The F2-Isoprostane:Thromboxane-prostanoid receptor signaling axis drives persistent broblast activation in pulmonary brosis

While transient broblast activation is a normal and adaptive aspect of injury-repair in many contexts, persistent broblast activation is a hallmark of idiopathic pulmonary brosis (IPF) and other chronic brotic lung diseases. The mechanisms regulating persistent broblast activation in IPF have not been fully elucidated. In the lungs of IPF patients and in mice with experimental lung brosis, we observed that expression of the thromboxane-prostanoid receptor (TBXA2R) was increased in broblasts. Genetic deletion of TBXA2R, but not inhibition of thromboxane synthase, protected mice from bleomycin-induced lung brosis, suggesting an alternative ligand activates probrotic TBXA2R signaling. We found that F2-isoprostanes (F2-IsoPs), nonezymatic products of arachidonic acid metabolism generated in the setting of reactive oxygen species (ROS), are persistently elevated in experimental lung brosis and act as an alternative TBXA2R ligand. Further studies demonstrated that F2-isoprostanes signal through TBXA2R to activate Smad signaling, revealing TBXA2R as a previously unrecognized regulator of the transforming growth factor beta (TGF-β) pathway. Further, treatment with the small-molecule TBXA2R antagonist ifetroban protected mice from lung brosis in three pre-clinical models: bleomycin treatment, Hermansky-Pudlak Syndrome, and radiation-induced brosis. Importantly, treatment with ifetroban during the brotic phase of bleomycin injury markedly enhanced brotic resolution. Together, these studies implicate TBXA2R signaling as a crucial mediator linking oxidative stress to broblast activation and indicate that TBXA2R antagonists could be ecacious for treatment of IPF and other chronic brotic disorders.


Introduction
Idiopathic Pulmonary Fibrosis (IPF) is a chronic, progressive syndrome that is characterized by destruction of the gas-exchange units of the lung and pathologic accumulation of extracellular matrix (ECM). 1,2 Despite decades of work, the underlying mechanisms driving progressive pulmonary brosis remain incompletely understood, and most patients with IPF succumb to their disease within 3-5 years of diagnosis. 1 Injury to and dysfunction of the lung epithelium is hypothesized to play a prominent role in disease initiation, 3,4 while broblasts are the primary effector cell type producing pathologic ECM 5,6 and are the target of FDA-approved IPF therapies. 7,8 The mechanisms underlying persistent broblast activation in IPF remain incompletely understood.
Oxidative stress and elevated production of reactive oxygen species (ROS) have been implicated as key features of the dysfunctional lung/alveolar epithelium in IPF, 9,10 and have been associated with broblast activation. 11 Our group previously demonstrated that there is extensive accumulation of isolevaglandinadducted proteins in IPF lung tissue, 12 suggesting ROS generation is widespread and persistent in the brotic lung. In the setting of elevated ROS, arachidonic acid metabolites undergo nonenzymatic conversion to a series of compounds known as F2-isoprostanes (F2-IsoPs). 13 F2-IsoPs have been widely used as a biomarker of oxidative stress, prior studies have demonstrated the F2-IsoP's may also play direct signaling roles, including activation of hepatic stellate cells through the thromboxane-prostanoid receptor (TBXA2R). [14][15][16] In these studies, we determined that TBXA2R expression is increased during lung brosis, particularly in broblasts, where ligand binding by F2-IsoPs mediates broblast activation. Together, these investigations de ne a novel mechanism by which ROS contribute to brogenesis and implicate TBXA2R as a regulator of transforming-growth-factor beta (TGF-β) signaling. Our ndings highlight TBXA2R as a promising new therapeutic target for treatment of pulmonary brosis.

