Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Histologic examination of biopsy or postmortem lung tissue from patients with asthma usually reveals thickened airway walls (1). This change is also seen in patients when airway wall thickness is assessed in vivo by high-resolution computed tomography (2). Remodeling changes that contribute to this thickening include subepithelial fibrosis (SEF), goblet-cell hyperplasia (GCH) (3, 4) with the accompanying epithelial layer thickening, smooth-muscle hypertrophy (5), and increased vascularity (6). Airway wall thickening is exaggerated by vasodilation and by edema due to microvascular leakage from postcapillary venules (7). SEF does not involve the epithelial basement membrane itself (8) but represents an increased deposition particularly of collagen III, but also of collagens I and V and fibronectin in the reticular layer (lamina reticularis) beneath the basement membrane. Amounts of tenascin are also increased (9). The increase in the thickness of the reticular layer is usually no more than a doubling. SEF is possibly the result of the activity of myofibroblasts (10, 11) whose number correlates with levels of collagens III and V and tenascin, while eosinophil numbers correlate with levels of matrix metalloproteinase-9 and tissue inhibitor of metalloproteinase-1 (9). A thickened subepithelial layer can make airway closure more marked during airway smooth-muscle contraction (12).

We have previously reported the presence of SEF as one of several asthma-like pathologic features reproduced in a murine model of the disease (13). Here, we describe the quantitation of SEF in this model using a stereologic point-counting method to assess the amount of reticulin in the subepithelial layer of small-to-medium-sized airways. Reticulin is not a single entity. Classical "reticulin" has been reported in the literature as consisting chiefly of collagen III but in association with smaller, and variable, amounts of proteoglycans (14). The absence of spontaneous resolution of established allergen-induced SEF at 50 d after the last allergen challenge, and the ability of treatment with dexamethasone (DEX) or an anti-interleukin (IL)-5 antibody to inhibit induction of SEF, is reported. Resolution of existing SEF is shown to be initiated by DEX. These studies highlight distinct differences in the characteristics of two concurrent features of airway remodelingSEF and GCHin terms of the ability of allergic or nonallergic inflammation to induce them, in the time required for their resolution, and in the apparent need for eosinophil infiltration of the airways in their development.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

This research complied with national legislation and with company policy on the care and use of animals.

Sensitization to Allergen

Sensitization of male BALB/c mice (Charles River UK Ltd, Margate, Kent, UK) was performed as previously described (15). Briefly, mice (16 to 18 g, 4 to 5 wk old) were immunized, using an adjuvant-free protocol, by the intraperitoneal injection of 10 µg ovalbumin (OVA) (grade V, Sigma No. A-5503; Sigma Chemical Co., Poole, Dorset, UK) in 0.1 ml endotoxin-free saline (Sodium Chloride for Injection BP, 0.9% weight/volume; Evans Medical Ltd., Langhurst, Horsham, West Sussex, UK) on each of seven alternate days. Systemic levels of OVA-specific immunoglobulin (Ig) E were allowed to increase for 40 d after the first sensitizing injection.

Intratracheal Challenge with Allergen

On or after Day 40, when the mice were 25 to 28 g in weight, intratracheal challenge was performed according to the standard protocol adopted for this model (challenge on 3 d, each 3 d apart). The animals were anesthetized with 0.2 ml Saffan (alphaxolone, 0.9% wt/vol plus alphadalone, 0.3% wt/vol; Vet Drug Ltd., Dunnington, York, UK) intraperitoneally, and 10 µl of saline alone (sham controls) or a solution of OVA in saline was instilled into the trachea by a nonsurgical technique. The effectiveness of this technique in achieving a widespread distribution of liquid throughout the pulmonary airways has been demonstrated (15). In this model, the degrees of cellular infiltration into the lung tissue and airway lumen, and of GCH in the airway epithelium, are directly related to the amount of OVA used for each of the three challenges (16). For most of the work, a challenge dose of 80 µg OVA was used to maximize the fibrotic response. Some sections were from earlier studies in which the challenge dose was 20 µg OVA. Group sizes varied from four to 13.

Intratracheal Treatment with Lipopolysaccharide

Lipopolysaccharide (LPS) (endotoxin from Escherichia coli serotype 055:B5, Sigma no. L-6529) was dissolved in endotoxin-free saline and administered intratracheally at a dose of 10 µg, in a volume of 10 µl, on each of 3 d, each 3 d apart. The mice were killed and examined 1 d after the last exposure to LPS.

