Nanoscale structural alteration of lung collagen in response to strain and bleomycin injury

Abstract

The link between the structural organization of the fibrillar components of lung extracellular matrix (ECM), local tissue stiffness and global viscoelastic behaviour is not known. Here we investigated the effect of injurious mechanical ventilation on the local lung tissue stiffness using 4D synchrotron phase-contrast micro-CT, in normal lung and 7 days after intratracheal bleomycin induced lung injury in anesthetized rats. Quantitative maps of local lung strain (ε) were computed within aerated lung acini, using a stepwise image registration method. Fibrillar organization of collagen and elastin at the nanoscale was measured using synchrotron small-angle x-ray scattering (SAXS). Local microscopic tissue ε was reduced in the aerated acini of normal lungs post injurious ventilation and in bleomycin-injured lungs and was associated with an increase in dynamic elastance (H). The scattering peak angle (q) which is inversely related to fibril D-spacing, was decreased by injurious ventilation indicating an elongation of the collagen fibril spacing in both normal and bleomycin-injured lung. There was a positive relationship between collagen periodicity and global tissue elastance, while an inverse relation was observed with tissue hysteresis. Our data demonstrate the effect of both bleomycin-induced lung injury and high-strain mechanical ventilation on the nanoscale fibrillar organization of collagen and for the first time, a link between collagen D-spacing and global lung tissue stiffening and viscoelastic behaviour.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-07218-9.

Introduction

Acute respiratory distress syndrome (ARDS) is characterized by heterogeneous mechanical abnormalities in the lungs¹. While mechanical ventilation remains the cornerstone of therapy for moderate to severe ARDS cases, it carries the risk of exacerbating lung injury by subjecting mechanically non-uniform lung tissue to excessive stress and strain. Exaggerated stretch of the lung tissue, through hyperinflation², cyclic alveolar recruitment and derecruitment or “atalectrauma”³, increased energy dissipation within the parenchyma⁴, and associated inflammation⁵ contribute to the progression of injury in ARDS⁶. The introduction of protective mechanical ventilation has allowed a significant reduction in mortality. Indeed, current protocols aim at reducing tidal volume and preventing atelectrauma through positive end-expiratory pressure (PEEP). Prone positioning, extracorporeal membrane oxygenation, and the judicial use of intravenous fluids⁷ are additional measures that allow improving clinical outcome in some patient categories. Despite all of these measures, the overall mortality of ARDS remains high⁸. In its later phase, repair of the alveolar epithelium is regulated by crosstalk between multiple alveolar cell types and the extracellular matrix. During this phase, fibroproliferation promoted by immune cells and inflammatory mediators can lead to fibrosis through excessive deposition of collagen in surviving ARDS patients⁶.

The lung extracellular matrix (ECM) provides the cells with structural support, but also transmits external physical forces to cells⁹. Local strain can trigger downstream cellular signalling cascades through mechanotransduction that modifies cellular structure and function, involving numerous processes including cell contractility and myosin light chain phosphorylation, cell rheology, focal adhesion assembly, angiogenesis and ECM remodelling and homeostasis⁹. However, data on the local microscopic deformation of the peripheral airspaces by mechanical ventilation are very limited despite their importance for understanding the mechanisms occurring at the onset of ventilator-induced lung injury (VILI). This is due to the technical difficulty of imaging deep lung tissue structures at microscopic resolution in vivo.

The lung tissue has dynamic viscoelastic properties. Unlike elastic materials that deform instantaneously in response to stress loading like many biological tissues, the response of the lung parenchyma to stress is delayed, as it is impeded by internal frictional forces¹⁰. These frictional forces are determined by the constituents of the lung ECM including collagen and elastin and a ground substance composed of proteoglycans (PGs) and glycosaminoglycans (GAGs)¹¹. The viscoelastic properties of the lung tissue are also influenced by adherent cells, the surface tension of the air-liquid interface and the organization of the parenchyma⁹. The relationship between the strain-induced changes in molecular organization of the main fibrillar constituents of the ECM, the collagen and elastin fibres, and the in vivo lung tissue mechanical properties is not known. Such a mechanistic link would be valuable to better understand the pathophysiology of VILI, but also of other lung diseases affecting the ECM such as emphysema and fibrosis⁹.

We previously developed a 4D synchrotron phase-contrast micro-CT (4DCT) technique using synchrotron radiation with gating to cardiac and respiratory motion¹² (Fig. 1A). The high intensity of synchrotron radiation allows improving both the spatial and temporal resolution, enabling dynamic in vivo microscopy in intact lungs. The high degree of coherence of this radiation allows phase-contrast imaging, elucidating small details in lung tissue morphology which poorly attenuate x-rays¹³.

Fig. 1.

Experimental setup description. A 4D microscopy of rat lung using a synchrotron x-ray source. High-intensity coherent x-rays are rendered monochromatic, and detected by a PCO Edge 5.5 camera with optics yielding an isotropic voxel size of 6 µm³. The in vivo anesthetized rat is mechanically ventilated while the electrocardiogram and respiration are monitored and recorded for cardiac and respiratory gating of image reconstruction; B SAXS: the lung sample is irradiated by monochromatic x-ray generated by a synchrotron. The ordered fibrillar organization of collagen in the sample produces an x-ray scattering pattern recorded on the detector; C azimuthal integration of the diffraction pattern produces an intensity vs. diffraction angle (q) curve, where the peaks are due to the d-periodicity of the collagen fibrils; D D-spacing of collagen fibrils; (d) is related to q by q = 2π/d.

