Gli signaling pathway modulates fibroblast activation and facilitates scar formation in pulmonary fibrosis

Abstract

Pulmonary fibrosis is characterized by progressive and irreversible scarring of alveoli, which causes reduction of surface epithelial area and eventually respiratory failure. The precise mechanism of alveolar scarring is poorly understood.

In this study, we explored transcriptional signatures of activated fibro- blasts in alveolar airspaces by using intratracheal transfer in bleomycin-induced lung fibrosis. Lung fi- broblasts transferred into injured alveoli upregulated genes related to translation and metabolism in the first two days, and upregulated genes related to extracellular matrix (ECM) production between day 2 and 7.

Upstream analysis of these upregulated genes suggested possible contribution of hypoxia- inducible factors 1a (Hif1a) to fibroblast activation in the first two days, and possible contribution of kruppel-like factor 4 (Klf4) and glioma-associated oncogene (Gli) transcription factors to fibroblast activation in the following profibrotic phase.

Fibroblasts purified based on high Acta2 expression after intratracheal transfer were also characterized by ECM production and upstream regulation by Klf4 and Gli proteins. Pharmacological inhibition of Gli proteins by GANT61 in bleomycin-induced lung fibrosis altered the pattern of scarring characterized by dilated airspaces and smaller fibroblast clusters.

Acti- vated fibroblasts isolated from GANT61-treated mice showed decreased migration capacity, suggesting that Gli signaling inhibition attenuated fibroblast activation.

In conclusion, we revealed transcriptional signatures and possible upstream regulators of activated fibroblasts in injured alveolar airspaces. The altered scar formation by Gli signaling inhibition supports that activated fibroblasts in alveolar airspaces may play a critical role in scar formation.

Introduction

Pulmonary fibrosis, such as idiopathic pulmonary fibrosis (IPF), is a disease in which persistent injury and tissue remodeling generate massive scars with excessive deposition of extracellular matrix (ECM) [1]. This fibrotic scar formation is usually progressive and intractable, eventually leading to respiratory failure.
Because there are currently very few therapeutic options for pulmonary fibrosis, elucidating the mechanism of scar formation and pro- gression is a pressing topic for the development of novel therapies. The lung has unique tissue structure, characterized by a reticular network of thin alveolar walls optimized for gas exchange [2].

Pulmonary fibrosis involves destruction of alveoli and replaces normal tissue with thick scars, which reduce the surface area for gas exchange [3]. The precise mechanism of alveolar scarring is unknown. It is thought that repetitive microinjury to epithelium triggers subsequent tissue remodeling [3]. Subepithelial resident fibroblasts migrate into injured area and differentiate into myofi- broblasts, followed by scar formation in the setting of wound healing [4].

Similar activation and migration of fibroblasts are implicated in pulmonary fibrosis [5]. In a previous study, we have shown that fibroblasts transferred into injured alveolar airspaces through an intratracheal route get activated to produce ECM and are eventually incorporated into scars [6]. This result suggests that activated fibroblasts in scars can originate from injured alveolar airspaces and scars can be formed by coalescence of injured alve- olar walls. It is possible that activated alveolar fibroblasts play a leading role in the formation of such scars as described in wound healing of other organs [4,7].

Therefore, exploring activation sig- natures of fibroblasts in injured alveolar airspaces may reveal the mechanism of scar formation in pulmonary fibrosis.

Glioma-associated oncogene (Gli) proteins are transcription factors which mediate hedgehog signalings and play important roles in fibroblast activation [8,9]. Growing evidence suggests that Gli proteins are also activated through non-canonical pathways such as transforming growth factor-b pathway [10].

Furthermore, a recent study has shown that pirfenidone, an approved drug for idiopathic pulmonary fibrosis, exerts its antifibrotic effects by suppressing the Gli signaling [11]. In this study, we utilized intra- tracheal transfer of fibroblasts in bleomycin (BLM)-induced lung fibrosis to investigate gene expression signatures of activated fi- broblasts after exposure to alveolar airspaces.

We identified Gli signaling as a possible upstream regulator for fibroblast activation after exposure to alveolar airspaces, and pharmacological inhibition of Gli signaling resulted in altered scar formation. These findings suggest an important role of Gli proteins in activated fibroblasts located at alveolar airspaces for scarring in pulmonary fibrosis.

Materials and methods

Mice

Collagen type 1 alpha 2-green fluorescent protein (Col-GFP) reporter mice were generated in a previous study and kept in C57BL/6 background [12]. Acta2-monomeric Kusabira Orange 1 (mKO1) transgenic reporter mice were generated with the 2.6 kb promoter and first intron of Acta2 as described in Mack et al. [13]. The detail of Acta2-mKO1 mouse generation is described in the supplementary method.

