Dopamine D1 receptor stimulates cathepsin K-dependent degradation and resorption of collagen I in lung fibroblasts

Approximately 3 million people worldwide are affected by idiopathic pulmonary fibrosis (IPF) (Martinez et al., 2017), a progressive interstitial lung disorder with a median survival from 2 to 4 years after diagnosis (Richeldi et al., 2017). A hallmark of this disease is the excessive and uncontrolled deposition of the extracellular matrix (ECM) fibrillar proteins, primarily collagen type I, by activated fibroblasts that contributes to scarring of the tissue, disruption of alveolar architecture and loss of respiratory function (Haak et al., 2018; Martinez et al., 2017; McKleroy et al., 2013; Richeldi et al., 2017). Intriguingly, evidence from experimental models and clinical case studies suggests that the pathological deposition of fibrotic scarring is reversible (Chitra et al., 2013; Della Latta et al., 2015; Haak et al., 2018; Jun and Lau, 2018). The most common rodent model for pulmonary fibrosis involves a single intratracheal administration of bleomycin in healthy young mice, which induces an inflammatory and a fibrotic phase characterized by overexpression of collagen I. At ∼30–60 days after bleomycin administration resolution of the injury, clearance of fibrotic ECM and repair of the lung architecture is observed (Della Latta et al., 2015; Moeller et al., 2008; Williamson et al., 2015). The reversibility of fibrosis in model systems is not limited to the lung; resolution of liver, kidney, skin and heart fibrosis is also observed in experimental animal models of each disease (Jeong et al., 2016; Jun and Lau, 2018; Kantari-Mimoun et al., 2015; Kim et al., 2013; Lemaire et al., 2016; Schuppan, 2015; Weiskirchen et al., 2019). In humans, multiple medications, including chemotherapies, promote interstitial lung scarring, collagen deposition, and reduced pulmonary function that resolves upon cessation of drug exposure (Schwaiblmair et al., 2012). Clinical cases of liver, kidney and heart fibrosis have also been observed to be reversible (Duffield, 2014; Friedman et al., 2013; Gourdie et al., 2016; Jun and Lau, 2018; Zoubek et al., 2017). Taken together, these observations suggest that the functional pathology of overabundant ECM deposition in fibrosis, even in IPF, may not be irreversible, and thus understanding the mechanisms that promote fibrosis resolution may lead to new and effective therapies for this disease.

Pulmonary ECM is a complex structure composed of fibrous proteins, glycoproteins and proteoglycans, which together sustain the structural and functional integrity of the tissue (Haak et al., 2018; Zhou et al., 2018). Matrix homeostasis is maintained by the balance in synthesis and degradation of the ECM proteins by resident cells (Chang et al., 2020; Humphrey et al., 2014). Following injury, fibroblasts deposit collagen I, among other proteins of the ECM, and then, with resident macrophages, participate in repairing and remodeling the lung architecture (Frantz et al., 2010; Wynn and Vannella, 2016), a process guided by the coordinated secretion and activation of matrix metalloproteases (MMPs) and cysteine cathepsins, and internalization of collagen through macropinocytosis, endocytosis or phagocytosis. Once internalized, collagen I is further degraded in acidic lysosomal compartments by cathepsin proteases (Bonnans et al., 2014; Fonović and Turk, 2014; Pakshir and Hinz, 2018). A substantial portion of collagen is degraded by cells before ever being deposited in the lung, a process that can be controlled by cAMP signaling (Rennard et al., 1982). Whether cAMP signaling also plays a role in clearance of already deposited ECM remains unclear.

The dopamine receptor D1 (DRD1) is a G_αs coupled G-protein coupled receptor (GPCR) that, in the lung, is expressed predominantly in fibroblasts and upon activation increases cAMP (Haak et al., 2020). Treatment with dihydrexidine (DHX), a DRD1 agonist, inactivates YAP/TAZ leading to reduced expression of ECM and ECM cross-linking genes, fibroblast contraction, proliferation and TGFβ-stimulated collagen I accumulation in vitro, and increased expression of cathepsin K (Haak et al., 2019), a gene associated with collagen I degradation and clearance (Bühling et al., 2004; Vidak et al., 2019). Furthermore, during the fibrotic ECM deposition phase of the bleomycin mouse model, transgenic mice overexpressing cathepsin K present reduced collagen deposition (Srivastava et al., 2008) and treatment with DHX accelerates the resolution of lung fibrosis (Haak et al., 2019).

Here, we aimed to determine whether signaling through DRD1 actively promotes the in vitro and in vivo extracellular clearance and degradation of collagen I by lung fibroblasts. We observed that treatment with DHX promotes the degradation of cell-deposited collagen I through the activity of cathepsin K and its internalization into lysosomal compartments, resulting in the reduced stiffness of cell-derived ECM and enhanced fibroblast-mediated collagen degradation in vivo. Our results demonstrate that dopamine D1 agonism switches fibroblasts from a state of matrix deposition to a state of matrix resorption, highlighting the dopamine D1 receptor as a target for reversing ECM accumulation in pulmonary fibrosis.

