INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: lung; remodeling; stress-strain; fluorescent labeling; volume reduction

Chronic obstructive pulmonary disease (COPD) is a leading cause of respiratory disability and death worldwide (1). Emphysema, one form of COPD, is characterized by permanent destruction of alveolar walls leading to airspace enlargement, loss of elastic recoil, decrease in surface area for gas exchange, lung hyperexpansion, and increased work of breathing (2, 3). The most accepted hypothesis of how emphysema develops is that an imbalance between protease and antiprotease activity exists within the lung leading to the degradation of elastin, a major component of the connective tissue fiber network (2).

Several animal models have been used to study the pathophysiology of emphysema (3). Intratracheal injection of pancreatic elastase is used to generate one such model, and the treatments produce pulmonary dysfunction similar to human emphysema (5). Biochemical and histological studies suggest that both the organization and the amount of connective tissue fibers are important determinants of the loss of elasticity in emphysema (6). Although neutrophils are commonly considered the primary source of elastase, experiments have shown that in mice, the presence of macrophage elastase is sufficient for the development of emphysema following chronic inhalation of cigarette smoke (7). Moreover, elastase per se does not appear to be essential for the pathogenesis of this disease. Animal studies have shown that lesions consistent with emphysema can also be induced by a variety of pathways that do not involve elastin breakdown at all. For example, one can induce parenchymal destruction indistinguishable from changes associated with smoking-related emphysema by genetic manipulations that result in increased interstitial collagenase activity (8), or by biochemical inhibition of proteoglycan synthesis (9). Hence direct alterations in the elastin fiber network do not represent the only pathway through which emphysema can develop. The common link between these different pathogenic processes and the development of morphological changes characteristic of emphysema is not well understood.

Insight regarding this important question is now emerging by considering outcomes among patients who have undergone lung volume reduction surgery (LVRS), a novel treatment for end-stage emphysema that involves resection and resizing the hyperexpanded lung to the chest wall (10). This therapy produces a rapid and frequently large increase in transpulmonary pressure (11). Although the majority of patients benefit transiently from LVRS, all lose lung function over time at a rate that exceeds that existing preoperatively (12). Because the remaining lung of these patients is stretched to fill the thoracic cavity, mechanical forces on the connective tissues are increased. The idea that mechanical forces are involved in the progression of certain pulmonary diseases was elegantly put forth by West as early as 1971 (13). He argued that because the uneven distribution of mechanical stresses in the lung is closely related to the distribution of disease, mechanical failure contributes to the development of centrilobular emphysema.

Based on these observations, we formulate the hypothesis that following the onset and initial progression of emphysema due to proteolytic injury, a critical point is reached at which mechanical forces generated during normal breathing become sufficient to gradually damage and promote stress failure of the remodeled alveolar walls leading to progressive morphological changes and lung dysfunction as seen in emphysema. To test these hypotheses, we have developed a novel technique to simultaneously measure the mechanical properties of thin, structurally intact samples and image the matrix components of the tissue during macroscopic deformation mimicking breathing. This technique was then applied to tissue strips obtained from normal and elastase-treated rat lungs.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal Preparation

Two groups of male Sprague-Dawley rats (Charles River Laboratories, Boston MA, weighing 280-350 g) were studied (protocol approved by Boston University and Harvard Medical School Animal Care and Use Committee). Animals were anesthetized (intraperitoneal xylazine 15 mg/kg and ketamine 30 mg/kg) and via an endotracheal tube were treated with either saline (n = 5) or porcine pancreatic elastase (PPE) (3 IU, Sigma Chemicals, St. Louis, MO) (n = 5). Four weeks later, the animals were anesthetized and tracheostomized, and the tracheal cannula (2 mm i.d.) was connected to a computer-controlled ventilator (Flexivent, SCIREQ, Canada). The static pressure-volume curve of the respiratory system was obtained during inflation from functional residual capacity (FRC). The animals were exsanguinated and the lung was perfused with saline and heparin (1000 IU/L) and removed for further processing.

Experimental Procedures

Thin tissue slices (0.3 × 4 × 5 mm) were prepared while preserving their structural and mechanical integrity. Freezing and cutting were not possible as thawing changes mechanics (14). The lung was infused with 2% agarose solution at 55° C and allowed to gel at room temperature. This temporarily fixed the tissue and allowed slicing with a vibratome. The samples were washed at 55° C to remove the agarose. To ensure that this did not alter the mechanical properties, we used larger pieces (4 × 4 × 10 mm) that could be sliced with or without agarose. The stress-strain curve of the agarose-treated strip was nearly identical to that of the agarose-free tissue (15).