The original BAC transgenic mouse for Acta2-mKO mouse is available from RIKEN center for Biosystems Dynamics Research (Accession No. CDB0539T, http://www2.clst. riken.jp/arg/TG%20mutant%20mice%20list.html). C57BL/6J mice were purchased from CLEA Japan (Tokyo, Japan). Mice were maintained in specific pathogen free facilities at the University of Tokyo.

All animal experiments were performed in accordance with the guidelines of the Animal Care and Use Committee of the Uni- versity of Tokyo and the Institutional Animal Care and Use Com- mittee of RIKEN Kobe Branch.

Purification of transferred fibroblasts

BLM treatment, isolation of primary fibroblasts, and intra- tracheal transfer into BLM-treated wild type mice were performed as described previously [6]. Briefly, primary lung fibroblasts iso- lated from Col-GFP or Col-GFP/Acta2-mKO1 double transgenic mice by negative selection for CD31, CD45, CD146, Ter119, and EpCAM were transferred into wild type mice, which received 2 mg/kg BLM 7 days before the transfer. Col-GFP+ cells from untreated lungs and transferred Col-GFP+ cells from host lungs were sorted with a MoFlo Astrios cell sorter (Beckman Coulter, Brea, CA). Samples were pooled from 4 mice at each time point. Total RNA was isolated by RNeasy mini kit (QIAGEN, Venlo, Netherlands).

Serial analysis of gene expression (SAGE) and data analysis

Whole transcriptome of purified transferred cells was amplified and acquired by SAGE as described previously with minor differ- ences [14e16]. Purified total RNA was dissolved in lysis buffer (1% LiDS, 100 mM Tris-HCl pH 7.5, 500 mM LiCl, 10 mM EDTA, 5 mM DTT) and mRNA was trapped on Dynabeads M270 streptavidin (Thermo Fisher Scientific, Waltham, MA) labeled with biotinylated oligo-dT- containing DNA adapters as described previously [16].

On-beads reverse transcription and second strand synthesis was performed by adding RT mix (1X First-strand buffer, 0.5 mM dNTP, 2.5 mM DTT, 1U/ml RNaseIn Plus (Promega, Fitchburg, WI), 1.2 mM SMART-LNA oligo (QIAGEN), 20U/ml Superscript II (Thermo Fisher Scientific)) and incubating 37 ◦C for 1 h, 70 ◦C for 10 min, and hold at 4 ◦C.

Then, whole transcriptome was amplified using PrimeSTAR DNA poly- merase (Takara, Kyoto, Japan) by performing the following program; 98 ◦C 2min, 18 cycles of [98 ◦C 20 s, 60 ◦C 15 s, 68 ◦C 4 min], 68 ◦C 4 min, and hold at 4 ◦C. SAGE libraries were constructed by using each amplified whole transcripts as described previously [14].

Sequencing was performed by using Miniseq and Miniseq High output kit (75 cycles) (Illumina, San Diego, CA). Mapping of SAGE tags and between-sample normalization of expression value were performed as described previously [14].

Sequence of SMART-LNA oligo was as follows; GCGGCTGAAGACGGCCTATGTrGrG + G. Raw data and processed file have been deposited in the NCBI Gene Expression Omnibus (accession number GSE126704).

Upstream regulator analysis

Relative gene expression changes between each time point or between Acta2-mKO1 high and low cells were calculated, and genes up-regulated more than three folds were used for subse- quent upstream regulator analysis.

Enrichment of transcription factor binding motifs in promoter regions of the upregulated genes was analyzed by the GeneXplain platform (GeneXplain GmbH, Wolfenbüttel, Germany). Promoter length was defined from +2000 to —100 base pairs around the transcription start site and the default settings were used for the other calculation parameters.

Gene ontology (GO) biological process analysis of upregulated genes was performed by using GeneXplain.

Histology

Col-GFP mice received subcutaneous injection of 25 mg/kg GANT61 (Abcam, Cambridge, MA) every other day from day 5 to day 13 after BLM treatment and the lungs were harvested 14 days after BLM treatment. Immunohistochemistry was performed as described previously [14].

Briefly, 10 mm frozen sections of the lungs were made after fixation with 4% paraformaldehyde and dehydra- tion with 30% sucrose. Sections were stained with hematoxylin (Wako Pure Chemical Industries, Osaka, Japan) and eosin (Wako) (H&E), or antibodies, followed by mounting with Entellan New (Millipore Sigma, Burlington, MA) for H&E staining or Prolong Gold (Thermo Fisher Scientific) for immunofluorescent staining. Anti- collagen 1 antibody (LSL-LB-1102) was purchased from LSL (Tokyo, Japan).

