INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary fibrosis is a disease process characterized by abnormal lung physiology and by the excess production of extracellular matrix molecules such as collagen, elastin, and proteoglycans (1). The bleomycin-induced lung fibrosis model has been used extensively in an attempt to better characterize the alterations in lung structure and function in this disease process (1).

Although the changes in the static mechanical properties of the lung have been relatively well defined, less is known about the changes in the dynamic mechanical behavior. As tissue resistance (Rti) accounts for a major proportion of the energy dissipated during breathing (5), defining the effects of a disease process on dynamic tissue behavior is key. Some data are available on changes in tissue resistance in fibrotic lung disease. A number of years ago Bachofen and Scherrer (6) used a technique that indirectly measured Rti, to show that Rti was increased in patients with lung fibrosis. Verbeken and colleagues (7) made measurements of Rti in autopsy lungs of patients with pulmonary fibrosis using the alveolar capsule technique and demonstrated that Rti was increased in these lung specimens. We have recently studied the bleomycin-induced rat model of pulmonary fibrosis, and determined that Rti is increased both in intact animals and in isolated lung parenchymal strips oscillated in the organ bath (8). Changes in vitro were of a magnitude similar to those measured in vivo.

The changes in lung structure in pulmonary fibrosis have been incompletely characterized. Increases in collagen and elastin have been well documented (1). More recently, investigators have become interested in changes in the "ground substance" component of the matrix. The ground substance is largely composed of proteoglycans that are macromolecules consisting of a protein core and glycosaminoglycan side chains. Proteoglycans subserve many different biological functions; they affect mechanical behavior, influence collagen fiber formation and assembly, modulate cell migration, and interact with various cytokines and growth factors (9). Some data are now available showing that these molecules are altered in fibrotic lung disease. Bensadoun and coworkers (4, 10) have shown in both granulomatous and nongranulomatous forms of human pulmonary fibrosis that the large proteoglycan versican is increased in the lung interstitium. These authors propose that versican is required as an early component for the remodeling process to proceed. In studies conducted in the bleomycin-exposed rat, Westergren-Thorsson and coworkers (11) showed that the small proteoglycans fibromodulin, decorin, and biglycan were altered, both in terms of message and protein expression, in the early stages of the disease process.

The link between changes in lung function and lung structure in the fibrotic lung has been somewhat difficult to establish. Sansores and coworkers (12) showed a modest correlation between the degree of fibrosis on lung biopsy in patients with idiopathic pulmonary fibrosis or pigeon breeder's disease and the exponential constant, k, that describes the shape of the lung pressure-volume curve. Goldstein and colleagues (3) were unable to document a correlation between collagen or elastin content and static lung compliance in the bleomycin model. They did, however, show a relationship between static compliance and total protein and DNA. In our recently completed study in bleomycin-induced fibrosis in rats, we demonstrated a positive correlation between volume proportion of collagen and parenchymal strip resistance and elastance (the latter also measured during dynamic oscillations). There was no correlation between collagen or elastin and hysteresivity, a measure of the mechanical friction in the system.

One question that arises is the role of proteoglycans (PGs) in contributing to the alterations in lung tissue mechanical behavior. These molecules contain many hydrophilic side chains that have the ability to attract ions and water into the matrix and thereby affect tissue turgor (9). Moreover, alterations in PG have been shown to affect the mechanical behavior of structures such as cartilage and tendon (13). Insofar as these molecules are altered in fibrotic disease, we reasoned that they may contribute to the mechanical changes we have previously observed. Further, fibrosis can be an ongoing, chronic process. We thought it would be of interest, therefore, to characterize the disease during the early stages, as well as after fibrosis was well established. Hence, we provoked lung fibrosis in rats using bleomycin and studied animals at 7, 14, and 28 d after bleomycin exposure. Isolated parenchymal strips were studied in the organ bath to define mechanical changes. We then fixed the strips with formalin and, using histochemical and immunohistochemical techniques, quantitated changes in collagen, elastin, and the small PGs, biglycan and fibromodulin. We examined whether changes in oscillatory mechanics were correlated with changes in the amounts of these different proteins, especially as the disease progressed.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal Preparation