Intratracheal instillation of bleomycin in rodents in its early phase (7days) induces lung inflammation and injury that shares many features with ARDS including neutrophilic infiltration, high-permeability oedema and extensive alveolar micro atelectasis that are followed by a later fibrogenic phase^14,15. These changes drastically modify the ECM structure and the macroscopic tissue viscoelastic behaviour¹⁶.

Here we hypothesized that high-stretch mechanical ventilation alters the nanoscale organization of the lung ECM fibrillar components collagen and elastin. Thus, in vivo 4DCT images of the lung tissue were acquired at 6 µm³ voxel resolution to obtain quantitative maps of local lung tissue deformation (strain) within aerated lung acini, under mechanical ventilation in anesthetized rats. This approach was used to assess the effect of the alterations in lung tissue biomechanics induced by short-term high-tidal volume (VT) mechanical ventilation in both normal and bleomycin-injured lung on local strain, as well as global respiratory tissue viscoelasticity. This was subsequently followed by exploring the supramolecular organization of collagen and elastin in unfixed lung samples using synchrotron small-angle x-ray scattering (SAXS) (Fig. 1B–D).

Results

Respiratory parameters under imaging conditions are summarized in Table 1. There was no statistically significant difference between the experimental conditions, except for driving pressure (ΔP) which was significantly higher after injurious ventilation in controls. The peak inspiratory pressure during injurious ventilation was: 41.3 ± 2.1 cmH₂O, which corresponded to a tidal volume of ~ 65 ml/kg. At a PEEP of 5 cmH₂O, the change in tissue fraction within selected ROI’s from end-expiration to end-inspiration was small: 4.5 ± 4.5% and 1.3 ± 4.2% in control and bleomycin rats at baseline, and 7.9 ± 12.1% and 4.2%±4.0% after injurious ventilation, respectively. The differences were not statistically significant.

Table 1.

Baseline respiratory parameters.

	Pmax	PEEP	ΔP	RR	VT
	cmH2O	cmH2O	cmH2O	bpm	ml
Control	10.5±1.2	5.1±0.5	5.3±0.8	74.9±14.4	5.7±1.5
Control-VILI	13.8±7.1	5.2±0.7	8.6±7.7*	81.9±8.3	5.6±1.2
Bleo	12±2.8	4.7±0.7	7.3±2.6	85.1±6.7	5.9±2.0
Bleo-VILI	11±1.9	4.6±0.5	6.4±2	83±12.8	5.1±0.6

Fmax: peak inspiratory flow, Pmax: peak inspiratory pressure; PEEP: positive end-expiratory pressure; ΔP: Pmax – PEEP; VT: tidal volume at baseline and during imaging; RR: respiratory rate. Bleo: bleomycin; VILI: injurious ventilation. *: p<0.05 vs. Control BL.

Figure 2 shows representative hematoxylin & eosin stains in the four experimental groups. Figure 2A represents the control animal group with normal appearing tissue architecture. Injurious ventilation induced mild leukocyte infiltration and extracellular matrix augmentation with occasional thickening of alveolar septa interspersed with normal-appearing regions, without almost any hyaline membrane formation or alveolar oedema (Fig. 2B). Bleomycin-instilled animals (Fig. 2C) showed extensive matrix deposition and thickening of alveolar walls at baseline, some intra-alveolar oedema, neutrophilic infiltrates and widespread microatelectases at baseline. These changes were amplified after injurious ventilation, with occasional occurrence of patchy alveolar oedema (Fig. 2D).

Fig. 2.

Light microscopic imaging. Representative hematoxylin & Eosin stains in; A control lung; B control after 20 min of injurious mechanical ventilation. Note signs of injury with neutrophilic cellular infiltration, alveolar wall thickening and occasional hyaline membranes; C day 7 post bleomycin intratracheal instillation with gross inflammation and alteration in the alveolar architecture and interstitial matrix thickening; D bleomycin-injured lung after additional 20 min of injurious mechanical ventilation shows greater inflammation and alveolar hyaline membranes.

Representative maps of local microscopic tissue ε in the 4 experimental conditions and sample strain vs. time curves within each map are shown in Fig. 3. Animations showing the time-evolution of local microscopic tissue strain throughout the respiratory cycle are included in the online supplement (see: supplemental video animation). Maximal parenchymal strain (ε_max) at the peak of inspiration, under protective mechanical ventilation settings within the aerated lung acini is shown in Fig. 4. The parenchymal strain was significantly reduced in bleomycin animals compared to controls. Following injurious ventilation, ε_max did not significantly change compared to baseline in controls, nor in bleomycin injured animals.

Fig. 3.

A–D: representative single slice phase-contrast CT images in a Control (A), injurious ventilation (B), Bleomycin-injured (C) and Bleomycin after injurious ventilation (D) rat; E–H: corresponding strain (ε) maps; I–L: corresponding dynamic ε curves as a function time relative to the beginning of the respiratory cycle (t/T) within ROI’s of same color as in A–D. Note the reduced tissue strain and slope in bleomycin-injured lung and further reduction after injurious ventilation.

Fig. 4.

Maximal local inspiratory lung parenchymal strain (ε_max). BL: baseline;  Ctrl-VILI: control after short-term injurious ventilation; Bleo: bleomycin injury; Bleo-VILI: bleomycin injury after short-term injurious ventilation.

