Versican expression from lung fibroblasts suppresses pulmonary fibrosis

Abstract

The activation and accumulation of lung fibroblasts, leading to excessive ECM deposition, is a pathogenic hallmark of Idiopathic Pulmonary Fibrosis, a lethal and currently untreatable disease. In this report, increased expression of Versican, a multifunctional ECM proteoglycan, is detected in both human and mouse pulmonary fibrosis, mainly in monocytic cells and fibroblasts. Ubiquitous genetic reduction of Versican expression in mice promotes collagen expression and polymerisation, alters pulmonary ECM composition and structure, and exacerbates pulmonary fibrosis, delaying its resolution. Moreover, the decrease in Versican in the ECM and the ensuing reorganisation stimulate Tenascin-C expression from fibroblasts, which is further shown to be a potent Toll-like receptor 4-dependent podosome inducer, promoting ECM invasion. Thus, fibroblast-expressed Versican regulates the underlying ECM composition and structure and suppresses autologous podosome formation, limiting ECM invasion and pulmonary fibrosis.

Idiopathic pulmonary fibrosis is characterised by the accumulation of fibroblasts, which deposit excessive extracellular matrix impairing respiratory functions. Here, the authors show that fibroblast-expressed versican, a chondroitin sulphate proteoglycan, suppresses fibroblasts’ ability to invade and further grow the underlying matrix, thus limiting their accumulation and attenuating pulmonary fibrosis.

Introduction

Idiopathic pulmonary fibrosis (IPF) is a chronic and progressive interstitial lung disease with a poor prognosis and limited therapeutic options. IPF is characterised by the accumulation and activation of lung fibroblasts, resulting in the excessive deposition of extracellular matrix (ECM) components in the interstitium, which leads to the distortion of lung architecture and impairment of respiratory functions^1,2. The accumulation of fibroblasts, beyond proliferation, resistance to apoptosis and senescence, is heavily regulated by the underlying ECM, which controls growth factor bioavailability and bioactivity and perpetuates fibroblast activation through mechanical cues³. In this context, it was recently shown that TGF and the fibrotic ECM promote the formation of podosomes on lung fibroblasts, stimulating ECM invasion and pulmonary fibrosis⁴. IPF fibroblasts presented with prominent podosome rosettes that could be retained ex vivo in the absence of any stimulation, suggesting podosome formation as an inherent property and a significant pathological attribute of IPF fibroblasts^4,5.

The composition of the ECM varies considerably between different organs, reflecting the specific structural and functional requirements of each tissue microenvironment^6–8. In the lung, ECM is organised into two spatially distinct structures with differing compositions and functions: the interstitial ECM and the basement membrane. The basement membrane is a dense, highly stable layer underlying the basal surface of epithelial cells. It is indispensable for tissue compartmentalisation and the establishment of cell polarity, while also functioning as a critical barrier to cancer metastasis^6,7. The interstitial ECM is a highly dynamic macromolecular structure comprising over 300 proteins (the core matrisome), predominantly fibrous proteins and proteoglycans (PGs)^8–10. Collagens are the most abundant fibrous ECM proteins (50–60% of total ECM) that form intermolecular connections through direct interactions or covalent cross-linking, forming higher-order fibrillar macromolecular structures that act as structural scaffolds for cells and thereby provide the mechanical stability and elastic recoil necessary for proper lung function. Collagen abundance and structural conformation are sensed by different cell receptors, such as integrins and discoidin domain receptors (DDR), which activate numerous pro-fibrotic signalling pathways influencing survival, cell migration and invasion^8–10.

The fibrillar ECM networks are surrounded by PGs (10–20% of total ECM), which consist of a core protein to which one or more glycosaminoglycan (GAG) chains are covalently attached^9,10. GAGs facilitate interactions with various ECM components, filling up ECM-cell structural gaps and fine-tuning the overall tissue architecture^9,10. Moreover, the negative charge of the PG core-attached GAGs regulates cation binding and water retention, thus influencing the bioavailability and bioactivity of various growth factors and chemokines⁹. Versican (VCAN), a large chondroitin sulphate PG (CSPG), is the most abundant PG in the fibrotic lung. VCAN consists of three protein domains (G1–G3) and exists in five different isoforms (V0, V1–4); V0–V1 are the most abundunt^11,12. The VCAN domain organisation allows interactions with a diverse array of binding partners modifying their structure and activity, such as ECM components (collagens, the GAG Hyaluronan/HA), chemokines (CCL2, CXCL10), cytokines and growth factors (TNF, IL-6 and TGF-β), and cell surface receptors (CD44, TLR2/4, LRP-1, and integrins)^11–13, many of which have been implicated in the pathogenesis of IPF^2,3. This wide range of VCAN’s interactions leads to many differential, both pro- and anti-inflammatory, effects in different cell types, characterising VCAN as a highly versatile molecule, as reflected in its name¹¹.

In the lung, increased VCAN expression has been reported in chronic obstructive pulmonary disease (COPD), asthma, bronchiolitis obliterans syndrome (BOS) and lung transplantation¹⁴, as well as non-small cel; lung cancer patients¹⁵ and its animal models¹⁶. In the fibrotic lung, early studies have indicated VCAN expression in the thickened interstitium in the fibroproliferative phase of adult respiratory distress syndrome (ARDS), in the intraluminal buds in bronchiolitis obliterans organising pneumonia (BOOP), and in early fibroblast foci in IPF^17,18. However, the possible mode of action of VCAN in the pathogenesis of lung diseases remains largely unknown.

Therefore, in this report, we explored the possible role of VCAN in pulmonary fibrosis and IPF pathogenesis through in silico analysis of transcriptomic profiles from human patients and immunostaining of lung tissues. Human studies were complemented by animal studies employing bleomycin (BLM)-induced pulmonary fibrosis, the most widely used IPF animal model^19–21, and multiple readouts, including the measurement of respiratory functions, multi-colour staining and confocal imaging, 3D μCT imaging, mass spectroscopy (MS) proteomics and atomic force microscopy (AFM). The role of VCAN, specifically in fibroblasts, was further dissected with the ex vivo culture of primary lung fibroblasts isolated from wt or genetically modified mice, post BLM administration and/or under the stimulation of TGF-β, the prototype profibrotic factor. The effect of VCANECM abundance on fibroblast function was studied using acellular ECM (aECM) prepared from the same mice or from corresponding primary fibroblasts, whose composition and structure were analysed by MS proteomics and transmission electron microscopy (TEM).

Results

Increased VCAN expression in pulmonary fibrosis

To examine a potential role for VCAN in the pathogenesis of IPF, we first queried its mRNA expression levels in silico in the lung tissue of IPF patients in publicly available IPF transcriptomics datasets, re-analysed in parallel at Fibromine²². The employed published datasets represent the largest and most controlled publicly available datasets, querying lung mRNA expression in hundreds of patients (Supplementary Table 1). In all eight datasets examined, increased VCAN mRNA levels were detected in lung tissue from IPF patients compared with control samples (Supplementary Fig. 1a and Supplementary Table 1). Notably, the increased VCAN expression was associated with an impairment of respiratory functions in two of the most extensive datasets that contained clinical data (Supplementary Fig. 1b–d). In all eight datasets examined, where COLLAGEN expression was found to be massively deregulated (Supplementary Fig. 1f), as expected, VCAN was among the highest-expressed PGs in the fibrotic lung and the most differentially regulated (Supplementary Fig. 1e, g). In the same samples, increased expression of HA synthases (HAS1-2), HA-mediated motility receptor (HMMR), and HA and proteoglycan link protein 3 (HAPLN3) was also detected (Supplementary Fig. 1e, h), suggesting a concurrent activation of the HA synthesis and signalling pathways.

To identify the VCAN-expressing cells in the fibrotic lung, we then queried VCAN mRNA expression in publicly available IPF single-cell RNA sequencing (scRNAseq) datasets (Supplementary Table 2), which were downloaded locally from public repositories and re-analysed in parallel, retaining the original subpopulation annotation. VCAN was found to be expressed predominantly by monocytic lineage cells and fibroblasts (Fig. 1a, b); similar results were obtained from all four analysed datasets (Supplementary Fig. 2a, b), which are among the largest and most widely reused available, querying cell-specific gene expression in thousands of single cells. Focusing on mesenchymal cells, VCAN expression was detected in COL1A1/HAS/CD44-expressing fibroblasts and myofibroblasts (Fig. 1c and Supplementary Fig. 2c) and was found upregulated in IPF myofibroblasts (Fig. 1d and Supplementary Fig. 2d). In agreement with the in silico results, intense VCAN immunostaining was noted in fibrotic areas of IPF lung tissue samples compared to control samples (Fig. 1e). VCAN staining was higher within regions of the thickened alveolar septa and fibroblastic foci (Fig. 1e), in overall agreement with previous studies^17,18,23. VCAN staining was particularly high at the rims of well-formed foci. Weak staining was also detected in endothelial cells, whereas intense staining was observed in macrophages (Fig. 1e).

Fig. 1. Increased VCAN expression in IPF.

Single-cell data re-analysis of VCAN expression in a publicly available IPF scRNAseq dataset (PMID: 32832599; Supplementary Table 2); see also Supplementary Fig. 1-2. a UMAP plot of VCAN mRNA expression in the indicated identified lung cell populations. b Single-cell specificity of VCAN mRNA expression in the human lungs. Red arrows indicate cell types in which VCAN is a statistically significant marker gene (*adj.p = 2.68E⁻²⁴⁴/4.22E⁻¹³⁹/<1E⁻²⁵⁰/<1E⁻²⁵⁰ from top to bottom). c Relative mRNA expression levels in lung mesenchymal cell types of VCAN and other profibrotic genes. d Differential expression analysis of VCAN in mesenchymal cell types at the single-cell level (*adj.p = 1.59E⁻¹³). Marker and differentially expressed genes at the single cell level were identified using the two-sided Wilcoxon Rank Sum test. Significance thresholds were set at FC > 1.2 and Bonferroni-adjusted p < 0.05 (*). e Representative images from immunohistochemistry for VCAN (brown) in IPF and healthy (CTRL) lung tissue; scale bars 50,100 μm. A representative experiment out of two successful independent ones is shown. In all panels, all samples are biologically independent; Source data for all panels are provided as a Source Data file.

BLM-induced pulmonary fibrosis was also found in silicoto stimulate Vcan mRNA expression in the lung in all related transcriptomic datasets examined (Supplementary Table 1 and Fig. 2a). To validate in silico mouse findings, pulmonary fibrosis was induced with BLM in C57Bl/6 mice (Fig. 2b), the most widely used model of IPF^19–21. The dose (0.8 U/Kg) and administration route (oropharyngeal aspiration; OA) were selected after extensive local testing over the years to minimise lethality while preserving a robust, reproducible fibrotic profile. In these settings and in our animal facilities, BLM-induced inflammation peaks at day 7 post-BLM, while fibrosis peaks at day 14 and resolves by day 21^4,21,24,25. BLM-induced fibrosis stimulated Vcan mRNA expression in the lungs (Fig. 2c), which returned to baseline levels upon the resolution of fibrosis (Fig. 2c), following the expression pattern of Col1a1 (Fig. 2d). The expression levels of all V0, V1 and V2 isoforms were found upregulated (Fig. 2e, f), whereas the V0–V1 isoforms were the most abundant (Fig. 2g); profibrotic gene expression was also evident in the same samples (Fig. 2h). Accordingly, increased staining for PGs was noted in fibrotic areas (Fig. 2i), where intense VCAN immunostaining was also detected (Fig. 2j), localised in fibrotic areas, epithelial cells and macrophages, in overall agreement with the human samples.

