**3. Discussion**

The ECM has important biological functions such as regulating wound healing responses and tissue remodeling through cellular interactions [7,29]. The implementation of decellularized IPF and healthy lung tissues as sca ffolds represents a promising approach to study the biological function of the ECM and how vital changes in matrisome properties, both in composition and biomechanically, influence cell behaviors [12]. In this study, we performed an in-depth characterization of the structural properties of these acellular lung matrices derived from healthy individuals and IPF patients, with regards to morphology, tissue density, and sti ffness. Alterations of these important features, linked to the pathophysiological changes seen in IPF sca ffolds, were sustained following decellularization. IPF is thought to be the result of an aberrant wound healing process involving abnormal deposition of matrix proteins e.g., collagens [1,30], and as seen in this study leading to almost a three-fold increase in tissue density compared to healthy, accompanied by a five-fold and 60% increase in sti ffness for native IPF tissue and IPF sca ffolds, respectively. Decellularization seemed to solely a ffect the biomechanical properties of healthy sca ffolds. Essentially, collagen content is retained after the decellularization process in healthy lung tissue, while elastin content is a ffected to some degree and even more the proteoglycans along with the ECM glycoproteins [31]. Our data sugges<sup>t</sup> that removal of these charged proteins most likely led to changes in biomechanical properties due to lost electrostatic interactions, leading to an entangled micro-structure, ultimately increasing the tensile strength of healthy sca ffolds, which might influence cell-matrix interactions. The IPF sca ffolds, on the contrary, had a higher content of collagens compared to healthy sca ffolds and a largely absent BM along with the removal of surfactant proteins, which gives specific biomechanical properties of the sca ffolds. Consistent with morphological di fferences of more or less fibrotic sca ffold samples, the biomechanical properties of the sca ffolds demonstrated large variations. Despite this, both within IPF patients and within biological replicates, the IPF sca ffolds were distinctly separated from healthy sca ffolds, actually exhibiting rather homogenous tissue characteristics, strengthening the results from our limited number of biological replicates (Figure S5).

One of the important findings was that there were di fferences in the abundance of distinct matrisome proteins between healthy and IPF sca ffolds, with nine out of 20 of these being BM components. Visualized with collagen type IV antibody labeling, the BM showed large spatial di fferences between healthy and IPF sca ffolds. The loss of BM integrity of the alveolar-capillary membrane along with an accumulation of collagen type VI, without normal structure reconstruction, causes an abnormal lung architecture, thought to promote fibrosis [30], where fibroblasts and especially myofibroblasts are known to be the main matrix producers and key players in fibrosis [4]. With that in mind, our aim was to examine how changes in ECM properties a ffected cellular responses in IPF. When IPF sca ffolds were repopulated with healthy fibroblasts, we demonstrated a significantly reduced production of important BM complexes such as nidogens, laminins, and collagen IV in IPF sca ffolds, results that are in support of a previously reported study [32]. In our study, laminin α3, α5, and β3 were not produced at the same level in IPF sca ffolds as in healthy sca ffolds. Data, which is in line with an in vivo study, where the loss of laminin α3 augmented the progression of lung fibrosis, is suggestive of its contribution to IPF disease progression [33]. Underlying the BM are anchoring matrix components perlecan and collagen type VI chains α1 and α3, which were elevated in repopulated IPF sca ffolds, indicating an imbalance in ECM turnover with a build-up of matrix underlying the fragmented BM. Furthermore, an early induction in synthesis of proteoglycans decorin, lumican, biglycan, and versican (Figure 6C, Figure S3), as well as the ECM regulator, TIMP-3, (Figure S3) in IPF sca ffolds compared to fibroblasts cultured on healthy sca ffolds further strengthen the picture of a promotion of a profibrotic feedback loop. Proteoglycans are multifunctional proteins involved in wound healing responses and shown to be elevated in lungs from IPF patients [34–37]. Fibroblasts from lung fibrosis patients have shown an increased production of small proteoglycans, with decorin as the major proteoglycan produced with implications in pulmonary fibrotic responses [35,38]. These

findings were replicated in our study and further support the hypothesis that fibroblast activity is modified by certain elements of the ECM.