Migration assay

Col-GFP mice were treated with 25 mg/kg GANT61 7 days after BLM treatment. GFP+ cells were purified by cell sorting 2 days after GANT61 treatment and their migration capacity was analyzed by fibronectin-coated Oris cell migration assay kit (Platypus Technol- ogies, Madison, WT) as described previously [6].

Statistical analysis

Data are expressed as mean ± SEM where applicable. Statistical comparisons were performed by one-way analysis of variance with the Tukey-Kramer posttest for multiple groups. A P value < 0.05 was considered statistically significant. Statistical analysis was performed using Prism 5.01 software (GraphPad Software, La Jolla, CA).

Results

To investigate serial gene expression changes of lung fibroblasts which are exposed to injured alveolar airspaces, we performed intratracheal transfer of Col-GFP+ lung fibroblasts in BLM-induced lung fibrosis and acquired whole transcriptome of purified donor cells on day 2, 4, and 7 after the transfer (Fig. 1).

Lineage markers (CD31, CD45, CD146, EpCAM, and Ter119)-negative Col-GFP+ cells from untreated mice were used as day 0. We then examined upregulated (>3 folds) genes between each time point and sought upstream regulators by analyzing enrichment of transcription factor binding motifs in the promoter regions of those upregulated genes (Fig. 1).

The top 6 upstream transcription factors are shown in Table 1. A notable upstream regulator between day 0 and 2 was hypoxia-inducible factors 1a (Hif1a). Upstream regulators of later time points (day 4/day 2, day 7/day 4) were similarly represented by kruppel-like factor 4 (Klf4) and Gli signaling molecules.

Of note, Klf4, Gli1, and Gli2 are identified in our recent report as hub tran- scription factors in lung fibroblast activation in BLM- and silica- induced pulmonary fibrosis [16]. We also performed GO enrich- ment analysis of the upregulated genes (Supplementary Table 1). Between day 0 and 2, GO terms related to translation or metabolic process were enriched.

Consistent to the upstream regulators, GO terms for day 4/day 2 and day 7/day 4 were similar and marked by terms related to ECM organizations (Supplementary Tables 2 and 3).

These data suggest that fibroblasts exposed to injured alveolar airspaces first underwent metabolic changes driven by Hif1a fol- lowed by activation of ECM production driven by Klf4 and Gli signaling molecules.

Next, we asked how a common activation marker of fibroblasts, Acta2, defined gene expression signatures in alveolar airspaces by using Acta2-mKO1 reporter mice (Fig. 2A). In the lungs of Acta2- mKO1 mice, mKO1 was expressed in smooth muscle cells and mesenchymal cells in alveoli (Fig. 2B).

By sorting mKO1 low, in- termediate, and high population, we confirmed that mKO1 expression correlated with Acta2 mRNA expression (Supplementary Fig. 1). We crossed Col-GFP mice with Acta2-mKO1 mice, and transferred lung fibroblasts isolated from those mice into BLM-treated wild type mice.

We sorted mKO1 high and low pop- ulation from the host lungs 4 days after the transfer, and performed transcriptome analysis (Fig. 2C). We calculated fold gene expression changes between mKO1 high and low populations and examined upstream regulators for genes upregulated (>2 folds) in the mKO1 high population (Table 2).

Similar to the later time points of the previous experiment, Klf4 and Gli signaling molecules were enriched as upstream regulators. GO term analysis also showed higher expression of genes related to ECM organization in mKO1 high population (Supplementary Table 4).

These data suggest that Acta2 high activated fibroblasts highly produce ECM and those activation genes are possibly regulated by Klf4 or Gli signaling molecules.

To test how upregulated genes after exposure to alveolar air- spaces affect scar formation in fibrosis, we utilized GANT61, an inhibitor of Gli1 and Gli2, in the BLM-induced lung fibrosis model. GANT61 was previously shown to partially reduce ECM deposition in BLM-induced lung fibrosis [17] and unilateral ureteral obstruc- tion kidney fibrosis [9].

We treated Col-GFP mice with BLM and injected GANT61 every other day from day 5e13 after BLM treat- ment (Fig. 3A). We found that GANT61 treatment altered the pattern of scars on day 14 (Fig. 3B and C). The lungs treated with GANT61 were characterized by dilated airspaces with partially thickened interstitial regions, which is in stark contrast to the dense scars with less airspaces in vehicle-treated mice.