Fig. 2. Increased Vcan expression in BLM-induced pulmonary fibrosis.

a Vcan mRNA expression levels in lung mouse tissue post-BLM, as compared to controls (FC > 1.2 and FDR < 0.05), in publicly available transcriptomics datasets (n = 3; Supplementary Table 1). b Schematic representation (biorender.com) of the locally employed BLM-induced pulmonary fibrosis model. c, d Vcan and Col1a1 mRNA expression at different time points post-BLM were interrogated using Q-RT-PCR. Values were normalised to the expression of the housekeeping gene B2m and presented as fold change over the control (n  =  4, 5, 6, 5 biologically independent mouse lungs). Statistical significance was assessed with one-way ANOVA followed by Welch’s correction and coupled with a post-hoc Games-Howell test; ^*p = 0.0136/0.0480/0.0145/0.0473 (c) and a Kruskal–Wallis one-way ANOVA coupled with a post-hoc Dunn’s test; ^*p = 0.0335/0.0280 (d). e Schematic representation of the different Vcan mouse isoforms (biorender.com) and the location of the Q-RT-PCR primers. f Vcan isoform mRNA levels in lungs of saline (SAL; n = 5) and BLM (n = 8) treated mice were detected with Q-RT-PCR; Values were normalised to the expression of B2m and presented as fold change over control; statistical significance was assessed with a two-tailed unpaired t-test (V0, V1); ^*p = 0.0061 (V0), 0.0182 (V1), with a two-tailed unpaired t-test with Welch’s correction (V2); ^**p = 0.0011 or with a two-tailed unpaired Mann–Whitney test (V3). g Isoform relative expression in BLM-treated lungs (n = 8 biologically independent mouse lungs) was detected with Q-RT-PCR; values were normalised to the expression of B2m and presented as 2^−ΔCq values; statistical significance was assessed with Kruskal–Wallis one-way ANOVA coupled with a post-hoc Dunn’s test; ^**p = 0.0057/0.0026, ^****p < 0.0001. h Col1a1, Has2 and Tnc mRNA expression in lungs of SAL (n = 4) and BLM (n = 6) treated mice, as detected with Q-RT-PCR; Values were normalized to the expression of B2m and presented as fold change (log2) over control; statistical significance was assessed with two-tailed unpaired t-test with Welch’s correction; ^*p = 0.0272 (Col1a1), 0.0358 (Has2) and ^**p = 0.0029 (Tnc). i Representative images of Alcian blue staining (blue) in the lungs of the indicated mice. j Representative images from double immunofluorescence staining in WT mouse lungs post-SAL/BLM administration for VCAN (green) and aSMA (Acta2; red) counterstained with DAPI. Scale bars 50 μm; a representative experiment out of three successful independent ones is shown. In all panels, all samples are biologically independent; boxplots visualise the median of each distribution; upper/lower hinges represent 1st/3rd quartiles; whiskers extend no further than 1.5 × IQR from the respective hinge. Source data for all panels are provided as a Source Data file.

Fibrosis stimulates Vcan expression in lung fibroblasts

To focus on fibroblasts, primary lung fibroblasts were isolated from wt mice at the peak and resolution phases of BLM-induced pulmonary fibrosis (14 d, 21 d; pBLM), as well as from control saline-treated mice (pSAL); fibroblasts were cultured ex vivo for exactly three passages in the absence of any additional stimulation, to obtain sufficient numbers and to deplete cultures from macrophages, but also to avoid the observed senescence in longer (>3) passages. pBLM^d14 fibrotic fibroblasts were found to express significantly increased mRNA levels of Vcan, as compared with control fibroblasts and fibroblasts isolated at the end of the resolution phase (Fig. 3a); Vcan expression followed the expression pattern of Col1a1 and other profibrotic markers (Fig. 3b, c). pBLM^d14 fibroblasts were found to proliferate faster (Supplementary Fig. 3a), to adhere stronger (Supplementary Fig. 3b), and migrate more efficiently (Supplementary Fig. 3d) than control fibroblasts, as previously shown⁴. pBLM^d14 fibroblasts presented ex vivo with prominent F-actin stress fibres (Fig. 3d, e), indicating an activated status, and prominent podosome rosettes (Fig. 3d, e and Supplementary Fig. 3d, e), as previously reported⁴. As shown with immunocytochemistry, these activated, podosome-bearing, fibrotic pBLM fibroblasts expressed significantly higher levels of VCAN (Fig. 3d) and COL1A1 (Fig. 3e), confirming mRNA results (Fig. 3a, b).

Fig. 3. Fibrosis stimulates Vcan expression in lung fibroblasts.

a–e Primary normal mouse lung fibroblasts (NMLFs) were isolated from wt mice post Saline (pSal) or Bleomycin (pBLM) administration (14 d and 21 d) and cultured ex vivo. a–c Vcan, Col1a1, and Tnc mRNA expression levels were interrogated using Q-RT-PCR; values were normalised to the expression of B2m and presented as fold change over the control (n  =  4, 5, 4); statistical significance was assessed using one-way ANOVA coupled with a post-hoc Tukey’s test; ^***p = 0.0004, ^****p < 0.0001 for a, ^**p = 0.0016, ^***p = 0.0002 for b and ^*p = 0.0490 for c. d, e Representative composite images from double immunofluorescence staining for F-actin (red) and VCAN (d; green) or COL1A1 (e; green), counterstained with DAPI (blue); arrows indicate representative podosomes. f–k Serum-starved, sub-confluent primary NMLFs were stimulated with recombinant human TGFβ (10 ng/mL) for 24 h. mRNA levels of Vcan (f; cumulative result from 2 different experiments), Col1a1 (h), Has2 (i) and Tnc (k) were interrogated with Q-RT-PCR; values were normalised to the expression of B2m and presented as fold change (log2) over control; (n: f =  6,11; h/i = 8,7; k = 4,5); statistical significance was assessed with a two-tailed unpaired Mann Whitney test; ^*p = 0.0145 (f) or a two-tailed unpaired t-test with Welch’s correction; ^***p = 0.0002 (h), ^***p = 0.0001 (i) and ^*p = 0.0180 (k). Representative composite images of VCAN (g) and HABP (j) immunofluorescence staining of TGF-β-induced NMLFs, counterstained with DAPI. l–n Serum-starved, sub-confluent primary NMLFs were isolated from WT mice and cultured on acellular Extracellular Matrix (aECM) isolated from mouse lungs post SAL (pSal) or BLM (pBLM) administration. Vcan (l) and Col1a1 (m) mRNA expression were interrogated with Q-RT-PCR; values were normalised to the expression of B2m and presented as fold change over control (n = 8, 9); statistical significance was assessed with a two-tailed unpaired t-test with Welch’s correction; ^**p = 0.0010 (l) and ^**p = 0.0234 (m). n Representative composite images from double immunofluorescence staining against VCAN (green) and COL1A1 (red), counterstained with DAPI (blue). Scale bars = 50 μm; a representative experiment out of three successful independent ones is shown. In all panels, all samples are biologically independent; f, h, i, k n refers to biologically independent lung fibroblast isolations from three different mice per condition; boxplots visualise the median of each distribution; upper/lower hinges represent 1st/3rd quartiles; whiskers extend no further than 1.5 × IQR from the respective hinge. Source data for all panels are provided as a Source Data file.

TGF-β, the major profibrotic factor driving fibroblast activation and disease development in vivo, was found to induce Vcan expression in primary normal mouse lung fibroblasts (NMLFs) in vitro (Fig. 3f,g), accompanied by increases in Col1a1 mRNA expression (Fig. 3h), as expected. TGF-β also induced Has2 mRNA expression (Fig. 3i) and thus the synthesis of HA, visualised through the immunostaining of the HA binding protein (HABP) (Fig. 3j). TGF-β primarily stimulated the most abundant V0-V1 isoforms in both human and mouse fibroblasts (Supplementary Fig. 4a–d), as observed in the case of BLM-induced pulmonary fibrosis (Fig. 2e–g), suggesting that V0/V1 are the fibroblast’s fibrotic VCAN isoforms. Identical results were obtained in TGF-β-stimulated clones of normal human lung fibroblasts (Supplementary Fig. 4e–j), confirming TGF-β-induced VCAN expression in fibroblasts.

To examine the possible effects of the fibrotic interstitial ECM on Vcan expression, acellular ECM (aECM) was prepared, through decellularisation and proteolytic digestion, from the lungs of mice post-BLM administration (14 d; pBLM) or control littermate mice administered saline (pSAL). The lung aECM was prepared simultaneously from pSAL and pBLM lungs, and an identical amount (μg) of aECM was used for cell culture; the employed protocol^26,27 is known to preserve the basic core matrisome²⁶, including collagens, as we have recently shown⁴. The culture of wt primary fibroblasts on the pBLM aECM stimulated both Vcan and Col1a1 expression (Fig. 3l–n), indicating that, beyond profibrotic growth factors, the fibrotic ECM itself also induces Vcan expression in fibroblasts.

Vcan haploinsufficiency exacerbates BLM-induced pulmonary fibrosis

To genetically validate and further dissect the possible role of VCAN in pulmonary fibrosis, BLM was then administered as previously (Fig. 2b)^4,21,24,25, to Vcan heterozygous knockout mice (Vcan^+/−), which are healthy and fertile, and with no apparent lung abnormalities²⁸. As expected, Vcan^+/− mice post BLM expressed 50% of normal Vcan levels (Supplementary Fig. 5a–c), including all Vcan isoforms (Supplementary Fig. 5a, b).

