell apoptosis and necrosis. We speculate that this abnormal epithelial-mesenchymal interaction contributes to the pathogenesis and/or exacerbation of fibrotic lung disease by preventing normal epithelial repair and thereby deepening the commitment to fibrogenesis.