To assess the capacity of fibroblasts to deposit and degrade ECM, we adapted methods for studying cell-derived matrices using fibroblasts seeded at high density (Cukierman, 2002; Franco-Barraza et al., 2016). We stimulated collagen deposition by treating normal human lung fibroblasts (NHLFs) with TGFβ for 3 days and collected total protein (both cellular and extracellular) to observe collagen abundance at this time point (Fig. 1A). At day 3, we then treated the cells with DHX, a dopamine D1 receptor agonist, for an additional 24 h and again collected total protein (both cellular and extracellular) at day 4. In prior work, we confirmed the specificity of DHX effects acting via D1 receptor using both receptor-specific antagonists as well as D1 receptor siRNAs (Haak et al., 2019). Here, we observed an increase in collagen I deposition between control and TGFβ day 3, and no significant change between TGFβ at day 3 and 4, verifying the presence of a cell-derived collagen-rich matrix prior to cell culture treatment with the DRD1 agonist. However, treatment with DHX reduced the collagen I content previously deposited by day 3, indicative of loss or degradation, not simply halting of production (Fig. 1A).

Collagen I is most commonly found as a triple helix formed by two α1 chains and one α2 chain that undergo several post-translational modifications prior to their assembly for fibril formation (Kadler et al., 1996). After leaving the intracellular compartment as pro-collagen, propeptide cleavage takes place at both ends of the protein, mediated by procollagen proteinase, resulting in formation of the telopeptide (Kadler et al., 1996). The resulting collagen peptide undergoes fibril formation and cross-linking to form mature fibers that build the extracellular matrix (McKleroy et al., 2013). As a way to specifically assess the extracellular collagen content in our system, we used an antibody that binds to the telopeptide and analyzed the same experimental time course (Fig. 1A). The changes in the collagen I α1 (Col1α1) telopeptide at ∼150 kDa strongly correlated with the observed changes in total collagen I at ∼150 kDa, confirming that the changes we observe with DHX and TGF-β are consistently observed for both total and extracellular deposited collagen I. In addition, enhanced degradation products were also observed in telopeptide western blots for samples treated with DHX for 24 h, consistent with collagenolytic activity induced by activation of DRD1 (Fig. 1A). We also treated cells on day 3 with forskolin to directly activate adenylyl cyclase, one of the downstream effectors of D1 receptor activation, and observed similar reductions in collagen I (Fig. 1B), consistent with a role for cAMP in these observations. We have previously shown that DHX promotes resolution of bleomycin-induced lung fibrosis; however, DHX treatment has no measurable effect on lung collagen content in healthy mice (Haak et al., 2019), suggesting that D1 agonism effects are dependent on the activation state of lung fibroblasts. Here, we treated unstimulated (no TGFβ) cultured lung fibroblasts with DHX for 24 h and observed no reduction in total or extracellular collagen I (Fig. 1C). These results are consistent with DHX promoting collagen I degradation selectively in TGFβ-activated cells.

To functionally confirm the collagenolytic activity following DHX treatment, we used DQ Collagen, a fluorescein-labeled collagen substrate that generates a fluorescent signal upon cleavage (Della Porta et al., 1999; Sameni et al., 2009). Treatment of fibroblasts with DHX for 24 h stimulated the collagenolytic activity of lung fibroblasts, generating a ∼2.5-fold increase of the fluorescent signal (Fig. 1B). This finding further supports the capacity for DRD1 agonism in fibroblasts to promote cleavage of extracellular type I collagen.

To analyze the functional outcome of D1 receptor signaling on cell-derived matrices, we adapted a decellularizing protocol from previously published methods (Cukierman, 2002). In order to assess the integrity of the ECM and the effectiveness of the de-cellularization methodology, immunofluorescence labeling of collagen type I was performed on intact fixed cultures and cell-derived matrices after decellularization (Fig. 2A). Image analysis confirmed the absence of cell nuclei and presence of collagen I with preserved fiber alignment and no loss of staining intensity after decellularization (Fig. 2A).

Using atomic force microscopy (AFM), we extended a previously developed assay for characterizing in vitro cell-derived matrix elasticity (Haak et al., 2019). First, cell-derived matrices were produced by stimulating lung fibroblasts for 3 days with TGFβ and ascorbic acid, as previously described (Franco-Barraza et al., 2016). The cells were then treated with DHX or vehicle control for 24 h. At day 4, decellularization was performed just before AFM microindentation. The mean and individual Young's modulus for all 50 measurements performed per independent experiment is shown in Fig. 2B (bottom left), and the average Young's modulus value determined for each independent experiment is also shown (bottom right). Consistent with our prior observations of DHX-stimulated collagen I degradation (Fig. 1), we found that DHX stimulation alters the stiffness of the cell-derived ECM after 24 h of treatment. While the magnitude of the changes in stiffness we observed are relatively modest, they span a large fraction of the difference in normal and fibrotic lung ECM stiffness previously observed in experimental and human lung fibrosis (Liu et al., 2015, 2010), consistent with a physiologically relevant effect size. To support that these results were an effect on ECM degradation and not altered collagen cross-linking, we measured lysyl oxidase (LOX) activity from cell culture conditioned medium treated with TGFβ and with or without DHX. Consistent with previous findings (Wei et al., 2017), we observed an increase in LOX activity with TGFβ treatment, but D1 agonism had no effect on LOX enzymatic activity (Fig. S1).

