INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Idiopathic pulmonary fibrosis (IPF) is a progressive interstitial lung disease of unknown etiology and is associated with significant morbidity and mortality despite aggressive therapies (1, 2). The clinical features of IPF are quite variable (3), and pathological review of lung specimens from patients with IPF shows a variety of histologic patterns. Katzenstein and coworkers (4, 5) reviewed the clinical relevance of the pathological classification of idiopathic interstitial pneumonias. Four histologically distinct forms of idiopathic interstitial pneumonia, usual interstitial pneumonia (UIP), nonspecific interstitial pneumonia (NSIP), desquamative interstitial pneumonia (DIP)/respiratory bronchiolitis-associated interstitial lung disease (RB-ILD) (6), and acute interstitial pneumonia (AIP) (7), comprise the morphological spectrum that has been traditionally included under the designation of IPF (5). UIP exhibits a specific histological pattern of fibrosing interstitial pneumonia seen in the majority of patients with IPF. The other patterns of idiopathic interstitial pneumonia have been delineated as potential mimics of IPF with UIP histology, and identified as unique clinicopathological subgroups with distinct natural histories. Pulmonary structures are extensively remodeled in UIP, whereas severe architectural remodeling is not present in NSIP and BOOP, which responds to corticosteroid therapy. For NSIP and BOOP, the prognoses are fairly good (4, 5, 8), and the histological changes are reversible and curable (8).

The processes observed in IPF are associated with the production, deposition, and proteolysis of extracellular matrix (ECM), which may lead to irreversible pulmonary structural remodeling or to appropriate repair without fibrosis. Recently, matrix metalloproteinases (MMPs) and the specific tissue inhibitors of metalloproteinases (TIMPs) have been shown to participate in the parenchymal destruction and repair processes resulting in ECM remodeling (9). Basement membrane is a specialized form of ECM, and a complex of type IV collagen occupies a substantial portion of the basement membrane structure in the lung, which is of critical importance in maintaining the structural integrity of the alveolar wall. In the pathogenesis of IPF, disruption of the epithelial basement membrane is known to be associated with fibroblast migration into the alveolar spaces and is considered a key event resulting in intraluminal fibrosis (14). MMPs are a family of ECM-degrading, zinc-dependent enzymes comprising at least 18 members with different, albeit overlapping, substrate specificities. Among these MMPs, two collagenolytic MMPs have been reported to possess substrate specificity to type IV collagen and can degrade basement membrane structures via collagenolytic actions. The two are MMP-2 (72- kD type IV collagenase, also named gelatinase A), preferentially secreted by fibroblasts and epithelial cells, and MMP-9 (92-kD type IV collagenase, also named gelatinase B), preferentially expressed by inflammatory cells including macrophages (17). Thus, we focused on the roles of MMP-2 and -9 in the pathogenesis of IPF.

In this study, we hypothesized that the pathophysiologies of pulmonary parenchymal remodeling differ among UIP, NSIP, and BOOP and that this may influence the distinct natural histories of these diseases and their therapeutic outcomes. We, therefore, assessed the expressions of MMP-2 and -9 and TIMP-2 in bronchoalveolar lavage fluid (BALF) and lung tissues from patients with UIP, NSIP, and BOOP. We demonstrate the expression patterns of MMPs and TIMPs and their relation to the remodeling processes and clinical courses in UIP, NSIP, and BOOP.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects and Tissue Collection