The qualitative examination of dynamic imaging data showed extensive zones of unaerated parenchyma in bleomycin-instilled animals but did not evidence significant tidal recruitment at PEEP of 5 cmH₂O. The change in image segmentation-based aeration fraction, induced by a tidal breath, was 4.5 ± 4.5% at baseline and 7.9 ± 12.1% after injurious ventilation in Controls (n.s.); and 1.3 ± 4.2% and 4.2 ± 4.0% (n.s.) respectively before and after injurious ventilation in bleomycin-injured animals.

Oscillatory respiratory mechanical parameters are summarized in Fig. 5. Respiratory tissue elastance (H) and damping (G), the latter representing tissue frictional energy dissipation, were significantly increased in bleomycin rats compared to controls, while hysterisivity (h = G/H) and airway resistance (Raw) were unchanged. Injurious ventilation significantly increased H in both bleomycin and control animals. Raw and η significantly decreased after injurious ventilation in both bleomycin and control animals. Regression analysis showed a significant inverse relationship between respiratory tissue elastance and maximal parenchymal ε: R = 0.63, p < 0.001, (Fig. 6).

Fig. 5.

Oscillatory mechanical parameters. H: tissue elastance; G: tissue damping; Raw: airway resistance; h: hysterisivity Bleo: bleomycin.

Fig. 6.

Relationship between respiratory tissue elastance (H) and maximal local parenchymal strain (ε_max). Ctrl: Control; Ctrl-VILI: Control after high-tidal volume ventilation; Bleo: bleomycin-induced lung injury; Bleo-VILI: bleomycin after high-tidal volume ventilation.

Representative maps of azimuthally integrated SAXS signal intensity (in greyscale), and corresponding maximum I(q) peak intensities for the 3rd -order collagen (green) and 1st -order elastin (red) peaks in representative samples of each experimental condition are shown in Figure S1. As expected, both collagen and elastin signals were strongest around bronchial and vascular structures.

Figure 7 shows the changes in peak location (q) for collagen and elastin with the experimental condition. The 3rd -order collagen peak location decreased significantly was significantly lower in bleomycin-injured lungs. Bleomycin-injured q significantly decreased after injurious ventilation, both in initially normal and bleomycin-injured lungs. The q values vary inversely with D-spacing, which is the regular axial repeat distance along a collagen fibril that arises from the end-to-end packing of tropocollagen molecules. Here, D-spacing corresponded to 65.17 ± 0.09 and 65.23 ± 0.14 nm in Control and Control post injurious ventilation; 65.34 ± 0.13 and 65.49 ± 0.10 nm in bleomycin and bleomycin post injurious ventilation groups, respectively. Unlike collagen, much less significant changes were observed in elastin q, decreased only after injurious ventilation in the bleomycin group.

Fig. 7.

Diffraction q peak position of 3rd -order collagen and 1st -order elastin; q is inversely related to D-spacing. Ctrl-VILI: control after short-term injurious ventilation; Bleo: bleomycin injury; Bleo-VILI: bleomycin injury after short-term injurious ventilation.

The relation between D-spacing and overall respiratory tissue elastance (H) and hysterisitivity (η) is shown in Fig. 8. While an increase in tissue elastance was associated with an increase in collagen fibril D-spacing (Fig. 8A), the opposite relation was observed with h (Fig. 8B). Due to a large scatter in the data, the differences in the CV of ε_max between groups were not statistically significant. There was nevertheless an association between increasing CV(ε_max) and D-spacing and (Fig. 8C).

Fig. 8.

A Relation between collagen fiber D-spacing and tissue elastance (H); B relation between collagen fiber D-spacing and tissue hysterisivity (η); C relation between collagen fiber D-spacing and local strain heterogeneity at maximum intra-breath lung expansion (CV(ε_max)). Note that tissue stiffness increases with D-spacing, conversely hysteresis is reduced with D-spacing. Increased heterogeneity of local ε_max is associated with increased collagen D-spacing.

Discussion

In this study, we used 4D synchrotron phase contrast micro-CT to dynamically map local lung microscopic tissue strain (ε) in mechanically ventilated rats with normal lungs and inflammatory lung injury 7 days after intratracheal instillation of bleomycin. Our main findings were that: (1) local maximal strain (ε_max) was reduced in the aerated acini in response to both injurious ventilation and bleomycin instillation; (2) the reduced local ε_max was associated with a larger increase in dynamic elastance (H) than dissipation measured by the oscillatory damping coefficient (G), resulting in a reduced hysteresis (η); (3) Injurious ventilation significantly decreased 3rd -order collagen peak intensity (enlarged the D-spacing) of collagen fibrils; (4) Collagen q value was reduced 7 days after bleomycin-induced lung injury. Injurious ventilation in this condition further reduced q; (5) far less significant changes were observed in elastin than in collagen.