Our prior work demonstrated significantly increased resolution of experimental lung fibrosis with prolonged treatment with DHX (Haak et al., 2019). However, whether acute treatment specifically increases fibroblast-mediated collagen degradation was not assessed. In order to determine whether DHX treatment can acutely promote fibroblast-mediated collagen I degradation in vivo we studied tissue from prior experiments in which bleomycin was administered intratracheally to Col1α1–GFP⁺ mice at day 0 to initiate lung injury. On day 10 and 11, these mice were then treated during ongoing fibrosis with two doses of DHX (5 mg/kg of body weight intranasal) or vehicle control, 24 and 2 h prior to collecting lungs for sectioning and staining (Haak et al., 2019). Collagen-hybridizing peptide (CHP) is a synthetic peptide designed to identify cleaved collagen products. It contains a repeating sequence of glycine, proline and hydroxyproline that specifically binds to unfolded collagen chains denatured by proteolytic activity, thereby reconstructing the triple-helical structure. In previous studies, CHP was shown to abundantly stain lung sections during the resolution phase of bleomycin-induced fibrosis in young mice, consistent with the hypothesis that collagen proteolysis is an endogenous component of lung fibrosis resolution (Hwang et al., 2017). Based on these findings, we incubated the lung slices from our acute DHX study with CHP (Fig. 3A). In these mice collagen type I-producing cells express GFP, a convenient means for predominantly identifying fibroblasts in situ. We observed a statistically significant enhancement in colocalization of CHP with GFP⁺ cells in DHX-treated compared to vehicle-treated groups (Fig. 3C) (no CHP control staining is shown in Fig. S2A). Thus at day 10 post bleomycin administration, acute treatment with DHX appears to mimic the previous results where CHP staining was increased in mice during the spontaneous resolution phase of bleomycin-induced fibrosis at day 28 post bleomycin (Hwang et al., 2017). Interestingly, we did not observe an overall enhanced staining of CHP in the lung sections in the DHX-treated group (Fig. 3B), consistent with the selective expression of the D1 receptor on lung fibroblasts and DHX specifically enhancing fibroblast-mediated collagen degradation in response to D1 receptor activation. Equally, we did not observe enhanced colocalization of CHP with GFP⁺ cells in sham-treated healthy mice (Fig. S2B), consistent with DHX not promoting collagen degradation in unstimulated cultured fibroblasts (Fig. 1C).

Canonical ECM resorption by fibroblasts involves internalization into lysosomes (Bonnans et al., 2014). To test whether DRD1 agonism stimulates this canonical resorption pathway, we used super-resolution microscopy to image primary lung fibroblasts immuno-labeled for Col1α1 telopeptide and lysosomal membrane protein 1 (LAMP1), a major transmembrane protein in the lysosomal compartment (Fig. 4A). DHX enhanced uptake of extracellular Col1α1 telopeptide, and, as assessed from 3D reconstruction of z-stacks, compartmentalized consistently and significantly with LAMP1-labeled lysosomes (Fig. 4B). We also repeated this experiment labeling collagen I degradation with CHP and observed similar colocalization between LAMP1 and CHP in cells treated with DHX (Fig. S3). These findings confirm that D1 receptor agonism mediates internalization of extracellular collagen into lysosomes.

One of the main effector proteases involved in lysosomal collagen I degradation is cathepsin K (Bühling et al., 2004; Sprangers and Everts, 2019). Based on our findings of DHX promoting lysosomal compartmentalization of collagen I , we investigated the effects of D1 agonism on cathepsin K expression, maturation and its functional role in ECM degradation. Stimulation of lung fibroblasts with TGFβ reduced transcript levels of the CTSK gene. However, dopamine receptor activation promoted enhanced transcript levels of the CTSK gene (Fig. 5A). At the protein level, cathepsin K is synthesized and secreted in a pro-format that is ∼43 kDa in size and upon cleavage for activation, the molecular mass is reduced to ∼29 kDa (Brömme, 2013). Treatment with DHX enhanced the expression of the low molecular mass active form of cathepsin K in cells stimulated with TGFβ (Fig. 5B). Aside from cathepsin K, the major proteases which degrade triple-helical collagen I are MMPs. To observe changes in MMP protein expression after D1 receptor agonism, we measured expression of MMP1, MMP2, MMP3, MMP8, MMP 9, MMP10 and MMP13, and their endogenous inhibitors TIMP1, TIMP2 and TIMP4 using an MMP array. We did not observe any appreciable changes in MMPs or TIMPs after DHX treatment (Fig. S4), enhancing our focus on CTSK.