The study population consisted of 37 patients with idiopathic interstitial pneumonia, 26 patients with UIP (IPF-UIP), and 11 patients with NSIP. Six patients with BOOP as a disease control and 10 healthy nonsmokers (having never smoked) and 10 healthy current smokers (NC-NS and NC-S, respectively) as normal control subjects were included in this study. Diagnosis was established in each of the patients on the basis of standard clinical criteria and of histopathological study by video-assisted thoracoscopy-guided lung biopsy or open lung biopsy (1, 4, 5, 8). UIP patients comprised 19 men and 7 women (16 smokers and 10 nonsmokers): mean age 62.2 yr, range 30 to 78 yr. NSIP patients comprised 2 men and 9 women (2 smokers and 9 nonsmokers): mean 60.8 yr, range 45 to 70 yr. BOOP patients comprised 3 men and 3 women (2 smokers and 4 nonsmokers): mean age 64.8 yr, range 40 to 73 yr. Normal control lung tissues were obtained from 7 individuals at lobectomy for removal of primary lung cancer. The IPF-UIP patients were divided into two groups, a rapidly progressive-IPF-UIP (R) group comprised of patients who died of the disease within 3 yr after initial diagnosis and a slowly progressive IPF-UIP (S) group. No significant difference was found in the duration from onset of symptoms to initial diagnosis between the IPF-UIP (R) and IPF-UIP (S) groups (12.9 ± 4.9, 14.5 ± 5.1 mo, respectively). None of the patients had received steroids or other immunosuppressive drugs at the time of this study. All of the subjects gave appropriate informed consent, and the study design was approved by our institutional ethics committee.

Bronchoalveolar Lavage

Bronchoalveolar lavage (BAL) was performed in all study subjects under local anesthesia with 2% lidocaine. A fiberoptic bronchoscope was gently wedged into the segmental bronchus of the middle lobe of the right lung. Sterile 0.9% saline totaling 150 ml was instilled in aliquots and recovered by gentle hand suction. Mucus was removed from the fluid by filtration with two sheets of gauze. Lavage fluid was centrifuged at 400 × g for 10 min at 4° C to separate cells from cell-free fluid. The cell-free lavage fluid was stored at 80° C until further analysis. Subsequent measurements of gelatinolytic activity evaluated by gelatin zymography demonstrated no decrease in concentration. Repeat freezing and thawing was avoided. The number of cells in the BALF was then determined using a hemocytometer. Cell differentials were determined by Giemsa staining for 500 cells prepared by cytocentrifugation (Cytospin-2; Shandon Instruments, Sewickey, PA). Cell numbers and differentials are expressed as mean ± SEM.

Gelatin Zymography

To detect the gelatinolytic activity in BALF, the samples were evaluated by gelatin zymography. Unconcentrated BALF samples (20 µl) were treated under nonreducing conditions and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) in 10% acrylamide gels containing 0.8 mg/ml gelatin (gelatin monomer; Serva Feinbiochemica GmbH & Co., Heidelberg, Germany) at a constant current of 25 mA at 0° C. After electrophoresis, the gels were washed three times with 40 mM Tris-HCl (pH 7.6), 10 mM CaCl₂, 2 µM ZnCl₂, and 0.1% Brij 35 containing 3% Triton X-100 for 15 min to remove all trace of SDS and then incubated for 24 h at 37° C in the same buffer without Triton X-100. Following incubation, the gels were fixed in a solution of 10% acetic acid and 50% methanol and stained with Coomassie brilliant blue R-250 for 60 min. Gelatin digestion was identified as a clear lytic zone against a blue background. The gels were scanned and converted to digitalized images by scanner (GT 6500 ART2; EPSON Co., Tokyo, Japan), and the images were inverted by a Macintosh G3 computer system (Apple Computer Inc., Cupertino, CA). The gelatinase activity was expressed as the density of the inverted-lytic zone measured by NIH image v.1.61. Each pixel of inverted-lytic band was taken as 1 unit of MMP activity.