Surprisingly, given the increased Vcan expression in human and mouse pulmonary fibrosis, Vcan^+/− mice, compared with wt littermates, lost more weight post-BLM (Fig. 4a) indicative of exacerbated disease burden, and presented with increased pulmonary oedema, as quantified by total protein measurements in the bronchoalveolar lavage fluid (BALF) (Fig. 4b). Inflammatory cells in the BALF, as measured with the haematocytometer, were only marginally increased (Fig. 4c). Multicolour FACS analysis confirmed the results (Supplementary Fig. 6), indicating no statistically significant changes in inflammation upon the genetic reduction of Vcan levels in Vcan^+/− mice, but in the infiltration of T-cells. However, increased soluble collagen levels were found in the BALF, as quantified with the Direct red assay (Fig. 4d). Accordingly, increased collagen deposition was noted in the lungs of Vcan^+/− mice stained with Fast green-Sirious Red staining (Fig. 4e), a combined stain where Sirius Red stains collagen and Fast Green stains non-collagenous proteins, allowing for a more sensitive and quantitative assessment of total collagen compared to Masson’s Trichrome²⁹. The increase in collagen deposition upon Vcan haploinsufficiency led to further stiffening of the lung tissue (Fig. 4f), as determined with AFM (Supplementary Fig. 7a) in decellularised lung sections, which preserve the overall lung structure (Supplementary Fig. 7b) and most core matrisome components^30,31; four totally separate regions from each lobe were assessed (Supplementary Fig. 7c), yielding 170–180 data points per condition. As a result, pulmonary fibrosis was found to be exacerbated in Vcan^+/− mice as shown with H&E histology (Fig. 4g, h) and μCT 3D analysis of the lungs (Fig. 4i and Supplementary Movies 1–3). Moreover, 21 d post-BLM, where fibrosis has been largely resolved in wt mice, fibrosis was still prominent in Vcan^+/− mice (Fig. 4j, k), suggesting that VCAN promotes the resolution of fibrosis.

Fig. 4. Vcan^+/− haploinsufficiency exacerbates BLM-induced pulmonary fibrosis.

BLM or saline (SAL) was administered to Vcan^+/− and control WT littermates. a Weight change (%) post SAL and BLM administration. b Total protein concentration in BALFs, as determined with the Bradford assay (n = 4,5,5,5); statistical significance was assessed with one-way ANOVA test, coupled with Tukey’s test; ^*p = 0.0136, ^****p < 0.0001. c Inflammatory cell numbers in BALFs, as counted with a hematocytometer (n = 3,5,3,5); statistical significance was assessed with a one-way ANOVA test; ^***p = 0.0006, ^****p < 0.0001. d Soluble collagen levels in the BALFs were detected with the direct red assay (n = 4,5,5,5); statistical significance was assessed with a one-way ANOVA followed by Welch’s correction and coupled with a post hoc Games-Howell test, ^*p = 0.0187/0.0454, ^**p = 0.0013. e Representative images of lung sections post SAL or BLM administration, stained with Fast Green/Sirius Red (F.G/S.R; green/red). f Lung stiffness (kPa) was evaluated using AFM to measure the Young’s modulus (E) of decellularised lung slices. Each data point represents a different probed locus (n = 171, 177, 170, 177); statistical significance was assessed using a Kruskal–Wallis one-way ANOVA coupled with a post hoc Dunn’s test; ^**p = 0.0073, ^****p < 0.0001. g Representative images of H&E-stained lung sections. h Quantification of fibrosis severity with the Ashcroft score (n = 4,5,4,5). Statistical significance was assessed with one-way ANOVA coupled with a post hoc Dunn’s test; ^*p = 0.0377, ^****p < 0.0001. i Representative high-resolution 3D reconstructions of μCT scans; see also Supplementary Movies 1–3. j Representative images from H&E-stained lung sections 21 days post BLM administration. k Quantification of fibrosis severity via Ashcroft scoring (n = 7,4). Statistical significance was assessed with a two-tailed unpaired t-test; ^**p = 0.0023; Scale bars 50 μm (e,g,j), 500 μm (i); a representative experiment out of three successful independent ones is shown. In all panels, all samples are biologically independent; f, h, i, k n refers to biologically independent lung fibroblast isolations from three different mice per condition; boxplots visualise the median of each distribution; upper/lower hinges represent 1st/3rd quartiles; whiskers extend no further than 1.5 × IQR from the respective hinge. Source data for all panels are provided as a Source Data file.

To exclude a developmental effect in the observed phenotype of Vcan^+/− mice, given the embryonic lethality of the complete knock-out Vcan^−/− mice^28,32,33, an identical experiment was performed in mice where Vcan was ubiquitously deleted upon tamoxifen administration in Vcan^fl/flR26-Cre^ERT2 adult mice; all control mice also received tamoxifen. Identical results were obtained (Supplementary Fig. 8a-g) with the heterozygous mice, while the exacerbated pulmonary fibrosis in Vcan^fl/flR26-Cre^ERT2 mice was additionally shown to deteriorate respiratory functions (Supplementary Fig. 8h). Restricted by the anti-inflammatory effects of higher tamoxifen doses in our protocol and settings, Vcan inducible deletion was confined to an efficiency of ~50% (Supplementary Fig. 8i, j). However, the results ruled out a developmental defect and strongly confirmed the studies in Vcan^+/− mice, supporting a suppressive role for mesenchymal VCAN in BLM-induced pulmonary fibrosis.

Vcan haploinsufficiency reorganises the lung ECM and promotes the expression of collagens and Tnc

To obtain structural and mechanistic insights into the exacerbated fibrosis in Vcan^+/− mice, lung tissue from wt or Vcan^+/− mice post BLM or SAL administration was enzymatically digested, separated by nano-liquid chromatography, and analysed with a Q Exactive HF-X Orbitrap mass spectrometer; raw MS files were analysed using MaxQuant with the Andromeda search engine (Fig. 5a, b). BLM administration to wt mice led to the upregulation of several ECM proteins  including many collagens (Supplementary Data 1), VCAN (in agreement with Figs. 2 and 3) and notably its protein interactors Fibulin 1–2 (FBLN1–2) and Fibronectin 1 (FN1), as well as Tenascin-C (TNC) (Fig. 5c and Supplementary Data 1). Gene set enrichment analysis (GSEA) indicated that the identified differentially expressed proteins (DEPs) in wt lungs post BLM are highly enriched with proteins that bind PGs (Fig. 5d and Supplementary Data 2), including Tnc.

Fig. 5. Mass spectrometry analysis of WT and Vcan^+/− mice lungs post BLM administration.

a Schematic representation (biorender.com) of experimental design. b Principal component analysis (PCA) for all four treatment groups indicates proper sample clustering. c Volcano plot graphically depicting DEPs between BLM- and SAL-treated mice. d Gene enrichment analysis (GSEA) of DEPs identified in wt mice post BLM. e Venn diagram comparing DEPs in wt and Vcan^+/− mice post-BLM. f GSEA of DEPs identified in Vcan^+/− mice post-BLM. In both (d, f) gene ontology terms were ranked by normalised enrichment score (NES) and were characterised by an adjusted p < 0.05. Data are derived from a single mass spectrometry experiment, and all samples are biologically independent.

Many more DEPs were identified in Vcan^+/− mice upon BLM (Fig. 5e, Supplementary Data 3) compared to BLM-treated wt mice (Supplementary Data 1), indicating extensive proteomic changes. TNC was again the most deregulated protein (Supplementary Data 3), and PG binding was among the top deregulated molecular functions (Fig. 5f and Supplementary Data 4). Several ECM/collagen-related ontologies were also highly enriched (Supplementary Data 4), indicating significant changes in the ECM upon reduction of VCAN levels. Comparison of the DEPs that were commonly found significantly upregulated in both wt and Vcan^+/− mice post BLM (Fig. 5e and Supplementary Data 5), indicated several genes that were further deregulated post BLM in Vcan^+/− mice compared to wt mice post BLM. Among them were found the Fibrinogens a, b and g (FGA, FGB, FGG), supporting a previously suggested role for VCAN in coagulation³⁴, as well as, most notably, TNC (Supplementary Data 5).

TNC is a large multifunctional ECM glycoprotein that interacts with many ECM components, including the identified DEPs FN1 and FBLN2³⁵, which also interact with VCAN^11,13. In agreement with the proteomic data (Supplementary Data 5), increased Tnc mRNA expression was detected, retrospectively, in the lung tissue upon BLM-induced pulmonary fibrosis (Fig. 2h), in primary lung fibroblasts isolated at the peak of the disease (Fig. 3c), and in TGF-β-induced primary lung fibroblasts (Fig. 3k). The increased Tnc mRNA expression post-BLM in wt mice was found to be exacerbated in Vcan^+/− mice (Supplementary Fig. 5d), a finding confirmed with immunostaining (Supplementary Fig. 5e). In agreement with the biochemical, histological and proteomic studies, increased Col1a1 mRNA expression and COL1A1 immunostaining were detected in the lungs of Vcan^+/− mice (Supplementary Fig. 5f–g). In the same samples, HA synthesis and/or distribution were also found to be upregulated (Supplementary Fig. 5h, i), suggesting that VCAN also regulates HA synthesis.

In translation, increased TNC mRNA expression was detected in the lung tissue of IPF patients in silico, in all eight datasets examined (Supplementary Table 1 and Fig. 6a). In agreement, increased TNC immunostaining was observed in the lungs of IPF patients (Supplementary Table 3), especially at fibrotic areas (Fig. 6b), confirming a previous study³⁶. scRNAseq analysis of all four selected datasets (Supplementary Table 2) indicated that in the fibrotic lung, TNC is predominantly expressed by fibroblasts and basal cells (Fig. 6c). High TNC expression was noted in IPF fibroblasts, almost missing from the corresponding healthy cells (Fig. 6d), as previously shown with Q-RT-PCR³⁶. Therefore, TNC expression is strongly upregulated in pulmonary fibrosis in both humans and mice, suggesting a pathologic role regulated by VCAN.

Fig. 6. Increased TNC expression in IPF.

a Increased TNC lung tissue mRNA average expression in different cohorts of IPF patients compared to controls (FC > 1.2 and FDR < 0.05) in publicly available transcriptomic datasets (n = 8). b Representative images from immunohistochemistry for TNC (brown) in IPF and CTRL lung tissue; scale bars 100 μm. A representative experiment out of two successful independent ones is shown. c Single-cell specificity of TNC mRNA expression in the human lungs (from top to bottom per dataset PMID:32832599 adj.p = < 1E⁻²⁷⁵/9.63E⁻¹⁴⁸/2.27E⁻⁴⁰/2.10E⁻¹⁴⁰/1.06E⁻³⁹/1.99E⁻²²⁹/2.16E⁻²⁷⁵/<1E⁻²⁷⁵/2.57E⁻¹⁴⁷/<1E⁻²⁷⁵/<1E⁻²⁷⁵/1.62E⁻³⁶; PMID: 32832598 adj.p = 1.50E⁻¹⁷⁴/<1E⁻²⁹⁷/<1E⁻²⁹⁷/3.56E⁻³²/6.65E⁻²⁹⁷/<1E⁻²⁹⁷/1.72E⁻²²/1.15E⁻⁶¹/1.08E⁻³²; PMID: 33650774 adj.p = <1E⁻²¹⁵/3.66E⁻¹¹⁷/<1E⁻²¹⁵/1.91E⁻¹⁰¹/<1E⁻²¹⁵/2.33E⁻⁴⁰/7.16E⁻²¹⁵/<1E⁻²¹⁵/7.23E⁻³⁸/6.09E⁻⁴⁵; PMID: 30554520 adj.p = <1.01E⁻¹³⁹/4.98E⁻⁴⁶/<1E⁻¹³⁹/4.52E⁻¹⁵). d Differential expression analysis of TNC in mesenchymal cell types at the single-cell level (PMID: 32832598 myofibroblasts *adj.p = 0.00027). Marker and differentially expressed genes at the single cell level were identified using the two-sided Wilcoxon Rank Sum test. Significance thresholds were set at FC > 1.2 and Bonferroni-adjusted p < 0.05 (red arrow or *). Boxplots visualise the median of each distribution; upper/lower hinges represent 1st/3rd quartiles; whiskers extend no further than 1.5 × IQR from the respective hinge.

