Static and dynamic mechanics of the murine lung after intratracheal bleomycin

Pulmonary fibrosis (PF), the progressive destruction of functional lung parenchyma and its conversion to scar tissue, can lead to severe decline in lung function, dyspnoea and death [1‐3]. Among the various methods available to model PF in laboratory animals, the bleomycin model is the most widely used. Systemic or local pulmonary administration of bleomycin elicits an initial inflammatory reaction, followed by fibrosis at later time-points [4‐7]. Despite its extensive use, little is known on the magnitude and time-course of lung functional alterations, which take place in this type of injury, owing partly to the technical difficulty of measuring lung function in mice. In addition to gaining important pathophysiological insight on the effects of collagen deposition in the lung, precise knowledge of lung function would be important in order to quantify the severity of physiological impairment imposed on the animals, as well as to measure treatment responses.

Quantification of lung fibrosis induced by bleomycin is challenging, largely relying on histology and determination of lung collagen content. Both methods, however, suffer from limitations, including user-dependency and stringent technical requirements [4, 6‐12]. Thus, objective assessment of lung dysfunction would be of benefit. Reports of lung function measurements in this model have yielded conflicting results. Using the technically challenging and highly invasive alveolar capsule technique in open-chested rats, Dolhnikoff et al. observed increased lung elastance in bleomycin-exposed animals, which correlated with collagen content of lung tissue sections but not with α-smooth muscle actin or elastin [8]. Conversely, in the study by Goldstein et al., lung collagen or elastin did not correlate with lung mechanics, as assessed by quasi-static Pressure-Volume (PV) curves [13]. Although collagen accumulation and histologic alterations imply the presence of lung dysfunction, the corresponding mechanical impairment of the respiratory system, a major determinant of the severity of the experimental intervention, cannot be precisely quantified, unless some lung function measurement has been conducted.

Highly sensitive airway pressure transducers developed during the past fifteen years have enabled objective lung function determinations on live mice and rats [8, 14‐16]. The forced-oscillation technique (FOT), a dynamic measurement of lung impedance to airflow, is based on the observation that lung elastance and airway resistance vary within the physiological respiratory frequency range [17, 18]. To accurately describe lung mechanical properties, pulmonary visco-elastic behaviour is assessed over a range of respiratory frequencies. Using a mathematical tool known as the constant phase model, lung impedance measurements obtained in the above fashion can be broken down to components attributable to airway resistance R _N and inertance I in addition to lung tissue elastance H and tissue damping or energy dissipation G [17]. With technical advancements, the method became available for mouse studies and is expected to offer the greatest degree of accuracy and specificity, not only since in vivo lung function determinations are carried out under precise experimental control [8, 14, 15], but also since lung mechanical properties are described by a set of four variables encompassing a wide spectrum of respiratory frequencies. Using this methodology, Cohen et al. were able to elucidate previously unrecognized lung dysfunction in cystic fibrosis trans-membrane conductance regulator-deficient (cftr ^-/-) mice, [19] while other groups have used it to quantify lung pathology in asthma [20, 21]. However, the advantage of the technique over the more widely available measurement of lung compliance using quasi-static PV-curves has not been experimentally demonstrated. In addition, changes in lung mechanics after bleomycin delivery to mice are incompletely understood.

The present study addresses lung structure-function relationships in the bleomycin model. To this end, we used two different methodologies to describe the temporal aspects of lung functional derangements and compared these to established biochemical and morphological indices of lung pathology. To induce fibrosis, bleomycin was delivered intratracheally (it) to C57BL/6 mice, which were sacrificed after seven (early time-point; predominance of inflammatory changes) or 21 days (late time-point; predominance of fibrotic changes). Saline-treated C57BL/6 mice as well as Balb/c mice, which are known for their resistance to bleomycin [7, 22‐24], served as negative controls. Lung functional parameters obtained at both time-points were complemented by biochemical and morphological assays, in order to characterize the different phases of the model from a structural and functional perspective.

We undertook a systematic survey of lung mechanical behaviour in the bleomycin model in order to describe the time-course of lung functional alterations and establish relationships between structural and functional parameters of lung injury and fibrosis. We used local pulmonary delivery of bleomycin to sensitive (C57BL/6) and resistant (Balb/c) mice. Negative controls, which received saline, were enrolled in each study arm. Pulmonary function determinations using the forced oscillation technique and PV curve analysis were easy and rapid and did not compromise subsequent biochemical assays. In fibrosis-prone C57BL/6 mice, bleomycin exposure led to an early inflammatory response characterized by a modest increase in lung elastance. The establishment of fibrosis led to a further, substantial increase in lung stiffness. The FOT more specifically detected the presence of fibrosis, while quasi-static PV-curves were more sensitive indicators of lung dysfunction in global terms, yielding pathological values in C57BL/6 mice at the early (inflammation) time-point as well as in Balb/c mice.