Sprague-Dawley rats were anesthetized with pentobarbital sodium (30 mg/kg intraperitoneally) and intubated with a tracheal cannula (i.d. = 2 mm, length = 2.5 cm). Bleomycin sulfate (BM) (Bristol-Myers Squibb, Princeton, NJ) (1.5 U) was dissolved in 0.3 ml of saline and instilled into the animal's lungs via the tracheal cannula. Saline was administered to the control group (n = 18 in both groups). The animals were shaken vigorously to facilitate distribution of the BM or saline. Animals were monitored daily and weights were recorded. At 7, 14, and 28 d after instillation, animals were studied (n = 6 at each timepoint). Rats were anesthetized with pentobarbital sodium (30 mg/kg intraperitoneally), the thorax was opened, and animals were killed by severing the inferior vena cava. The heart, lungs, and trachea were carefully resected en bloc and rinsed with a modified Krebs solution (in mM: 118 NaCl, 4.5 KCl, 1.2 KH₂PO₄, 25.5 NaHCO₃, 2.5 CaCl₂, 1.2 MgSO₄, and 10.0 D-(+)-glucose [Sigma Chemical, St. Louis, MO]) with pH 7.40. Lung parenchymal strips (7 × 2 × 2 mm) were cut from the subpleural portion of the lung, and the pleura removed from the strips. Resting (unloaded) length (L_r) and wet weight (W) of each strip were measured. Strips were placed in iced Krebs solution, continuously bubbled with 95% O₂/5% CO₂ at pH 7.40.

Experimental Apparatus

Metal clips were glued to either end of the tissue strip with cyanoacrylate. Steel music wires were attached to the clips, and the strip was suspended vertically in a Krebs-filled organ bath maintained at 37° C and continuously bubbled with 95% O₂/5% CO₂. A mercury bead was placed in the bottom of the organ bath, allowing the wire to pass through the bath but preventing the Krebs solution from leaking out. One end of the strip was attached to a force transducer (model 400A; Cambridge Technologies, Watertown, MA) that had an operating range of ± 10 g, resolution of ± 200 mg, and compliance of 1 mm/g. The other end was attached to a servo-controlled lever arm (model 300A; Cambridge Technologies), which was capable of peak-to-peak length excursions of 8 mm and a length resolution of 1 mm. The lever system was connected to a function generator (model 3030; BK Precision, Chicago, IL) that controlled the frequency, amplitude, and waveform of the oscillation. The resting tension was set by means of a thumbscrew system that effected changes in the vertical displacement of the force transducer. Length and force signals were lowpass filtered, digitized with an analog-to-digital converter (Model DT2801-A; Data Translation, Marlborough, MA) and recorded on an AT-compatible computer.

The linearity and hysteresis of the system were tested by measuring the moduli of a steel spring of stiffness comparable with that of the tissue strip. The spring was suspended in the bath by music wire in the same manner as the strip. The frequency and amplitude dependence of the system were assessed over a range of frequencies (0.1-10 Hz). The spring stiffness did not show any dependence on oscillation frequency below 10 Hz. The hysteresivity of the system was independent of frequency and had a value of < 0.01.

Measurement of Oscillatory Mechanics

Parenchymal strips were preconditioned by slowly cycling the tissue from 0 to 5 kPa lagrangian stress three times. Lagrangian stress () was calculated from the formula  = T/A₀, where T is the tension and A₀ is the unstressed cross-sectional area of the strip. A₀ was calculated from the formula A₀ = W(g)/ × L_r, where  is the mass density of the tissue taken as 1.06 g/cm³. After three cycles of preconditioning were performed, tension was adjusted to approximately 10 ~ 20% larger than the value that corresponded to a stress of 2 kPa, and stress relaxation was allowed to occur for 45 min. After stress adaptation, tension was adjusted again to a stress of 2 kPa and left to stabilize for 5 min, during which time we considered that a plateau tension had been reached. Sinusoidal length oscillations with an amplitude ( ) of 1% of L_r at different frequencies (f = 0.3, 1, 3, and 10.0 Hz) were applied. Twenty-second recordings of tension and length were collected at each frequency. The oscillatory amplitude was changed to 3 and 10% of L_r, and recordings at each frequency were repeated.