GPCR signaling through cAMP has previously been shown to enhance acidification of lysosomal compartments in a variety of cell types (Coffey et al., 2014; Guha et al., 2012; Liu et al., 2008), and cathepsin K is known to be spontaneously activated at low pH within lysosomes (Fonović and Turk, 2014; McQueney et al., 1997). We measured lysosomal pH using LysoSensor Yellow/Blue DND-160, which exhibits predominantly yellow fluorescence in acidic environments and blue fluorescence in less-acidic environments. Dual-emission measurements with this probe allow ratiometric quantification of lysosomal pH (Liu et al., 2012, 2008). We validated the methodology using compounds known to increase lysosomal pH, namely tamoxifen and chloroquine (Fig. S5), and then confirmed that both DHX and forskolin acidify the lysosomal environment (Fig. 5C).

To assess the effect of cathepsin K on collagen I degradation, we inhibited cathepsin K activity with odanacatib (ODN), a highly selective and potent inhibitor developed for the treatment of postmenopausal osteoporosis (Chapurlat, 2015). As observed previously, TGFβ enhanced collagen I expression, which was reduced by treatment with DHX. However, when we inhibited cathepsin K activity with ODN, the effect of DHX on reducing the collagen I levels was blocked, and the amount of collagen I present was greater than that observed in fibroblasts stimulated only with TGFβ (Fig. 5C). A similar trend was observed for Col1α1 telopeptide, demonstrating that ODN also prevented the DHX-mediated decrease in the extracellular collagen I content. Furthermore, ODN prevented the appearance of collagen I degradation products induced by DRD1 agonism with DHX (Fig. 5D). ODN treatment of cells in the absence of DHX did not enhance collagen I deposition (Fig. S6). These data are consistent with cathepsin K playing a primary effector role in the collagen I degradation stimulated by DRD1 receptor agonism in lung fibroblasts. Finally, we repeated the DQ collagen analysis and confirmed ODN largely prevented DHX mediated collagen degradation (Fig. 5E).

Taken together, our results thus demonstrate that DRD1 agonism stimulates extracellular degradation and intracellular resorption of collagen I by fibroblasts, representing a potentially important fibroblast-mediated mechanism for clearance of fibrillar ECM during injury repair and fibrosis resolution.

Fibrosis progression is associated with an increased deposition of fibrillar collagens, with collagen I representing 80% of the collagen content (McKleroy et al., 2013). As it is consolidated in the ECM, this protein transforms the extracellular compartment into a denser and more-rigid structure (Karsdal et al., 2017). In IPF, treatment options that halt matrix deposition have been developed, but mechanisms that promote matrix resorption have not been fully explored. In addition, degradation of collagen I without disrupting tissue homeostasis is an important challenge that must be considered. There is strong experimental support for collagen degradation being an essential component of resolution in patients with liver, kidney and heart fibrosis, as well as in the animal models for each disease (Duffield, 2014; Friedman et al., 2013; Gourdie et al., 2016; Iredale et al., 2013; Jeong et al., 2016; Jun and Lau, 2018; Kantari-Mimoun et al., 2015; Lemaire et al., 2016; Weiskirchen et al., 2019; Zoubek et al., 2017). Recently, we have identified DHX, acting through DRD1 expressed predominantly on fibroblasts, to have beneficial effects in mouse models of lung and liver fibrosis, resulting in reduced levels of collagen I in vitro and in vivo (Haak et al., 2019). Here, we found that agonism of DRD1 receptor through DHX promotes fibroblast-mediated degradation of cell-derived collagen I through the activity of cathepsin K, a major cysteine protease responsible for extracellular and intracellular collagen I degradation. DHX enhanced both expression and activation of cathepsin K, consistent with coordinated engagement of a matrix resorption program stimulated by DRD1 agonism (Fig. 6).

Degradation of mature collagen is challenging due to its tight triple-helical structure and crosslinking with other ECM proteins, which obstructs access to proteolytic sites (Kafienah et al., 1998; Sprangers and Everts, 2019). ECM degradation involves membrane-bound and secreted enzymes; cathepsin K and members of the MMP family have been identified as proteases capable of degrading fibrillar collagen (Sprangers and Everts, 2019). A recent finding directly implicated cathepsin K in fibroblast-mediated maintenance of collagen homeostasis in ex vivo mice tendons (Chang et al., 2020). In addition, protein expression of cathepsin K in tendon fibroblasts was associated with increased collagen α1 and α2 protein expression (Chang et al., 2020), suggesting a feedback mechanism between collagen deposition and degradation that maintains homeostasis of the extracellular compartment. Our results support previous findings that fibroblasts synthesize and deposit collagen I in the extracellular compartment (Chang et al., 2020) and demonstrate that acute stimulation of the dopamine D1 receptor with DHX rapidly engages cathepsin K activity to shift fibroblasts toward net degradation of collagen I, a potentially important step in resorption of excess ECM.