Vcan^+/− fibroblasts overexpress HA, Collagen and TNC, promoting autologous podosome formation and ECM invasion

To examine specifically the role of fibroblasts in the observed exacerbation of BLM-induced fibrosis in Vcan^+/− mice in vivo, primary lung fibroblasts were isolated from wt and Vcan^+/− mice post BLM administration (14 d; pBLM) and cultured ex vivo. Vcan^+/− pBLM fibroblasts expressing 50% VCAN as expected (Fig. 7a–c), expressed significantly increased levels of COL1A1 (Fig. 7d–f), HA (Fig. 7g–i), and TNC (Fig. 7j–l) without any pro-fibrotic stimulation. Moreover, pBLM Vcan^+/− fibroblasts presented with a significantly higher number of podosomes (Fig. 7m, n) and exhibited enhanced adhesion (Fig. 7o), migration in Boyden chambers (Fig. 7p) and aECM invasion in Boyden chambers layered with aECM (Fig. 7q) compared to wt fibroblasts.

Fig. 7. Vcan^+/− lung fibroblasts express increased levels of Col1a1, HA, and Tnc post-BLM, and form more podosomes, leading to increased ECM invasion.

Primary lung fibroblasts were isolated from WT and Vcan^+/− mice post-BLM and were cultured without any additional stimulation. a Vcan, d Col1a1, g Has2, and j Tnc mRNA expression was interrogated with Q-RT-PCR; values were normalised to the expression of B2m and presented as fold change over control (n = 5, j n = 4); statistical significance was assessed with a two-tailed unpaired t-test with Welch’s correction (a, d, g) or unpaired t-test (j); ^**p = 0.0018 (a), ^*p = 0.0179 (d), ^**p = 0.0026 (g), ^*p = 0.0134 (j). Representative composite images from immunofluorescence staining for b VCAN (green), e COL1A1 (green), h HABP (red), and k TNC (green); cells were counterstained with DAPI (blue); scale bars 50 μm. c, f, i, l Quantification of staining intensity (px: pixels) was performed in different optical fields using ImageJ (n = 5; f n = 3). Statistical significance was assessed with a two-tailed unpaired Mann–Whitney test (c) or a two-tailed unpaired t-test (f, I, l); ^**p = 0.0079 (c), ^**p = 0.0029 (f), ^*p = 0.0240 (i), ^*p = 0.0190 (l). m Representative images from immunofluorescence staining against F-actin (red) counterstained with DAPI; arrows indicate representative podosomes. n Quantification of the number of podosome-containing cells (%; n = 5) and the number of podosomes per cell (n = 6), respectively; statistical significance was assessed with a two-tailed unpaired t-test; ^*p = 0.0438/0.0208. o Cell adhesion was assessed with the Crystal Violet assay (n = 7); statistical significance was assessed with a two-tailed unpaired t-test; ^*p = 0.0496. p Migration was assessed using the Transwell assay (n = 4); statistical significance was assessed with a two-tailed unpaired Mann–Whitney test; ^*p = 0.0286. q Invasion capacity of NMLFs, as detected with the transwell invasion assay. Statistical significance was assessed with a two-tailed unpaired t-test (n = 8, 7); ^****p < 0.0001. A representative experiment out of two successful independent ones is shown. In all panels, all samples are biologically independent; boxplots visualise the median of each distribution; upper/lower hinges represent 1st/3rd quartiles; whiskers extend no further than 1.51.5 × IQR × IQR from the respective hinge. Source data for all panels are provided as a Source Data file.

In agreement, TGF-β stimulation of primary fibroblasts ex vivo phenocopied the observed BLM effects, and Vcan^+/− fibroblasts (Fig. 8a) were found to express more Col1a1 and Tnc (Fig. 8b) and to adhere more strongly than wt cells (Fig. 8c). Moreover, Vcan^+/− fibroblasts adhered more to different ECM components, most notably collagens 1–2 (Fig. 8d), and contracted a collagen gel more efficiently than wt cells (Fig. 8e), indicating a close interaction of VCAN and collagen. In agreement, VCAN was shown to modulate collagen polymerisation in a cell-free in vitro turbidity assay, where high VCAN concentrations were found to inhibit polymerisation (Fig. 8f). In contrast, lowering VCAN levels promoted collagen polymerisation (Fig. 8f), indicating a direct dose-related effect of VCAN levels on collagen polymerisation, and suggesting that in the lungs of Vcan^+/− mice post BLM, collagens are likely more polymerised than in the lungs of wt mice. Moreover, TGF-β, as in the case of BLM (Fig. 7m–q), stimulated more podosomes on Vcan^+/− fibroblasts (Fig. 8g, h), resulting in increased TGF-β-induced migration (Fig. 8i).

Fig. 8. Diminished VCAN expression in fibroblasts stimulates COL1A1 expression and polymerisation, promotes podosome formation and enhances ECM invasion.

Serum-starved, primary lung fibroblasts isolated from WT and Vcan^+/− mice (PMLFs) were stimulated with recombinant human TGFβ (10 ng/mL) for 24 h. mRNA expression levels of a Vcan (V0, n = 5,6,4,6; V1, n = 4,5,4,6), b Col1a1 (n = 5,6,4,6) and Tnc (n = 3,4,4,6) were interrogated with Q-RT-PCR; normalization to B2m expression and presentation as fold change over control; statistical significance was assessed with one-way ANOVA coupled with Tukey’s test (a.V0, b.Col1a1, Tnc) or Kruskal–Wallis coupled with Dunn’s test; ^**p = 0.0013/0.0095 (V0), ^*p = 0.0397 (V1), ^**p = 0.0046/0.0064 (Col1a1), ^****p < 0.0001 (Col1a1), ^*p = 0.0462 (Tnc), ^***p = 0.0007 (Tnc). c Cell adhesion assessed with the Crystal Violet assay (n = 7); statistical significance was assessed with one-way ANOVA coupled with Tukey’s test; ^**p = 0.0062, ^****p < 0.0001. d Cell adhesion on different substrates as assessed with the ECM Cell Adhesion Array Kit. e Collagen contractile activity of cells was assessed using a collagen gel contraction assay. Contraction area was quantified with ImageJ (n = 4); statistical significance was assessed with one-way ANOVA coupled with Tukey’s test; ^*p = 0.0273/0.0343, ^**p = 0.0014/0.0011, ^***p = 0.0008, ^****p < 0.0001. f The effect of VCAN (various concentrations) on collagen polymerisation was evaluated using a turbidity assay; statistically significant difference of area under the curve (AUC) values was assessed with Kruskal–Wallis coupled with Dunn’s test; Data are presented as mean values ± SEM ^****p < 0.0001, ^*p = 0.0116. g Representative composite images from double immunofluorescence staining for F-actin (red) and CTTN (green) counterstained with DAPI; arrows indicate podosomes; scale bars 50 μm. h Quantification of podosome-containing cells (%; n = 10,8) and the number of podosomes/cell/optical field (n = 10,7); statistical significance was assessed with two-tailed unpaired t-test (cells with podosomes) or two-tailed unpaired Mann–Whitney test (podosomes/cell); ^*p = 0.0242/0.0124. i Cell migration capacity assessed with Transwell assay (n = 3,4,4,4); statistical significance was assessed with one-way ANOVA coupled with Tukey’s test; ^*p = 0.0192, ^***p = 0.0006, ^****p < 0.0001. j Cell adhesion of 3T3 fibroblasts on lung aECM, as assessed with the Crystal Violet assay (n = 7,5,7,6); statistical significance was assessed with one-way ANOVA coupled with Tukey’s test; ^***p = 0.0009, ^****p < 0.0001. k PMLF invasion capacity from the indicated mice into lung aECM, as assessed with transwell invasion assay (n = 4); statistical significance was assessed with one-way ANOVA coupled with Tukey’s test; ^*p = 0.0338/0.0101, ^***p = 0.0002/0.0006. A representative experiment out of two successful independent ones is shown. In all panels, all samples are biologically independent; boxplots visualise the median of each distribution; upper/lower hinges represent 1st/3rd quartiles; whiskers extend no further than 1.5 × IQR from the respective hinge. Source data for all panels are provided as a Source Data file.

To confirm that the observed characteristics of Vcan^+/− fibroblasts were due to their decreased deposition of VCAN in the ECM, and not just decreased expression, acellular ECM (aECM) was isolated from the lungs of wt and Vcan^+/− mice and used as a substrate for the culture of fibroblasts. The decrease of VCAN in the aECM, which is indirectly supported by the decreased VCAN expression in the lungs of Vcan^+/− mice (Supplementary Fig. 5a–c) and from Vcan^+/− fibroblasts (Fig. 7a–c), resulted in increased TGFβ-induced adhesion of 3T3 fibroblasts (Fig. 8j), confirming VCAN as an anti-adhesive ECM molecule. Moreover, TGF-β-induced primary wt lung fibroblasts invaded the aECM with diminished VCAN levels more efficiently (Fig. 8k), an effect further exacerbated in Vcan^+/− fibroblasts (Fig. 8k), indicating that VCAN restricts ECM invasion of fibroblasts, in agreement with the developmental role of VCAN.

TNC was found to be a very potent podosome inducer in primary fibroblasts, even in the absence of TGF-β (Fig. 9a, b), stimulating their ECM invasion and potentiating TGF-β effects (Fig. 9b). The stimulation of aECM invasion by TGF-β and TNC could be decreased by the pharmacologic inhibition of DDR-1 (Fig. 9c), the primary collagen receptor, confirming that reverse collagen signalling is involved in podosome formation and, thus, the stimulation of ECM invasion. The pharmacologic inhibition of TLR4, one of the suggested TNC receptors, also attenuated the TGF/TNC-induced ECM invasion (Fig. 9c), suggesting a Tnc/TLR4 axis in the pathologic activation of fibroblasts and ECM invasion. The central role of podosomes in fibroblast ECM invasion has been recently proven by the pharmacologic inhibition of SRC, the master regulator of podosomes, which attenuated podosome formation and ECM invasion⁴. Therefore, the exacerbated disease in Vcan^+/− mice can, at least in part, be attributed to increased podosome formation and ECM invasion of Vcan^+/− lung fibroblasts, resulting from the increased expression of collagen and Tnc.

Fig. 9. Tnc is a potent, TLR4- and DDR1-dependent, podosome inducer in lung fibroblasts.