In a previous study, where primary fibroblasts were cultured on healthy lung scaffolds freely floating in culture medium [13], cells contracted the surface area of the scaffold from approximately 1 cm<sup>2</sup> to 1 mm<sup>2</sup> in less than 9 days. To prevent this and to more closely mimic the physiological conditions, we introduced custom-made holders to mount the scaffolds in order to sustain a stretched and organized lung tissue structure during cell culture. This approach clearly demonstrated the importance of imposing a static stretch of the scaffolds, sustained by the holders, to transduce forces similar to the native situation. It has been shown that cells sense resistance to pulling as well as the local environment due to protein conformation, substrate rigidity, and architecture [39]. Repopulated scaffolds showed equivalent numbers of cells attached in both healthy and IPF scaffolds, verified by cellular viability and histone levels over time. In addition, patterns of focal adhesions, shown with vinculin staining, did not appear to be different in the groups. Although, the cell morphology appeared to be similar in both types of scaffolds, fibroblasts on IPF scaffolds seemed to have a higher accumulation of vimentin, indicating a shift in the cellular response due to an increased stiffness, also seen by others [40]. Compositional alterations of the ECM affect mechanical properties of tissues, which in turn influence how the cells perceive its local environment in terms of forces and ECM tension through integrin-ECM interactions that in turn will have an impact on the intracellular signaling [24,39]. An enhanced matrix stiffness with reduced tissue compliance is known to promote fibroblast activation and fibrosis [41]. We demonstrated an increased tissue stiffness in IPF scaffolds, as recognized in other studies of native lung tissue from IPF patients [42], which in turn had an effect on fibroblast activity. The stiffness of the ECM is increased in areas of fibrosis [43] and stimulates fibroblast migration, differentiation, and other cellular events that are associated with tissue remodeling [6]. Invasive migratory fibroblasts degrade and disrupt surrounding barriers to propagate its migration toward stiff and fibrotic areas of the lung [44]. In our study, repopulated IPF scaffolds synthesized lower amounts of the MMP (metalloproteases) inhibitor TIMP-3 (Figure S3). The shift in ECM regulators connect to our previous study by Ahrman et al. and also to other studies showing decreased levels of tissue inhibitors and increased levels of proteases in IPF [8,10,45]. Deprived balance of MMPs and TIMPs, enzymes necessary for matrix reorganization, contribute to a pathological turnover rate of the ECM [45,46]. These factors with both direct and indirect regulation of ECM structures, including activation of growth factors, cytokines, and chemokines, have been suggested to have an important role in the development of fibrosis, however, with diverse and complex functions [47]. We also saw that ECM regulators and secreted factors decreased at day 9 only in the IPF light matrisome, indicative of a high enzymatic activity in fibroblasts cultured on IPF scaffolds, thus resulting in enhanced release and/or removal of ECM components to the medium compared to fibroblasts cultured on healthy scaffolds. We saw that the fibroblasts filled up the spaces in less dense areas and covered dense areas with a compact cell sheet in the IPF scaffolds, while the alveolar structure was maintained in the healthy scaffolds. This feature may be explained by the loss of an intact BM in combination with an increased stiffness in the IPF scaffolds. Collectively, these characteristics may contribute to a dysregulated and imbalanced proteolysis of matrix proteins, changes that we saw in the temporal expression of light labeled BM proteins in between the two types of scaffolds (Figure S4).

Interestingly, when cultured on IPF scaffolds, fibroblasts showed an increased production of tenascin and periostin, proteins which are upregulated in IPF patients, with the latter recognized as a disease marker for IPF progression [25,26]. Tenascin is a large ECM glycoprotein transiently expressed during wound healing and involved in several tissue remodeling processes, which was reflected in our system where the fibroblasts responded to the altered milieu in the IPF scaffolds. This protein stimulates migration of fibroblasts and increased mechanical stiffness, seen in vitro, and is upregulated in patients with IPF, especially at fibroblastic foci [26,27,48]. The matricellular protein periostin is able to bind to tenascin facilitating its incorporation to the ECM [49]. In seemingly healthy looking areas in the IPF scaffolds we found strong staining for periostin, whereas in heavily remodeled and

fibrotic areas the staining was absent. The spatial distribution of periostin may direct the progression of fibrosis by acting as an early trigger for matrix build-up, seen with an initial high production in IPF scaffolds compared to healthy, which supports our observation that fibroblast migrate to less dense areas in IPF scaffolds.