The set-up consisted of a servo-controlled lever arm (Series 308B, Cambridge Technologies, MA) and a force transducer (Model 403, Aurora Scientific, Canada) attached to an acrylic base. The sampleglued to metal plates that were attached to the lever arm and the force transducer with steel wireswas suspended horizontally in Krebs solution. The apparatus was fit atop an inverted fluorescent microscope (Zeiss Axiovert 100) with the sample above the objective. The lever arm stretched the sample at a rate of 1%/s, displacement and force were sampled, and images were captured using a CCD camera. From the dimensions of the sample, the stress and uniaxial strain () were calculated.

Immunohistochemical Labeling

The fibers were immunohistochemically labeled. The manufacturer's protocol was optimized to achieve acceptable image quality without compromising the mechanical properties of the sample as judged from comparisons of the stress-strain curves with and without labeling. Anticollagen Type I (Sigma Chemical, St. Louis, MO) was diluted 1:2000 with phosphate-buffered saline (PBS). Antielastin (Sigma Chemical, St. Louis, MO) was diluted 1:5000 with PBS. The samples were immersed into 800 µl of diluted anticollagen or antielastin antibodies and incubated at 37° C for 1 h. The samples were washed three times in 1 ml of PBS at 20 min intervals, immersed in 800 µl of anti-mouse IgG FITC conjugate (Sigma Chemical, St. Louis, MO) at a concentration of 100 µg/ml, and then incubated at 37° C for 1 h. The samples were washed three times in 1 ml of PBS at 20 min intervals.

Statistical Analysis

Features including diameters and aspect ratios were manually measured as a function of strain from the images. The data were grouped according to animals and two-way analysis of variance (ANOVA) was applied to determine the effects of strain and condition (elastin, collagen, normal, or emphysematous). Fiber thickness at zero strain was also measured and two-way ANOVA was used to determine the effects of condition on fiber diameter. Differences between various groups were determined using the Student-Neuman-Keuls test.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Example stress-strain curves for normal and elastase-treated tissues are plotted in Figure 1a. The stress-strain curve of the normal tissue displayed a nonlinear behavior, whereas that of the treated tissue was nearly linear and shifted to the right. This suggests a considerably reduced elasticity at the level of the connective tissue structure similar to the pressure-volume curve that characterizes lung function in emphysema (Figure 1b).

View larger version (17K): [in this window] [in a new window]   Figure 1.   (a) Examples of stress-strain curves for normal and emphysematous tissue strips obtained by slow stretching of tissue sections and measuring the force. (b) Examples of pressure-volume curves of the respiratory system in a rat before and after treatment with PPE.

To examine the origin of this behavior, we visualized how the fibers in the matrix stretch under increasing uniaxial strains () along the stress-strain curve of the sample under an inverted fluorescent microscope. At  = 0%, both the elastin and collagen fibers in the normal tissue appeared as a network of hexagonal-like structures (Figure 2a and 2c, respectively). There was considerable heterogeneity in the size and shape of the unit cells of both networks. The cell diameters ranged from 40 to 100 µm suggesting that the fibers contoured the folded alveoli and some alveolar ducts in the tissue sample. When the tissue was stretched to  = 40%, both the elastin and collagen networks became elongated with an ellipsoidal shape of the unit cells (Figure 2b and 2d, respectively).

View larger version (56K): [in this window] [in a new window]   Figure 2.   Examples of micrographs from a normal elastin fiber network at 0% (a) and 40% (b) strain, a normal collagen fiber network at 0% (c) and 40% (d) strain, an emphysematous elastin fiber network at 0% (e) and 40% (f) strain, and an emphysematous collagen fiber network at 0% ( g) and 40% strain (h). Bar = 100 µm. Direction of macroscopic strain is horizontal.