Pulmonary fibrosis, regardless of whether idiopathic or occurring in the setting of systemic disease [1], leads to destruction of lung architecture and lung functional decline. Among the various predictors of outcome studied, serial lung function testing is so far the most useful predictor of death [32]. It would thus appear that, for a bleomycin experiment to adequately model human fibrosis, some degree of functional impairment should be present. Yet few studies have investigated functional alterations in response to bleomycin in mice [9, 33‐35] and only a small minority of bleomycin studies report lung function results, possibly due to the required technical expertise and cost. Thus, despite their substantial drawbacks lung histology and collagen content determinations remain the established means of assessing bleomycin-induced dysfunction. Histological measurements are not only subject to observer bias and sampling errors, but also require the application of expensive and/or time-consuming processing, such as morphometry [36]. Collagen measurement, the gold standard for evaluating murine lung fibrosis, does not give spatial information [37] and, so far, has not been clearly shown to correlate with lung mechanics [4, 8, 13]. In addition, the amount of collagen may not be the sole determinant of lung mechanical properties; rather, the spatial organization of collagen fibres and the degree of cross-linkage as well as connections to other matrix components may be critical [38, 39]. Importantly, treatment interventions may have a positive impact on lung dysfunction in the bleomycin model without changing collagen content or histology [9]. Finally, even if collagen deposition and histological alterations consistent with fibrosis can be detected, their functional importance cannot be assessed without lung function testing.

The FOT is gradually emerging as an objective and detailed means of assessing lung function in normal and diseased lungs. Because lung tissue elastance and airway resistance vary with respiratory rate [27], measuring frequency-dependent total lung impedance and breaking it down to its resistive and elastic components may provide physiological information, which would be missed if only static PV-curves were to be determined [40]. By fitting frequency-dependent lung impedance data into the constant-phase model, lung mechanical properties can be described using four parameters: airway resistance R _N , inertance I, lung tissue energy dissipation G, and lung elastance H [14]. This approach has provided valuable insights in the pathophysiology of pulmonary functional impairment in various experimental models. For example, using FOT it was possible to show that the mechanism of exaggerated metacholin response in ovalbumin-challenged mice is primarily due to airway closure rather that smooth muscle contraction [41]. However, the utility of this methodology in the bleomycin model is less clear, since the studies so far have not described the temporal evolution of lung (dys)function in the model, nor have they examined the correlation of individual lung function parameters to morphological and biochemical indices. In addition, despite its theoretical advantages compared to classical PV-curves, which can be obtained with a variety of rodent ventilators, its superiority over the existing method has not been demonstrated experimentally in a manner convincing enough to justify the added cost. To address some of these issues, our study compared both methods to the biochemical and histological fibrosis indices of the bleomycin model.

In the early stages following bleomycin administration, we found, as expected, histological evidence of acute inflammation, which was associated with relatively mild changes in lung function, depending on the methodology used and the strain. In C57BL/6 mice, despite clear trends toward increased H and G by the FOT, statistical significance was not reached, possibly due to limited sample size. When lung mechanics were assessed using PV curves, however, the increase in PV-A was in fact statistically significant, implying that this method may have an advantage over the FOT in terms of sensitivity. At later time-points, inflammation is succeeded by fibrosis in the lungs of bleomycin-susceptible mice. Lung matrix expansion gradually leads to bridging of increasingly large areas of the lung with collagen fibres; beyond a certain threshold, a sharp rise in lung stiffness is thought to occur [38, 39]. In functional terms, this process is expressed as a dramatic rise predominantly in G, H and to a lesser extent in PV-A. The observed correlation of G and H with fibrosis scores indicates that these functional perturbations are likely explained by the presence of matrix remodelling, with collagen deposition as its major feature. Intuitively, the rise in H could be attributed to increased tension in the alveolar walls during mechanical deformation, in conjunction with reduction of lung volume due to fibrosis. In the absence of lung volume measurements, the relative importance of these two factors cannot be quantified. The rise in G is less intuitive. Even though G reportedly represents inhomogeneity in airflow within the peripheral lung and has also been associated with increased peripheral airway resistance, [41] its meaning remains somewhat obscure [42] and G values are less frequently reported. In these studies, the increase in G may represent the tissue inhomogeneity inherent to fibrosis development. The ratio G/H, known as hysteresivity, was aimed at improving on some of the shortcomings of H or G alone. However, in these experiments due to the proportionate increase of G and H, hysteresivity did not change in any of the groups.

Lung mechanics assessment using quasi-static PV-curves yielded, to some extent, similar results with the data obtained by the FOT. PV-A increased significantly at the early time-point in C57/B6 mice and a further increase was recorded at the late time-point. However, this increase was of a lesser magnitude compared to the FOT data. Thus, correlation with fibrosis indices was weak at best but mostly absent. Taken together, these data indicate that the FOT may be more specific when it comes to detecting the presence of fibrosis in the mouse lung, whereas the PV-curve, although sensitive at recording lung dysfunction, may not allow the differentiation between the cause i.e. inflammation/edema vs. fibrosis.