Resistance (R) and elastance (E) were estimated by fitting the equation of motion to changes in tension (T) and length (L): T = EL + R(L/t) + K, where L/t is the length change per unit time, and K is a constant. Results were standardized for strip size by multiplying the values of R and E by L_r/A₀.

Hysteresivity (), a dimensionless variable coupling the dissipative and elastic behaviors, was calculated with the equation  = (R/E) 2f (14).

Measurement of Quasistatic / Curve

After measurement of oscillatory mechanics, parenchymal strips were left at 0 kPa for 15 min. Five cycles of 0.02-Hz triangular wavelength change from 0 to 5 to 0 kPa were then applied to the strip. We analyzed data from the third cycle and calculated quasistatic elastance (d/d) as a function of . A change in length from L_r was designated as L, and stretch ratio () was defined as L/L_r, where L is the operating length (L = L_r + L).

Histology and Immunohistochemistry

After physiological experiments were completed, strips (n = 4 in each subgroup) were fixed with 10% formalin and embedded in paraffin for histology and immunohistochemistry. Five-micrometer-thick sections were cut with a microtome and stained with Verhoff's stain for elastic fibers, or Van Gieson's picric acid fuchsin stain for collagen. Immunohistochemistry was done using the APAAP method. Sections were rinsed with Tris-buffered saline (TBS, 0.5 M Tris, pH 7.6, 1.5 M NaCl) and incubated with primary antibodies for biglycan and fibromodulin (rabbit antihuman, 1:500) for 1 h at room temperature. Sections were then rinsed with TBS and incubated with a biotin-labeled swine antibody (antirabbit, 1:500) (DAKO, Missisauga, ON, Canada). After further washing with TBS, sections were incubated with alkaline phosphatase-conjugated avidin (1:200) (DAKO). Sections were washed again with TBS, and then developed with Fast Red salt (1 mg/ml in alkaline phosphatase substrate) for 10 min at room temperature. Finally, sections were counterstained with Mayer's hematoxylin for 45 s.

Morphometric Study

A semiquantative analysis was performed by applying point counting in one strip per animal. Using a 42-point grid, the volume fraction of collagen, biglycan, and fibromodulin was calculated on nonoverlapping fields at ×400 magnification. The entire parenchymal strip was sampled; the total number of points counted was approximately 3,000 per strip. Volume fraction for each component was calculated as the number of test points falling on positive staining, divided by the total number of test points falling on the alveolar wall. Positive stainings located in blood vessels or bronchial wall were excluded from analysis.

The length of elastic fibers per unit volume of alveolar wall (L_VA) was determined using a 21-bar grid on one strip per animal. The number of intersections of elastic fibers with a test line was counted on nonoverlapping fields sampling the entire strip at ×400 magnification. Elastic fibers located in the blood vessels or bronchial wall were excluded. We also determined the volume proportion of alveolar wall (pr_alv) by point counting. L_VA was calculated as L_VA = N_tot/(L_test × d × pr_alv), where N_tot is the number of intersections of elastic fibers, L_test is the total length of test line, and d is thickness of sections (5 µm).

Intra- and interobserver variance were assessed, the former by repeated measurements of 24 biglycan-stained samples, the latter by measurements on eight samples by a second observer. Both intra- and interobserver correlation coefficients equaled 0.95 (p < 0.0001).