Conventionally, cathepsin K is thought to degrade intracellular collagen I in the lysosomal compartment. However, it has also been suggested to play a major role in the degradation of fibrillar collagen in the extracellular compartment (Bonnans et al., 2014; Chang et al., 2020; Haak et al., 2018; Kafienah et al., 1998; Sprangers and Everts, 2019). Transgenic mice globally overexpressing CTSK are protected against collagen deposition and fibrosis in the lung following bleomycin injury (Srivastava et al., 2008), in line with the beneficial effect of treating bleomycin-injured mice with DHX (Haak et al., 2019). Interestingly, transcript levels of CTSK are highly expressed in fibroblasts from patients with IPF (Bühling et al., 2004), suggesting that, potentially, mechanisms that regulate maturation and activation of cathepsin K might be crucially important to regulate collagen degradation and resorption in IPF. While we have not studied the link between DHX treatment and enhanced cathepsin K activity in detail, our previous work has demonstrated that DHX, acting through DRD1, stimulates cAMP signaling in lung fibroblasts (Haak et al., 2019). To investigate whether DHX exerts a collagenolytic effect through cAMP elevation, we treated fibroblasts with forskolin, a positive stimulator of cAMP production and observed similar effects to those seen upon treatment with DHX. Furthermore, GPCR signaling through cAMP has previously been shown to enhance acidification of lysosomes in a variety of cells (Coffey et al., 2014; Guha et al., 2012; Liu et al., 2008), and cathepsin K is known to be spontaneously activated at low pH within lysosomes (Fonović and Turk, 2014; McQueney et al., 1997), providing a plausible mechanism linking DRD1 agonism to the enhanced collagenolytic activity observed. Here, we report reversal of the collagenolytic effects of a D1 receptor agonist when we co-treated the cells with a cathepsin K inhibitor, both in total collagen I and in extracellular Col1α1 telopeptide, indicating a role for cathepsin K in extracellular collagen degradation stimulated by DRD1 agonism. We also observed colocalization between collagen telopeptide and LAMP1 in lysosomal compartments, a finding that could be explained by two mechanisms, phagocytosis of collagen fibrils and degradation in the lysosomes, or degradation of collagen I in the extracellular compartment (by either cathepsin K alone or with other enzymes) and uptake into the lysosomal compartment. Extracellular cathepsin K might be activated in situ, or transported in active form from lysosomal compartments to the ECM. These mechanisms are not mutually exclusive, and future work will need to further delineate their relative importance.

In addition, our results do not rule out important contributions from other collagenolytic pathways, which likely play important roles in unfolding the triple-helix and cleaving collagen peptides, notably MMPs and plasminogen activators. Interestingly, urokinase plasminogen activator (uPA), which itself is activated extracellularly by multiple cathepsin proteins (Vidak et al., 2019), was recently shown to promote degradation of established lung fibrosis (Horowitz et al., 2019). An important extension of the current work will be to further demonstrate the in vivo role of fibroblasts in collagen resorption during fibrosis resolution. Prior work has demonstrated that depletion of macrophages during the resolution phase of mouse models of liver and lung fibrosis delays the clearance of collagen I (Duffield et al., 2005; Gibbons et al., 2011). However, work in dermal and lung tissues also suggests an important role for fibroblasts in proteolytic degradation of collagen to maintain tissue homeostasis (Podolsky et al., 2020; Zigrino et al., 2016). In the in vivo study performed here, we only treated mice for 24 h with DHX, which we had previously shown selectively targets lung fibroblasts (Haak et al., 2019). We found enhanced colocalization of CHP, indicative of collagen degradation, with Col1a1–GFP⁺ cells, supporting a contribution from fibroblasts to collagen degradation and resorption. Importantly, future work will need to ascertain the relative contributions of fibroblasts, through cathepsin K and other collagenolytic enzymes, to maintenance of ECM homeostasis and resolution of fibrosis in the lung. However, our work strongly suggests that fibroblasts stimulated by D1 agonists play an important direct role in collagen degradation and fibrosis resolution, emphasizing the potential utility of targeting this pathway therapeutically in IPF.

Dimethyl sulfoxide (DMSO, BP231-100) and L-ascorbic acid (BP351-500) were purchased from Thermo Fisher Scientific (Waltham, MA, USA). Ammonium hydroxide (LC11080-2) was acquired from LabChem Inc (Zelienople, PA, USA). Formalin solution (HT501128-4L), Triton X-100 (T8787) and tamoxifen (T5648-5G) were obtained from Sigma-Aldrich (St Louis, MO, USA). Dihydrexidine (884) was purchased from Tocris Bioscience (Bristol, UK). Forskolin was purchased from Tocris Bioscience (1099) and Cayman Chemical (11018, Ann Arbor, MI, USA). Odanacatib (24166) and chloroquine (14194) were acquired from Cayman Chemical. TGFβ1 was purchased from eBioscience (14-8348-62; San Diego, CA, USA) and PreproTech (100-21C; Rocky Hill, NJ, USA). Biotin–CHP (b-CHP, Bio60) was acquired from 3Helix (Salt Lake City, UT, USA).