Bronchoalveolar Lavage

Mice were killed by an intraperitoneal injection of 0.1 ml pentobarbitone sodium, 200 mg/ml (Euthatal; Rhone Merieux, Harlow, Essex, UK). Bronchoalveolar lavage (BAL) was performed by introducing 1 ml phosphate-buffered saline (PBS) into the lungs via a tracheal cannula, withdrawing this fluid into a test tube on ice, and repeating the procedure a further four times. This procedure has been found to recover all the intraluminal cells that can be recovered. Additional 1-ml washes merely serve to dilute the BAL cell suspension. The recovered fluid (approximately 4.5 ml) was centrifuged (300 × g for 6 min) and the cells were resuspended in 0.5 ml PBS. Total cells were counted using an improved Neubauer hemocytometer chamber or a Sysmex K-1000 automated hematology analyzer (Toa Medical Electronics Co., Ltd., Kobe, Japan). An air-dried slide preparation was made of each sample (Cytospin 3; Shandon Scientific, Runcorn, Cheshire, UK) and stained with May-Grunwald-Giemsa stain. Differential counts of at least 200 cells were made according to standard morphologic criteria. The numbers of cells recovered per mouse were then expressed as the mean and standard error of the mean (SEM) for each treatment group.

Histologic Analysis of the Lung

Analysis was performed as previously described (13, 15). Lungs were fixed by slow in situ inflation with 1 ml 10% phosphate-buffered neutral formalin, pH 7.0, via a tracheal cannula and, after immediate removal from the thorax, immersion in the fixative for a minimum period of 24 h. After fixation of the lung tissue and processing to paraffin wax, sections (3 to 4 µm thick) were cut longitudinally through the left lung (one lobe) and right lung (cranial, medial, accessory, and caudal lobes) so as to include all lobes (sometimes with the exception of the right accessory lobe). Sections were stained with hematoxylin and eosin for general morphology, or Gordon and Sweets (G&S) stain (17) for identification of reticulin by the deposition of silver.

Quantitation of SEF

Using the same lung sections, we initially compared three different methods of two-dimensional area analysis for the amount of reticulin around tranversely sectioned airways: (1) stereologic point-counting using a Weibel graticule in the eyepiece of a light microscope (18); (2) computerized image area measurement using a Sony 3CCD color camera and a Leica Quantimet Q600S computer with image analysis software; and (3) point-counting of a microscopic image projected onto a stereologic grid. The latter proved to be the most suitable method in terms of time, accuracy, and ease of use, and was adopted for routine measurement of SEF as follows.

A slide of a G&S-stained longitudinal section of one pair of lungs was placed onto the stage of a Leitz Neo-Promar light microscope fitted with a projection tube and prism. The image from the section (×40 objective) was deflected through 90 degrees by the prism and projected onto a vertical stereologic grid (grid P2, enlarged to 200 × 130 mm [19]). This was placed at a fixed distance from the microscope, producing a total magnification of ×520. The grid contained a total of 384 equally spaced points. The total subepithelial reticulin point count from 12 fields containing an airway was taken as the count for each pair of lungs (total possible count = 12 × 384 = 4,608). Group mean counts for sham-challenged airways were approximately 170; those for allergen-challenged airways were approximately 240 (expressed in the text as mean ± SEM). Only reticulin that coincided with points on the grid, that was associated with the subepithelial reticular layer, and that lay parallel to the basement membrane of the epithelium was counted. The selection of small-to-medium-sized airways for measurement was standardized by matching them to defined size criteria: the entirety of the airway transverse section had to fit within the borders of the 200 × 130 mm point grid (equivalent to a rectangular tissue area of 385 × 250 µm before magnification), and airways with a diameter below 115 µm (60 mm when magnified and projected) were excluded. Only airways that had been cut transversely were measured; any airway cut obliquely was excluded. All microscope slides were coded before counting to eliminate observer bias. Close agreement was found between point counts of the same samples by different operators.

Some measurements were made on lungs from studies in which GCH and cell recruitment data have already been published. Where this was the case (the studies on spontaneous resolution and those using LPS or TRFK5), the relevant reference is indicated in the text.