In accordance with previous transcriptome studies by Parker et al. [29], our results support the notion of the ECM being a key driver and regulator of fibrosis, causing a positive feedback loop between fibroblasts and the diseased ECM (Figure 7), which warrants further investigation. We hypothesized that the biomechanical properties and the composition of the ECM dictate the cellular response in human primary fibroblasts as reflected by the overall cellular response to a healthy and diseased matrix.

**Figure 7.** Cell-matrix interactions in IPF. Fibroblasts cultured on stiff IPF scaffolds secrete increased amounts of periostin, known to stimulate myofibroblast differentiation and migration. Increased synthesis of tenascin and reduced levels of metalloprotease inhibitors (TIMP-3) support migration and movement toward stiffer tissue. Fibroblasts become activated and generate increased deposition and build-up of collagens and proteoglycans (decorin, versican, and biglycan). Alveolar epithelial cell (AEC) damages causes basement membrane disruption and the loss of structural barriers. Fibroblasts on IPF scaffolds reduced their production of BM complexes (laminins, nidogens, and collagen IV), potentially hindering the rebuild of a functional BM for anchoring AEC.

We saw that fibroblasts cultured on sti ff IPF sca ffolds secreted increased amounts of periostin, known to stimulate myofibroblast di fferentiation and migration. This in combination with an increased synthesis of tenascin and reduced levels of metalloprotease inhibitors (TIMP-3) supports the migration of fibroblasts towards a sti ffer ECM. The cells become activated and generate increased deposition and build-up of collagens and proteoglycans including decorin, versican, and biglycan. With the elevated levels of periostin, the incorporation of tenascin-C into the matrix may be facilitated as well as the formation of collagen fibrils, assisted by decorin. The IPF sca ffolds have a clear disruption of the BM, which is thought to arise from alveolar epithelial cell (AEC) damages leading to the loss of structural barriers. This may promote transdi fferentiation of epithelial cells to mesenchymal cells which in addition activates the progression of remodeling. Interestingly, fibroblasts on IPF sca ffolds reduced their production of BM complexes (laminins, nidogens, and collagen IV), potentially hindering the rebuild of a functional BM for anchoring AEC. The continuation of fibroblast activation in combination with a disorganized BM seem to propagate changes in the matrisome properties, further promoting disease progression.

This study is separate from previously performed studies on pulmonary fibrotic lung tissue [12,29] as we focused on the distal lung properties of IPF, where this disease typically manifests itself with subpleural fibrotic formations. Furthermore, the advantage of our human 3D-model excludes the e ffect of resident cellular components of the parenchymal tissue characteristics and focuses on the cellular response of the ECM.

We have clearly demonstrated that the biomechanics and the matrisome composition of the IPF sca ffolds are closely connected, which make up an intricate biological system controlling cellular behavior with the ability to sustain a profibrotic lung environment. To mimic the physiological conditions more closely and to maintain the complex structure of decellularized lung tissue during repopulation, a novel approach was implemented in this study through the application of sca ffold holders, manufactured to mount lung tissue in order to impose a static stretch. By combining the biomechanical properties of a sca ffold, linked with its own proteomic profile, unique matrisome properties were identified in IPF patients in comparison to healthy individuals. In a novel way of analyzing proteomic data, tissue density adjustments enabled an in-depth study of ECM turnover in IPF and healthy sca ffolds by recognizing structural heterogenetic di fferences, which thereby separated the two types of tissues. The cellular responses studied in repopulated sca ffolds identified the cell-matrix interactions as essential in the progression of IPF, emphasizing the BM and the underlying composition of ECM proteins as a possible disease mechanism in the induction of normal versus fibrotic tissue remodeling. We demonstrated that the IPF sca ffolds had an enhanced content of proteoglycans and after the repopulation with healthy fibroblasts, we also distinguished a shift in the synthesis of proteoglycans, accompanied with a distinct localization of the tissue deposition. Together, these results further support that the existing cellular milieu alters fibroblast activity, promoting a profibrotic phenotype when cultured in a diseased matrix. More in-depth examinations with a larger number of patients on how specific ECM components may direct cellular activity are warranted to further elucidate how healthy cells become programed to synthesize a disease-like protein profile in a diseased ECM environment, studies which may lead to the unveiling of potential targets and biomarkers for IPF.