Local lung tissue strain (ε)

The average maximal dynamic ε measured in normal lung is in line with our previous observation in this model¹². The lung tissue is a complex network of fibrillar protein constituents embedded in a ground substance or ECM composed of GAG and PG infiltrated by cellular components, some of which have contractile properties¹¹. Several alterations in the parenchymal micro-architecture could contribute to stiffening in lung injury either due to VILI or bleomycin. One is the increase in collagen and elastin content, as well as increases in their cross-linking¹⁷. However, at 7 days post-bleomycin, previous studies have shown that the volume fraction of collagen or elastin has not yet increased¹⁸. Increases in the ECM ground substance mainly composed of glycosaminoglycans, and proteoglycans may also be involved¹⁹. The volume fraction of PGs such as biglycan have been shown to be increased in Sprague-Dawley rats already at 7 days after tracheal instillation of bleomycin¹⁸. Surface tension of the air-liquid interface lining the alveoli and small airways also plays an important role in determining parenchymal micromechanics¹⁶. Increased interfacial surface tension due to surfactant dysfunction is a hallmark of bleomycin-induced lung injury in its early phase²⁰ and may promote widespread atelectasis and stiffening of the aerated acini due to an increased elastic recoil. Moreover, neighbouring less injured areas can exhibit compensatory stretch and hyperinflation²¹, due to parenchymal interdependence.

The significant association of H with ε_max shows that the stiffening and consequent reduction in strain within aerated lung acini is a strong determinant of the overall changes in dynamic elastance of the respiratory system as a result of both bleomycin-induced lung injury and injurious ventilation. In both, a significant part of the increase in H is likely due to atelectasis and inflammation or oedema of the airspaces which acts to reduce the aerated lung volume and increase the micromechanical heterogeneity. It is noteworthy that the reduced local strain in bleomycin-instilled rats reflects the increased stiffness due to changes in lung parenchymal mechanics in the aerated regions, the collapsed alveolar or oedematous regions being excluded by the image segmentation process. Therefore, even the aerated alveoli have altered micromechanical properties, a point that could not be elucidated based on overall respiratory mechanical measurements alone. Indeed, we observe an association of D-spacing with CV(ε_max). The spatial heterogeneity of local strain within the aerated alveoli is significant because non-uniformities in lung tissue expansion magnify the local distending stress that may largely exceed transpulmonary pressure locally²².

Nanostructural arrangement of collagen and Elastin

Small-angle x-ray scattering has been extensively utilized to characterize the nano-structural features of periodically arranged extracellular matrix fibrillar proteins, collagen^23,23–29 and to a lesser extent elastin^30–34. To our knowledge this is the first observation of the alteration in the nanostructural arrangement of collagen fibrils induced by strain due to mechanical ventilation, in normal lung and following injury due to bleomycin-induced inflammation. The organization and molecular ordering of collagen fibrils at the nanoscale has been studied under tensile loading, in tissues such as rat tail tendon³⁵ skin²⁵ and cartilage²⁹. The current understanding is that small strains or the “toe” region of the stress-strain curve corresponds to the unfolding of crimps along the fibrils³⁶. At higher strains, in the “heel” and “linear region” of the stress-strain curve, thermally activated molecular kinks thought to occur within the gap region (Fig. 1D) are straightened leading to the elongation of the fibrils. Also, lateral ordering increases, whereby the fibril orientation shifts towards the direction of the strain^25,37. Beyond this point, further increased strain leads to stretching of the triple helices and cross-links between them allowing side-by-side gliding of the neighboring collagen molecules, causing increased D-spacing. In small-angle x-ray scattering, the position of the azimuthally integrated collagen peak q is inversely proportional to fibril D-spacing, and decreases with increasing tensile strain³⁸.

In this study, we found that injurious ventilation as well as bleomycin-induced lung injury decreased the 3rd-order collagen diffraction q value, indicating larger D-spacing of collagen fibrils. It should be noted that the lungs were inflated to the same degree upon sample collection, therefore the observed differences in the SAXS profiles are independent of static tensile strain on collagen and elastin within the ECM.

The association we observed between increased collagen D-spacing and higher tissue elastance (H) is consistent with several converging pathways, but cannot alone establish causality. The increased D-spacing of the collagen fibrils post injurious ventilation observed in both normal and bleomycin-injured lung may be a consequence of cyclic stretch applied to the collagen fibres during injurious ventilation³⁹, leading to straightening and elongation of the fibrils. While the change in d-period appears small, it is comparable to changes observed in other tissues such as cartilage, under strain²⁹. At baseline, εe locally reached ~ 20% in normal lung with baseline tidal volumes and ~ 10% in bleomycin injured lungs indicating that the lung tissue was stiffer in this condition (Fig. 3). The strain of the fibrils is typically less than that of the tissue³⁶, which suggests that the proteoglycan rich matrix between the fibrils also undergoes a considerable strain²⁷.

Bleomycin-induced lung injury also increased the D-spacing of collagen fibrils even before injurious ventilation. The reason for this increase is not known, however, both inflammation and the accelerated turnover of ECM collagen may be involved. First, the interstitial oedema and consequent ECM swelling due to inflammation can increase interfibrillar water molecules, which may affect collagen periodicity. Indeed, there is evidence that collagen D-spacing is dependent upon the water content of the interfibrillar proteoglycan network^29,40. Second, ECM collagen turnover is significantly increased even at 7 days post bleomycin instillation⁴¹. However, it is unclear whether newly formed collagen has a different periodicity than older collagen.

An alternative interpretation of the association between collagen D-spacing and global lung tissue elastance is surface-tension–driven stiffening with purely hydration-induced D-spacing changes. Distinguishing among the different potential mechanisms will require interventions that selectively block LOX (e.g., β-aminopropionitrile), deplete proteoglycans (chondroitinase ABC), or normalise surface tension (exogenous surfactant) while repeating coupled SAXS–mechanics measurements. Such experiments will more firmly elucidate the causal chain from biochemical remodelling through fibrillar nano-architecture to whole-organ viscoelastic behaviour.