There were noticeable visual differences in the unstrained elastin structure between the normal (Figure 2a) and the elastase-treated tissue (Figure 2e). In general, images from the emphysematous tissue often showed an incomplete fiber network, which is consistent with tissue destruction. When the elastin network from the treated tissue was stretched to  = 40% (Figure 2f), the unit cells also became ellipsoidal but on a scale larger than in normal tissue. The example in Figure 2g shows a collagen network from the treated tissue at  = 0% that was quite similar to the normal tissue (Figure 2c) in accord with the notion that it is the destruction of elastin that leads to the characteristic lesions seen in emphysema (3). Surprisingly, however, the stretching of the collagen network from the elastase-treated tissue led to alveolar unit organization and deformation grossly different (Figure 2h) than in the normal tissue (Figure 2d). These images suggest that at  = 40%, the collagen network from the elastase-treated tissue was more like the elastin network from the same lung. Over 2,600 such structural units were identified from the images and various features of the structures were used in statistical analysis as follows.

Independent of fiber type (collagen or elastin), enzymatic exposure used to produce emphysema caused a highly significant (p < 10⁹) increase in fiber thickness (average diameter of fibers in the focal plane) in the unstressed tissues. Individual comparisons of the groups showed that collagen fibers had a thickness 24% higher (p < 0.05) in tissues exposed to elastase (8.7 ± 3.4 µm) compared with normal tissue (7.0 ± 1.4 µm). Elastin fiber thickness was also significantly higher (p < 0.05) in elastase-treated lungs (8.2 ± 2.1 µm) by 22% than in the normal tissue (6.7 ± 1.5 µm). Independent of condition (treated or normal tissue), the fiber thickness of collagen was not different from that of elastin. These results indicate that elastase treatment led to a noticeable remodeling of both elastin and collagen.

During stretching, the fiber networks reoriented in alignment with the macroscopic stress field and the unit cells became elliptical in shape (Figure 2b, 2d, 2f, and 2h). To determine whether these distortions were characteristic of emphysema, we measured the aspect ratio (Ar) of the unit cells, defined as the ratio of the cells' diameter in the direction of the macroscopic strain to that perpendicular to the strain (Figure 3). Thus, Ar as a function of strain is a measure of how the structure changes shape or distorts during uniaxial stretching. Independent of whether the tissue was from treated or normal lung, the effect of strain on Ar was highly significant (p < 10¹⁶; two-way ANOVA). Independent of strain, the effect of condition (collagen, elastin, elastase, or normal) was also highly significant (p < 0.0005). Individually, Ar of collagen was different (p < 0.05) from the Ar of elastin in the normal tissue but not in the elastase-treated tissue. The Ar of normal elastin was smaller than the Ar of elastin from treated tissue (p < 0.05). The standard deviation (SD) of Ar estimated as the mean of the SD values obtained from the individual animals did not depend on strain, however, it was significantly higher (p < 0.05) in the treated tissue for both elastin and collagen. To better understand the dependence of Ar on strain, we calculated linear regressions between strain and the means of Ar at a given strain (Table 1). It can be seen that the intercepts differ by less than 18%. In agreement with the ANOVA test results, there was a significant difference between the intercepts of the normal and treated elastin (p < 0.02) and the normal elastin and collagen (p < 0.02). The slopes for the normal collagen and elastin were similar. Following treatment, the slope increased for collagen and decreased for elastin, but these changes were not significant.

View larger version (50K): [in this window] [in a new window]   Figure 3.   Bar chart of the mean and standard deviation (SD) of aspect ratios of the unit cells of the elastin fiber networks from normal (EN) and emphysematous tissue (EE) and collagen fiber networks from normal (CN) and emphysematous tissue (CE) plotted as a function of macroscopic strain. The means and SD values represent the mean of the five average aspect ratios and SD values from the five individual animals.

To further examine the micromechanical basis for the changes in organ level function, we compared how the diameters of the unit cells in the direction of strain changed with stretching (Figure 4). Similar to Ar, the effect of strain on diameter was highly significant independent of whether the sample was from treated or normal tissue (p < 10⁸; two-way ANOVA). Independent of strain, the effect of condition (collagen, elastin, elastase treatment, or normal) was also highly significant (p < 10⁷). Independent of strain, the diameters of the unit cells in the treated tissue were significantly higher than in the normal tissue for both elastin and collagen (p < 0.05). The diameters of the elastin and collagen networks were not different independent of whether the tissue was from normal or elastase-treated lung. The SD of diameters was significantly larger in the treated animals than in the normal animals for both elastin and collagen (p < 0.0001). The linear regression results between the average diameter (d_m) of the networks and strain are included in Table 1. The intercepts in the normal tissue were around 72 and 67 µm for collagen and elastin, respectively, and they increased to 80 and 78 µm, respectively, in the treated tissue. The slopes describing the rate of change of diameter with strain were 77 and 79 µm for normal collagen and elastin, respectively, which increased to 123 and 109 µm, respectively, in the treated tissue. The increases in intercept and slope following treatment were close to the significant level (p = 0.12 for collagen and p = 0.055 for elastin), but they did not reach it.