Primary NMLFs were isolated from wt mice. Serum-starved cells were then stimulated with recombinant human TGF-β (10 ng/mL) and/or TNC (4 μg/mL) for 24 h. a Representative composite image from the double immunofluorescence staining of NMLFs against F-actin (red) and Cortactin (CTTN; green); arrows indicate podosomes. b Quantification of the number of podosomes per cell (n = 5,11,4,12), statistical significance was assessed with a Kruskal–Wallis one-way ANOVA coupled with a post hoc Dunn’s test; ^*p = 0.0396, ^***p = 0.0002. c Cell invasion capacity of NMLFs into lung aECM, as assessed with the transwell invasion assay (n = 4,4,5,4,4,4); statistical significance was assessed with one-way ANOVA coupled with Tukey’s test; ^**p = 0.0034/0.0026/0.0070, ^****p < 0.0001. d, h Cell proliferation in the presence of the DDR-1 inhibitor (n = 8) (d) or the TLR-4 inhibitor (n = 8,6,7,7,8) (h); statistical significance was assessed with one-way ANOVA coupled with post-hoc Tukey’s test. e, i Col1a1 mRNA expression levels was interrogated with Q-RT-PCR; values were normalized to the expression of B2m and presented as fold change over control (n = 4,5,5); statistical significance was assessed with an one-way ANOVA followed by Welch’s correction and coupled with a Games-Howell test (e) or an one-way ANOVA coupled with Tukey’s test (i); ^*p = 0.0136, ^**p = 0.0056 (e), ^***p = 0.001, ^****p < 0.0001 (i). f, j The collagen contractile activity of cells was assessed using a collagen gel contraction assay, and the contraction area was quantified with ImageJ (n = 5,4,4,5,4,4,5,4,3 for f; n = 5,4,3,5,4,3,5,4,3 for j); statistical significance was assessed with one-way ANOVA coupled with Tukey’s test; ^*p = 0.0269 (f), ^**p = 0.008, 0.0049 (f), ^****p < 0.0001 (f, j), ^*p = 0,0192 (j), ^**p = 0.0013 (j), ^***p = 0.0002 (j). g, k Vcan mRNA expression levels were interrogated with Q-RT-PCR; values were normalised to the expression of B2m and presented as fold change over control (n = 4,5,5); statistical significance was assessed with one-way ANOVA coupled with post-hoc Tukey’s test; ^*p = 0.0199 (g), ^***p = 0.0006 (g), ^*p = 0.0216/0.0205 (k). Scale bars 50 μm; a representative experiment from two independent experiments are shown, except for (a, b, c), which are derived from a single experiment. In all panels, all samples are biologically independent; boxplots visualise the median of each distribution; upper/lower hinges represent 1st/3rd quartiles; whiskers extend no further than 1.5 × IQR from the respective hinge. Source data for all panels are provided as a Source Data file.

Blocking collagen reverse signalling with non-toxic concentrations (Fig. 9d) of a DDR1 inhibitor, which suppresses Col1a1 expression (Fig. 9e) and collagen gel contraction (Fig. 9f) as expected, also suppressed TGF-β-induced Vcan expression (Fig. 9g), further supporting that increased collagen expression and deposition induce VCAN expression from lung fibroblasts, and indicating a bidirectional relationship of collagens and VCAN. Similar results were obtained with a TLR4 inhibitor (Fig. 9h–k), suggesting that TNC also upregulates VCAN expression in fibroblasts, which, in turn, suppresses TNC expression in a negative feedback regulatory loop.

Vcan regulates ECM homeostasis and pathologic fibroblast activation

To focus specifically on the ECM secreted from fibroblasts per se, equal numbers of primary wt and Vcan^+/− lung fibroblasts were cultured ex vivo in the presence of TGF-β and ascorbic acid to stimulate fibroblast activation and ECM deposition. Cultures were grown to confluency, ensuring equal output cell numbers, followed by decellularisation, which preserves the core matrisome proteins³⁷. The remaining fibroblast-produced aECM (aECM^F) was then used as a substrate for the autologous culture of primary lung fibroblasts. The diminished VCAN levels in aECM^F (Fig. 10a) promoted Col1a1 and Tnc expression from lung fibroblasts (Fig. 10b) and stimulated the formation of podosomes (Fig. 10c, d). Moreover, F-actin imaging analysis indicated that although the overall F-actin content is similar, stress fibres in wt fibroblasts grown on Vcan^+/− aECM^F were shorter, less curved and better aligned (Fig. 10e), reflecting a migration-supportive cytoskeletal organisation³⁸.

Fig. 10. Vcan in the ECM regulates collagen fibrillation and the formation of podosomes on fibroblasts.

Acellular ECM (aECM) was prepared from WT and Vcan^+/− PMLFs and used as a substrate for WT fibroblasts. a Representative images from double immunofluorescence staining for VCAN (green) in the aECM from the indicated genotypes; scale bars 50 μm. b mRNA expression levels of Col1a1 (n = 6,9) and Tnc (n = 7,8) interrogated with Q-RT-PCR; values normalized to B2m expression and presented as fold change over control; statistical significance was assessed with two-tailed unpaired t-test (Col1a1) and two-tailed unpaired t-test with Welch’s correction (Tnc); ^***p = 0.0004 (Col1a1), ^**p = 0.0026 (Tnc); n refers to independent lung fibroblast samples seeded on acellular ECM, derived from wild-type or Vcan + /– fibroblasts. c Representative composite images from double immunofluorescence staining for F-actin (red) and Cortactin (CTTN; green) counterstained with DAPI (blue); arrows indicate podosomes; scale bar 50 μm. d Quantification of podosome-containing cells (%; n = 10) and the number of podosomes/cell (n = 4); statistical significance was assessed with a two-tailed unpaired t-test; ^***p = 0.0005 (podosome-containing cells), ^*p = 0.0263 (number of podosomes/cell). e Quantification of F-actin quantity, fibre length, chirality and alignment (n = 59,62); statistical significance was assessed with a two-tailed Mann–Whitney test; ^****p < 0.0001 (fibre length), ^*p = 0.0289 (chirality), ^*p = 0.0154 (fibre alignment) n refers to independent measurements performed over 20 optical fields per condition. f Representative TEM images of longitudinal sections from the ECM generated by WT and Vcan⁺/⁻ PMLFs. Arrows indicate collagen fibres. g Quantification of the number of collagen fibres/μm², performed on TEM images (n = 15) of different fields (25,000×; scale bar: 500 nm). Statistical significance was assessed with a two-tailed unpaired t-test; ^***p = 0.0001. h Quantification of collagen fibre width (n = 40,39) assessed in 20 different images at original magnification 25,000× (scale bar: 500 nm), with evaluations conducted at two different optical fields within each image. Statistical significance was assessed with a two-tailed unpaired t-test with Welch’s correction ^*p = 0.0198; n refers to collagen width measurements quantified over 20 independent optical fields per condition (image j). i, j Representative TEM images of longitudinal sections from the ECM generated by WT and Vcan⁺/⁻ PMLFs. Arrows indicate elastin fibres (i) and PGs (j). k Representative TEM images of double immunolocalisation for VCAN (25 nm gold particle; red arrows) and TNC (15 nm gold particle; green arrows) at original magnification 25,000× (scale bar: 500 nm). l Quantification of VCAN and TNC particles (n = 60); statistical significance was assessed with two-tailed Mann–Whitney test; ^****p < 0.0001. Representative data from two independent experiments are shown, except for (e, k, l), which are derived from a single experiment. In all panels, all samples are biologically independent; boxplots visualise the median of each distribution; upper/lower hinges represent 1st/3rd quartiles; whiskers extend no further than 1.5 × IQR from the respective hinge. Source data for all panels are provided as a Source Data file.

MS proteomic analysis of Vcan^+/− aECM^F (Supplementary Fig. 9a) confirmed the reduction of VCAN and identified increased expression of 17 different chains of different COL species (including 1, 3, 4, 6, 12, 16, 18; only COL5A3 was found downregulated (Supplementary Data 6 and Supplementary Fig. 9b). Moreover, increased expression of procollagen C-endopeptidase enhancer (PCOLCE; PSPE) was identified (Supplementary Data 6 and Supplementary Fig. 9b), whose genetic deficiency has been reported to attenuate liver fibrosis³⁹. The expression levels of lysyl oxidase homologue 1 (LOXL1), but not LOXL2, were also found to be upregulated, suggesting increased collagen and elastin crosslinking (Supplementary Data 6 and Supplementary Fig. 9b) upon VCAN reduction; the expression of Elastin (ELN) itself was also found to be upregulated. The expression of several A disintegrin and metalloproteinase with thrombospondin motif (ADAMTS) proteases was found upregulated (1, 2, 4, 5, 12, l3, l4; Supplementary Data 6 and Supplementary Fig. 9b), suggesting deregulated ECM proteolysis in the absence of VCAN. Among them, Adamts1 and 5 have been proposed to cleave VCAN, generating the bioactive Versikine peptide. TNC (and TNXb) was also upregulated in the Vcan^+/− aECM^F (Supplementary Data 6 and Supplementary Fig. 9b), further supporting that VCAN regulates TNC expression and its profibrotic effects. Gene enrichment analysis (GSEA) indicated that the identified DEPs are highly enriched for proteins involved in collagen homoeostasis, ECM organization, as well as in elastic fibre assembly (Supplementary Data 7 and Supplementary Fig. 9c). Surprisingly, DEPs were also enriched in proteins involved in the SNARE complex and SNAP receptor activity, suggesting a possible effect of Vcan in membrane trafficking, which can include exosome secretion or endocytosis. Moreover, DEPs were also enriched in proteins involved in mitochondrial and Golgi homeostasis, suggesting increased biosynthetic activity of Vcan^+/− fibroblasts.

Structural characterisation of the ECM produced by Vcan^+/− fibroblasts with TEM indicated an increased density of intertangled collagen fibres (Fig. 10f), composed of a higher number of collagen fibres/μm² (Fig. 10g) of increased thickness (Fig. 10h), in agreement with the proteomic and AFM studies. Elastic fibres in the ECM from Vcan⁺/⁻ fibroblasts formed aggregates, a phenomenon rarely observed in ECM from wt fibroblasts (Fig. 10i), suggesting that VCAN also regulates elastin aggregation. Finally, lower PG levels, and differentially distributed, were observed in Vcan⁺/⁻ ECM (Fig. 10j), indicating disturbed PG metabolism, in agreement with the lung proteomic analysis.

Immunogold labelling of VCAN confirmed Vcan reduction upon haploinsufficiency and indicated that VCAN was primarily localised in collagen fibres (Supplementary Fig. 10a), confirming the physical interaction of VCAN with collagens, as suggested with the turbidity assay, and as previously suggested in extrahepatic bile ducts⁴⁰. Immunogold labelling of TNC confirmed the increased TNC expression upon the reduction of VCAN levels (Supplementary Fig. 10b), in agreement with the lung proteomic analysis. Moreover, double immunogold VCAN and TNC labelling, utilising immunogold particles with different diameters (15 and 25 nm, respectively), provided the ultimate confirmation of the single immunogold labelling and TEM results (Fig. 10k), and allowed quantification (Fig. 10l). Therefore, VCAN profoundly affects ECM organization and structure, including collagen fibrillation, elastin aggregation, and PG homoeostasis, thus controlling the activation and pathological functional properties of the overlying fibroblasts.