Mechanistic significance of the fall in hysterisivity (η)

We found increases in both respiratory tissue dynamic elastance (H) and dissipation or damping (G) due to early bleomycin-induced lung injury. This finding is in agreement with previous data in the literature^16,18,42, although the changes in respiratory mechanics in bleomycin lung injury evolve over time^17,18,43. High stretch ventilation resulted in further small but significant increases in H. Both injurious ventilation and bleomycin injury at 7 days significantly reduced tissue hysterisitivity (η). In the structural-damping framework of lung tissue⁴⁴, η represents the ratio of energy dissipated (G) to energy stored (H) per breathing cycle; a fall in η therefore denotes a shift toward more relative elastic behaviour with less intrinsic damping. The reduction in η on day 7after bleomycin is in agreement with Ebihara et al.,¹⁸ who showed that lower η in rat tissue strips correlated with the local volume fraction of the small proteoglycan biglycan¹⁸. Small leucine-rich proteoglycans (biglycan, fibromodulin) accumulate during the inflammatory/oedematous phase and act as molecular lubricants between collagen fibrils⁴⁵, lowering interfibrillar friction and, hence, dissipative losses. Concomitantly, bleomycin up-regulates lysyl-oxidase–mediated cross-linking and promotes early straightening of crimped collagen⁴⁶.

High-stretch ventilation imposed on the bleomycin-injured lung further decreased η. Cyclic over-stretch can realign collagen bundles and introduce additional cross-links, making the matrix stiffer yet relatively less dissipative⁴⁷. Functionally, a lower η reduces the tissue’s relative ability to absorb mechanical power; more of the ventilator-supplied energy is stored elastically and returned at end-inspiration, potentially amplifying regional stress and accelerating ventilator-induced lung injury⁴. Thus, the parallel rise in H and fall in η that we report likely reflects early proteoglycan-rich, cross-link–stabilised matrix remodelling that stiffens the parenchyma while relatively diminishing its damping capacity. Our data suggest that these mechanical changes in the ECM can translate into an increased cyclic stretch of collagen fibrils and an increased elongation due to slippage within fibrils or groups of fibrils as suggested by the increased D-spacing^25,36. As a result, the collagen fibres become stiffer, and less hysteretic. This is illustrated by the observed relation between collagen fibrillar D-spacing, and global tissue elastance on one hand, and the inverse relation with η on the other hand (Fig. 8).

Contribution and structural stability of elastin

Unlike collagen, the SAXS 1st-order peak assigned to elastin showed only a modest, non-significant shift across conditions (Fig. 7). This is consistent with the entropic-spring behaviour of cross-linked elastin: single tropo-elastin molecules can extend up to eight-fold and recoil with virtually zero energy loss, producing negligible changes in axial repeat distance even at strains far greater than those applied here⁴⁸. At the microscopic tissue strains we measured (≈ 10–20%), molecular extension is accommodated mainly by uncoiling and domain rotation, processes that do not alter the periodicity that gives rise to the SAXS peak.

In the early inflammatory phase of bleomycin injury (≤ 7 days), mature elastic fibres are largely preserved; nevertheless, elastin gene transcription is already up-regulated in peri-bronchiolar regions⁴⁹. Such de-novo tropo-elastin requires days to be cross-linked into insoluble fibres, so its presence would not yet affect lattice spacing. Likewise, proteolytic pathways that degrade and remodel elastin—e.g., neutrophil elastase or A disintegrin and a metalloprotease 9 (ADAM-9)–mediated cleavage—reach appreciable activity only after the first week⁵⁰.

Our ventilation episode (20 min, PIP ≈ 40 cmH₂O) is too brief to modify fibre ultrastructure. In newborn mice, detectable changes in elastin deposition appears only after ≥ 8 h of mechanical ventilation and are preceded by a surge in tropo-elastin mRNA without parallel changes in assembly proteins⁵¹.

Functionally, an intact elastin scaffold dominates the pressure–volume curve at low strain, buffering the initial load before collagen engages at higher stretch⁵². Because elastin dissipates little energy, the relative increase in collagen engagement after bleomycin and high-stretch ventilation raises H more than G, contributing to the fall in η we report. Recent reviews highlight how elastin–microfibril interactions and lysyl-oxidase–dependent cross-linking maintain this low-hysteresis behaviour across repeated cycles⁵². The relative stability of elastin D-spacing in our study therefore supports the view that early ventilator-induced stiffening is driven primarily by collagen straightening and new cross-links, while elastin continues to provide reversible recoil with minimal energy loss.

Translational significance for VILI and future research directions

Whether or not the observed changes in collagen nanostructure have adverse effects cannot be determined directly based on our data and requires further investigation. However, data in the literature suggest that altered collagen is not inert. Collagen stiffening due to mechanical stress or covalent cross-linking interacts with macrophages and fibroblasts and may promote inflammation⁵³. Whether it appears as bioactive fragments, becomes glycated/oxidised, or is physically remodelled and stiffened, the ECM acquires new biochemical epitopes and mechanical properties that actively recruit and program inflammatory cells⁵⁴. These pathways are increasingly being targeted therapeutically. For example, Pan-lysyl-oxidase inhibitors attenuate bleomycin-induced cross-linking and fibrosis in rodent lungs, and are entering early-phase clinical evaluation⁵⁵. Therefore, our findings may have translational significance and provide a rationale for future research in matrix-directed pharmacotherapy in VILI.