View larger version (44K): [in this window] [in a new window]   Figure 4.   Bar chart of the mean and standard deviation (SD) of diameters in the direction of macroscopic strain of the unit cells of the elastin fiber networks from normal (EN) and emphysematous tissue (EE) and collagen fiber networks from normal (CN) and emphysematous tissue (CE) plotted as a function of macroscopic strain. Means and SD values are calculated as in Figure 3.

To interpret the slopes and intercepts in Table 1, we note that it is expected that d_m follows a linear relationship with  as the sum of all diameters in the direction of strain must be equal to the length of the sample. Specifically, it is easy to show that for a contiguous set of objects having random diameters in the direction of strain:

(1)

where d_m() and d_m(0) are the average diameters at strains  and 0, respectively. Thus, the slope (a) and intercept (b) of the linear regression between d_m and  should have similar numerical values. Table 1 shows that applying this simple continuity relationship, there was good agreement for both the normal elastin (a = 79, b = 67) and normal collagen (a = 77, b = 72). Indeed, the differences between the slopes and intercepts were not statistically different. In the emphysematous tissue, however, there was considerable and significant discrepancy between the slopes and intercepts for both elastin (a = 109, b = 78; p < 0.01) and collagen (a = 123, b = 80; p < 0.05). Equation 1 was developed based on pure geometric considerations describing a contiguous set of objects. Thus, the apparent failure of Equation 1 to correspond to the data following treatment is most likely due to a failure of the connectivity of the fiber network. To test this possibility, we imaged the same region of the emphysematous collagen network from  = 0% to 30%. Figure 5 shows images at  = 10% and  = 20%, respectively. The arrow labels an anatomic landmark, the same bifurcation, at each strain. Notice that the fibers that are clearly seen at  = 10%, are no longer apparent at  = 20%. This sample failed at approximately  = 30%, that is, it separated into two pieces. We sometimes observed such failure in emphysematous tissue but not in normal tissue. Thus, mechanical forces caused the collagen fibers and consequently the alveolar wall from the emphysematous tissue to break. Due to the lack of anchoring, these fibers moved out of the focal plane rendering them no longer visible. The fact that fibers can break and disappear from the images causes two unit cells to merge as one, giving rise to the failure of the diameters to conform to Equation 1.

View larger version (50K): [in this window] [in a new window]   Figure 5.   Collagen network from emphysematous tissue at 10% (a) and 20% (b) macroscopic strains. Notice the fibers connected to the branching point in (a) marked by the arrow. In (b), these fibers are no longer in focus, indicating breaking of fibers.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main findings of this study are that in an elastase-induced rat model of emphysema, the micromechanical properties of the connective tissue are considerably influenced by the progression of emphysema. In particular, we found morphological evidence of remodeling of the collagen and elastin network and that these networks exhibited much larger deformation and distortion during macroscopic stretching compared with normal tissue. Most importantly, however, in addition to the current view that the process of eliminating the alveolar wall is biochemical, whereby the tissue is slowly degraded by the activity of enzymes, our observations provide direct evidence that mechanical forces alone are sufficient to cause tissue destruction and contribute to the progression of emphysema. Before interpreting these results, we first discuss the relevance of our animal model to human emphysema and the limitation of the experiments.