Discussion

VCAN is highly expressed during embryonic development, particularly in tissues undergoing rapid growth and remodelling. VCAN expression is necessary for embryonic development, as homozygous knockout mice (Vcan^−/−) die in utero due to severe cardiovascular defects^28,32,34, attributed to aberrant neural crest cell migration⁴¹ and ventricular septal formation³⁴. Several ECM components, including HSPGs, are expressed along the neural crest cell migration routes and have been suggested to act as guiding cues, while tissues with high Vcan expression have been proposed to act as a migration barrier³³, suggesting a suppressive role for VCAN in developmental cell migration.

In adult tissues, VCAN expression is more restricted, but it has been reported to be higher in different tissues upon inflammation or cancer¹¹. In this report, increased VCAN mRNA expression was detected in silico in the lung tissue of IPF patients compared to controls, and VCAN immunostaining was noted in the lungs of a small cohort of IPF patients, confirming and complementing previous studies^17,18,23. Early studies have suggested that the VCAN-rich areas contained little mature collagen¹⁷, while in a different study, VCAN expression was associated with proliferating (PCNA⁺) myofibroblasts in early lesions, but was absent from areas of dense collagen¹⁸. In our study, the most characteristic staining was noted at the outer rim of well-defined fibroblast foci. In some cases, a Vcan-rich region completely encapsulates the foci, possibly restricting its further invasion and growth, in line with Vcan’s developmental role.

VCAN staining was also prominent in different lung regions and in (or around) other cell types. However, the relative contributions of each cellular compartment to the overall lung VCAN levels, as well as their impact on pathology, remain unknown. The increased levels of VCAN in IPF were found to be associated with the impairment of respiratory functions, suggesting that the overall VCAN levels in human patients are pathological. However, a degradation product of VCAN in the serum (VCANM) was reported to be inversely correlated with mortality following acute exacerbations of IPF⁴²; a different proteolytic N-terminal fragment of VCAN (versikine) has also been detected⁴³. Therefore, the possible diagnostic or prognostic value of VCAN expression and/or its proteolysis should be examined in larger cohorts. Proteolysis is a critical component in IPF pathology through its regulatory influence on matrix remodelling, cell behaviour, and fibrosis progression. Proteolytic fragments from other ECM components (matrikines), including collagens and HA, have also been reported⁴⁴. Their combined profiling in pulmonary fibrosis, as in the INMARK trial⁴⁵, could provide further insights into the interconnection between Vcan, ECM, proteolysis, and IPF pathogenesis.

In further agreement with the developmental role for VCAN in migration and invasion, the genetic reduction of VCAN expression and deposition in Vcan^+/− mice promoted the formation of podosomes in lung fibroblasts and stimulated the invasion of lung fibroblasts into the interstitial ECM. BLM-induced pulmonary fibrosis was found to be exacerbated in Vcan^+/− mice, reaffirming the importance of podosomes and ECM invasion to fibroblast accumulation and pulmonary fibrosis^4,5,46. In support, pharmacologic inhibition of SRC, a central pro-fibrotic signalling hub in IPF and the master regulator of podosomes, was recently shown to attenuate BLM-induced pulmonary fibrosis^4,47,48, and Saracatinib, a selective SRCinhibitor, has entered IPF clinical trials (NCT04598919).

It should be noted that VCAN is constitutively expressed in healthy tissue and normal fibroblasts, as shown in the scRNAseq analysis, in agreement with the homoeostatic roles of VCAN to maintain ECM elasticity and to support normal tissue repair. The absence of prominent collagen increases in the normal lung upon VCAN haploinsufficiency suggests that VCAN effects are context-specific, requiring the upregulation of profibrotic factors (collagens, HA), which we have shown to stimulate VCAN expression. Moreover, as evident in the turbidity assay and confirmed by TEM, the relative levels of VCAN are equally crucial for the observed effects, suggesting that VCAN effects are also dose-specific, a hypothesis that could reconcile previously published contradictory findings.

Different cell types have been reported to express VCAN, including epithelial cells, macrophages, smooth muscle cells, and fibroblasts¹¹. In the fibrotic lung, VCAN expression was detected predominantly in fibroblasts and myofibroblasts, as well as in monocytes, macrophages, and a population of aberrant basaloid cells. The latter cells are located at the edge of myofibroblast foci in the IPF lung⁴⁹, and a keratin17^high subpopulation was shown to induce fibroblast proliferation and ECM deposition in vitro, and src-dependent pulmonary fibrosis in vivo upon their intratracheal injection into immunocompromised mice⁴⁸. Moreover, genetic deletion of Vcan from epithelial (SPC⁺) cells was reported to promote airway inflammation during viral (RSV) infection⁵⁰, suggesting a role for VCAN in epithelial homeostasis and pathogenicity that warrants further exploration in future studies.

Significant VCAN expression was detected in silico in monocytes and macrophages, validated by VCAN immunostaining of macrophages in the fibrotic lungs of both humans and mice. Macrophages are well recognised as essential players in IPF pathogenesis, although their exact mode of action remains controversial, since they are highly heterogeneous and exhibit remarkable temporal plasticity⁵¹. Beyond their well-established roles in apoptotic cell clearance and the production of pro-fibrotic mediators such as TGF-β and IL-13, macrophages secrete numerous cytokines and chemokines, defining the inflammatory response⁵¹. Moreover, macrophages secrete ECM components and ECM-modifying enzymes, actively participating in tissue remodelling⁵¹. PMA has been reported to induce VCAN expression⁵², suggesting a role for VCAN in monocyte-to-macrophage differentiation. Moreover, LPS has been reported to induce VCAN expression in bone marrow-derived macrophages⁵³, a response completely abrogated in Tlr4^−/− mice⁵⁴, and human peritoneal macrophages⁵⁵. Accordingly, VCAN has been reported to induce pro-inflammatory macrophage polarisation (“M1”)⁵⁶. Therefore, Vcan^+/− macrophages are likely shifted towards a profibrotic (“M2”) phenotype, possibly contributing, to an unknown extent, to the overall exacerbated pathology of Vcan^+/− mice driven by fibroblasts. However, no significant differences were detected in macrophage subpopulations with multiparametric FACS analysis of BALF cells. Several immune mechanisms, as encoded in GO terms, were found to be deregulated in the lungs of Vcan^+/− mice, including antigen binding and B-cell immune response, suggesting possible specialised effects of VCAN on adaptive immunity. In this context, an intriguing role for VCAN in antigen presentation has recently been proposed in cancer⁵⁷, which may also apply to pulmonary fibrosis. Perturbed antigen presentation could account for the observed increase in T-cell infiltration upon VCAN haploinsufficiency. In support, VCAN has recently been suggested to have a major effect on T-cell trafficking in solid tumours^58,59. However, macrophage-specific VCAN deletion will be necessary to fully understand the effects of VCAN on macrophage homeostasis and pulmonary fibrosis.

Lung fibroblasts, the bona fide ECM-expressing cells and the primary effector cells in pulmonary fibrosis⁶⁰, were shown here to express VCAN in response to pro-fibrotic stimuli. Fibroblast-secreted VCAN was shown to have significant effects on collagen homeostasis, suppressing its expression, and regulating its polymerisation and fibrillation. The relative abundance of collagen, degree of fibrillation and cross-linking, in concert with other fibrous proteins and PGs, are well known to induce tissue stiffness, and to perpetuate fibroblast activation and promote pulmonary fibrosis^61,62. The degree of collagen fibrillation has been reported to have, among others, a decisive effect on podosome formation⁶³, previously shown to be essential for ECM invasion and pulmonary fibrosis⁴. Reducing VCAN levels in the ECM, and the associated overexpression and increased fibrillation of collagens, promoted podosome formation on fibroblasts, stimulating their ECM invasion. Collagen effects on podosomes and ECM invasion are likely mediated by the DDRs that bind collagen and regulate its reverse signalling⁶⁴. Vcan^+/− lung fibroblasts’ ECM invasion was shown to be partly suppressed by the pharmacologic inhibition of DDR1, suggesting that the VCAN effects on podosome formation and ECM invasion are partly mediated by altered collagen homoeostasis resulting from the reduction of Vcan.

HA is one of the GAGs found attached to the VCAN core protein, and VCAN has been proposed to act as an HA “linker”, modulating its structural properties and signalling effects¹¹. The reduction of VCAN in Vcan^+/− mice, fibroblasts and ECM, was shown to promote HA synthesis from fibroblasts, while HA signalling has been previously suggested to affect the invasive properties of fibroblasts in IPF⁴⁶. CD44, one of the primary HA receptors and a possible VCAN interactor, has been localised in invadopodia in breast cancer cells⁶⁵. Genetic deletion or pharmacologic blockage of CD44 attenuated BLM-induced pulmonary fibrosis, while fibroblasts isolated from IPF patients exhibited an invasive phenotype that was also dependent on HAS2 and CD44 expression⁴⁶. Therefore, the observed effects on podosome formation and ECM invasion in Vcan^+/− mice can be partly due to exacerbated HA signalling.

Beyond collagen and HA homoeostasis, VCAN deletion was shown here to simulate TNC expression. TNC is highly expressed in embryonic tissues, around motile cells and sites of branching morphogenesis, and is suggested to promote neural crest cell migration³⁵, a decisive developmental pathway that was suggested to be inhibited by VCAN³³. TNC expression is much more restricted in adult tissues, mainly found in regions of active neurogenesis⁶⁶. TNC is also expressed during wound healing, chronic inflammation, and cancer. TNC expression has been localised at the intratumor stroma but also the invasive front and has been associated with poor prognosis in several malignancies^67,68. TNC expression was shown here to be aberrantly reactivated in mouse and human pulmonary fibrosis, suggesting yet another aberrant recapitulation of a developmental program in pulmonary fibrosis⁶⁹. Consistent with a role for TNC in IPF pathogenesis, Tnc knock-out mice were reportedly resistant to BLM-induced pulmonary fibrosis^70,71. As shown here, TNC is a very potent inducer of podosomes, as previously only shown for invadosomes in Ewing sarcoma cells⁷². Therefore, the promotion of podosome formation and ECM invasion, as well as the exacerbated disease in Vcan^+/− mice, can also be partly attributed to the stimulation of TNC expression. TNC is an endogenous activator of TLR4⁷³, and TLR4 inhibition was shown here to inhibit TGF-β/Tnc-induced podosome formation in fibroblasts and ECM invasion, suggesting that TLR4 is one of the cell receptors mediating TNC effects on fibroblasts. The conditional deletion of TLR4 in fibroblasts attenuated BLM-induced pulmonary fibrosis⁷⁴, suggesting that the Tnc/TLR4 axis is a key driver of podosome formation and fibroblast ECM invasion. In this context, the presented results suggest TNC as a promising therapeutic target in IPF, as previously suggested for cancer⁶⁸, given its relatively low expression in healthy tissues and high specific expression in the fibrotic stroma.