Primary normal human lung fibroblasts (NHLFs) were purchased from Lonza (Allendale, NJ, USA) or provided by Peter Bitterman and Craig Henke from the University of Minnesota (Minneapolis, MN, USA); NHLFs were isolated by explant culture from donors whose organs were rejected for transplantation, under a protocol approved by the University of Minnesota Institutional Review Board. NHLFs were cultured in Eagle's minimal essential medium (EMEM) (ATCC, Manassas, VA, USA) containing 10% fetal bovine serum (FBS) and antibiotic-antimycotic (Thermo Fisher Scientific) unless otherwise noted. Cells were routinely characterized by PCR or western blotting using specific fibroblast markers including PDGFRα and vimentin. Mycoplasma contamination was monitored routinely by PCR. Experiments were performed with cells between passages four and eight.

NHLFs were plated on tissue culture dishes (Thermo Fisher Scientific, 60×15 mm) in EMEM containing 10% FBS and allowed to attach and grow for 48 h. Cells were stimulated with 2 ng/ml TGF-β and 10 µM DHX in EMEM containing 0.1% FBS for 24 h prior to RNA isolation using the RNeasy Plus Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Isolated RNA (165 ng) was then used to synthesize cDNA using SuperScript VILO (Invitrogen, Carlsbad, CA, USA). Quantitative PCR (qPCR) was performed using FastStart Essential DNA Green Master and analyzed with LightCycler 96 (Roche, Basel, Switzerland). Data are expressed as a fold change by ΔΔCt relative to the level of the GAPDH housekeeping gene, and normalized to control. All wells were treated with a final concentration of 0.1% DMSO. Primer sequences were: GAPDH (forward, 5′-ACATCGCTCAGACACCATG-3′; reverse, 5′-TGTAGTTGAGGTCAATGAAGGG-3′), CTSK (forward, 5′-CTCCTTCCAGTTTTACAGCAAAG-3′; reverse, 5′-TTTCCCCAGTTTTCTCCCC-3′).

A clear-bottom 96-well plate (Thermo Fisher Scientific) was coated with DQ™ Collagen (Thermo Fisher Scientific) diluted to 50 µg/ml in PureCol^® (Advanced Biomatrix, Carlsbad, CA, USA) and incubated for 1 h at 37°C and 5% CO₂. The coating solution was then removed, and wells were gently washed with PBS. NHLFs were plated in EMEM containing 10% FBS and allowed to attach for 6 h. Medium was removed and cells were treated for 24 h with 2 ng/ml TGFβ, 10 µM DHX and 3 µM odanacatib in EMEM containing 0.1% FBS; all wells were treated with a final concentration of 0.1% DMSO. After 24 h, cells were fixed with 10% formalin and incubated for 1 h at room temperature with DAPI solution (Thermo Fisher Scientific) diluted at 1:1000 in DPBS (Life Technologies, Carlsbad, CA, USA). Images were taken and analyzed using a Cytation 5 imager (BioTek, Winooski, VT, USA). Results are expressed as a fold change relative to a control group.

NHLFs were plated at 80% confluency in six-well plates (Thermo Fisher Scientific) in EMEM containing 10% FBS, and allowed to attach for 24 h. Medium was then exchanged with EMEM containing 0.1% FBS, and cells were stimulated for 72 h with 2 ng/ml TGFβ, prior to treatment with 10 µM DHX or 10 µM forskolin for 24 h. All wells were treated with a final concentration of 0.1% DMSO. Total protein was isolated using RIPA buffer pH 7.6 containing Pierce phosphatase inhibitor and Halt™ protease and phosphatase inhibitor cocktail (Thermo Fisher Scientific). Lysate protein concentration was determined using the Pierce™ BCA protein assay kit (Thermo Fisher Scientific) and samples were run on 4–15% Mini-PROTEAN^® TGX™ gels (Bio-Rad, Hercules, CA, USA) at 100 V for 80 min. Proteins were transferred onto a PVDF membrane using the Trans-Blot^® Turbo™ Transfer System (Bio-Rad). Membranes were incubated overnight with primary antibodies against the following proteins: collagen I (Novus Biologicals Inc., Centennial, CO, USA; NB600-408) diluted 1:1000, cathepsin K (Santa Cruz Biotechnology, Dallas, TX, USA; sc-48353) diluted 1:200, Col1α1 telopeptide (Invitrogen; PA5-35380) diluted 1:1000 and GAPDH (Cell Signaling, Danvers, MA, USA; 14C10) diluted 1:1000 in 5% nonfat dry milk (Bio-Rad). Membranes were washed with TBS with 0.1% Tween 20 before 60 min incubation with HRP-conjugated anti-rabbit IgG or anti-mouse IgG (Promega, Madison, WI) diluted 1:3000 in 5% nonfat dry milk. Membranes were imaged via a ChemiDoc™ Imaging System (Bio-Rad) and protein quantification was performed via densitometry using Image Lab v6.0 (Bio-Rad). Raw values were normalized against housekeeping gene GAPDH and are presented as a fold change relative to a control group.