Limitations of the Study

Numerous animal models have been used to test various hypotheses about the development and progression of emphysema. These models, which have included chemical treatments to digest elastin (3), cigarette exposure to simulate the human condition (7, 16), and genetic manipulations (8), all share the critical physiological properties of emphysema such as enlarged air spaces and loss of elastic recoil. In this study we used a rat model of PPE-induced emphysema. The primary difference between the pathophysiology of this model and human emphysema is that PPE causes an acute inflammation in the tissue as opposed to a more chronic inflammation in human patients. The immunological response of the animal and the topological distribution of the disease may also be different from those observed in human patients. Thus, this model is not able to reproduce certain features of human emphysema. However, we believe that the major findings at the fiber level still have important implications for the human disease, and are consistent with observations in patients with emphysema. In this study, we did not aim to reveal the molecular mechanisms of how a given treatment triggers inflammation and how it induces injury. Instead, our goal was to provide a possible link between tissue injury and the end result of emphysema, progressive tissue destruction, which is invariably characterized by an increase in mean linear intercept on histological slides. Although the different emphysema models all lead to similar pathology (increased mean linear intercept), the actual process of how the alveolar wall is destroyed is unknown. To shed light on this process, we visualized the deformation of the major load-bearing components of the alveolar wall and studied how their micromechanical behavior changes in a model of emphysema. To the extent that changes in micromechanics of the alveolar wall lead to fiber weakening, stress failure, and produce changes in mean linear intercept, our results can help us better understand the persistent progression of emphysema in human patients.

The technique introduced in this study allows us to visualize the elastin or collagen networks of the parenchyma using immunohistochemical labeling and to simultaneously measure the macroscopic stress-strain curve of the sample. Because the samples were processed in vitro, we paid careful attention to ensure that the mechanical and structural properties of the samples are similar to those of the lung in situ. At every step of the processing, we analyzed the stress-strain curve of the sample. For example, freezing that facilitates slicing but damages tissue ultrastructure was replaced by inflating the lobe with agarose. Also, drying the samples before adding antibodies, which changes the stress-strain properties of the sample, was left out of the labeling procedure. These resulted in a somewhat compromised quality of the images, but the stress- strain curves were nearly identical to those obtained in fresh tissues without the use of agarose or drying the sample. To ensure that the networks studied were indeed elastin or collagen fibers, negative controls were also done in which only the fluorescent secondary antibodies were added to the samples. In these experiments no fluorescent signal was seen on the microscopic images. Following the slicing and labeling procedures, the samples were not viable. However, in an earlier study we found that the mechanical properties of lung tissue strips are largely determined by the connective tissue matrix and not by the viability or contractility of the cells in the matrix (17). Because the samples were in a tissue bath, surface tension effects were excluded.

The fiber networks seen in our images correspond to elastin or collagen running in the alveolar walls. The average air space size of the rat lung at a distending pressure of 20 cm H₂O as measured by the mean linear intercept on histological slides is approximately 100 µm (18). The average diameters of the hexagonal-like unit cells in our images from normal tissue at  = 0% were 72 and 67 µm for collagen and elastin, respectively. These numbers are much larger than the interfiber distances within the alveolar wall and hence the fluorescent images likely constitute dense fiber bundles contouring an entire alveolus. To better understand these images, we carried out numerical simulations mimicking the sectioning of the alveoli with a microscope as follows. We sectioned a sphere serially along its diameter and recorded the diameters of the circles. The distribution of these diameters is the same as the amplitude distribution of a sine wave that has a U-like shape. The measured distribution of the diameters of the hexagonal-like cells in our images at  = 0% is, however, very different and has a shape similar to a log normal distribution. Next, we simulated the heterogeneity of the diameter of the alveoli by sampling 100 spheres as above whose diameters had a coefficient of variation (CV) of 40%. The corresponding distribution of the diameters of the cross sections was skewed, closely mimicking the distribution of diameters obtained from the images. This suggests that the observed variability of the diameters at  = 0% is partly due to sectioning the alveoli with the microscope at varying distances from their geometric center. The analysis also suggests that the true CV of the alveolar diameter in the lung may be as large as 40%. With increasing stretch it is also possible that the alveolar walls fold into the direction of macroscopic strain and diameters of the smaller structures actually represent interfiber distances within the alveolar wall. Nevertheless, the absolute size of the network pores is not as important as the way the networks stretch and undergo distortions and structural changes during macroscopic deformation of the samples.

Macro- and Micromechanics of the Lung Tissue

The macroscopic pressure-volume behavior of these lungs indicated a considerable reduction in lung elasticity (Figure 1) similar to what has been reported in patients with emphysema (19). The stress-strain curve of our tissue samples was qualitatively similar to the pressure-volume curve (Figure 1). To relate the stress-strain curve to strains on individual fibers in vivo, we note that the lung at FRC is in a prestressed state. Assuming that uniaxial strain varies as the cube root of lung volume, and that 40% of the total lung volume at FRC is tissue (20), we estimate that breathing including sighs corresponds approximately to uniaxial strains between 25% and 60%. These numbers are close to those (20% and 67%, respectively) estimated by Sata and coworkers (21). Thus, our images were obtained under conditions in which the strains on the elastic fibers during uniaxial testing were similar to the strains on the fibers that occur during normal breathing.

