Background
Pulmonary fibrosis results from accumulation of fibroblasts, scar-forming myofibroblasts, and extracellular matrix proteins including collagen, often leading to irreversible loss of lung function. It can be caused by various factors including toxins, radiation exposure, autoimmune disorders, and infection. Idiopathic Pulmonary Fibrosis (IPF) is a severe form of fibrotic lung disease that can progress to respiratory failure and has a prognosis worse than lung cancer. There are currently few effective therapies. Usual interstitial pneumonia (UIP) is the histopathology underlying IPF and is characterized by heterogeneity of disease and accumulation of fibroblast foci and collagen with an emphasis on collagen type I (col1) over type III (col3) [
1,
2], and abnormalities in other matrix molecules including elastin [
3]. IPF is one of many diseases associated with significant collagen and other matrix protein accumulation. It is the most common of the idiopathic interstitial pneumonias, is increasing in prevalence, and it is a progressive disease that causes significant morbidity and mortality. The median duration of survival from the time of diagnosis is only 2.9 years [
4,
5]. There are currently few effective FDA approved treatments for IPF (for review and overview of current and targeted therapies for IPF, please see: [
6‐
10]), making research into IPF pathogenesis critical.
Cryptogenic organizing pneumonia (COP) is another common fibrotic lung disease. It is also characterized by accumulation of matrix components resulting in organized areas of granulation tissue in the lung. Components of this pathologic matrix accumulation in COP also include col1 and col3 (with an emphasis on col3 over col1, in contrast to UIP), fibronectin, and proteoglycans [
1,
11]. In contrast to UIP, the granulation tissue found in COP accumulates in the airspaces and small airways rather than in the interstitial spaces and importantly, COP is a treatable disease with most cases responding to corticosteroids. Although the matrix components of UIP and COP have some similarities, it is unknown why the excess matrix in COP can be reabsorbed or cleared after treatment with corticosteroids while the matrix in UIP is resistant to treatment and resolution [
1].
A growing body of literature supports the roles of matrix organization and structure as important effectors of fibrotic lung disease. Extracellular matrix (ECM) components have important mechanobiological properties including the abilities to activate pro-fibrotic cytokines; regulate cell trafficking; and modulate cell activation, proliferation, survival and differentiation [
12,
13]. The organization and structure of the ECM, including collagen, also helps regulate availability of and interactions with a large variety of cell-matrix binding sites critical for controlling lung function. These findings further reinforce the notion that in biology, structure is a key determinant of function. Indeed, other data suggests that ECM stiffness regulates key cellular activities that may contribute to IPF [
14], as well as endogenous lung function [
15]. Hence, there is heightened interest in the content and structure of the matrix, and how abnormal content and structure may impact lung pathophysiology. For these reasons, we hypothesized that differences in ECM structure, and collagen microstructure in particular, underlie the different natural histories, prognoses, and responses to treatment of UIP and COP.
To explore this question, we used Second Harmonic Generation (SHG) Microscopy (SHGM) to compare the matrix of UIP and COP to that of healthy lung tissue. SHGM is a variant of two photon (2P) microscopy that can detect the fibrillar collagens (FCs) without exogenous labels. The fibril-forming collagens include collagen types 1–3, 5, 11, 24, and 27 [
16], and at least several of these FCs such as types I, III, and V are key players in lung fibroses including usual interstitial pneumonia (UIP) and cryptogenic organizing pneumonia (COP) [
1‐
3,
11]. SHGM can be used to interrogate changes in collagen’s macrostructural properties (e.g. collagen fiber density, arrangement, and organization), as well as collagen’s subresolution microstructural properties (e.g. the diameter, order versus disorder, and/or packing density of collagen fibrils within larger collagen fibers) [
17‐
22]. These microstructural features of individual collagen fibers, as they can influence SHG directionality from that fiber (i.e. F
SHG/B
SHG, defined below), are herein collectively referred to as collagen “microstructure”. In this aspect, SHGM is unique in its ability to interrogate subresolution structure of FCs (e.g. col1 and col3) in intact and potentially live samples without exogenous labels, abilities which also make SHGM an attractive potential clinical and investigational diagnostic tool. Thus this technique can utilize intrinsic properties of matrix components to characterize the content and organization of the ECM in these fibrotic lung diseases.