The presented genetic, proteomic and AFM/TEM studies suggest that VCAN expression specifically from fibroblasts suppresses pulmonary fibrosis, suggesting VCAN as an autologous fibrotic brake. Fibrotic brakes are cellular or molecular mechanisms that limit excessive scarring and promote the resolution of fibrosis, such as PGE₂, PPARγ, PTEN and KLF4, and are often found suppressed in IPF^5,75. However, the expression of other proposed fibrotic brakes has been paradoxically reported to be upregulated in pulmonary fibrosis. MFGE8, a soluble glycoprotein that binds collagen, acts as a fibrotic brake by facilitating the removal of accumulated collagen, although its expression is upregulated in IPF⁷⁶. FGF21 levels have also been reported to be elevated in IPF, while its genetic deletion exacerbated BLM-induced pulmonary fibrosis⁷⁷, as seen here with VCAN. FGF23 has also been proposed as a fibrotic brake, although its expression is upregulated in fibrosis⁷⁸. However, both FGF23 and FGF21 signalling and effects have been proposed to be regulated by Klotho proteins (KLA-C), transmembrane coreceptors of FGFs, whose levels are downregulated in pulmonary fibrosis⁷⁸. Klotho haploinsufficiency exacerbated BLM-induced pulmonary fibrosis⁷⁸, possibly by alleviating endocrine FGF anti-fibrotic effects. Therefore, some fibrotic brakes act in concert with other complementary mechanisms, likely in a cell-specific manner. In the case of VCAN, TGF-β-induced collagen expression and polymerisation seem to be among the prerequisites for the manifestation of VCAN’s suppressive functions, which are, nevertheless, dose-, cell-, and context-specific. The VCAN protein interactors FBLN1-2 and FN1, whose expression was also found to be upregulated post BLM following VCAN expression, are also candidates for the modulation of VCAN functions. Understanding the regulation of fibrotic brakes will be essential to unravel the mechanisms governing the resolution of fibrosis, the holy grail of fibrosis research, which will be instrumental in designing effective therapeutics.

In conclusion, increased VCAN expression was observed in IPF and BLM-induced pulmonary fibrosis, mainly in macrophages and fibroblasts. VCAN expression in fibroblasts was shown to be induced by TGF-β and a collagen-rich fibrotic ECM. The reduction in VCAN expression and deposition in the ECM from haploinsufficient Vcan^+/− lung fibroblasts led to the upregulation of collagen production and polymerisation, HA synthesis, and TNC secretion. The altered ECM, both in composition and structure, containing thicker, intertwined collagen fibres and aggregated elastin fibres, along with increased HA and TNC synthesis, promoted the autologous formation of podosomes in fibroblasts. Podosomes promoted ECM invasion, thereby exacerbating BLM-induced pulmonary fibrosis in Vcan^+/− mice and delaying its resolution (Supplementary Fig. 11). However, the increased synthesis and polymerisation of collagen, the increased HA and TNC synthesis upon VCAN reduction, can have additional effects in fibroblast activation that remain to be explored. Moreover, and beyond the shown suppressive role of fibroblast-derived VCAN in autologous profibrotic functions, the overall effects of VCAN in the lung, which are dose-, cell-, and context-specific, remain to be further investigated. Notably, the pro-inflammatory, pathogenic role of macrophage-derived VCAN in pulmonary fibrosis warrants further study, fully supporting VCAN’s name and its versatile functions and suggesting that, in fibrosis, VCAN can act as a “double-edged sword”.

Methods

All experimentation was conducted in accordance with the respective ethical regulations outlined below.

Datasets

All analysed, re-normalised transcriptomics datasets were sourced from Fibromine²² and are listed in Supplementary Table 1. Utilised, publicly available, scRNAseq datasets are listed in Supplementary Table 2. Both Tables include accession numbers and hyperlinks to the original publications.

Patients

Studies using human patient samples (Supplementary Table 3) were conducted in accordance with the Helsinki Declaration principles and approved by the Ethics Committee of the Sotiria Chest Diseases Hospital (1662/16-01-2024). All patients provided written consent for the use of their samples for research purposes; no compensation was offered for their participation.

Mice

Mice were bred at the animal facilities of Biomedical Sciences Research Center ‘Alexander Fleming’, under SPF conditions, at 20–22 °C,55 ± 5% humidity, and a 12 h light-dark cycle; food (Mucedola diet #4RF21: humidity 12%, protein 18,5%, fat 3%, carbohydrate 53,5%, crude fibres 6%) and water were provided ad libitum. The genotypic strategy for Vcan^+/− and Vcan^fl/fl mice have been reported previously in ref. ²⁸. Both strains of mice are healthy and fertile, with no apparent lung abnormalities²⁸. Mice were bred and maintained in their genetic backgrounds for over 10 generations. All randomly assigned experimental groups consisted of littermate age-matched mice. Both male and female mice were included in all experiments. The mice’s health was monitored daily; no unexpected deaths were observed. Euthanasia was humanely performed in a CO₂ chamber with gradual filling at predetermined time points. All experimentation was approved by the Protocol Evaluation Committee (PEC) of the Biomedical Sciences Research Centre “Alexander Fleming”, and the Veterinary Service of the governmental prefecture of Attica, Greece (# 541668/2024).

BLM-induced pulmonary fibrosis

Pulmonary fibrosis was induced by a single oropharyngeal administration (OA) of 0.8 U/kg BLM hydrogen chloride (Nippon Kayaku Co., Ltd., Tokyo, Japan) and analysed following published protocols^4,21,24,25; the detailed protocol is deposited at protocols.io. Briefly, following body weight measurement, respiratory functions were evaluated using the FlexiVent system. Lungs are then flushed, and the retrieved BALF, following centrifugation, is stored at −80 °C; cell pellets are counted using a Neubauer hemacytometer, and/or analysed with multiparametric FACS Analysis (PI, CD45, CD11b, CD11c, MHCII, Ly-6G, Ly-6C, CD64, Siglec-F, CD24 and CD206; Supplementary Table 5). Total protein concentration in BALF, indicative of pulmonary oedema and vascular permeability, is determined using the Bradford protein assay, and total soluble collagen is quantified using the Direct Red assay.

The right lung lobe, following fixation and embedding in paraffin, is sectioned (5 μm) and stained with Hematoxylin and Eosin (H&E; Papanicolaou’s solution HX16967353 Sigma Aldrich/Eosin G CI45380 ROTH) or Fast Green–Sirius Red (F.G/S.R). For Immunofluorescence staining, paraffin-embedded lung sections are first deparaffinized in xylene, rehydrated through a graded ethanol series, and rinsed with distilled water. For VCAN staining, sections are incubated with chondroitinase ABC (Sigma-Aldrich, C3667) at 37 °C for 20 min to degrade chondroitin sulphate chains and enhance epitope accessibility. Antigen retrieval is performed by autoclaving the slides in sodium citrate buffer (pH 6.0) for 20 min. Νon-specific binding is blocked by incubating the sections at room temperature for 1 h in a blocking buffer containing 10% normal goat serum and 2% BSA. Slides are then incubated overnight at 4 °C with primary antibodies diluted in blocking buffer. The following day, sections are washed and incubated with fluorophore-conjugated secondary antibodies, also diluted in blocking buffer. After three washes with PBS-T, the slides are mounted using a DAPI-containing mounting medium (Sigma-Aldrich, F6057) to visualise nuclei. Utilised antibodies are indicated in Supplementary Table 5. Brightfield imaging was performed using an Olympus VS200 slide scanner, and fluorescent imaging was performed on a Zeiss LSM900 confocal microscope.

Primary NMLFs

NMLFs were isolated from 8- to 10-week-old mice, including both untreated controls and BLM-treated animals. After sacrifice, lungs were perfused, excised into DMEM, minced, and digested with 0.7 mg/mL collagenase type IV (Sigma-Aldrich C5138) for 1 h at 37 °C. The digested tissue was filtered, centrifuged at 1200 g for 5 min, and the resulting cell pellet was resuspended in DMEM (GIBCO 41966-029) supplemented with 10% foetal bovine serum (FBS, GIBCO 10437-028). All experiments were performed using cells at passage 3, to avoid cultures from macrophages, and to avoid senescence at higher passages.

Cells were cultured in DMEM containing 10% foetal bovine serum (FBS), penicillin/streptomycin (GIBCO 15140-122), and amphotericin B (GIBCO 15290-018), and maintained at 37 °C in a humidified 5% CO₂ atmosphere. Once cells reached 60–80% confluence, they were serum-starved overnight in DMEM with 0.1% BSA. For stimulation experiments, cells were treated with 10 ng/mL recombinant human TGF-β1 (R&D Systems, 240-B-002) for 24 h, while control samples received vehicle (7.5% BSA in H₂O). For pharmacological assays, cells were pre-treated with the indicated compounds (or vehicle) for 1 h before TGF-β1 exposure. Fibroblasts isolated from BLM-treated lungs were cultured under identical conditions but were not further stimulated in vitro.

The murine embryonic fibroblast cell line NIH/3T3 was obtained from ATCC (CRL-1658), and human control fibroblasts were obtained with informed consent and in accordance with institutional ethical guidelines, as previously described⁴. All cell lines were maintained under identical conditions (DMEM, 10% FBS). Further details can be found at protocols.io.

Fibroblast proliferation was assessed using the MTT assay in 96-well plates. A working solution of OPTI-MEM (GIBCO 11058-021) medium containing 0.7 mg/mL MTT (ACROS ORGANICS158990010) was added to each well. After a 4-hour incubation, the formation of purple formazan crystals was visually confirmed. The medium was then aspirated, and acidified isopropanol was added to solubilise the crystals. Absorbance was measured at 570 nm with background correction at 660 nm using an OPTImax Microplate Photometer (Molecular Devices).

Assessment of fibroblast migration and invasion was performed using Boyden chamber assays (Costar, REF3422) according to the manufacturer’s instructions. For migration assays, cells were seeded in the upper chamber without ECM coating, while invasion assays involved chambers pre-coated with ECM substrate derived from mouse lungs. Cells were allowed to migrate or invade for 6 h toward the lower chamber containing starvation medium. After incubation, non-migratory cells remaining on the upper side of the membrane were removed. Cells that had migrated or invaded to the lower surface were fixed, stained with crystal violet, and then lysed using lysis buffer for 20 min. Absorbance was recorded at 550 nm using a TECAN Sunrise Microplate Photometer to quantify cell migration/invasion.

For immunofluorescence staining of fibroblasts, cells were plated onto coverslips at a density of 20,000 cells per well. After overnight serum starvation, subconfluent (<60–70%) cells were treated with TGF-β1. Fixation was carried out using 4% paraformaldehyde (PFA) for 15 min, followed by permeabilisation with 0.1% Triton X-100 (Sigma Aldrich, T8532) for 10 min. Cells were then blocked with 2% bovine serum albumin (BSA) in PBS for 1 h at room temperature. Cells were incubated with primary antibodies overnight at 4 °C. The following day, after washing, cells were incubated with appropriate secondary antibodies along with conjugated phalloidin diluted in 1% BSA/PBS for 1 h at room temperature. After final washes, coverslips were mounted using Fluoroshield mounting medium containing DAPI to label nuclei. Utilised antibodies are indicated in Supplementary Table 5. Fluorescence images were obtained with a Zeiss LSM900 confocal microscope.