Thickened inter- stitial regions in GANT61-treated mice encompassed collagen matrix but with smaller Col-GFP+ clusters compared to vehicle-treated mice (Fig. 3C). Increased airspaces in GANT61-treated mice compared to vehicle was also illustrated by image quantifi- cation (Fig. 3D). These data may suggest that inhibition of Gli sig- nalings affect fibroblast activation and alter scar formation.

To explore how GANT61 affects fibroblast activation, we treated col-GFP mice with GANT61 on day 7 after BLM treatment, and measured migration capacity of Col-GFP+ cells purified from the lungs on day 9, which is the phase that alveolar fibroblasts migrate into alveolar airspaces (Fig. 4A) [6].

As expected, Col-GFP+ cells from GANT61-treated mice showed reduced migration capacity compared to those from vehicle-treated mice (Fig. 4B and C). These data suggest that one of the mechanisms of altered scar formation by Gli signaling inhibition is the reduction of migration capacity in activated fibroblasts in alveolar airspaces.

Discussion

Our previous study demonstrated that the intratracheal transfer of lung fibroblasts in bleomycin-induced lung fibrosis is a useful model for studying fibroblast activation [6]. Since the migration of interstitial fibroblasts into alveolar airspaces is a crucial process in forming a fibrotic niche at injured sites [5], we utilized intratracheal transfer to investigate the molecular changes occurring in fibroblasts within alveolar airspaces.

We found that Hif1a binding motifs were enriched in the promoter regions of genes upregulated during the initial phase following the transfer. Hif1a plays a critical role in responding to hypoxia and has been reported to have various functions in multiple cell types involved in fibrosis [18].

Fibroblasts undergo significant morphological changes after exposure to injured alveolar airspaces, including an increase in cell size and a rounded shape [6]. Since Hif1a activation in fibroblasts has been shown to enlarge cell size and rearrange actin filaments [19], it is possible that metabolic and translational changes driven by Hif1a contribute to these morphological alterations in the initial activation phase.

An open question remains regarding how Hif1a is activated in injured alveolar airspaces. While the canonical hypoxia pathway may play a role, the extent of hypoxia in injured alveolar airspaces is not yet known. Another possibility is that iron sequestration mediated by ferritin activates Hif1a [20], as altered iron metabolism has been associated with lung diseases [21]. These questions warrant further investigation.

Gli signaling pathways play critical roles in both embryonic and postnatal lung development [22]. Numerous studies suggest that wound healing and fibrotic diseases involve the activation of pathways that are also utilized during development [23].

In the adult lung, Gli signaling activity, as indicated by Gli1 expression, is restricted to peribronchial and perivascular regions within bronchovascular bundles, while it is absent in alveoli [24,25].

However, in bleomycin-induced lung fibrosis, activated fibroblasts within scars exhibit Gli signaling activity, suggesting that the Gli signaling pathway is reactivated in scar-forming fibroblasts within alveoli after injury [24]. This finding aligns with the present study, which indicates that Gli proteins function as upstream regulators of fibroblast activation during the fibrotic phase.

The mechanism by which Gli signaling is activated in fibroblasts remains unknown. Notably, inhibiting the canonical Hedgehog pathway does not attenuate bleomycin-induced lung fibrosis [17,24], suggesting that Gli pathway activation in fibrotic lesions is primarily driven by a non-canonical mechanism.

A previous study demonstrated that inhibiting Gli signaling with GANT61 reduces collagen deposition in lung fibrosis [17]. In the present study, we found that GANT61 treatment also alters the pattern of scar formation. Pulmonary fibrosis is characterized by the spatially dynamic remodeling of alveoli, leading to a loss of alveolar surface area and a reduction in total lung volume [3]. Understanding how alveoli undergo such remodeling after injury is crucial to comprehending the pathology of pulmonary fibrosis.

Our data suggest that attenuating fibroblast activation in injured alveolar airspaces suppresses excessive scar formation and helps preserve alveolar surface area. This highlights the potential of an alternative therapeutic approach—focusing on preserving alveolar airspaces rather than solely reducing extracellular matrix (ECM) deposition. It would be valuable to investigate whether the beneficial effects of pirfenidone on respiratory function are mediated by alveolar space preservation through Gli signaling inhibition [11].

In conclusion, we identified the transcriptional signatures of activated fibroblasts in injured alveolar airspaces using the intratracheal transfer model. Investigating the roles of potential upstream regulators identified in this study, such as Hif1a, Klf4, and Gli proteins, may provide valuable insights into the mechanisms of alveolar scarring caused by activated fibroblasts.

Furthermore, Gli pathway inhibition with GANT61 attenuates fibroblast activation and modifies scar formation, suggesting that targeting fibroblast activation in injured alveolar airspaces could serve as a novel therapeutic strategy for treating pulmonary fibrosis.