Using SHGM, herein we describe important differences in matrix content and organization in UIP/IPF and COP compared to healthy lung tissue. Specifically, we found differences in collagen’s subresolution structural properties in UIP compared to COP and healthy lung as assessed by SHGM and the F
SHG/B
SHG ratio. Importantly, even adjacent normal UIP tissue exhibited these differences in collagen microstructure compared to healthy lung, thus introducing the compelling possibilities that altered collagen microstructure might lead to or correlate with fibrosis in the relatively intractable disease UIP, but not in the more treatable COP. We also report different col1:col3 ratios in UIP versus COP and healthy lung tissue, which is important especially in the context of our F
SHG/B
SHG data, because others have reported that altered col1:col3 ratios can drive (or perhaps be driven by) changes in FC microstructure such as fibril diameter [
23‐
25], which is one aspect of collagen microstructure interrogated by the F
SHG/B
SHG ratio [
17‐
22]. Finally, we show both UIP and COP have differences in mature elastin fiber content, and elastin:collagen ratio, suggesting that both fibrotic disease have identifying physiological differences in matrix structure suggestive of lung disease, but only the less tractable disease, UIP, exhibits differences in underlying collagen microstructure. These results are important because they provide new insights into the potential biological and biostructural underpinnings of refractory versus “treatable” lung fibroses, with an emphasis on subresolution collagen microstructure, and demonstrate SHGM’s potential as a powerful new tool for aiding in the diagnosis and treatment of lung fibrosis.
Discussion
Pulmonary fibrosis is characterized by accumulation of ECM proteins in lung tissue. The mechanisms leading to pathologic (or non pathologic) accumulation and organization of matrix proteins remain poorly understood. Although we have some insight into the composition, structure and/or organization of the matrix, many properties of the matrix remain uninvestigated. Numerous matrix proteins likely contribute to organ dysfunction in pulmonary fibrosis, however, we are only beginning to understand how homeostasis and organization of these proteins impact cellular function.
Collagen, produced and organized mainly by fibroblasts and scar-forming myofibroblasts, is one of the most abundantly studied matrix proteins. At least twenty-eight different collagen subtypes have been described to date. All collagen species contain three alpha peptide sequences forming a triple helix. Collagen type is determined by the type(s) of alpha peptides and post translational modifications, hydroxylation, and/or glycosylation. Further modification of collagen structure occurs after release into the extracellular space. Here, crosslinking and joining of the helices occur to form collagen fibrils and larger collagen fibers, and fibrosis (aberrant excess deposition of collagen) may occur.
Due to their ability to provide information on the intrinsic content and structure of collagen and other endogenous ECM proteins without exogenous labels or tissue destroying procedures (see description of SHGM and TPEF in
Background and
Results, respectively), there has been increasing interest in using SHGM and TPEF to provide insights into the matrix structure of healthy and diseased lung, both for fibrotic as well as non-fibrotic lung diseases [
28,
36,
44‐
48] (and for lung cancers, not discussed herein). One group reported that combined TPEF and SHG identified “characteristic features of fibroblastic foci in human Idiopathic Pulmonary Fibrosis samples” [
47], whereas another report used F
SHG and B
SHG signals to differentiate Col1 from Col3 in lung tissue from patients with chronic obstructive pulmonary disease (COPD) [
48]. Also in COPD, Tjin et al. found the F
SHG/B
SHG ratio was different in COPD versus non-diseased lung tissue [
28]. However, reports demonstrating a diagnostic capability of quantifiable SHGM and TPEF parameters to discriminate between different clinical lung fibroses, or perhaps more importantly to provide insights into the underlying etiology or structure-function origins of lung disease, are still scarce.
In this report, we extend these works by using SHGM and TPEF imaging to identify key differences in the ECM of UIP compared to COP and healthy control lung tissue. UIP and COP were chosen because they are both characterized by increases in matrix proteins, particularly FCs, yet they have contrasting natural histories, responses to corticosteroids, and prognoses. The reasons why UIP is progressive and difficult to treat are not clear. One possible explanation is that there may be a fundamental difference in collagen’s content, structure, and/or organization in the UIP ECM that renders collagen more structurally more resistant to degradation in UIP versus COP. We tested this hypothesis using SHGM, a microscopy approach that is sensitive to the intrinsic FC organization and microstructure within the matrix, to confirm whether FC in UIP has different microstructural properties versus COP or healthy lung.