Acellular extracellular matrix (aECM)

Two types of aECM were used in this study: fibroblast-derived matrices and lung tissue-derived matrices; further details are available at protocols.io. Fibroblast-derived aECM was prepared by culturing primary lung fibroblasts from wild-type (WT) and Vcan^+/− mice in DMEM supplemented with 10% FBS, 1% penicillin–streptomycin, and 0.1% amphotericin B. Cells were seeded at a density of 3 × 10⁵cells/mL and maintained for 7–10 days until confluence, with medium supplemented with 50 μg/mL ascorbic acid (A4403-sigma aldrich) to promote ECM deposition and changed every 2 days. Once confluent, cells were decellularised using 0.2 M ammonium hydroxide (NH₄OH; Carlo Erba-419993) for 10–15 min, with real-time monitoring by light microscopy to confirm complete removal of cellular material. The resulting ECM was washed three times with PBS and either used immediately or stored at 4 °C.

Lung-derived aECM was generated with published protocols⁴. Briefly, intact lungs were excised and sequentially treated with increasing concentrations of SDS (Fisher Bioreagents BP166)—0.01%, 0.1%, and 1%—each diluted in PBS and applied for 24 h per step. Following SDS treatment, decellularised lung tissue was thoroughly rinsed in PBS for a minimum of three days to remove residual SDS. The tissue was then sectioned into small fragments, snap-frozen, and stored at −80 °C. Frozen samples were subsequently lyophilised and ground under liquid nitrogen into a fine powder. To create an ECM substrate suitable for cell culture, the powdered matrix was enzymatically digested. Pepsin (Sigma-Aldrich, P6887) was dissolved in 0.1 M HCl to prepare a 1 mg/mL digestion solution. Approximately 10 mg of ECM powder was added to 1 mL of pepsin solution and incubated for ~48 h to solubilise the ECM proteins. The resulting digest was then diluted in 0.1 M acetic acid to achieve a final concentration of 5 mg/mL.

Real-time PCR analysis

Quantitative real-time polymerase chain reaction (Q-RT-PCR) was performed according to a published protocol⁴, using sSoFast™ EvaGreen Supermix (Bio-Rad Laboratories, CA, USA) according to the manufacturer’s instructions on a Bio-Rad CFX96 Touch™ Real-Time PCR Detection System. Gene expression levels were normalised to β2-microglobulin (B2M). Primer sequences are listed in Supplementary Table 4.

Turbidity assay

Collagen turbidity assays were performed according to a published protocol⁴⁰. Briefly, type I bovine collagen (1.5 mg/mL; Corning 354249) was prepared on ice and mixed with PBS, NaOH, and deionised water to adjust the pH to 7.4. Versican (origene #TP762337) was added at the indicated concentrations, and BSA was included as a protein control. Solutions were incubated on ice for 1 h, then transferred to 96-well plates. Gelation was monitored at 400 nm at 37 °C until a plateau was reached, with all conditions in each dataset tested simultaneously.

Collagen gel contraction

Following a published protocol⁷⁹, 12-well plates were coated with 1% BSA for 1 h at 37 °C to prevent gel adhesion. Cells were trypsinised, counted, and embedded at 1 × 10⁵ cells/mL in 0.5 mg/mL type I collagen. The cell-collagen mixture (1 mL per well) was polymerised for 45 min at 37 °C, after which fresh growth media was added. Gel contraction was monitored every day, and gel surface area was quantified using ImageJ software.

Micro-computed tomography (μCT)

Formalin-fixed lungs were dehydrated via a gradient ethanol dehydration protocol followed by 2 h incubation in Hexamethyldisilazane (HMDS; Sigma Aldrich 440191-100ML) and 2 h air-drying⁸⁰. The dehydrated lungs were then scanned in a Bruker SkyScan 1172 scanner.

Fluorescent microscopy

Fluorescence images were acquired using a Zeiss LSM 900 confocal laser scanning microscope with identical exposure and acquisition settings across all conditions. For quantification, 6–10 randomly selected optical fields per sample were analysed, each containing multiple cells. Experiments were performed in technical replicates using fibroblasts pooled from at least three mice per group. The ImageJ software was used for quantification, and regions of interest were selected in a blinded manner to minimise bias. Podosome quantification was independently performed by two observers, and the average values were used for analysis. F-actin image analysis was performed on fluorescence images of phalloidin-stained fibroblasts at 20x magnification, quantified using a custom-made pipeline (CSKMorphometrics) coded in Matlab⁸¹. Further details can be found in the Supplementary Information.

AFM

AFM was performed on 30-micron-thick decellularised lung ECM slices⁸², and as described in detail in Supplementary Information. Briefly, the mechanical properties of the decellularised lung sections were assessed using a custom-built atomic force microscope (AFM) integrated with an inverted optical microscope, using cantilevers with a nominal spring constant of 0.01 N m⁻¹ equipped with a 4.5 µm diameter silicon nitride bead for indentation. Force–indentation curves (d-z) were analysed to calculate the apparent Young’s modulus (E). Further details can be found in the Supplementary Information.

MS

Homogenised, heated, and sonicated lung tissue samples were digested with trypsin according to the Sp3 protocol⁸³. Nano-liquid chromatography mediated separation of the resulting tryptic peptide mixture using an Ultimate3000 RSLC system⁸⁴. Data analysis was performed in MaxQuant⁸⁴. The LFQ MaxQuant search engine results were analysed using the Perseus computational framework (version 1.6.10.43)⁸⁵. Further details can be found in the Supplementary Information.

TEM

TEM was performed on the ECM secreted from Wt or Vcan⁺/⁻ lung fibroblasts, with established protocols^86,87. The generated ECM (wt or Vcan⁺/⁻) with the corresponding lung fibroblasts were fixed in 2.5% glutaraldehyde with 0.05% tannic acid in 0.1 M cacodylate buffer. After fixation, the specimens were washed with 0.1 M cacodylate buffer and embedded in 4% gelatin aqueous solution, which was then solidified by cooling on ice⁸⁸. Ultrathin epoxy sections (70–80 nm thickness) were cut using a Diatome diamond knife, mounted onto 200-mesh copper grids and stained with alcoholic uranyl acetate and lead citrate. Finally, they were observed in a JEOL JEM-2100 Plus TEM, operated at 80 kV and equipped with a GATAN OneView CMOS camera.

Immunoelectron microscopy (EM)

Samples were fixed in 3% PFA and 0.5% glutaraldehyde in 0.1 M PB for 30 min, and they were embedded in acrylic resin according to the Progressive Lowering of Temperature method and were further processed for immunocytochemistry^88,89. After blocking solution, sections were incubated with different primary antibodies, i.e TNC monoclonal mouse antibody (Invitrogen, 4C8MS) and anti-Versican monoclonal rabbit antibody (Abcam, ab269270), diluted 1:200 and 1:100, respectively, overnight at 4 °C. Immunogold labelling was applied with different gold-conjugated secondary antibodies (goat anti-mouse IgG (Gam)—15 nm (Aurion, cat. no. 815.022) and goat anti-rabbit IgG (Gar)—25 nm (Aurion, cat. no. 825.011), diluted 1:70. Afterwards, all sections were stained with uranyl acetate and lead citrate and observed in a JEOL JEM 2100Plus TEM (Japan), operated at 80 kV and equipped with a GATAN OneView CMOS camera (Gatan, USA). TEM electron micrographs were analysed using ImageJ to quantify the number of 15 and 25 nm immunogold particles per ECM area (μm²). Images were calibrated before use based on the micrograph's scale bar. A defined area of the ECM was selected, in each electron micrograph.

In silico data analysis was performed following published protocols⁴. Briefly, Seurat (v.5.1.0) R package functionalities were used for single-cell data mining processes⁹⁰. The Wilcoxon Rank Sum test was utilised for the identification of both markers and differentially expressed genes at the single cell level (FC > 1.2 and Bonferroni-adjusted p < 0.05). Gene Set Enrichment Analysis (GSEA) was implemented in the Cluster Profiler (v.4.12.6) R package⁹¹. Further details can be found in the Supplementary Information.

Statistics and reproducibility

Beyond transcriptomics data, statistical significance was evaluated using Prism (GraphPad) software, following the program’s built-in guidelines, as specified in each figure legend. In general, and unless noted otherwise, all datasets were tested for normality using the Shapiro-Wilk test, and all measurements were taken from distinct, independent samples. For comparisons between two groups with normally distributed data, we used a two-tailed unpaired t-test (assuming equal standard deviations) or Welch’s t-test (for unequal SDs). Non-normally distributed data were assessed with the two-sided Mann–Whitney test. For normally distributed, multi-group comparisons with equal SDs, unpaired one-way ANOVA followed by Tukey’s post hoc test was applied; if SDs were unequal, Welch’s ANOVA with Games-Howell post hoc test was used. For non-normally distributed multi-group data, the Kruskal–Wallis test followed by Dunn’s post hoc test was performed. Correlation analyses were conducted using Pearson’s test for normally distributed data and Spearman’s test for non-normally distributed data. No statistical method was used to predetermine sample size. Data are presented as box-and-whisker plots showing the median and all individual data points (n), and no data were excluded from the analyses.

Image creation

Third-party images were created at bioRender.com (under the relevant agreements; 26/11/2025-26/12/2025): Fig. 2b (AJ291L58LM), Fig. 2e (VL291L6Q8G), Fig. 5a (OI291L5KEC), Supplementary Fig. 7a (UE291L4Y2S), Supplementary Fig. 7c (QF295V1SL5), and Supplementary Fig. 11 (ZH291WJ0FH).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

V.A. conceptualised, supervised and funded the study. P.K. performed most of the experiments presented, assisted by I.B., E.K., D.N., S.S., M.S., and C.M. A.G. and S.G. performed FACS analyses. V.R. performed and analysed μCT. D.F. analysed transcriptomics data. M.Sa. performed and analysed MS proteomics. A.M.B. and N.G. performed and analysed AFM and F-actin image analysis. I.T., S.H., and V.G. performed and analysed TEM. I.T. and I.V. provided human lung samples and analysed related IHC. H.W. provided resources and co-evaluated results. P.K. and V.A. wrote the manuscript, which was then critically read and edited by all authors.

Peer review

Peer review information

Nature Communications thanks Fuquan Yang, Libang Yang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The scRNA-seq data utilised in this study are listed in Supplementary Table 2, including accession numbers and hyperlinks. The raw MS proteomics data generated in this study have been deposited in the ProteomeXchange Consortium via the PRIDE⁹² partner repository under the dataset identifiers PXD060583 and PXD063546. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-68377-5.