Strengths and limitations

The strength of this unique study set-up is the use of a well-established animal model and employing state-of-the art 4D synchrotron phase contrast micro-CT to dynamically and quantitatively image local tissue deformation in vivo, in intact lungs, at spatial resolutions allowing investigations of deep acinar structures¹². We combine in vivo 4D synchrotron phase-contrast micro-CT imaging with ex-vivo SAXS measurements to link changes in collagen nanostructure to macroscopic mechanical behaviour. Indeed, the lung can be perceived as a complex system where the gross mechanical behaviour emerges from the ensemble behaviour of its constituents at multiple length scales¹⁰. Overall measurements of respiratory mechanics however, give no insight into the local tissue behaviour that could optimally be measured in vivo by synchrotron 4DCT. Yet, an obvious limitation is that the spatial resolution and the correction of motion blurring can still be improved. Synchrotron 4DCT has been performed in vivo with a voxel resolution of ~ 1 μm in mice. This technique allowed the morphometric characterisation of individual alveoli, albeit in the very apical regions of the lung⁵⁶. Deep pulmonary acini in larger animal models are very challenging to image in vivo due to multiple and non-linear sources of motion blurring. Here we were able to image pulmonary acini at a voxel resolution of 6 μm in rats¹², which is sufficient to depict subacinar structures but somewhat limited for the assessment of individual alveoli (12). The animals were imaged in upright position, which can impact the overall respiratory mechanics⁵⁷. Our study was underpowered to detect differences in ε between baseline and VILI conditions (Fig. 4). No quantitative analysis of the histologic slides was performed; histology data were used mainly to assess signs of inflammation and injury in the lung tissue. We did not partition lung and chest wall mechanics due to the complexity of the experimental setup. Approximately half of the overall closed-chest G and H are due to chest wall contributions⁵⁸. However, the changes in lung viscoelastic tissue parameters induced by bleomycin and high VT are predominantly pulmonary. We did not observe a significant amount of recruitment/derecruitment. This may have been because a PEEP level of 5 cmH₂O was sufficient to reduce tidal recruitment and the early timing of imaging after injurious ventilation. Another strength of our study is that synchrotron SAXS was performed in hydrated and unstained or fixed lung samples. Indeed, chemical tissue fixation can significantly affect cross-linking of the tissue ECM components⁵⁹. We did not analyse fibre orientation data from SAXS profiles, which will be the subject of further study.

Conclusions

In conclusion, local microscopic tissue deformation imaged dynamically in intact in vivo rat lungs and global respiratory viscoelastic behaviour can be linked to nanoscale alterations in collagen fibril organization, in both VILI and bleomycin induced lung injury. This finding is significant because collagen mechanically interacts with the cellular components of the ECM and modulates numerous vital processes. Importantly, collagen offers numerous targets for drug development⁶⁰. Our findings shed new light on the pathophysiology of ventilator-induced lung injury, and mechanistic insight into the association between the ECM and global respiratory mechanical behaviour.

Materials and methods

Animal preparation

The in vivo experiments were performed on 24 Sprague-Dawley rats, average weight: 399 ± 26 g. The care of animals and the experimental procedures were in accordance with the Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes and complied with the ARRIVE guidelines⁶¹. Experimental procedures were evaluated and approved by the local institutional ethical review board and the French Ministry of Higher Education and Research (authorization number: APAFIS#31021-2021040617424365). The animals were housed in a facility with 12 h light/dark cycles, fed ad libitum, and were allowed to acclimatize to the housing conditions for a minimum of 7 days before any experimental procedures. One group of the animals underwent imaging (n = 14) while the other group served as controls for a separate post-mortem imaging study not included here.

Bleomycin challenge

Rats received a single intratracheal (i.t.) dose of bleomycin (Sigma-Aldrich, St. Louis, MO, United States), with a concentration of 1000 iU in 200 µl saline. Control animals received the same volume of saline only. Prior to i.t. instillation, the animals were lightly anesthetized with inhaled isoflurane through a nose mask (IsoFlo^® vet, Orion Pharma, Sweden). The procedure of i.t. administration was performed by placing the rat in a supine position on a slanting board as previously described¹⁵.

4D synchrotron phase-contrast micro-CT

Anaesthesia was induced by intra-peritoneal injection of 1.5–2 ml kg^− 1 body weight of a solution containing Ketamine (40 mg ml^− 1) and Xylazine (20 mg ml^− 1). The animal was tracheostomized and a 14 G polyethylene catheter (Venflon; Becton Dickinson, Le Pont-de-Claix, France) was inserted and secured with a gas-tight seal. A 26 G polyethylene catheter (Neoflon; Becton Dickinson, Helsingborg, Sweden) was inserted in the jugular vein for fluid replacement (Lactated Ringer’s, B.Braun, France, 1–2 ml/kg/h). After surgery, the animal was immobilized in the vertical position in a custom-made plastic holder and placed on a remote-controlled rotation stage in the experimental hutch for imaging. After verifying adequate depth of anaesthesia (heart rate stability, inhibition of response to limb stimulation), muscle relaxation was induced by Atracurium injection (4 mg kg^− 1) to avoid motion and suppress spontaneous breathing. Anaesthesia was maintained during the experiment at 1% inhaled isoflurane (Isoflurane Belamont, Piramal Critical Care), while animals were mechanically ventilated using a custom pressure-controlled mechanical ventilator¹². The nominal tidal volume (VT) of 10 ml·kg^− 1 was used with FiO2 set to 0.5 with the positive end-expiratory pressure (PEEP) of 5 cmH₂O. The electrocardiogram (ECG) was recorded by placing subcutaneous needle electrodes on the paws, and the neck, connected to a differential amplifier (Bio Amp, Adinstruments, Dunedin, New Zealand). ECG, airway pressure and air flow (Fig. 1A) were continuously sampled at 10 kHz and recorded using a Powerlab 16/35 data acquisition device (Adinstruments, Dunedin, New Zealand).