The microscopic consequences of the treatment included significantly increased fiber width and altered micromechanics. Despite the large variability of fiber thickness, the difference between fiber thickness in the normal and emphysematous tissue was significant. Because the fluorescent signal was likely coming from dense fiber bundles within the alveolar wall, the increased thickness indicates that remodeling of both the elastin and collagen networks took place in these animals in agreement with previous studies (16, 22, 23).

The aspect ratios, Ar, for both elastin and collagen significantly changed in the emphysematous tissue. To interpret the strain dependence of Ar, we first need to consider the deformation of a unit hexagonal cell. Hexagonal structures are inherently unstable. The reason is that the hexagon collapses under uniaxial stretching or shear deformation because the structure cannot resist deformation as it can stretch by collapsing the angles between two joining line elements (24). Because our images provide evidence that the structures stretch but do not collapse under uniaxial deformation, there must be some mechanism that stabilizes the structure. In the inflated lung the prestress on the fibers can stabilize the structures (24). However, the stress-strain curves in our samples were measured starting from the unstressed state. We have recently reported that at low strains, proteoglycans in the interstitium at the junction of the alveolar walls can serve to stabilize the elastin and collagen networks (25). The elastic resistance of the network has therefore two components: the tensile stiffness of the fibers and the compressive stiffness of the proteoglycan matrix related to its ability to impede rotation of the fibers relative to each other at their junction. Thus, the aspect ratio at a given strain is determined by the relative magnitudes of the fiber and matrix stiffness. If the fibers are much stiffer than the matrix then the hexagon stretches by changing the angles, which results in a high Ar. Because elastin has a Young's modulus about two orders of magnitude smaller than collagen (26), one would expect that the Ar as a function of strain is different for the collagen and elastin networks. Although the intercepts between the normal and emphysematous elastin and between the normal elastin and collagen were significant, the slopes characterizing how the networks stretch were not different (Table 1). As noted already above, this suggests that the labeling visualizes nearly the entire alveolar wall instead of individual fibers within the wall. Because our data suggest morphological evidence of collagen remodeling, it is likely that collagen and elastin stiffness was also altered. Thus, the fact that the slopes did not change significantly with treatment may imply that the stiffness of the proteoglycans was also affected by the repair process in emphysema (27).

Morphologic Evidence of Stress Failure of the Alveolar Wall

The observation in Figure 5 that the remodeled collagen breaks under the influence of mechanical force is critical to understanding the possible role of elastin and collagen in the breakdown of the alveolar wall and hence in the development of emphysematous lesions. The breakdown is also supported by statistical analysis. Equation 1 was developed based on pure geometric considerations describing a contiguous set of objects. Although the data come from a three-dimensional network, this cannot explain the discrepancy between the predictions of Equation 1 and the results in Table 1 that the slopes and intercepts of the diameter-strain relationship are significantly different in the treated tissue. Indeed, numerical simulations suggest that a two-dimensional network also obeys Equation 1 during stretching (25). Thus, the apparent failure of Equation 1 to correspond to the data in Table 1 is due to a true failure of the connectivity of the fiber network. In other words, fibers in the alveolar walls as well as the walls of the alveoli break and two or more alveoli form a larger cluster, which then appears in our images as those shown in Figure 2f and 2h. These findings therefore indicate that some of the alveolar walls ruptured during mechanical testing.

West (13) proposed that tissue failure due to mechanical stresses caused by the weight of the lung may contribute to the development of certain diseases including centrilobular emphysema. In addition, the alveolar wall tissue is continuously undergoing cyclic straining during normal respiration. Moreover, quiet breathing is regularly interrupted by deep sighs in which lung volume at least doubles and forces on the tissue significantly increase. It is natural to assume that following a chemical injury to the fibers and the matrix in the alveolar wall, the fatigue and failure thresholds are reduced. Our data suggest that following elastase exposure, the connectivity of the fibers in the alveolar wall becomes weak and mechanical forces during breathing can rupture the major load-bearing elements. In human disease, it is perhaps more appropriate to think of this process as mechanically induced, but specifically made possible by preexisting alterations to the fiber network due to injury.