In the present study, gelatin zymography was used for quantitative measurement of proMMP-2 and -9 activities. To confirm the ability of our gelatin zymographic analysis to estimate MMP-2 and -9 activity quantitatively, the following study was performed. Serial concentrations (4 to 4,000 ng/ml) of purified human proMMP-9 (isolated from human neutrophils according to our previously described method [18]) and proMMP-2 (Biogenesis Inc., Kingston, NH) were subjected to gelatin zymographic analysis described above. As shown in Figure 1, good correlation was observed between the activity of proMMP-2 and -9 measured by the above mentioned method and the concentrations of purified human proMMP-2 and -9 loaded. Furthermore, two concentrations of purified proMMP-9 and proMMP-2 (133 and 400 ng/ml) were used simultaneously as positive controls for each gelatin zymographic analysis of BALF samples to standardize the measurement. The specificity of the gelatinolysis produced by proMMP-2 and -9 was confirmed by disappearance of the lytic band by pretreatment of samples with o-phenanthroline or with immunoprecipitation with a specific antibody for human MMP-2 or -9 (Fuji Chemical Industries Ltd., Takaoka, Japan) (19). Based on these experiments, we determined that the amount of proMMP-2 and -9 in BALF samples could be quantitatively and specifically assessed by our analysis.

View larger version (45K): [in this window] [in a new window]   Figure 1.   Quantitative estimation of gelatinolytic activities of proMMP-2 and -9 by gelatin zymography. (A) SDS-PAGE analysis of purified proMMP-2 and -9 used as controls in gelatin zymography. Purified proMMP-9 from human neutrophils shows a homogeneous band with a molecular size of 92 kD under the reducing condition by dithiothreitol (DTT). Under the nonreducing condition, purified proMMP-9 from human neutrophils showed three forms with molecular weights of 92, 134, and 220 kD. These correspond to proMMP-9, proMMP-9-lipocalin complex, and proMMP-9 dimer, respectively. Purified proMMP-2 showed a major 72-kD band and a minor 66-kD band, the latter band corresponding to the autodigestive active form of MMP-2. (B) Quantitative measurement of proMMP-2 and -9 activity by gelatin zymographic analysis. Serial concentrations (4 to 4,000 ng/ml) of purified human proMMP-9 and MMP-2 were subjected to gelatin zymographic analysis. Good correlation was observed between the activity of proMMP-2 and -9 measured by the above mentioned method and the concentrations of purified human proMMP-2 and -9 loaded.

Histological Evaluation

Immediately after biopsied tissues were obtained, paraffin and frozen sections were prepared. For paraffin sections, tissues were fixed in 10% neutral buffered formalin (Nacalai Tesque, Kyoto, Japan) for 72 h, dehydrated through a graded series of ethanol and xylene, and embedded in paraffin. Three-micron-thick paraffin sections were cut, deparaffinized, washed with phosphate-buffered saline (PBS), and subjected to immunostaining (20). For frozen sections, tissues were snap-frozen in liquid nitrogen, cut by cryostat into 6-µm-thick serial sections, air dried, and fixed in acetone for 10 min at 4° C. The sections were rehydrated in PBS.

Primary antibodies used for immunohistochemical studies included a mouse anti-human MMP-2 monoclonal antibody (mAb) (clone 42-5D11), a mouse anti-human MMP-9 mAb (clone 56-2A4), and a mouse anti-human TIMP-2 mAb (clone 67-4H11) (all from Fuji Chemical Industries, Ltd., Takaoka, Japan). The mAbs against the MMPs and TIMP-2 were well characterized and previously used for immunolabeling. After inhibition of endogenous peroxidase activity by treatment with 0.3% hydrogen peroxide in methanol, the cryostat sections were stained with the mAbs by an indirect immunoperoxidase method using peroxidase-labeled anti-mouse immunoglobulin [F(ab')₂] (Amersham, Amersham, UK) diluted 1:100. Peroxidase activity was visualized using 3,3'-diaminobenzidine (DAB; Dojin, Kumamoto, Japan) as the substrate. After immunolabeling, sections were counterstained with hematoxylin. For control, sections were incubated with nonimmunized mouse immunoglobulin (IgG) (DAKO, Glostrup, Denmark) instead of specific mAbs, and processed according to the same procedure. The distribution and intensity of the immunohistochemical positive reactions were evaluated by three independent observers (K.I., Y.G., T.O.) without knowledge of the kinds of tissues and antibodies used.