TBXA2R expression in broblasts is increased during lung brosis in mice and humans
To determine the role of TBXA2R in pulmonary brosis, we rst investigated TBXA2R expression in the lungs of IPF patients. In lung tissue lysates, quanti cation of TBXA2R by western blot demonstrated signi cantly higher TBXA2R levels in IPF compared to control lungs (Fig. 1a). Examining single-cell RNA sequencing (scRNA-seq) data generated from pulmonary brosis and non-brotic control (declined donor) lungs that our group has recently reported, 6 we observed that TBXA2R expression was detected in broblasts in addition, to endothelial cells and smooth muscle cells (Fig. 1b). Consistent with transcriptomic data, dual immuno uorescence staining showed TBXA2R co-localized with a broblast marker (S100A4) in areas of brotic remodeling in IPF lung tissue sections, but co-staining was not observed in lung parenchyma from non brotic controls (Fig. 1c). In mouse lung tissue, TBXA2R expression increased following bleomycin challenge as measured by western blot (Fig. 1d), and similar to our observation in IPF lung tissue, TBXA2R expression was localized to areas of brosis and colocalized with the broblast marker S100A4 (Fig. 1e). Together, these data raised the possibility of a previously unrecognized role for TBXA2R signaling in regulating broblast activity during pulmonary brosis in mice and humans.
To test the effects on TBXA2R signaling on lung brosis, we generated inducible TBXA2R de cient mice by crossing TBXA2R-oxed mice (TBXA2R f/f ) with a universal tamoxifen inducible Cre-recombinase line (see Methods) (TBXA2R iKO hereafter). These mice develop normally and have no spontaneous phenotype except for a mildly increased bleeding time. 17 Tamoxifen treatment (400 mg/kg chow ad libitum for 14 days) led to ~75% reduction in TBXA2R protein in lung tissue (see Extended Data Fig. 1a,b). The resulting TBXA2R iKO and age-matched adult WT controls were treated with tamoxifen and then challenged with IT bleomycin (0.04 units). At day 21 post-bleomycin, TBXA2R iKO mice had a striking reduction in lung brosis as determined by morphometric analysis (Fig. 1f,g and see Extended Data Fig. 1c) and measurement of hydroxyproline content (Fig. 1h). In contrast to genetic deletion of TBXA2R, treatment with ozagrel, a small-molecule inhibitor of the enzyme responsible for generation of TXA 2, 18 failed to protect mice from bleomycin-induced brosis (Fig. 1i,j,k), thereby indicating that an alternative ligand is responsible for pro brotic TBXA2R signaling.
F2-isoprostanes induce TBXA2R signaling in pulmonary brosis F 2 -isoprostanes (F 2 -isoPs) are a non-enzymatic product of free radical-induced peroxidation of arachidonic acid 13 that are increased in the lungs of IPF patients, 19 as well as other conditions where reactive oxygen species (ROS) are produced. 13 F 2 -isoPs have structural similarities to TXA 2 and can activate TBXA2R signaling. 16 We measured F2-IsoPs and thromboxane B2 (TxB2), the major stable metabolite of TxA2, in the lungs of mice at baseline and after IT bleomycin. TxB2 was increased in lung tissue by day 1 post-bleomycin and subsequently returned towards baseline by day 7 (Fig. 2a). In contrast, F2-IsoPs were increased in the lungs throughout the 21-day time course (Fig. 2b).
To determine whether F2-IsoPs are responsible for TBXA2R-driven pro-brotic phenotypes in lung broblasts, we isolated mouse-lung broblasts (MLFs) from WT and tamoxifen-treated TBXA2R iKO mice and cultured these cells in serum-containing media, which contains isoprostanes 20 . When grown on tissue culture plates in serum-containing media, TBXA2R iKO MLFs had ~50% reduced proliferation compared to WT MLFs (Fig. 2c). Under low serum conditions (2.5%), however, TBXA2R iKO MLFs had similar BrdU incorporation compared to WT. In low serum conditions, addition of F2-IsoPs (100 nM, Cayman Chemicals) to the culture medium enhanced proliferation of WT but not TBXA2R de cient MLFs ( Fig. 2d). In addition, treatment with F2-IsoPs induced α-SMA expression in WT MLFs but not TBXA2R iKO MLFs (Fig. 2e,f). We also measured Collagen 1 expression by western blot (Fig. 2g) and collagen accumulation in media (Fig. 2h) and found that F2-IsoPs upregulated collagen synthesis and secretion by WT MLFs but not TBXA2R iKO MLFs. Together, these studies indicated a speci c role for F2-isoprostaneinduced TBXA2R signaling in broblast activation.