If mechanical failure is important in the progression of emphysema then the next question is: What is the process by which the alveolar wall itself fails? The primary load-bearing elements in the alveolar wall are the elastin and the collagen. Compared with elastin, normal collagen is very strong and protects the alveoli from rupture at high lung volumes. The cells, proteoglycans, and elastin are attached to the collagen fiber that reinforces the alveolar wall (28). Using in vitro digestion of elastin and collagen, we have found that collagen contributes to lung tissue elasticity at lung volumes as low as FRC (29). Thus, even if elastin is damaged, the alveolar wall cannot possibly rupture during breathing unless collagen is weakened to an extent that it, too, is prone to mechanical failure. The images in Figure 5 provide direct visual confirmation of this possibility. This means that the failure threshold of collagen fibers must have been reduced, likely due to the remodeling process. It is important to note that the strain range over which collagen failure was observed is within the range of strains that the fibers in the intact lung would experience during normal breathing.

Possible Link between Injury and Collagen Failure

Based on our results, we can now speculate how the fibers become weak and prone to failure, which in turn allows us to propose a possible link between proteolytic injury and tissue destruction in this emphysema model. Following the first exposure to elastase, degradation of elastin in the tissue is initiated. This alone does not appear to lead to a destruction of the alveolar walls at stresses that occur during normal breathing. The link between primary injury and tissue destruction may be related to other structural elements of the connective tissue matrix. First, it is known that there are several matrix metalloproteinases (MMP) that are capable of degrading type I collagen (30, 31). Because alveolar macrophages are certainly involved in at least the early phase of emphysema (32), it is important to realize that alveolar macrophage-derived MMP, interstitial collagenase (MMP-1), neutrophil collagenase (MMP-8), as well as collagenase-3 (MMP-13) can all cleave fibrillar collagen helices. Most notably, however, a recent study suggests that even neutrophil elastase is capable of slowly degrading type I collagen (33). Thus, the presence of various interstitial collagenases or other enzymes may directly weaken or cleave the collagen fibers. Second, it is known that elastolytic injury immediately triggers collagen and elastin remodeling (22, 23), due most likely to the involvement of proinflammatory cytokines (34, 35). Collagen assembly is influenced by the composition of the extracellular matrix (28, 36). Because elastin (22) and matrix proteoglycans (27) are altered in emphysema, extracellular collagen assembly becomes abnormal producing thicker but weak fibers that are prone to failure. Because the emphysematous collagen network in Figure 2 appears as a continuous network in the unstressed state and the statistics on the fiber width suggest that there was significant remodeling, it is likely that most of the remodeled fibers were not cleaved. Instead, Figure 5 provides evidence that mechanical forces can separate the fibers. Thus, we suggest that during normal breathing, cyclic mechanical forces are sufficient to cause fatigue failure of the remodeled collagen fibers, which eventually break at foci of stress concentrations. Failure of collagen eventually leads to the failure and destruction of the alveolar wall. The failure of one fiber can lead to the failure of neighboring fibers, thus propagating the defect in the fiber network. This cascade of coalescence of small bullae grows until the defects increase in size to reach the macroscopic level. The end result of such a process is readily observable on CT images of patients with COPD (37).

Finally, we note that there is in fact evidence to support the notion that collagen breakdown and/or remodeling does occur in patients with COPD. In human emphysematous tissue, the mean linear intercept directly correlated with the increase of septal wall thickness, which was accompanied by significant increases in both elastin and collagen content in the alveolar tissue (23). Stone and coworkers (38) assayed for urinary desmosine (DES) and hydroxylysylpyridinoline (HP), specific markers for the degradation of crosslinked elastin and collagen, respectively, both in normal subjects and in subjects with COPD. They found that DES and soft tissue-related HP levels were elevated in smokers.

Summary

Our results are the first to show direct evidence that mechanical forces are capable of causing failure of the remodeled collagen network and so contribute to the progression of emphysema. Thus, a key element of emphysema is not elastic fiber destruction per se, but rather extracellular repair of which elastic fiber damage is only one type. For this repair process to result in emphysema, enzymatic alterations in the collagen fiber network must also occur, which can lead to a weak fiber network prone to mechanical failure and hence the development of the emphysema phenotype. Future investigations of the molecular interactions within fibers and among the constituents of the alveolar wall will be essential for understanding the basis for tissue destruction in emphysema.