Our aim in not routinely quantifying SEF by measuring the thickness of the reticular layer was to avoid mistaking increases in thickness, which theoretically could have been caused by deformation of interfiber space by edema, as increases in the amount of reticulin present. In fact, when the values for the reticulin point counts and the thickness of the reticular layer around the same airways in a series of lungs were compared, there was a highly significant correlation between these parameters (95% confidence limits, P < 0.0001). The thickness of the reticular layer was measured directly from the projected image of the G&S-stained lung section and then corrected for magnification (×520). The mean of four reticular layer thickness measurements across each transversely cut airway (in north-south, east-west, northeast-southwest and northwest-southeast directions) was calculated.

Treatment of Mice with DEX

DEX-21-phosphate, disodium salt (Sigma No. D-1159) was dissolved in endotoxin-free saline and administered daily at a dose of 1 mg/kg (expressed in terms of the base) by intraperitoneal injection in a volume of 0.1 ml. The treatment was given either for 8 d during the triple-challenge period (beginning on the day before the first challenge) or for 7- or 15-d periods immediately after the last allergen challenge. The animals were killed 1 d after the last dose of DEX.

Treatment of Mice with Anti-IL-5 Antibody

TRFK5 (rat antimurine IL-5) antibody has been shown to have a serum half-life in mice of 2.4 wk and, at 1 mg/kg, to inhibit allergen-induced lung eosinophilia for at least 12 wk, and to inhibit the release of eosinophils from bone marrow for at least 8 wk (20). In the studies reported here, it was injected intravenously 30 min before each of the three intratracheal OVA challenges. This time point was chosen arbitrarily and used to ensure that a defined and consistent interval elapsed between the administration of antibody and OVA challenge. Control animals received saline or an isotype control antibody (rat IgG1, product No. PRP02; Serotec, Kidlington, Oxford, UK). The antibodies were administered at 2 mg/kg at each injection. The preparation of TRFK5 was shown, using the Limulus amoebocyte lysate assay (E-toxate kit, Sigma No. 210-C1), to contain no detectable endotoxin. The mice were examined 48 h after the third OVA challenge. In these animals, TRFK5 selectively reduced blood eosinophil numbers in allergen-challenged mice to below those in sham-challenged control mice, and virtually abolished the infiltration of eosinophils into the airway lumen (13). The IgG1 isotype control antibody had no effect on numbers of eosinophils in either the blood or airway lumen.

Statistical Methods

The statistical significance of differences between group mean reticulin point counts (shown in the text ± SEM) was determined by Student's paired t tests. Data from the study showing suppression of SEF induction by DEX were subjected to analysis of variance and correlation analysis.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Induction of SEF with Allergen

In a typical study, the amount of subepithelial reticulin in the airways of mice challenged three times with 20 µg OVA (mean count of 180.8 ± 10.5) was 1.24 times that in airways from sham-challenged mice (145.5 ± 16.5) when measured after the third challenge. With small numbers of mice per group (as here), this increase was usually not statistically significant. However, by using more mice per group and a larger challenge dose (80 µg OVA), the increase in reticulin was larger (1.4-fold) and acquired statistical significance (Figure 1A). It occurred in airways containing a significantly increased number of eosinophils, neutrophils, lymphocytes, and macrophages (Figure 1B).

View larger version (27K): [in this window] [in a new window]   View larger version (25K): [in this window] [in a new window]   Figure 1.   (A) Induction of SEF by allergen in OVA (OA)- sensitized mice. A highly significant 1.4-fold increase in subepithelial reticulin had been induced when measured at 1 d after a third intratracheal instillation of 80 µg OVA. Mean counts were 169.18 ± 10.96, n = 11, and 238.36 ± 12.07, n = 13, for saline and OVA, respectively. (B) Cells present in the lumen of airways in these lungs, showing significant increases in all four types (*P < 0.05, **P < 0.001). Closed bars = eosinophils, open bars = neutrophils, striped bars = lymphocytes, cross-hatched bars = macrophages.

Effect of Treatment with TRFK5 Anti-IL-5 Antibody

Intravenous treatment of mice with the anti-IL-5 antibody TRFK5 (2 mg/kg) before each allergen challenge produced a statistically significant reduction (16.7% less reticulin) in the induction of SEF by triple challenge with 20 µg OVA (Figure 2A) and virtually eliminated eosinophil recruitment into the airways (Figure 2B). The effect of TRFK5 was selective for eosinophils, the numbers of other infiltrating cells (neutrophils, lymphocytes, and macrophages) in the airways being unchanged (Figure 9B in Reference 13). An IgG1 isotype control antibody (2 mg/kg intravenously) had no effect on SEF or eosinophil numbers.