TBXA2R-induced calpain activity mediates downstream activation of TGF-β
We next sought to determine the mechanisms connecting TBXA2R signaling to TGF-β activation. First, we treated TBXA2R iKO and control MLFs with recombinant human TGF-β1 (10 nM, R&D Systems) and evaluated SBE luciferase reporter activity at 4 hours. TGF-β1-induced Smad activation was not attenuated in TBXA2R iKO MLFs (Fig. 4a). Similarly, there was no attenuation of TGF-β1-induced gel contraction in TBXA2R iKO MLFs (Fig. 4b). These results showed that TBXA2R is not required for TGF-β pathway activation when active TGF-β is added directly to cells. We next tested whether TBXA2Rdependent activation of the TGF-β pathway by F2-IsoPs requires TGF-β receptors. Treatment of MLFs with F2-IsoPs induced robust induction of TGF-β targets SerpinE1 and Timp1, which was prevented by addition of the TGF-β receptor inhibitor LY2109761 (Fig. 4c). These ndings suggested that TBXA2R signaling does not activate TGF-β signaling independent of the TGF-β receptor complex.
In broblast cultures, TGF-β is primarily bound by a latent TGF-β binding protein (LTBP), either as part of a latent extracellular pool, or, in some cell types, intracellularly 21 . In this context, the calcium-dependent cysteine protease calpain-4 has been shown to activate TGF-β and induce Smad2/3 phosphorylation, likely by activating TGF-β in endosomal vesicles 21 . In light of the well-established effects of TBXA2R signaling on calcium ux 22 , we investigated whether F2-IsoPs could induce calpain activation in broblasts. Treatment of primary lung broblasts from WT mice with F2-IsoPs (100 nM) signi cantly increased calpain activity (Fig. 4d). Subsequent studies showed that addition of a calpain inhibitor (Z-LLY-FMK, 30 μM) blocked F2-IsoP induction of Smad2/3 signaling, collagen type I protein expression, and Serpine1/Timp2 gene expression ( Fig. 4e-g). Cumulatively, these studies support the conclusion that F2-IsoP induction of TBXA2R signaling stimulates calpain-induced activation of latent TGF-β, thus leading to downstream activation of the TGF-β pathway.
Ifetroban treatment attenuates brosis in mouse models of lung brosis.
To test whether pharmacological inhibition of TBXA2R could be used therapeutically to prevent or treat lung brosis, we treated wild-type (WT) mice (C57Bl/6 background) with Ifetroban (25 mg/kg/day in drinking water) or standard drinking water (vehicle) beginning 3 days before a single intratracheal (IT) injection of bleomycin (0.04U) or beginning on day 7 post-bleomycin and continuing until harvest at day 21. As shown in Figure 6, Ifetroban treatment blocked bleomycin-induced brosis when administered for the entire 21-day course or administered only during the brotic phase (day 7-21), as assessed by morphometric analysis (Fig. 6a,b) and hydroxyproline content (a major component of collagen) (Fig. 6c).
While epithelial apoptosis and in ammation, have been correlated with subsequent brosis [24][25][26][27] in the bleomycin model, Ifetroban did not affect epithelial apoptosis (Extended Data 3a-b) or in ammatory cell recruitment (Extended Data Fig. 4). To determine whether Ifetroban treatment could hasten resolution of brosis, we performed additional experiments where drug or vehicle was started on day 14 and continued until day 28 (Fig. 6d). In these studies, Ifetroban attenuated weight loss (Fig. 6e) and accelerated resolution of brosis ( Fig. 6f-h). Cumulatively, studies in the bleomycin model showed that Ifetroban reduces brosis and enhances resolution when given during the brotic phase of bleomycin-induced brosis without affecting bleomycin-induced in ammation or epithelial apoptosis, thereby suggesting direct effects of TBXA2R signaling on brogenesis. Proteins associated with brosis, including Tissue Inhibitor of Metalloproteinase-1 (Timp-1) and Smad2 phosphorylation downstream of canonical TGF-β signaling, were both strongly induced by bleomycin and blocked by Ifetroban treatment (Fig. 6i-k).
We next tested whether Ifetroban could prevent brosis in a model of genetic susceptibility to lung brosis, Hermansky-Pudlak Syndrome (HPS). HPS is an autosomal recessive disease in which several of the genetic subtypes (HPS1, HPS2, and HPS4) have highly penetrant pulmonary brosis with onset in early adulthood. 28 Naturally occurring mutations in HPS mice reliably model important features of the human disease, including susceptibility to pro-brotic stimuli. [29][30][31] To induce brosis in HPS mice, low doses of bleomycin (0.025 units) delivered by IT injection result in a rapidly progressive pulmonary brosis phenotype. [29][30][31] We treated HPS1 mice and HPS2 mice (with loss-of-function mutations in the HPS1 gene or the HPS2 gene, respectively) with Ifetroban beginning on the day prior to IT bleomycin and continuing until day 7 post-bleomycin, when lungs were harvested. Ifetroban treatment signi cantly attenuated lung brosis in HPS1 (Fig. 7a) and HPS2 mice (Fig. 7b,c) as determined by measurement of lung collagen.
We also used a model of radiation-induced brosis to investigate the anti-brotic effects of Ifetroban. Immediately following thoracic irradiation (17 Gy), mice were administered Ifetroban in drinking water or vehicle and continued treatment until lungs were harvested 4 months post-irradiation. Compared to placebo, Ifetroban-treated mice had signi cantly reduced lung collagen content following exposure to ionizing radiation (Fig. 7d,e). Together, these data indicate that TBXA2R is a promising therapeutic target for lung brosis.