Using this approach, we have demonstrated that FC microstructure in the ECM of UIP is significantly different from FC microstructure in either COP or healthy control lung tissue, as evidenced by the F
SHG/B
SHG ratio. Changes in this F
SHG/B
SHG ratio suggest that there is a significant difference in the density, structure, and/or organization of FC in UIP compared to COP and healthy lung tissue, particularly with regard to the effective diameter or packing arrangement/density of collagen fibrils in the ECM [
18‐
22,
26]. These results are compelling because while previous studies (discussed above) have elegantly demonstrated the utility of SHGM for investigating lung fibroses, or have shown expression changes in several collagen subtypes in fibrotic lung diseases, to our knowledge this is the first report of abnormalities in ECM and FC microstructure in UIP as being quantifiable and differentiable from other lung fibroses (and from healthy) by SHGM, specifically F
SHG/B
SHG. Still more compelling is the fact that only the intractable fibrosis (UIP) demonstrated significant differences in FC microstructure versus healthy lung, whereas the treatable fibrosis (COP) did not, thus providing compelling and to our knowledge seminal evidence that alterations in collagen’s fundamental underlying structure may contribute to whether or not pulmonary fibroses are treatment responsive. These results provide previously unavailable insights into the biological underpinnings of treatment-resistant pulmonary fibrosis, and also highlight the potential of SHGM as a novel clinical diagnostic and investigational tool for distinguishing between intractable and treatable lung fibroses.
We also found that lung tissue with preserved alveolar architecture adjacent to UIP fibroblastic lesions and honeycomb areas all have different FC microstructure (i.e. F
SHG/B
SHG) versus healthy lung. Moreover, there was no difference in F
SHG/B
SHG between UIP fibroblastic foci, honeycomb areas, and adjacent areas with preserved alveolar architecture (Fig.
4). Together these results suggest the possibility that pre-existing alterations in FC structure even in “normal” lung tissue may foreshadow or precipitate (or at minimum, associate with) development of UIP. In other words, underlying collagen structural deficits – perhaps present in pre-UIP tissue well before the onset of UIP – might be a biomarker that predicts or predisposes development of future UIP.
As expected, both col1 and col3 were elevated in UIP and COP versus healthy lung, with col1 deposition being predominant to col3 in UIP, and vice-versa in COP, as has been previously reported [
1]. These results are significant in the context of our other results reported herein because it is known that changes in FC ratios, particularly col1:col3 ratios, plays a significant role in regulating collagen fibril diameter [
23‐
25] (one component of collagen microstructure). Similarly, by regulating the availability of fibroblast (or other effector cell type) binding sites on collagen fibrils, changes in collagen’s subresolution fibril microstructure could in turn control relative levels of FC expression. In other words, different col1:col3 ratios may in turn drive or be driven by altered collagen microstructure in UIP. Together with the earlier data, these results demonstrate that the ECM of UIP not only contains more collagen (particularly more col1) than the ECM of COP and/or healthy lung tissue, but also that there are significant differences in the subresolution microstructure of these collagen fibrils (i.e. fibril diameter, density, and/or organization, as interrogated by F
SHG/B
SHG) in UIP versus COP and healthy, independent of the absolute amount of collagen deposition in each disease.
Finally, we demonstrated that mature elastin content in both UIP and COP is reduced compared to healthy controls (Fig.
6). Elastin’s intrinsic autofluorescence originates from pyridoxine-based pyridolamine cross-links [
35,
42] found primarily in mature elastin fibers [
43], therefore TPEF of endogenous lung elastin preferentially identifies mature elastin fibers in lung tissue. These results are consistent with the concept that breakdown of mature elastin fibers in the lung, and their “replacement” with often excess deposition of immature elastin fibers and elastin precursors, is believed to contribute to reduced lung function in a variety of pulmonary diseases [
49]. In other words, increased elastosis (i.e. breakdown of mature elastin fibers), as has been demonstrated for both UIP and COP [
50], most likely leads to a compensatory increase in elastin production in an (ultimately unsuccessful) effort to restore the mature elastin fibers which have been lost.