Statistical Analysis

To compare resistance, elastance, and hysteresivity in BM and control groups, four-way (group, week, frequency, and amplitude) analysis of variance (ANOVA) with a Bonferroni correction was applied. Subsequently, three-way ANOVA was applied at each week. For comparison of quasistatic elastance, two-way (group, week) ANOVA was performed to compare the quasistatic elastance versus  relationship. To compare morphological data between groups, two-way (group, week) ANOVA with a Bonferroni correction was applied. Finally, to investigate the correlation between mechanical parameters and morphological data in BM and control groups, a Pearson correlation was performed. Significance levels were adjusted for multiple correlations at a probability level of 5%. All values are given as mean ± standard error (SE). p Values < 0.05 were taken as significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Oscillatory Mechanics

Results of dynamic oscillation are shown in Figures 1-3. Resistance demonstrated negative frequency dependence at all amplitudes (p < 0.001), as well as negative amplitude dependence (p < 0.001). E showed positive frequency dependence at all amplitudes and negative amplitude dependence (p < 0.001). Hysteresivity demonstrated negative dependence on frequency and positive dependence on amplitude (p < 0.001). These findings were consistent in both control and BM-treated animals. Within the control group, E decreased with age (p < 0.001).

View larger version (20K): [in this window] [in a new window]   Figure 1.   Resistance (R) as a function of frequency and amplitude ( ) in control (C) and bleomycin (BM)-exposed parenchymal strips at 7, 14, and 28 d postexposure.

View larger version (20K): [in this window] [in a new window]   Figure 2.   Elastance (E) as a function of frequency and amplitude ( ) in control (C) and bleomycin (BM)-exposed parenchymal strips at 7, 14, and 28 d postexposure.

View larger version (20K): [in this window] [in a new window]   Figure 3.   Hysteresivity () as a function of frequency and amplitude ( ) in control (C) and bleomycin (BM)-exposed parenchymal strips at 7, 14, and 28 d postexposure.

Resistance and elastance in the BM group at all time points were greater than control (p < 0.001). Within the BM group, the greatest increases in R and E were seen at 14 d post-BM (p < 0.001). R and E at 7 d were also higher than at 28 d (p < 0.001). Hysteresivity of parenchymal strips in the BM group was significantly less than that in control group (p < 0.001). Within the BM group the decrement in  was greatest at 7 d, as compared with 14 and 28 d (p < 0.005).

Quasistatic - Curve

Mean quasistatic elastance (d/d) versus  curves for parenchymal strips from BM and control animals are shown in Figure 4. Quasistatic E decreased with age in the control group (p < 0.001). Quasistatic elastance in all BM groups was significantly greater than the time-matched control group (p < 0.001). The greatest increase in quasistatic elastance was seen in the 14 d post-BM group (p < 0.001).

View larger version (15K): [in this window] [in a new window]   Figure 4.   Quasistatic elastance (d/d) versus  in lung parenchymal strips from control (C) and bleomycin (BM)-exposed animals at 7, 14, and 28 d postexposure. *p < 0.001 versus C; +p < 0.01 versus BM 7 and 28 d; #p < 0.001 versus other control groups.

Histology and Immunohistochemistry

Photomicrographs of parenchymal strips from control and 28 d post-BM animals are shown in Figure 5. Collagen stained pinkish with Van Gieson picric acid fuscin (Figures 5A and 5B). Note in the BM parenchymal strip the extensive accumulation of collagenous material (Figure 5B). In control tissue, elastic fibers were evident on the blood vessel wall and, to a lesser extent, on the alveolar wall (Figure 5C). Areas of fibrosis showed extensive staining for elastic fibers (Figure 5D). Results of immunohistochemical staining similarly show modest staining for biglycan in tissue from normal animals as compared with tissue from fibrotic animals (Figures 5E and 5F). Although fibromodulin was identified in the bronchial wall of control tissue we were unable to localize fibromodulin in the alveolar wall of control tissues. Conversely, fibromodulin was present in substantial amounts in areas of fibrosis (Figures 5G and 5H).

View larger version (96K): [in this window] [in a new window]   Figure 5.   Photomicrographs of parenchymal strips. (A) Control strip stained for collagen with Van Gieson picric acid fuscin, collagen stains pinkish; (B) BM strip; (C ) control strip stained with Verhoff's elastic stain; arrowheads denote elastic fibers in alveolar walls; (D) BM strip; (E ) immunostaining for biglycan staining in control strip; arrowhead denotes positive staining for biglycan within alveolar wall; (F ) BM strip. Note extensive staining in area of fibrosis; (G) immunostaining for fibromodulin in control strip; (H ) BM strip. All strips were obtained at 28 d postexposure. Original magnification: ×400.