View larger version (23K): [in this window] [in a new window]   View larger version (18K): [in this window] [in a new window]   Figure 2.   (A) Inhibition of allergen-induced SEF by treatment with anti-IL-5 antibody (TRFK5). A 1.24-fold increase in reticulin was induced by three challenges with 20 µg OVA (OA) (column B), and significantly reduced (by 16.7%) by treatment with 2 mg/kg TRFK5, given intravenously before each challenge (column D versus column B, *P < 0.05). An IgG1 isotype control antibody (2 mg/kg intravenously) had no effect on SEF (column C). (B) Suppression of eosinophil infiltration into these airways by TRFK5 (**P < 0.001).

Effect of Treatment with DEX

Preliminary work indicated that the induction of SEF by three challenges of mice with 20 µg OVA could be reduced significantly (P < 0.05, n = 5) by daily treatment with 1 mg/kg DEX for 8 d during the challenge period (mean count of 252.2 ± 20.38 in vehicle-treated mice, and 181.6 ± 6.15 in DEX-treated mice). This finding was confirmed by showing similarly significant inhibition by DEX of SEF induced by a higher challenge dose (80 µg OVA). This was shown using two methods of assessing SEF in the same samples of lung: (1) two-dimensional area measurement, and (2) thickness measurement.

Two-dimensional area measurement of subepithelial re-ticulin. Daily treatment of mice with 1 mg/kg DEX during the period of allergen challenges (80 µg OVA each time) produced a highly significant (P < 0.001) reduction (15.4%) in the reticulin counts in the airways. Airways of OVA-challenged mice treated with vehicle alone had a group mean count of 266.75 ± 4.06 (n = 12) compared with 225.67 ± 4.81 (n = 9) in airways of mice treated with the steroid (Figure 3A).

View larger version (24K): [in this window] [in a new window]   View larger version (19K): [in this window] [in a new window]   Figure 3.   Inhibition of allergen (3 × 80 µg OVA)-induced subepithelial fibrosis by daily treatment with 1 mg/kg DEX, given intraperitoneally (i.p.) during the challenge period (**P < 0.001). Column A: daily treatment with intraperitoneal saline. Column B: daily treatment with 1 mg/kg DEX. Airways in the same pairs of lungs were assessed by point-counting of reticulin (A), and by measurement of reticular layer thickness (B). Similar degrees of reduction in SEF were seen with both methods of assessment (15.4 and 13.1%, respectively).

Measurement of thickness of reticular layer. Measurements of reticular layer thickness in these airways confirmed the reductions indicated by point-counting of reticulin. Airways in the DEX-treated group (reticular layer thickness of 6.59 ± 0.18 µm) had a significantly thinner (13.1%) reticular layer (P < 0.001) than did those in vehicle-treated mice (7.58 ± 0.17 µm) (Figure 3B).

The number of eosinophils (per mouse) recovered by BAL from the airways of these mice fell from 9.3 × 10⁴ ± 2.7 × 10⁴ in the saline-treated group (group A, Figure 3) to 0.4 × 10⁴ ± 0.2 × 10⁴ in the DEX-treated group (group B, Figure 3) (P < 0.01). However, unlike that of TRFK5, the suppressive effect of DEX on cell recruitment was not selective for eosinophils, inasmuch as the treatment also markedly reduced the numbers of lymphocytes (saline: 2.4 × 10⁴ ± 0.5 × 10⁴; DEX: 0.2 × 10⁴ ± 0.1 × 10⁴), and macrophages (saline: 40.7 × 10⁴ ± 6.5 × 10⁴; DEX: 12.8 × 10⁴ ± 3.4 × 10⁴) in the airways. Numbers of neutrophils were unchanged by DEX treatment (saline: 3.4 × 10⁴ ± 1.6 × 10⁴; DEX: 3.4 × 10⁴ ± 2.9 × 10⁴).

Absence of Spontaneous Resolution of Established SEF

There was no spontaneous reduction in the amount of established subepithelial reticulin at 50 d after a third challenge with 20 µg OVA (mean count of 212.4 ± 7.50, n = 5) when compared with the amount at only 14 d after the challenge (mean count of 192.0 ± 7.25, n = 4). This contrasted with the previously reported resolution of GCH in the same airways, which had completely resolved at 50 d, the ratio of ciliated cells to goblet cells in the epithelium having been restored to normal by this time (Figure 4C, Reference 15).