Statistical Analysis

Results are reported as mean ± SEM. The nonparametric Mann- Whitney test was used to detect differences between patients and control groups. Correlations between the BAL analysis data and MMPs activities were assessed using the nonparametric Spearman correlation test. To determine which constituents of BAL cells may participate in the production of MMPs in BALF, we performed a stepwise multivariate regression analysis. Differences were considered significant at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

General Characterization of BALF

General characteristics of all patient groups are summarized in Table 1, and comparisons of BALF components are summarized in Table 2. All patient groups and the NC-S group showed a significantly higher total cell count than did the NC-NS group (p < 0.01). The mean percentage of lymphocytes was significantly high in NSIP and BOOP samples, and the mean percentage of alveolar macrophages was low in IPF-UIP, NSIP, and BOOP samples in comparison to percentages in normal control (NC-NS and NC-S) (p < 0.01) samples. The mean percentage of neutrophils was significantly higher in all patient groups than in the NC-NS group (p < 0.01); it was highest in the IPF-UIP (R) group. When the IPF-UIP (S) and the IPF-UIP (R) data were compared, the mean percentage of neutrophils was significantly higher in the IPF-UIP (R) group (p < 0.005).

Detection of Gelatinolytic Activity in BALF

Gelatinolytic bands of approximately 134, 92, 85, 72, and 66 kD were detected in BALF. These bands correspond to the proMMP-9-lipocalin complex, proMMP-9, active form of MMP-9, proMMP-2, and active form of MMP-2, respectively. In IPF-UIP (R), intense lytic bands corresponding to pro MMP-9, coexpressed bands of the proMMP-9-lipocalin complex, active form of MMP-9, and proMMP-2 were detected. In IPF-UIP (S), lytic bands corresponding to proMMP-9 and proMMP-2 were barely detected. In this inactive group, no active forms of MMP-9 or MMP-2 were demonstrated. NSIP and BOOP cases showed a relatively uniform pattern of gelatinolytic activity in which proMMP-2 activity was predominant in comparison with proMMP-9 activity. In BALF from NC-NS and NC-S subjects, no gelatinolytic bands were apparent. Representative results of gelatin zymography are demonstrated in Figure 2.

View larger version (80K): [in this window] [in a new window]   Figure 2.   Gelatin zymograms of bronchoalveolar lavage fluid (BALF) from patients and controls. Gelatin zymogram of BALF from representative patients with rapidly progressive IPF-UIP (R), slowly-progressive IPF-UIP (S), NSIP, BOOP, and from normal control nonsmokers (NC-NS) and current smokers (NC-S). Gelatinolytic bands of 134 kD (proMMP-9-lipocalin complex), 92 kD (proMMP-9), 85 kD (active form of MMP-9), 72 kD (proMMP-2), and 66 kD (active form of MMP-2) are detected. Intense lytic bands corresponding to the proMMP-9-lipocalin complex, pro-MMP-9, the active form of MMP-9, and proMMP-2 are visible in IPF-UIP (R), but only proMMP-2 and -9 are barely visible in IPF-UIP (S). In NSIP and BOOP, proMMP-2 is predominantly detected. In NC-NS and NC-S, no apparent gelatinolytic bands are observed. Two concentrations of purified proMMP-9 and proMMP-2 (133 and 400 ng/ml) were used simultaneously as positive controls for each gelatin zymographic analysis of BALF samples to standardize the measurement results.