Morphometric Study

The morphometric analyses of collagen, elastic fibres, biglycan, and fibromodulin are shown in Figures 6-9. The volume fraction of collagen on the alveolar wall in BM-treated tissues was significantly greater than the time-matched control only at 28 d post-BM (p < 0.01) (Figure 6). The length of elastic fiber per unit lung volume of alveolar wall (L_VA) was greater than control at 14 and 28 d (p < 0.02 and 0.05, respectively) (Figure 7). L_VA in both the 14- and 28-d post-BM groups was greater than that in the 7-d post-BM group (p < 0.02 and 0.005, respectively). The volume fraction of biglycan on alveolar wall in the BM group was higher than that in the control group at all time points (p < 0.001). Furthermore, BM groups at 14 and 28 d had significantly greater biglycan than BM at 7 d (p < 0.001 and 0.002, respectively) (Figure 8). No fibromodulin was detectable on the alveolar wall in strips from control animals or animals 7 d post-BM. However, at 14 and 28 d post-BM, positive staining for fibromodulin was observed (p < 0.001 at 14 d post-BM versus control and p < 0.002 at 28 d versus control). In addition, FM at 14 and 28 d post-BM was significantly greater than at 7 d post-BM (p < 0.002).

View larger version (23K): [in this window] [in a new window]   Figure 6.   Volume fraction of collagen in alveolar walls for BM and C groups at the three time points. *p < 0.01 versus C group.

View larger version (25K): [in this window] [in a new window]   Figure 7.   The length of elastic fiber per unit lung volume (LVA) in alveolar walls for BM and C groups at the three time points. *p < 0.05 and **p < 0.02 versus C group; +p < 0.02 and ++p < 0.005 versus BM7 group.

View larger version (11K): [in this window] [in a new window]   Figure 8.   Volume fraction of biglycan in alveolar wall for BM and C group at the three time points. *p < 0.001 versus C group, +p < 0.01 versus BM7 group.

View larger version (8K): [in this window] [in a new window]   Figure 9.   Volume fraction of fibromodulin in alveolar wall for BM and C group at the three time points. *p < 0.01 versus C group; +p < 0.002 versus BM7 group.

Correlation between Mechanics and Morphology

Finally, we investigated whether there were correlations between the mechanical parameters, resistance, elastance, and hysteresivity, and the structural components, collagen, elastic fibers, biglycan, and fibromodulin. The only significant correlations were between volume fraction of biglycan and the three mechanical parameters (Figure 10). The volume fraction of biglycan was significantly correlated with R, E, and  (r = 0.87, 0.88, and 0.64, respectively).

View larger version (13K): [in this window] [in a new window]   Figure 10.   Correlation between the mechanical parameters, resistance (R), elastance (E), and hysteresivity (), and volume fraction of biglycan in alveolar wall in control (C) and bleomycin (BM)-exposed parenchymal strips at 7, 14, and 28 d postexposure. Linear regression for all data points is shown.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Bleomycin-induced lung fibrosis resulted in alterations in the viscoelastic properties of the lung parenchymal tissues. These physiological changes were coincident with important changes in the structure of the extracellular matrix. The maximal changes in physiology occurred at a time point prior to significant changes in collagen deposition, but when alterations in elastic fibers and the small proteoglycans, biglycan and fibromodulin, were most significant. Moreover, changes in the mechanical parameters correlated with changes in biglycan, suggesting that proteoglycan molecules are important in contributing to the alterations in tissue viscoelastic behavior observed in this model.

The oscillatory behavior of the parenchymal strips sampled in the current study is similar to that described by us and others previously (15). R, E, and  showed typical frequency and amplitude dependence. These characteristics were also observed in the strips from animals exposed to bleomycin. The major change with the induction of fibrosis was an increase in the values of R and E and a decrement in . These changes were apparent at all time points sampled. Previous data from this laboratory have shown an increase in R and E, both in vivo and in vitro, at 28 d postbleomycin (8). The change in R and E of the parenchymal strip was of similar magnitude in our two studies. Verbeken and colleagues reported on Rti and elastance in autopsy specimens from patients with pulmonary fibrosis. They showed that effective elastance and Rti measured at a frequency of 4 Hz using a complex signal and alveolar capsules to sample alveolar pressure were markedly increased as compared with that of lungs taken from patients with no discernible lung disease. Nava and Rubini (18) measured lung elastance, airway resistance, and tissue viscoelasticity using the interrupter technique in ventilated patients with end-stage pulmonary fibrosis. They found increases in all three mechanical parameters.