The in vivo imaging data were acquired at the biomedical imaging beamline (ID-17) of the European Synchrotron Radiation Facility (ESRF, Grenoble, France). Details of the methodology for image acquisition have been previously described in detail^12,62. Briefly, x-rays produced by a multipole wiggler source were monochromatized using a double bent Laue Si (111) monochromator⁶³, selecting a photon energy of 38 keV (Fig. 1A). Monochromatic radiation impinged on the sample 146 m from the source and was detected by a fast complementary metal-oxide-semiconductor (CMOS) PCO Edge 5.5 camera (PCO AG, Kelheim, Germany) coupled to a 250 μm thick Cerium-doped Lutetium Aluminum Garnet (LuAG: Ce) scintillator and optics yielding an isotropic pixel size of 6 × 6 µm². The field of view of the camera was reduced to 16.4 × 1.9 mm² (H x V) to achieve a frame rate of ~ 333 fps as a compromise between field of view and temporal resolution. The sample-to-detector distance was set to 3.5 m to utilize propagation-based phase contrast imaging⁶⁴. Mechanical ventilation was synchronized with the heartbeat, to resolve the periodic motion of the lung parenchyma. Inspiration was triggered by one R-wave out of two, thus obtaining two cardiac beats per respiratory cycle. A gating methodology was then used to reconstruct 80 sequential CT images per respiratory cycle, by retrospectively sorting individual image projections corresponding to specific phases of the cardiac and respiratory cycle. Sequential 3D CT images covering a field of view of 16.4 × 16.4 × 1.9 mm³ were thus reconstructed at 10 ms time intervals.

4DCT image processing

Aerated lung acini were segmented from the surrounding tissues using the Otsu algorithm⁶⁵. Larger non-alveolate conducting airways were then excluded using morphological erosion and dilatation operations¹². A non-rigid registration algorithm⁶⁶ was then applied to warp the image at each time step to perfectly match the immediately previous one. This process allowed us to compute the displacement field, hence the local volume change, at each time step from the beginning of inspiration¹². Local lung tissue deformation or strain (ε) was computed as:

1

Where ε_t is local strain at time t from the beginning of the respiratory cycle (t₀); Vt and Vt₀ the local volume of airspaces at time t and t₀. Average maximal strain (ε_max) is defined as the mean ε within the imaged lung region at end-inspiration. Strain heterogeneity was measured as the coefficient of variation (CV) of ε_max (standard deviation/mean). Gas fraction was determined by image segmentation within selected regions of interest (ROI, n = 12) in aerated acini.

Respiratory mechanics

The low-frequency forced oscillation technique (FOT) was used in order to separate airway and lung tissue contributions to the total respiratory system impedance, using the constant-phase model^67,68. Two sighs (inflation to 30 cm H₂O) were delivered before the beginning of data collection. At end-expiration, mechanical ventilation was paused and a small-amplitude (1 cm H₂O peak-to-peak) forcing signal was delivered into the respiratory system by a loudspeaker-in-box system connected to the tracheal cannula via a polyethylene tube (100 cm length, 2.0 mm inner diameter). All measurements were performed at a nominal PEEP of 5 cmH₂O. The loudspeaker was driven by a computer-generated pseudorandom signal ranging from 0.5 to 21 Hz. Lateral pressures were measured at the loudspeaker end (P1) and the distal end (P2) of the wave-tube with miniature sidearm transducers (ICS 33NA00D). These pressure signals were low-pass filtered (< 25 Hz) and digitized at a sampling frequency of 128 Hz. The pressure transfer function (P1/P2) was created by fast Fourier transformation from the 8 s recording. The input impedance of the respiratory system (Z_rs) was computed from the pressure transfer function as the load impedance of the wave-tube⁵⁷ by using the transmission line theory⁶⁹:

2

where L is the length of the wave tube, Z₀ is the characteristic impedance of the wave tube, γ is the complex propagation wave number, sinh is hyperbolic sine and cosh is hyperbolic cosine. Both Z₀ and γ depend on the geometrical parameters of the wave tube (diameter, material constants and gas inside the tube). Three to five Z_rs spectra were ensemble-averaged at each time point. A model that includes Newtonian resistance (Raw), inertance (Iaw) in series with constant-phase tissue compartments incorporating tissue damping (G) and elastance (H) was fitted to the averaged Z_rs data⁶⁷. Hysterisivity (η) was calculated as the G/H ratio. The data at frequencies coinciding with heart rate and its harmonics were often corrupted (as evidenced by poor coherence and a high SD) and they were omitted from the model fitting.