Discussion
Transient broblast activation is an adaptive process that is crucial for wound and other forms of injuryrepair. While there has been considerable progress made in understanding the signaling and molecular processes that activate broblasts in the lung and other organs during injury-repair, the mechanisms that lead to persistent and pathologic broblast activation remain less well-de ned. In these studies, we demonstrate a novel paradigm linking oxidative stress to persistent broblast activation through F2-IsoPinduced TBXA2R signaling, which is a previously unrecognized regulator of the TGF-β pathway.
In the context of lung brosis, ROS are produced by a variety of cells, including epithelial cells 10,32 and broblasts. 33,34 NADPH oxidase 4 has been shown to be an important source of ROS in lung brosis. 11,35 ROS have been shown to enhance broblast proliferation, induce collagen formation, 34 and promote apoptosis-resistance. [35][36][37] F2-IsoPs are produced by non-enzymatic conversion of arachidonic acid in the presence of ROS 13,19 and have been shown to function as an alternative ligand for TBXA2R. 38,39 F2-IsoPs are increased in serum, BAL, and exhaled breath condensate from IPF patients [40][41][42] and have previously been suggested to contribute to bleomycin-induced brosis in rats. 43 Our ndings indicate that F2-IsoPs can mediate the effects of ROS on broblasts and that F2-IsoP-induced broblast activation can be prevented by TBXA2R antagonism or genetic deletion.
Although TBXA2R performs diverse functions beyond platelet biology, its role in lung brosis has not been previously investigated. Mechanistically, we found TBXA2R signaling in broblasts induces TGF-β signaling through a calpain-dependent mechanism. TGF-β is a master regulator of mesenchymal responses in physiological and pathological conditions, and persistent activation of TGF-β signaling has been described as a hallmark of pulmonary brosis. While mechanical forces 44 and other factors likely play a role in broblast activation during pulmonary brosis, 45-47 the molecular mechanisms that underlie the persistence of TGF-β activity in this setting remain uncertain. It has previously been recognized that oxidative stress can be associated with activation of latent TGF-β activation, 48,49 but the speci c mechanisms have not been determined. After translation, TGF-β is bound by LTBPs to maintain an inactive state. TGF-β can be activated or released from LTBPs through a variety of mechanisms, 50 including cleavage of LTBPs by the calcium-dependent cysteine protease calpain-4, 21 and downstream signaling through the TGF-β pathway interacts with numerous other pathways that mediate and modulate the biological effects of TGF-β. 51 Our data extend knowledge in this area by showing that TBXA2R signaling induces calpain activity in broblasts, which in turn mediates down-stream TGF-β activation. Consistent with this paradigm, calpain inhibition in broblasts reduced F2-IsoP-induced Smad2/3 signaling and inhibited pro brotic gene expression.
While these data provide compelling evidence that TBXA2R signaling is involved in pulmonary brosis, there are important limitations of these studies. First, there are likely cell-type speci c effects of TBXA2R in endothelial cells, smooth muscle cells, and platelets that may also impact brogenesis in the lungs. Further studies will be required to discern the effects of TBXA2R antagonism in these cell types on brogenesis. Second, we have not conclusively shown that F2-IsoPs are the major ligand for activation of TBXA2R signaling during in-vivo lung brosis. This would require a strategy to reduce ROS, which would likely have additional effects on lung brosis not mediated directly through TBXA2R. Third, additional work is required to determine whether TBXA2R is an important therapeutic target in brotic lung diseases other than IPF and brotic conditions in other organs.
Together, these studies have demonstrated a key role of TBXA2R-signaling in pulmonary brosis, and TBXA2R represents an attractive target with translational potential. In these experiments, we demonstrate the small-molecule ifetroban effectively inhibits TBXA2R signaling and pulmonary brosis in multiple experimental models. Ifetroban and other TBXA2R antagonists have been or are currently in human studies for secondary prevention of coronary artery disease, 52 allergic asthma, 53 and other conditions, [54][55][56][57] and appear to be safe and well-tolerated.
In summary, we have implicated TBXA2R-signaling as a key pathway in pulmonary brosis. These studies offer the potential for rapid translation into clinical trials with the goal of improving outcomes in IPF.