Computer-aided densitometric analysis of the gelatinolytic bands performed with NIH image software is shown in a scatterplot (Figure 3A). To show the difference in MMP-2/ MMP-9 ratios between patients with IPF-UIP and NSIP/ BOOP, MMP-9 activity (vertical axis) and MMP-2 activity (horizontal axis) in each BALF sample are shown in a scatterplot (Figure 3B). The results of a nonparametric Mann-Whitney test are summarized in Table 3. A significant predominance of MMP-9 activity in the IPF-UIP sample, and of MMP-2 in the NSIP and BOOP samples was clearly demonstrated. Furthermore, IPF-UIP (R) samples showed significantly higher MMP-9 activity than IPF-UIP (S), NSIP, or BOOP samples (p < 0.005). NSIP and BOOP samples showed higher MMP-2 activity than IPF-UIP (S) samples (p < 0.005). The MMP-2/MMP-9 ratio was significantly higher in NSIP and BOOP samples than in IPF-UIP (S) (p < 0.005) or IPF-UIP (R) samples (p < 0.05).

View larger version (19K): [in this window] [in a new window]   Figure 3.   Relationship between MMP-2 and MMP-9 activity in BALF from all patients. (A) Densitometric analysis of gelatinolytic bands in BALF from slowly progressive IPF-UIP (S) (n = 19), rapidly progressive IPF-UIP (R) (n = 7), NSIP (n = 11), and BOOP (n = 6) was performed. MMP data shown are the total activity of both active- and pro-form of MMPs. (B) MMP-9 activity (vertical axis) and MMP-2 activity (horizontal axis) in each BALF sample from patients with disease are shown in a scatterplot. Predominant activity of MMP-9 in IPF-UIP and MMP-2 in NSIP and BOOP is clearly demonstrated. Furthermore, IPF-UIP (R) shows significantly higher MMP-9 activity than the activity seen in IPF-UIP (S), NSIP, or BOOP (p < 0.005). NSIP and BOOP in comparison to IPF-UIP (S) show high MMP-2 activity (p < 0.005).

Correlation between Gelatinolytic Activity in BALF and BAL Cell Analysis Data

We performed a multiple comparison analysis to determine which constituents of BAL cells may participate in the production of MMP-2 and -9 in BALF. Table 4 shows the standardized regression coefficient (r_st) of BAL cell constituents on MMP-2 and -9 activity according to stepwise multivariate regression analysis. After adjusting for other factors by stepwise methods, only neutrophils were significantly associated with MMP-9 activity (r_st = 0.419, p = 0.0051), and lymphocytes alone were significantly associated with MMP-2 activity (r_st = 0.730, p < 0.0001) in all patients with disease. In patients with IPF-UIP, only neutrophils were significantly associated with MMP-9 activity (r_st = 0.509, p = 0.0079). In contrast, only lymphocytes were significantly associated with MMP-2 activity (r_st = 0.688, p = 0.0023) in patients with NSIP/BOOP. No BAL cell constituents were significantly associated with MMP-9 activity in patients with NSIP/BOOP. Eosinophils in BAL were not significantly associated with MMP activity in BALF.

The correlation lines between the MMP-9 activity and BAL neutrophils data and MMP-2 activity and BAL lymphocytes data are shown in Figures 4-6. Regression coefficients (r_s) and p values were assessed with the nonparametric Spearman correlation test. When analysis was performed in all patients with disease, a significant positive correlation between MMP-9 activity and the percentage of neutrophils was observed (r_s = 0.5288, p = 0.0003) (Figure 4A). MMP-2 activity correlated significantly with the percentage of lymphocytes (r_s = 0.4356, p = 0.0039) (Figure 4B). In the IPF-UIP group, MMP-9 activity correlated significantly with the percentage of neutrophils in the BALF (r_s = 0.7505, p < 0.0001) (Figure 5A), whereas no correlation was obtained between MMP-2 activity and the percentage of lymphocytes (r_s = 0.0217, p = 0.918) (Figure 5B). In the NSIP/BOOP group, MMP-2 activity correlated significantly with the percentage of lymphocytes in the BALF (r_s = 0.5126, p = 0.0354) (Figure 6B), whereas no correlation was obtained between MMP-2 activity and the percentage of neutrophils (r_s = 0.2015, p = 0.4381) (Figure 6A). Between the IPF-UIP and NSIP/BOOP groups, a significant difference in MMP-2 and -9 activity was observed. Here, we want to emphasize that the elevated MMP-9 activity may reflect significant participation of activated neutrophils in the pathogenesis of IPF-UIP and that elevated MMP-2 activity may reflect significant participation of lymphocytes in that of NSIP and BOOP.