View larger version (27K): [in this window] [in a new window]   Figure 4.   Partial resolution of existing SEF (previously established with three challenges of 80 µg OVA) in mice treated with DEX for 15 d after the third allergen challenge. Column A: untreated mice killed 1 d after the third challenge. Column B: mice treated daily with saline for 15 d after the third challenge and killed on the following day. Column C: mice treated daily with 1 mg/kg DEX for 15 d after the third challenge and killed on the following day. (*P < 0.05). An 11.5% reduction in reticulin occurred after treatment with the glucocorticoid.

Initiation of Resolution of Established SEF by DEX

Daily treatment of mice with 1 mg/kg DEX for 7 d after a third challenge with 20 µg OVA caused a small (statistically nonsignificant) decrease (8.9%) in the amount of subepithelial reticulin (mean point counts 207.17 ± 9.48 in vehicle-treated mice and 188.83 ± 11.87 in DEX-treated mice, n = 6). When the period of treatment with DEX was extended to 15 d after a third challenge with 80 µg OVA, the reduction in reticulin (11.5%) became statistically significant (P < 0.05, n = 8) (Figure 4). At 1 d after the third challenge with 80 µg OVA, the mean reticulin count was 234.57 ± 10.25. On the day after the last of 15 daily treatments, the counts were 240.57 ± 7.63 (vehicle-treated) and 212.88 ± 6.79 (DEX-treated).

Steroidal treatment for 15 d was therefore capable of initiating a partial resolution of SEF, which otherwise would not have occurred even at 50 d after a third OVA challenge.

At 16 d after the third allergen challenge, very few eosin-ophils (saline: 0.05 × 10⁴ ± 0.03 × 10⁴; DEX: 0.005 × 10⁴ ± 0.005 × 10⁴) or neutrophils (saline: 0.1 × 10⁴ ± 0.07 × 10⁴; DEX: 0.07 × 10⁴ ± 0.02 × 10⁴) remained in the airways of mice in the saline-treated group (group B, Figure 4) and, as shown, these small numbers were reduced further by DEX treatment (group C, Figure 4). The numbers of lymphocytes (saline: 2.1 × 10⁴ ± 0.95 × 10⁴; DEX: 0.6 × 10⁴ ± 0.14 × 10⁴) and macrophages (saline: 23.0 × 10⁴ ± 3.7 × 10⁴; DEX: 13.7 × 10⁴ ± 2.11 × 10⁴) were also reduced in the DEX-treated group.

SEF Not Induced by Nonallergic Inflammation

SEF was not induced in mice that received three intratracheal instillations of 10 µg LPS, each 3 d apart (mean reticulin counts of 186.0 ± 9.17, n = 6, in saline-treated mice; and 189.5 ± 9.59, n = 6, in LPS-treated mice). This treatment did cause a sustained nonallergic neutrophilic infiltration and marked activation of macrophages in the airways of these animals, in addition to inducing a moderate degree of GCH in the airway epithelium (Reference 15).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

SEF is one of the features of airway remodeling seen in the airways of patients with atopic or nonatopic asthma. It is also seen, in a less marked form, in atopic nonasthmatics (21). Although some studies have found that its degree is unrelated to disease severity (22), others show correlation with severity but not with disease duration (23). Because it has been reported as being present in the pulmonary airways of nonasthmatics with allergic rhinitis (24, 26) and as being seen in bronchial biopsies from children 2 to 5 yr before the eventual clinical diagnosis of asthma (27), it may be one of the earliest structural changes in airways that are destined to manifest hyperresponsiveness at a later date (28, 29). SEF may represent part of the "wound-healing" process in which fibroblast and myofibroblast proliferation and collagen synthesis and deposition take place (30). Collagen deposition in asthma is not confined to the reticular layer but also occurs in deeper layers of the airway wall, and it is possible that fibrosis at these sites has a greater effect on airway function than that in the reticular layer (31). One altered function that may be important is the limited ability of thickened airway mucosa to form multiple small folds during airway constriction, necessitating the formation of fewer, but larger, folds and causing an exaggerated reduction in luminal volume (32). In this way, hyperresponsiveness to inhaled spasmogens may exist in airways that show no increased baseline resistance.

Our understanding of SEF may be advanced by the characterization of a disease model that reproduces this feature of asthma and in which the quantitation of reticulin is reasonably straightforward. In this paper we describe the use of a murine model of atopic asthma for the experimental study of allergen-induced SEF and its quantitation by the stereologic point-counting of reticulin. By using this method, rather than measuring the thickness of the reticular layer, we hoped to avoid mistaking increases in thickness, which theoretically could be due to edema within the layer, for increases in the amount of reticulin.