Subjects and Samples
IPF tissue samples were obtained from explanted lungs removed at the time of lung transplantation. Non brotic control tissue samples were obtained from lungs declined for organ donation.

Mouse Models
TBXA2R oxed mice 17 were obtained from Jackson Labs (#21985) . These mice have loxp sites surrounding TBXA2R exon 2, which under the in uence of Cre recombinase is excised, eliminating protein production. They had been backcrossed onto a B6 strain. These were crossed to mice with tamoxifeninducible cre expression under the control of the universally expressed Rosa26 locus (Rosa26-CreER, Jackson Labs #4847). 58 In combination, these Rosa26-CreER + TBXA2R f/f mice are referred to as TBXA2R iKO , and have global deletion of TBXA2R when induced by Tamoxifen. Hermansky-Pudlak Syndrome mice included "pale ear" mice (Jackson Laboratories #525) which carry a constitutive mutation in Hps1, 59 and "pearl" mice (Jackson Laboratories #3215), which carry a mutation in Hps2, also called Ap3b1. 60 Bleomycin-induced brosis models Bleomycin (Hospira Inc.) was purchased from Vanderbilt University Medical Center pharmacy. Bleomycin (0.04 units) in 100 μl saline was delivered by direct i.t. instillation under anesthesia as described previously. 25,26,61 Tamoxifen-inducible transgenic mice were treated with tamoxifen (400 mg tamoxifen citrate /kg of chow ad libitum) for 2 weeks prior to bleomycin instillation (4 weeks prior to bleomycin instillation). Lungs were harvested following euthanasia by pentobarbital at designated time points. Right lungs were tied off and snap frozen for estimation of collagen and extraction of RNA and protein, and left lungs were in ated to 25 cm H20 with 10% neutral buffered formalin for histology.
Mice were randomized to receive either 25 mg/kg/day CPI211 (ifetroban; Cumberland Pharmaceuticals, Nashville, TN) in drinking water or normal drinking water (vehicle). In a follow-up experiment, 50 mg/kg/day ozagrel HCl hydrate (CombiBlocks, San Diego, CA) was given to mice, and several mice were alternatively treated with ethanol vehicle as negative controls. Final concentration of ethanol in drinking water was approximately 0.04%. All drugs were pretested for palatability to ensure normal consumption of drinking water. Mice were weighed and water was changed once a week.
Radiation-induced brosis model 10-12 week-old male/female mice were randomly assigned to vehicle-or Ifetroban-treated groups. Iso urane anesthetized mice were placed on a 37°C recirculating water heating pad and the thorax was administered 17 Gy (300 kVp/10 mA X-rays) at 1.64 Gy per min as we previously described. 62 With the exception of the thorax the entire animal was shielded by a custom lead block 2.5 cm thick.