One discrepancy exists between the data of the current study and those we previously reported (8). In the previous study,  increased after BM, whereas in the current study there was a decrement noted in the value of . One possible explanation for this difference is that in the previous experiment, parenchymal strips were studied at only one frequency (1 Hz) and one amplitude of oscillation (2.5% L_r) at one time point (28 d post-BM instillation). Moreover, measurements were obtained at a given tension, rather than at a given stress (the latter corrects for cross-sectional area). If there were systematic differences in the size of the tissue strip from bleomycin-exposed animals, this could further complicate the comparison between the two sets of data.

In addition to the increase in E observed during dynamic oscillation, quasistatic elastance was increased in tissue from bleomycin-exposed animals as compared with control. This finding is similar to that previously reported by a number of authors who measured quasistatic elastance in both animal and human disease (3, 19). Of note in the control animals, both E and quasistatic E decreased as the animals aged. This decline in E with maturation has been previously documented both by us and others (20, 21).

The time frame of changes in the mechanical parameters is also of interest. The maximal increase in R and E occurred at 14 d post-BM instillation. Conversely, the decrement in  was greatest at 7 d post-BM. Quasistatic elastance was also most altered at 14 d. Although little information is available in the literature in this model regarding the time frame of changes in the dynamic mechanics, Goldstein and coworkers (3) reported data in animals 8 and 30 d post-BM exposure, showing that, at these two time points, chord compliance was decreased as compared with control, but was not different over time. If we compare our data at the two time points of 7 and 28 d, we show similar results.

We also characterized changes in collagen, elastic fibers, and the small proteoglycans, biglycan and fibromodulin, in response to bleomycin and how these structural components of the extracellular matrix were altered as the disease process developed. As expected, collagen content increased in response to bleomycin. The first time point at which the volume proportion of collagen was increased was at 28 d post-BM. Elastic fibers were increased as early as 14 d post-BM, an increase that persisted at 28 d. These findings are consistent with those previously reported by us and others. In our previous study in BM rats (8), we showed an increase in the volume proportion of both collagen and elastic fibers at 28 d post-BM. Goldstein and coworkers (3) reported on hamsters exposed to BM and showed that collagen content was unchanged 8 d post-BM and then increased by 40% above control levels at Day 30. A similar time frame for changes in elastin content was demonstrated. Thrall and coworkers (22) reported that 2 wk after BM instillation, there was a biochemically detectable increase in total lung collagen. At 4 wk, lung collagen doubled. Starcher and colleagues (2) demonstrated a doubling in collagen and elastin content at Day 30 in this disease model.

More recently, investigators have focused on changes in proteoglycans in this disease process. Proteoglycans are macromolecules that comprise the ground substance of the lung. Three major subclasses of proteoglycans have been described: hyalectins, which are large molecules such as versican that form aggregates with hyaluronic acid; basement membrane proteoglycans such as perlecan; and the small leucine-rich PGs (SLRP), which include lumican, biglycan, fibromodulin, and decorin (9). These molecules have all been identified in the lung, although the relative amounts and distributions vary (4, 10, 23). Bensadoun and coworkers (4, 10) have recently published data on human fibrotic disease, examining the role of versican in both granulomatous and nongranulomatous fibrotic processes. In specimens from patients with diseases such as idiopathic pulmonary fibrosis, bronchiolitis obliterans organizing pneumonia, and sarcoidosis, immunohistochemical analysis showed that there was prominent deposition of versican. In the areas of versican deposition, myofibroblasts stained positively for type I procollagen. Staining for biglycan was also observed in these specimens, particularly in the sarcoidosis specimens. Westergren-Thorssen and coworkers (11) studied the bleomycin-induced fibrosis model in the rat; they showed that biglycan mRNA was increased maximally at 10 d post-BM, and biglycan protein at 14 d post-BM. Fibromodulin message was unaltered. Both decorin message and protein were decreased as compared with normal controls. Our data focused on the SLRPs, biglycan and fibromodulin, both of which were increased in response to BM. Increases in biglycan were evident at 7 d, and a further increase was apparent at 14 and 28 d post-BM, similar to the findings cited above. The volume fraction of fibromodulin increased somewhat later, at 14 and 28 d post-BM. The lack of change in fibromodulin message in the Westergren-Thorrsen study suggests that posttranscriptional modulation is required to explain our observation of increased fibromodulin protein.