Experimental protocol

In control and bleomycin-injured rats, 7 days post-intratracheal instillation, FOT was performed at baseline, followed by 4DCT image acquisition for 9 min. After baseline data acquisition, injurious ventilation was initiated by increasing peak respiratory pressure to 41 ± 2 cmH₂O, with a PEEP of 0 while respiratory rate was reduced to 30 bpm for 20 min. The measurements including lung mechanics and image acquisition were then repeated post-injurious ventilation. At the end of the measurements, the animals were euthanized by intraperitoneal injection of pentobarbital sodium (Dolethal, 200 mg/kg, Vetoquinol, Lure, France) and the heart and lungs were dissected and removed en bloc for histological analysis.

Histological analysis

The left lungs were fixed in 4% paraformaldehyde at 20 cmH₂O and dehydrated with a graded ethanol series, then embedded in paraffin. Lung tissue sections of 4 μm were prepared and stained with hematoxylin and eosin for qualitative histological evaluation.

SAXS data acquisition and processing

The right lungs were inflated to 20 cmH₂O and frozen in isopentane cooled with liquid nitrogen. The samples were cut down to 1 mm thick slabs with a microtome and stored at -80 °C until data collection. Immediately before data acquisition, the samples were included in custom made Kapton film pockets and thawed at room temperature. The samples were measured with synchrotron radiation at the coherent small-angle x-ray scattering (cSAXS) beamline at the Swiss Light Source (SLS), Paul Scherrer Institute (PSI), Villigen Switzerland⁷⁰. The samples were raster-scanned with 25 × 25 µm² spot size and with scanning steps of 25 × 25 µm² in a continuous line-scan mode. A Pilatus 2 M detector⁷¹ was used to record the 2D scattering patterns. An exposure time of 100 ms and x-ray energy of 11.2 keV (wavelength of 1.1 Å) was used (Fig. 1B). The sample-to-detector distance and the beam center on the detector were calibrated using a Silver-Behenate (AgBH) powder standard. The beam flux (∼2 × 10¹⁰ photons/s) was measured using a glassy carbon standard. A beamstop was used to block the directly transmitted beam. A diode measuring sample transmission was mounted on the beamstop for transmission correction to account for differences in local sample thickness.

The I(q) scattering curves (Fig. 1C), that inform on collagen structure, were obtained by azimuthally integrating the detector images. Two q-intervals were analysed: the first ranged from 0.0278 to 0.0297 Å⁻¹, encompassing the 3rd -order collagen peak (at 0.029 Å⁻¹), while the second ranged from 0.138 to 0.142 Å⁻¹, containing an elastin peak (at ~ 0.14 Å⁻¹). To account for background signal, an exponential function I_bkg(q) = a q^k was fitted to the regions adjacent to the corresponding peak range using linear regression. After subtracting the fitted background, the remaining intensity within the peak regions were summed to determine the scattering intensity of the peak. The q-value corresponding to the maximum intensity within each peak region was selected to represent the peak position (q). The d-period of the collagen molecule is inversely proportional to the peak location q following Bragg’s law, e.g.: q = 3(2π/d), for the 3rd -order collagen peak (Fig. 1D), and q = 2π/d for the 1^st order elastin peak. For each subject and condition, the q-values were computed within 6 identical regions of interest in the 2D scatter maps.

Statistical analysis

We estimated that in order to detect a 50% increase in H, 4 observations per group are needed to reach 80% power at the 5% significance level. We increased the sample size to 5 to safeguard against data loss. Data are expressed as mean ± standard deviation, with the exception of boxplots of average maximal strain (e_max: median-interquartile range). Differences in the mean values of e_max, respiratory mechanical parameters and the SAXS parameter q, between the experimental conditions (Control, Control-VILI, Bleomycin, Bleomycin-VILI) were analyzed by Student’s t-test with Holm’s post-hoc correction for multiple comparisons. In vivo imaging data were obtained in 5 control and 5 bleomycin injured animals at baseline and post injurious ventialtion. SAXS data were averaged from 6 regions of interest per sample in 6 Control, 5 Control-VILI, 5 Bleomycin, and 3 Bleomycin-VILI animals. The relationship between e_max and respiratory elastance was assessed by simple linear regression. The statistical analysis was performed with R (Version 1.2.1335, https://www.R-project.org). Some were produced with Sigmaplot software (V14.0, Systat Software, Chicago, Illinois). A p < 0.05 was considered significant for all tests.

Supplemental Figure S1. Total scattering signal intensity (grey) of collagen (a) and elastin (b) and corresponding background-corrected intensity of 3rd-order collagen peak (green) and 1st order elastin peak (red), in representative lung regions of interest, at baseline and after injurious ventilation (VILI), in a control and day 7 after bleomycin-induced lung injury.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

RD, RK, LF, IMP, JLCP, GP, AM, MB, LEO, and SB made substantial contributions to study design, to data acquisition, study analysis, and interpretation of data, AND drafted the manuscript, AND approved the version to be published, AND agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Funding

This work was supported by the Swedish Research Council (grant 2018–02438), the Swedish Heart and Lung Foundation (20200877, 20200825, 20220681, 20230767), and the Alvar Gullstrand research grant (ALF-938050), the European Synchrotron Radiation Facility, and the French Institute of Health and Medical Research (INSERM UA07).

Data availability

All data are available upon reasonable request to the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.