Hence our results here together with these previous studies all support the concept of increased elastin turnover (i.e. synthesis and “deposition” of “immature” elastin components) consequent to breakdown and loss of mature elastin fibers in UIP and COP, with resultant deficits in pulmonary function. Indeed Enomoto et al. show a significant increase in very fragmented appearing elastin (i.e. likely to be fragmented mature elastin fibers and/or deposition of immature elastin precursors; see Fig.
1 in [
51]) associated with a decline in lung function in IPF [
51]. Others have also reported apparently increased elastin production, for example increased elastin gene expression and protein expression [
52], as well as increased enzymatic breakdown of mature elastin in COPD and IPF [
50], in these and other [
53] pulmonary fibroses. Finally Eurlings et al. recently reported increased collagen and decreased elastin in aveolar and small airway walls of chronic obstructive pulmonary disease (COPD) [
54], another disease with fibrotic pathology, similar to what we demonstrate herein for COP and UIP (Fig.
3 and Fig.
5).
Taken together with our findings on different FC microstructure in UIP but not COP versus healthy lung, these observations on elastin content are especially compelling because they demonstrate that compared to healthy lung, both fibroses (UIP and COP) have significant identifying physiologic disruptions in ECM structure and organization that are quantifiable with non-invasive and non-tissue destructive combined SHG and TPEF microscopy. Yet only the more intractable UIP fibrosis has disrupted FC microstructure identifiable by F
SHG/B
SHG, and thus together these techniques may represent novel clinical diagnostic tools for distinguishing between intractable and treatable lung fibroses. The continual pre-clinical advancement of SHG and TPEF endoscopic technology makes the prospect of such a diagnostic tool for distinguishing between intractable and treatable lung fibroses all the more compelling [
17,
55].
While both UIP and COP are fibrotic diseases, UIP is a fatal disease affecting the alveolar walls and subpleural areas, while COP is a treatable disease affecting the alveolar spaces and bronchiolar lumen. Thus while both have fibrosis, their topography and natural history are distinct, and therefore any inter-disease comparative study of this nature cannot completely exclude the possibility that some measured differences might be attributable at least in part in part to temporal or anatomical/topographical differences in the lung tissues examined between the patient groups. Moreover, incorporating additional corresponding clinical data such as symptoms or pulmonary function testing (unavailable for these current patient sets) will allow us to strengthen our findings and interpretations in future studies. Nonetheless the possibility that non-invasive and non-tissue destructive combined SHG and TPEF microscopy, utilized either ex vivo or perhaps ultimately in vivo, may be able to distinguish and/or predict onset or outcome of tractable versus fatal lung fibroses remains compelling.
In summary, using SHG and TPEF microscopy, herein we identify several previously unreported key differences between UIP, COP and healthy lung tissue. The collagen microstructure differences we observed in UIP ECM provide novel insights as to why this pathology may be resistant to many therapies. For example, an ECM and/or collagen fibrils that are more densely packed, more ordered or disordered, and/or more cross-linked may be more resistant to homeostatic turnover and exhibit differences in matrix stiffness that are key to modifying cellular activity of resident cells and activation of pro-fibrogenic cytokines such as transforming growth factor beta (TGF- β. Identifying all the microstructural changes present in UIP and/or the mechanisms that regulate them will be a critical part of our future research. These ongoing studies will seek to determine more specifically exactly what features of collagen’s microstructure (e.g. fibril diameter, fibril density, and/or hetero- or homo-typic fibril composition or organization) are different in UIP versus COP and healthy lung, and identify molecular targets that may effect these changes in collagen’s underlying microstructure. Although further studies are required to ascertain whether or not the altered FC microstructure as we demonstrate here is an underlying cause of (rather than just associated with) differences in natural history, treatment responsiveness, and/or prognosis between UIP and COP, at minimum these results introduce the intriguing possibility of using SHG microscopy as a novel clinical biomarker that may help predict treatment responsiveness of idiopathic fibrotic lung disease.
Competing interests
The authors declare that they have no competing interests.
Author’s contributions
R M K; study design and manuscript author; J S; SHG imaging, data acquisition and analysis; K O; data acquisition; P S; statistical data analysis; G Q X; data acquisition, study design; R P P; manuscript review and revision, study design; P J S; manuscript review and revision, study design; E B B; manuscript review and revision, study design; S W P; SHG study design, data analysis, manuscript author. All authors read and approved the final manuscript.