We correlated changes in function with changes in structure in this disease model. Of particular note was the change in oscillatory and quasistatic mechanics prior to a measurable change in either volume fraction of collagen or elastic fibers. Moreover, the maximal change in the mechanical parameters occurred prior to any discernible increase in collagen content. In our previous experiment in this model, changes in R and E were significantly correlated with changes in the volume proportion of collagen. However, all measurements were obtained at 28 d, a time point at which collagen had also increased in the current experiment.

We felt it was important to define changes in this system prior to 28 d, as human lung fibrotic disease is usually an ongoing process, rather than a response to a single insult. Understanding the pathophysiology of the process as it evolved over time, therefore, seemed especially pertinent.

The important positive correlations we observed between changes in structure and function were those between biglycan and the three different oscillatory parameters, R, E, and . That the amount of biglycan should correlate so strongly with the mechanical change is perhaps surprising, as biglycan is a small PG with relatively few glycosaminoglycan side chains and few ionic charges. One possible explanation for the effect of biglycan on lung mechanics is its influence on matrix assembly (27). If the matrix is altered, then R and E would likely be affected. Hysteresivity, which reflects the mechanical friction in the tissues, was however, negatively correlated with biglycan. One hypothesis that has recently been advanced is that PGs act as "lubricants," coating the surface of matrix fibers (28). As the amount of lubricant is increased, the ratio of the energy dissipated to that conserved as the fibers slide past one another may be lessened.

We did not correlate changes in fibromodulin with changes in the mechanical parameters, as in control tissues and in tissues from 7 d post-BM animals, fibromodulin was below the level of detection by antibody. Therefore, performing a linear correlation on these data was not appropriate. This does not preclude the possibility that changes in fibromodulin were also important in contributing to altered tissue viscoelasticity.

In this study, we were unable to sample all the PGs in the lung tissue. In particular, it would have been useful to study versican, a large PG with many hydrophilic chondroitin sulfate side chains. In the previous work of Bensadoun and coworkers (4, 10), versican and biglycan were increased in human fibrotic disease. We have reported data in this model showing an increase in versican protein as assessed by Western blotting (29). In this regard, biglycan might be considered as a proxy for the other PGs, which have been shown to be increased in this disease model, and it may be these other PG molecules that are influencing viscoelastic behavior.

PG and glycosaminoglycan (GAG) have been shown to be important in other systems in terms of altering mechanical behavior. Schmidt and coworkers (13) have shown that viscoelasticity of cartilage preparations is influenced by GAG content. In studies conducted on the viscoelasticity of solutions, adding the hyalectin, aggrecan, markedly changed the viscoelastic properties (30). Finally, we have reported preliminary data in lung parenchymal strips showing that degradation of GAG side chains results in altered viscoelastic behavior (31, 32). Hence, one could postulate that, at least in the early phase of the fibrotic response, PGs drive the altered mechanics.

In summary, we have shown that the viscoelastic properties of lung parenchymal tissues from animals exposed to bleomycin are altered early in the fibrotic response. This change occurs before collagen and elastic fiber content of the tissue is increased, but at a time when the molecules that comprise the ground substance, i.e., proteoglycans, are altered. In addition to understanding the contribution of these molecules to changes in the physiology, defining how these molecules are involved mechanistically in the fibrotic response warrants investigation.