View larger version (18K): [in this window] [in a new window]   Figure 4.   Correlation between MMP-2 or MMP-9 activity and BAL cell constituents in all patients with disease. A significant positive correlation is demonstrated between MMP-9 activity and the percentage of neutrophils (rs = 0.5288, p = 0.0003) (A) and between MMP-2 activity and the percentage of lymphocytes (rs = 0.4356, p = 0.0039) (B). Neu = neutrophils, Lym = lymphocytes.

View larger version (15K): [in this window] [in a new window]   Figure 5.   Correlation between MMP-2 or MMP-9 activity and BAL cell constituents from patients with IPF-UIP. MMP-9 activity correlates significantly with the percentage of neutrophils (rs = 0.7505, p < 0.0001) (A), whereas no correlation is observed between MMP-2 activity and the percentage of lymphocytes (rs = 0.0217, p = 0.918) (B). Neu = neutrophils, Lym = lymphocytes.

View larger version (17K): [in this window] [in a new window]   Figure 6.   Correlation between MMP-2 or MMP-9 activity and BAL cell constituents from patients with NSIP and BOOP. MMP-2 activity correlates significantly with the percentage of lymphocytes (rs = 0.5126, p = 0.0354) (B), whereas no correlation is observed between MMP-9 activity and the percentage of neutrophils (rs = 0.2015, p = 0.4381) (A). Neu = neutrophils, Lym = lymphocytes.

Immunohistochemical Analysis of MMP-2, -9, and TIMP-2 Expression

UIP. The presence of a patchy and highly variegated structure spanning the entire spectrum from normal alveolar walls to fibrotic end-stage lesions is characteristic of UIP. Variegated epithelial cells line the airspaces and range from rounded cells on less-damaged alveolar walls to cuboidal, columnar, ciliated, and squamous metaplastic cells on scattered alveolar walls (20, 21). Dense collagen deposition is observed. In the so-called honeycomb lesions, the lumens of airspaces are covered by bronchiolar metaplastic epithelial cells, a phenomenon termed bronchiolization.

In sections incubated with nonimmunized mouse IgG as negative controls, no significant reaction was observed (Figures 7A, 7E, and 7I). MMP-9 was intensely expressed by these regenerated cells, alveolar and interstitial macrophages, and neutrophils (Figure 7C). The intense expression of MMP-9 by metaplastic epithelial cells was especially characteristic of UIP. Interstitial fibroblasts, vascular endothelial cells, and smooth muscle cells were positive for MMP-2 (Figure 7B). TIMP-2 was widely detected not only in regenerated epithelial cells and macrophages but also in interstitial fibroblasts (Figure 7D). In the fibroblast foci, interstitial cells expressed TIMP-2 strongly.

View larger version (168K): [in this window] [in a new window]   Figure 7.   Immunohistochemical detection of MMP-2 and -9 and TIMP-2 in lung tissue from patients with IPF-UIP (A-D), NSIP (E-H ), or BOOP (I-L). In sections incubated with nonimmunized mouse IgG as negative controls, no significant reaction was observed (A, E, I ). In IPF-UIP, MMP-9 are intensely expressed by regenerated cells, alveolar macrophages, and neutrophils (C ). MMP-2 (B) and TIMP-2 (D) are widely detected in regenerated epithelial cells and macrophages and in interstitial fibroblasts of the fibroblast foci. NSIP shows a distinct expression pattern of MMP-2 and -9 and TIMP-2 from that of UIP. MMP-2 is intensely expressed by regenerated cuboidal epithelial cells (F ), although expression of MMP-9 is apparently absent (G), and that of TIMP-2 is weak (H ). In BOOP, the staining patterns of MMPs and TIMPs are consistent with that of NSIP. Regenerated type II cells express MMP-2 predominantly (J ), whereas no apparent MMP-9 expression is observed (K ). TIMP-2 expression is weak in macrophages and interstitial cells (L). Paraffin sections are shown.