Fibrosis in the reticular layer of mouse airways was consistently reproduced in this model by triple intratracheal challenge with OVA (80 µg) over a period of 8 d. The typical 1.4-fold increase in the amount of subepithelial reticulin in our model approximated to increases in the thickness of the reticular layer reported for human asthmatic airways (rarely more than a doubling). Examples of such values for the thickness of the reticular layer (in µm) in asthmatic and control human airways, respectively, followed by their ratios, are 7.95/4.17 = 1.91-fold (8), 7.55/4.68 = 1.61-fold (10), 17.19/7.80 = 2.2-fold (24), 8.09/4.02 = 2.01-fold (25), 6.01/3.19 = 1.88-fold (31), and 7.5/5.5 = 1.36-fold (22).

Unlike GCH, which resolved in this model to leave a normal epithelial cell population by 50 d after the third allergen exposure (15), SEF showed no indication of spontaneous resolution at this time. We do not know how much more time needs to elapse before resolution begins. SEF may remain as a permanent or semipermanent feature of diseased airways. This indicates, perhaps not surprisingly, that different features of allergen-induced remodeling resolve at different rates. There is clinical evidence to suggest that, in human asthmatics, SEF and the associated increases in numbers of subepithelial fibroblasts, mucosal mast cells, and lymphocytes can show spontaneous resolution when exposure to the inducing antigen is avoided for 6 to 21 mo (33).

We used triple exposure of airways to LPS to induce a severe, sustained, nonallergic inflammation characterized by a marked airway infiltration by neutrophils (but not by eosinophils) and a moderate degree of GCH (15). Despite these changes, SEF was not induced. This may suggest that one or more components of allergic inflammation (perhaps mast cells and/or eosinophils) are required for induction of fibrosis at this anatomic site.

The induction of SEF by allergen was suppressed by treatment of mice with an anti-IL-5 antibody that virtually eliminated eosinophil infiltration into the airways. This suggests that IL-5 itself, or more probably products of the eosinophil, may be required for triggering of the fibrotic processes in the reticular layer. It is especially significant that this treatment had no effect on the allergen-induced increase in goblet-cell numbers in these same airways (13). The eosinophil may therefore be implicated in, or at least associated with, the induction of one feature of remodeling (SEF) but not that of another (GCH).

The degree of allergen-induced SEF, whether assessed in terms of the amount of subepithelial reticulin or the thickness of the reticular layer, was shown to be reduced by the action of a glucocorticoid. The inhibition of SEF induction, and the initiation of its resolution, seen in this model by treatment with DEX, provides experimental evidence to support clinical findings that, in asthmatics previously untreated with steroids, reductions in reticular layer thickness can be detected after inhaled treatment with beclomethasone dipropionate (34, 35) or with low doses of a the more potent glucocorticoid fluticasone propionate for a shorter period (36). When data from patients who have previously been treated with glucocorticoids is included in analyses, the correlation between SEF and disease severity can be lost (22). Glucocorticoid-induced reductions in SEF have been correlated with reductions in airway levels of some growth factors (35). SEF should therefore not be described as a "steroid-resistant" feature of asthma, but may require treatment of greater duration for its reversal than do some of the other pathologic features of the disease.

The induction of SEF in mice by allergic, but not by nonallergic, inflammation and its prevention by two different treatments, both of which, in addition to other possible actions, abolish eosinophil recruitment to the airways, may help to explain why SEF is frequently seen in asthma but rarely in chronic obstructive pulmonary disease. Airway inflammation in the latter is characterized by neutrophilic rather than eosinophilic infiltration.

SEF has been reported as being present in another model of asthma in mice (37, 38) and has been produced in mice by inducing the local expression of IL-5 (39), IL-9 (40), IL-11 (41), or IL-13 (42) in airway epithelial cells. It is difficult at present to adequately explain the process that is triggered by the actions of these diverse mediators, but a common downstream event in the inflammatory cascade is likely to mediate the final deposition of collagen.

The present studies have shown that SEF can be routinely reproduced and quantitated in the mouse, that daily treatment with a glucocorticoid can inhibit its induction and initiate its resolution, and that it can be distinguished from GCH by its apparent dependence on allergic mechanisms (perhaps involving the eosinophil) and by its delayed resolution. Such studies may advance the development of new treatments to prevent SEF or to restore the thickened reticular layer of asthmatic airways to normal.