The difference in susceptibility of C57B/6 and Balb/c mice to bleomycin-induced fibrosis is established in the literature [4, 5, 7, 22‐25, 29, 30, 36] and thus Balb/c mice served as internal controls for sensitive C57B/6 mice. Studies comparing responses of these two strains to bleomycin report differences in gene expression [7], BAL cytokine profile [43] and ability for tissue repair [7]. Specifically, fibrosis-prone C57BL6 mice tend to have increased vascular protein leak [7], proteolytic enzyme activity [7], pro-inflammatory cytokine levels (MCP-1, TNF-α, IL-6) [7, 43], TGF-β receptors [43], and apoptosis indices (caspase-8 and-9 activity). With respect to gene expression profile, Pottier et al. conducted a microarray analysis and found a set of 25 genes differentially expressed between the two strains, many of which were related to inflammatory response and tissue repair [7].

Predictably, the Balb/c strain did not develop fibrosis in response to bleomycin and had normal lung function by FOT. This finding may seem to contradict a recently published study reporting increases in H in bleomycin-exposed Balb/c mice [34]. However, in this paper a different protocol was used to induce lung fibrosis, since, in addition to bleomycin, intraperitoneal cyclophosphamide was also administered. Since collagen values are not reported in this study, it is not possible to interpret this discrepancy in findings. A surprising finding in our dataset, however, was that PV-A and slopes were significantly higher in bleomycin-compared to saline groups, indicating that despite lack of evidence for collagen accumulation, subtle mechanical changes take place in this strain as well. Furthermore, even though Balb/c mice developed similar degree of inflammation at the early time point as C57B/6 mice by morphological criteria, this was not reflected in lung function indices, for reasons yet to be determined.

One point of controversy regarding invasive lung function measurements in mice is whether these measurements should be performed in the presence or absence of PEEP. Use of PEEP helps to prevent alveolar collapse and reduction in lung volume during mechanical ventilation. However, we reasoned that this is more likely to occur in the presence of lung damage, including acute lung injury. While in our hands mice with unperturbed lungs tend to maintain stable values for H and G for several minutes, this is not the case with mice subjected to insults injurious to the lung. Therefore, avoiding PEEP may help unmask subtle functional abnormalities, which may be missed if PEEP is applied.

Our findings in bleomycin-treated mice are similar to the ones observed in humans with end-stage fibrotic lung disease. Published data from mechanically ventilated humans with end-stage fibrosis demonstrate that the disease is characterized by extensive scarring and reorganization of lung architecture, resulting in diminished lung volumes [44]. With loss of lung volume, the pressure-volume curve of the lung becomes contracted on its volume axis, thus reducing static and dynamic compliance. In this regard, the findings of the present study are in line with data from human subjects, a fact further validating the intratracheal bleomycin mouse model of lung fibrosis as an analogue to the human disease.

Although in our hands murine lung function measurements proved valuable indicators of bleomycin-induced lung fibrosis, some shortcomings of the present work and the method need to be pointed out. Firstly, procuring the equipment requires a financial investment. Secondly, we did not study lung function in other models of pulmonary fibrosis (eg, using silica or in genetic models), to further expand the possible uses of the method. Thirdly, it may be problematic to adjust tidal volume to body weight in the presence of fibrosis-induced weight loss, but this issue has not been resolved experimentally to our knowledge. Tidal volume in diseased animals may depend on a variety of factors, including functional dead space and carbon dioxide production, which are difficult to quantify in experimental mice. Thus, pending further experimental evidence, adjusting ventilation to body weight may currently be a reasonable option. And finally, since functional residual capacity was not measured in our study, some of the mechanical alterations observed following bleomycin challenge may be related to loss of lung volume, which would be expected in the presence of both acute lung injury and fibrosis.

The present work demonstrates that functional decline in the course of bleomycin injury is time-dependent and associated with the presence of fibrosis to a much greater extent than inflammation. We further show that for a more accurate assessment of lung function in rodent lung pathology models, complementary methodologies may be more informative than individual ones. Therefore, lung function measurements may serve as a valuable adjunct to pulmonary fibrosis research. Recent reports strongly advocate in favour of re-evaluating traditional histologic and biochemical tools for quantifying pulmonary fibrosis, since human trials for pulmonary fibrosis mostly use lung function parameters [1, 25]. Based on our results, oscillatory measures of elastance and tissue damping are sensitive and reliable surrogates of fibrotic injury following bleomycin at later time-points, when fibrotic remodeling has had sufficient time to develop. These parameters reflect whole-lung visco-elastic behaviour and could serve as novel, user-independent end-points of lung fibrosis in future studies using this model.

- Lung damage due to intratracheal administration of bleomycin can be divided into two stages, the inflammation stage and the fibrosis stage, with distinct functional profiles, at least when the forced oscillation technique is used.

- Lung elastance measured by the forced oscillation technique strongly correlates with soluble collagen content in lungs and histological indices of lung pathology.

- Quasi-static Pressure-Volume curves of the respiratory system sensitively detect lung dysfunction in the bleomycin model but seem to be less specific for fibrosis.