NSIP. NSIP is characterized by varying degrees of temporally uniform inflammation and fibrosis in the alveolar walls. A chronic inflammatory-cell infiltrate containing lymphocytes and some plasma cells is seen, and cuboidal epithelial cells cover most of the airspaces.

In our study, NSIP demonstrated an MMP-2, -9, and TIMP-2 expression pattern that was distinct from that of UIP. In general, MMP-2 was intensely expressed by regenerated cuboidal epithelial cells (Figure 7F); MMP-9 expression was weakly detected in some macrophages, regenerated epithelial cells, and interstitial fibroblasts, although the intensity was very faint (compared with that of MMP-2) (Figure 7G). TIMP-2 was widely detected in regenerated epithelial cells and macrophages (Figure 7H). Some interstitial cells also stained for TIMP-2 weakly; however, the intensity was more weak compared with that of MMP-2.

BOOP. BOOP is characterized histologically by chronic inflammatory interstitial infiltrates associated with patchy granulation filling terminal and/or respiratory bronchioles, with extension into distal airspaces.

Apparent fibrosis with remodeling of the pulmonary architecture was absent according to light microscopy. Many of these intraalveolar buds were covered by regenerated type II epithelial cells. The staining patterns of MMP-2 and -9 and TIMP-2 were consistent with patterns in NSIP. Regenerated type II cells expressed MMP-2 (Figure 7J), whereas MMP-9 expression was absent or very weak (Figure 7K). TIMP-2 expression was detected in regenerated epithelial cells and macrophages. Some interstitial cells also stained for TIMP-2 but the intensity was very weak (Figure 7L). Immunohistochemical labeling results are summarized in Table 5.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

MMPs capable of degrading various components of connective tissue matrices and TIMPs are believed to play a significant role in remodeling after parenchymal damage, resulting in tissue destruction or the induction of repair processes in pulmonary diseases (9). In interstitial pneumonias, it is generally recognized that the integrity of thin alveolar walls depends on the basement membrane and that the destruction of the subepithelial basement membrane precedes the intraalveolar fibrosis process (8, 14, 22, 23). The discontinuity of the basement membrane is considered necessary to provide inflammatory cells, exudative factors, and interstitial cells with access to the alveolar space, resulting in further tissue destruction and intraalveolar fibrosis (15, 16). MMP family enzymes MMP-2 and -9 possess type IV collagenolytic activity, which is of importance in the destruction of the subepithelial basement membrane in early alveolar remodeling. In this study, gelatin zymography of BALF and immunohistochemical analysis of lung tissue demonstrated predominant expression of MMP-9 in association with IPF-UIP and MMP-2 in association with NSIP and BOOP. Furthermore, their active forms were increased in BALF from patients with rapidly progressive IPF-UIP and in some patients with NSIP and BOOP. These MMPs may play a role in intraalveolar fibrosis through destruction of the subepithelial basement membrane.

In IPF-UIP, macrophages, regenerated and metaplastic epithelial cells, and neutrophils were detected immunohistochemically as major sources of MMP-9. In NSIP and BOOP, macrophages, fibroblasts, and regenerated type II cells were considered the major source of MMP-2. However, the activity of MMPs in the tissue depends upon the relative concentration of TIMPs as well as their own expression and extracellular activation. Myofibroblasts situated at ongoing lesions of intraalveolar fibrosis in cases of BOOP and NSIP expressed predominantly MMP-2 without significant expression of TIMP-2, whereas these cells in cases of IPF-UIP expressed TIMP-2 intensely. These findings were supported principally by the findings of previous studies (9, 10). We suspect that TIMPs in IPF-UIP may contribute to ECM deposition resulting in irreversible structural remodeling at active fibrosing lesions (10). It is of interest that characteristic expression patterns of MMPs and TIMPs at active fibrosing lesions may relate to the distinct prognostic features of these diseases, which are reversible and curable in BOOP and NSIP and progressive and irreversible in IPF-UIP (1, 3, 8).