Measurement of collagen (in vivo)
Total collagen in right upper lobes or whole right lungs of the lung was measured using the hydroxyproline assay (Biovision, kit #K555) or Sircol assay (Biocolor S1000 kit) as per the manufacturer's instructions and as previously published 61 .

Treatment with F2-IsoPs or U46619 (in vitro)
Cells were seeded at a density of 8 × 10 4 cells/ml in DMEM supplemented with 20% FBS and allowed to grow to confluence. Twelve hours before the treatment by F2-IsoPs or U46619, the medium was changed to DMEM with 2.5% FBS and 50 μg/ml ascorbic acid. Three hours before the treatment by F2-IsoPs or U46619, Ifetroban (300nM) was added to the culture media. Cells were then treated with either 15-F2-IsoPs (100nM) (Cayman Chemical, USA) or with TBXA2R-agonist, U46619 (Cayman Chemical, USA) in the same concentration. A stock solution of 15-F2t-IsoPs and U46619 (both 1 mg/ml in ethanol) was diluted to a concentration of 10 −5 M and then further diluted to final concentrations with DMEM.

Measurement of collagen (in vitro)
Cells were grown to con uence in 60-mm dishes and the medium was replaced with DMEM containing 2.5% FBS. The cells were incubated with either F2-IsoPs or TxB2 (Cayman Chemical, Ann Arbor, MI) for 48 h at 37°C. In some experiments, Ifetroban or Z-LLY-FMK (Abcam) was added 3h before stimulation, respectively. The amount of total soluble collagen in culture supernatants was measured by the Sircol dye binding assay kit (BioColor Ltd.) according to the manufacturer's protocol.
Mouse lung broblast culture At the time of harvest, the lungs were perfused blood free with 30 ml PBS containing 10 U/ml heparin from the right ventricle, then minced and digested in an enzyme cocktail of DMEM containing 1% of BSA , 2 mg/ml of collagenase type IV, 100 ug/ml of DNase, and 2.5 mg of Dispase II (Roche Diagnostics, Mannheim, Germany) at 37°C for 30 min, then strained through a 100-um nylon cell strainer. Mouse lung cells were pretreated with anti-CD16/32 antibody (BioLegend) to block Fc receptors and then incubated with specific antibodies at 4°C in the dark. Lung broblasts were isolated by FACS Aria, de ned as CD140a+/CD31-/CD45-/CD326-. Cells were isolated from 3-5 mice per group for each experiment.
Isolation and culture of primary human lung broblasts Primary lung broblasts were isolated from IPF lungs removed at the time of lung transplant surgery as previously described 63,64 .

Ex-vivo TBXA2R deletion
Cre recombinase activity was induced by 4-hydroxytamoxifen (4-OH TAM) (Sigma) at a dose of 0.2 µM, which was added into cell culture medium for 8 days.

Measurement of F2-IsoPs and TxB2
The levels of F2-IsoPs and TxB2 in lung tissues were determined by gas chromatography-mass spectrometry as previously described 65,66 .

Immunoblotting
Western blotting was performed as previously described 67 Fig. 4b).
Single-cell RNA-sequencing analysis Single-cell RNA-sequencing (scRNA-seq) data generated from single-cell suspensions generated from pulmonary brosis and non brotic control lungs underwent quality control ltering and unbiased clustering followed by cell-type annotation as previously reported 6 . Raw and processed 10X genomics data can be found on GEO using accession number: GSE135893. The code used to analyze the data can be found at https://github.com/tgen/banovichlab/.

Statistics
Values are expressed as mean ± SEM, and sample size is given for each figure. Two-way analysis of variance (ANOVA) or 1-way ANOVA followed by the Sidak post-test was performed on Prism (GraphPad, San Diego, CA) to determine statistical significance.

Declarations
Study Approval: All studies involving human samples were approved by the Vanderbilt Institutional Review Board (IRB #'s 060165, 171657, 192004). Subjects provided informed consent prior to the collection of samples used in these studies. All animal studies were approved by the IACUC at Vanderbilt University.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. EXTENDEDDATAFIGURES.pdf