Adapting from previously published methods (Cukierman, 2002; Franco-Barraza et al., 2016), NHLFs were plated to confluence onto gelatin (Cell Biologics, Chicago, IL, USA)-coated tissue culture dishes (Thermo Fisher Scientific, 60×15 mm) in EMEM containing 10% FBS and allowed to attach for 24 h. Medium was then exchanged with EMEM containing 0.1% FBS, 2 ng/ml TGFβ and 20 µg/ml ascorbic acid to enhance the ECM deposition, and cells were cultured for 3 days prior to treatment with 10 µM DHX. After 24 h, medium was removed and cell-derived matrices were decellularized by incubating with 20 mM ammonium hydroxide for 5 min at room temperature. Subsequently, matrices were washed three times (1 min each time) with DPBS and incubated with 0.5% Triton X-100 for 30 s at room temperature. Finally, matrices were washed three times (1 min each time) with DPBS and maintained in EMEM containing 0.1% FBS for AFM microindentation experiment.

Col1a1–GFP transgenic mice were generated as previously described (Yata et al., 2003). All animal experiments were carried out under protocols approved by the Mayo Clinic Institutional Animal Care and Use Committee (IACUC).

The CHP analysis performed here made use of previously generated lung samples collected in our initial study of dopamine signaling in the context of pulmonary fibrosis (Haak et al., 2019). Briefly, 6–8-week-old Col1α1-GFP⁺ mice were anesthetized with ketamine/xylazine, and then treated with bleomycin (50 μl at 1.2 U/kg body weight) or PBS control on day 0, as described previously (Haak et al., 2019). Mice were subsequently treated with DHX (5 mg/kg body weight) twice more, at 24 h and 2 h prior to lung harvest. We fixed the right-side lung with 4% paraformaldehyde for 24 h before rinsing with PBS and adding optical tissue compound (OCT Tissue Plus, 4585, Fisher Scientific); 6 μm sections were cut for staining and imaging.

For the in vitro studies, adapting from previously published methods (Cukierman, 2002; Franco-Barraza et al., 2016), NHLFs were plated to confluence onto gelatin (Cell Biologics, Chicago, IL, USA)-coated tissue culture dishes (Thermo Fisher Scientific, 60×15 mm) in EMEM containing 10% FBS and allowed to attach for 24 h. Medium was then exchanged with EMEM containing 0.1% FBS, 2 ng/ml TGFβ and 20 µg/ml ascorbic acid to enhance the ECM deposition, and cells were cultured for 6 days. At day 6, medium was removed and cell-derived matrices were decellularized as previously described here. Subsequently, the collagen I intensity of intact (cell containing) and decellularized cell-derived matrices, were obtained using a confocal microscope at 20× magnification. Fluorescence imaging was performed for five independent experiments, acquiring between 3 and 5 images per condition. Images were processed using ImageJ; for each experiment, the color channels are split and the mean gray value for each color is measured, for each image acquired per experiment. The data is expressed as the average of the mean gray values for each condition. For the in vivo studies, Col1α1–GFP lung slices were permeabilized using 0.1% Triton X-100 (X100, Millipore Sigma) and stained with biotin-conjugated CHP (b-CHP) according to the manufacturer's instructions, to investigate collagen resorption adjacent to fibroblasts. Secondary antibody conjugated to Alexa Fluor 555 (S32355, Thermo Fisher) and Hoechst (33258, Tocris Bioscience), were used for visualization. Z-stack micrographs were taken using an LSM 780 with Zen software (Zeiss). Images were taken at the same laser power, gain and offset. Using ImageJ (Schindelin et al., 2012), images from each channel were processed together for brightness and contrast modifications, with the γ-values adjusted to 2.

Lung slices stained with b-CHP were imaged using a Cytation 5 Cell Imaging Reader (BioTek Instruments) at 10× magnification. Images were taken to avoid tissue folding or lung edges, within the entire imaging field containing tissue. For quantifying CHP and Col1α1–GFP colocalization, using ImageJ, thresholds were set to the top 5% of pixels for images taken in the b-CHP and Col1α1–GFP channels. CHP pixels colocalized with GFP pixels were measured and quantified as the percentage of colocalized pixels per image field. In addition, CHP images were analyzed to determine total amount of CHP per image field, as a representation of collagen degradation. Using ImageJ, thresholds were set for each individual image by two investigators that were blind to the sample identification; each point represents the average of two thresholds.