A recent in vitro study has demonstrated that neutrophils use MMP-9 and elastase to migrate across the basement membrane (24). In the present study, a significant positive correlation was obtained between MMP-9 activity and an increased percentage of neutrophils in IPF-UIP BALF. This result indicates the significance of neutrophils as one of the major sources of MMP-9, especially in active cases of IPF-UIP. It was recently reported that elevated levels of MMP-2 and -9 correlate with the number of neutrophils in BALF from patients with acute respiratory distress syndrome (11,12). In addition, Finlay and colleagues (13) reported on the significance of MMPs in addition to neutrophil elastase in the pathogenesis of emphysema. In these reports, significant roles of collagenolytic MMPs in the degradation of type IV collagens were suggested, and neutrophils were considered as a major source of collagenolytic MMPs. Using gelatin zymography, proMMP-9 purified from neutrophils showed three lytic bands, 92, 134, and 220 kD, corresponding to proMMP-9, proMMP-9-lipocalin complex (lipocalin: neutrophil gelatinase-associated lipocalin), and dimerized forms of proMMP-9, respectively (25). In our study, the pattern observed in BALF from active IPF-UIP cases showed an intense band corresponding to the complex of MMP-9 with lipocalin. These results strongly suggest significant participation of neutrophil-derived MMP-9 in the gelatinolytic activity in BALF, although BALF from most patients with slowly progressive IPF-UIP showed gelatinolytic activity in the cells in the absence of the MMP-9-lipocalin complex. Neutrophils may influence the disease activity of UIP through the production of additional neutrophil-derived MMP-9. The importance of neutrophils in this disease is supported by the clinical recognition that BAL neutrophilia is associated with a poor treatment outcome in IPF-UIP (26, 27).

The role of MMPs in the transmigration of T lymphocytes through the basement membrane structure has been reported. Leppert and coworkers (28) reported that T cells express MMP-2 and -9 in vitro during migration across a basal lamina equivalent and that MMP-2 expression by normal T cells is inducible, whereas that of MMP-9 is constitutive. Other investigators have shown that the expression of MMP-2 and -9 by T cells depends on their activation state following exposure to chemotactic factors or cytokines (29, 30) and that integrin-mediated adhesion induces MMP-2 expression by T cells (29). In view of these reports, it is of interest that a positive correlation was observed in our study between elevated MMP-2 activity and increased lymphocytes in BALF of NSIP and BOOP. We speculated that chemokines and cytokines, such as RANTES, monocyte chemoattractant protein-1, and tumor necrosis factor-, and integrin-mediated adhesion may up-regulate MMP-2 expression by T cells and that some MMP-2 may derive from T cells themselves. The predominant MMP-2 expression in NSIP and BOOP may indirectly reflect the important role of MMP-2 in the influx of T cells to active inflammatory foci.

In conclusion, we demonstrated potential roles of MMPs and TIMPs in the tissue remodeling that occurs in IPF-UIP, NSIP, and BOOP. MMP-9 in IPF-UIP and MMP-2 in NSIP/ BOOP may contribute to pulmonary structural remodeling through type IV collagenolytic activity. Our data also suggest significant participation of MMP-9 derived from activated neutrophils in the pathogenesis of IPF-UIP. The characteristic contributions of matrix-degrading proteins may relate to the distinct prognostic features of these diseases, which are irreversible in the case of IPF-UIP and reversible in the case of NSIP or BOOP. Appropriate regulation of these MMPs could provide a tool to treat irreversible tissue remodeling in IPF.