NHLFs were grown on a BioCoat™ eight-well culture slide in the presence of 2 ng/ml TGFβ and 20 µg/ml ascorbic acid for 5 days. At 24 h and 2 h prior to fixation, cells were treated with 10 μM DHX. At day 6, cells were fixed and stained (including antigen retrieval) for LAMP1 (NBP2-52721SS, Novus Biologics), and with biotin-conjugated collagen hybridizing peptide (b-CHP) according to the manufacturer's instructions, and anti-Col1α1 telopeptide antibody (PA5-35380, Thermo Fisher Scientific). Secondary antibody Alexa Fluor 647 (A31573, Thermo Fisher Scientific), Alexa Fluor 488 (21202, Thermo Fisher Scientific), Alexa Fluor 555 (S32355, Invitrogen), and Hoechst (33258, Tocris Bioscience), were used for visualization. We took z-stack super resolution images on a Zeiss Elyra PS.1 super-resolution microscope using a 63× oil immersion objective (NA 1.4). Images were batch processed using Structured Illumination from Zen (Zeiss). The output files were imported into Imaris 8 (Bitplane, version 8.2.0) for 3D reconstruction and we used the colocalization tool with default settings (automatic thresholding based on P-value) to quantify colocalization. To account for volumetric changes in cell size or number of images per field of view, the normalized number of voxels (volumetric pixels) of colocalized LAMP1 and collagen telopeptide were used to create plots. Results are expressed as the fold change relative to TGFβ. Representative images were produced by using the maximum intensity projections obtained from Zen (Zeiss). Images were imported into ImageJ, and channels were batch processed together using brightness and contrast adjustments for visualization, with the γ-value set to 2. Colocalization images were produced by determining pixels containing both LAMP1 and telopeptide staining, or LAMP-1 and bCHP staining, using the ‘AND’ procedure of the image calculator.

NHLFs were plated at 80% confluency in six-well plates (Thermo Fisher Scientific) in EMEM containing 10% FBS, and allowed to attach for 24 h. Medium was then exchanged with EMEM containing 0.1% FBS, and cells were stimulated for 72 h with 2 ng/ml TGFβ and 20 µg/ml of ascorbic acid, prior to treatment with 10 µM DHX for 24 h. After 24 h, medium from each condition was collected, and LOX activity was determined using the fluorometric LOX activity assay kit (ab112139, Abcam, Cambridge, UK), as per the manufacturer's instructions. Briefly, cell culture supernatant was collected and centrifuged at 13,000 g for 5 min at 4°C. Then, in a clear-bottom 96-well plate, reaction wells were set in duplicates with 50 μl of cell culture medium and 50 μl of LOX reaction mix; 50 μl of assay buffer was used for the blank control. Plates were incubated at 37°C and 5% CO₂ for 40 min. Fluorescence was read on a microplate reader at excitation (Ex) and emission (Em) of 540 nm and 590 nm, respectively, with a cut off at 570 nm. Background subtraction from the fluorescence intensity was performed by subtracting the blank control from the fluorescence intensity collected for each sample well. Data is presented as the average of the fluorescent intensity.

NHLFs were plated at 80% confluency in 96-well plates (Thermo Fisher Scientific) in EMEM containing 10% FBS, and allowed to attach for 24 h. Medium was then removed and fibroblasts were incubated for 1 min with the pre-warmed Lysosensor™ Yellow/Blue DND-160 (L7545, Thermo Fisher Scientific) at 5 µM diluted in PBS. Then, the Lysosensor probe was removed and cells were washed with PBS, prior to treatment with 10 µM DHX, 10 µM forskolin, 20 µM chloroquine and 30 µM tamoxifen, diluted in PBS. Fluorescence was measured during a time interval spanning between 1 and 60 min, in a microplate reader at a dual Ex and Em of 329 and 440 nm, and 384 and 540 nm. Data are presented as the ratio of 440 nm to 540 nm emission.

NHLFs were plated to confluence onto tissue culture dishes (Thermo Fisher Scientific, 60×15 mm) in EMEM containing 10% FBS, and allowed to attach and grow for 48 h. Medium was then exchanged with EMEM containing 0.1% FBS and cells were stimulated with 2 ng/ml TGFβ, and treated with 10 µM DHX for 24 h. The activity of human MMPs was measured from culture medium by using the Human MMP Array C1 (AAH-MMP-1-8, RayBiotech, Peachtree Corners, GA, USA) as per the manufacturer's instructions.

Data analysis and plotting were performed using Prism 8.0 (GraphPad Software, La Jolla, CA, USA). In vitro experimental results follow a normal distribution according to the Shapiro Wilk test, except for the graph ‘decellularized ECM’ in Fig. 2B. Data sets derived from in vivo experiments (Fig. 3) have a normal distribution according to the D'Agostino and Pearson's test. Statistical comparison between two groups was performed by paired and unpaired t-test, or Mann–Whitney test, according to the parametric or non-parametric behavior of the data, respectively. A statistical analysis of three or more groups was performed by repeated measures (RM) one-way ANOVA with Dunnett's multiple comparison test, where the mean value of each group was compared against a control group. Results are expressed as the mean±s.e.m. unless otherwise indicated in the figure legends, with statistical significance represented by P-value. The sample number (n) indicates the number of independent samples in each experiment.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.248278.reviewer-comments.pdf