Therapeutic targeting of 15-PGDH in murine pulmonary fibrosis

Idiopathic pulmonary fibrosis (IPF) is a progressive disease characterized by interstitial remodeling and pulmonary dysfunction. The etiology of IPF is not completely understood but involves pathologic inflammation and subsequent failure to resolve fibrosis in response to epithelial injury. Treatments for IPF are limited to anti-inflammatory and immunomodulatory agents, which are only partially effective. Prostaglandin E2 (PGE2) disrupts TGFβ signaling and suppresses myofibroblast differentiation, however practical strategies to raise tissue PGE2 during IPF have been limited. We previously described the discovery of a small molecule, (+)SW033291, that binds with high affinity to the PGE2-degrading enzyme 15-hydroxyprostaglandin dehydrogenase (15-PGDH) and increases PGE2 levels. Here we evaluated pulmonary 15-PGDH expression and activity and tested whether pharmacologic 15-PGDH inhibition (PGDHi) is protective in a mouse model of bleomycin-induced pulmonary fibrosis (PF). Long-term PGDHi was well-tolerated, reduced the severity of pulmonary fibrotic lesions and extracellular matrix remodeling, and improved pulmonary function in bleomycin-treated mice. Moreover, PGDHi attenuated both acute inflammation and weight loss, and decreased mortality. Endothelial cells and macrophages are likely targets as these cell types highly expressed 15-PGDH. In conclusion, PGDHi ameliorates inflammatory pathology and fibrosis in murine PF, and may have clinical utility to treat human disease.


15-PGDH is highly expressed and active in healthy murine lung tissue. (+)SW033291 is known
to increase systemic PGE2 levels and enhance tissue regeneration 26 . To determine if 15-PGDH may be targetable in the murine lung, we first examined its expression in the lungs of healthy mice, relative to other organs in which 15-PGDH inhibitors (PGDHi) have demonstrated therapeutic efficacy 26,27 . Immunohistochemical staining revealed subsets of PGDH+ hematopoietic cells in the bone marrow (BM) and numerous PGDH+ cells in the colonic epithelium (Fig. 1A,B). In contrast, 15-PGDH was highly expressed throughout the lung parenchyma (Fig. 1C), suggesting pulmonary tissue may also be responsive to PGDH inhibition. We further compared expression of Hpgd, the gene that encodes 15-PGDH, in colonic and pulmonary tissue relative to BM. Lung tissue homogenates displayed 25-and 3-fold higher Hpgd gene expression than BM and colon, respectively (Fig. 1D).
To confirm that pulmonary 15-PGDH is functional and therefore capable of regulating local PGE2 levels, we next measured specific enzymatic activity in homogenates from the same organs. Lung tissue demonstrated > 200-and 3.7-fold higher activity per milligram total protein than BM and colonic tissue, respectively (Fig. 1E).
Together these data demonstrate that 15-PGDH is abundantly expressed and highly enzymatically active in the murine lung and provide rationale for pharmacologic targeting of 15-PGDH as a strategy to increase local PGE2 and limit the pathogenesis of pulmonary fibrosis (PF).

PGDHi mitigates early bleomycin-induced inflammation.
In mice, administration of bleomycin results in lung injury that mimics key aspects of human IPF 28 , with an initial inflammatory phase followed by a subsequent fibrotic phase 29 . To study the effects of 15-PGDH inhibition in PF, we administered bleomycin intravenously and began twice daily treatment of mice with (+)SW033291 (PGDHi) or vehicle control ( Fig. 2A). www.nature.com/scientificreports/ PGDHi attenuated early pulmonary inflammation, as evidenced by greater than 50% reductions in Il1b and Il6 expression in lung tissue 7 days post-bleomycin exposure, in addition to moderate reductions in the expression of other inflammation-associated genes (Fig. 2B). These data indicate that inhibiting 15-PGDH in the context of bleomycin-induced lung injury may limit pathologic inflammation in the lung. Moreover, PGDHi treatment was associated with significantly lower levels of the neutrophil chemoattractant CXCL1, the cytokine TNFα, and a trend towards reduced IL-10 in the serum (Fig. 2C). CXCL1 and TNFα were also reduced in PGDHi-treated mice exposed to bleomycin intratracheally (Supplementary Fig. 1A-B). TNFα promotes TGFβ1 expression 30 , and although IL-10 limits inflammation in many contexts, it is thought to promote fibroblast proliferation in PF 31 , therefore these changes likely have an anti-fibrotic effect on injured pulmonary tissue. Of note, TGFβ1 was elevated in the serum of vehicle-treated mice 7 days post-bleomycin exposure and PGDHi treatment led to a moderate but statistically insignificant reduction (44 ± 5.5 vs. 36 ± 1.9 pg/mL). Additionally, while vehicletreated mice displayed alveolar wall thickening and an abundance of small alveolar spaces, a trend towards greater alveolar size was observed with PGDHi ( Fig. 2D). Analysis of the lungs of mice 14 days post-intratracheal bleomycin administration demonstrated similar morphological changes ( Supplementary Fig. 1C). These data therefore indicate that PGDHi limits both systemic and local inflammation during the early phase of bleomycininduced lung injury.
PGDHi improves survival in bleomycin-induced pulmonary fibrosis. The above results demonstrate that PGDHi mitigates acute inflammatory responses to bleomycin lung injury. To test whether PGDHi additionally protects against the severity of chronic fibrotic disease, we extended our analysis and continued to administer (+)SW033291 to bleomycin-exposed mice through day 35 (Fig. 3A). While vehicle-treated mice experienced severe weight loss in the first 8 days following bleomycin administration, PGDHi limited the severity and duration of weight loss ( Fig. 3B and Supplementary Fig. 2A-B). Specifically, vehicle-treated mice reached a nadir of 26.8% body weight loss on day 8 post-intravenous bleomycin treatment, versus PGDHi-treated mice, which experienced only 16.0% maximum mean body weight loss on day 6 with weight stabilization thereafter. PGDHi-treated mice also achieved weight recovery closer to baseline between 21 and 28 days post-bleomycin administration, suggesting that PGDHi attenuated the effects of bleomycin on overall health. Notably, 35% of vehicle-treated mice succumbed to death, compared to 10% of PGDHi-treated mice (Fig. 3C). These results were www.nature.com/scientificreports/ also observed in mice exposed to bleomycin intratracheally, where 40% of vehicle-and 16% of PGDHi-treated mice succumbed to death ( Supplementary Fig. 2C). Together these results suggest PGDHi therapy ameliorates systemic pathology during bleomycin-induced lung injury.
To determine the impact of PGDHi on bleomycin-induced lung fibrosis, ECM accumulation was evaluated in lung tissue 35 days post-bleomycin exposure. Importantly, pulmonary 15-PGDH expression was maintained following bleomycin administration ( Supplementary Fig. 3). Lung sections from vehicle-treated mice showed regional accumulation of Collagen 1a1 (COL1a1) and alpha-smooth muscle-actin (αSMA), whereas staining was reduced and non-focal in the lungs of PGDHi-treated mice (Fig. 3D), indicating that 15-PGDH inhibition protects against severe fibrosis in bleomycin-induced PF. In addition to COL1a1, total collagen and lung morphology was evaluated on Masson's trichrome-stained lung sections. Vehicle-treated mice displayed striking collagen deposition in the lung parenchyma, thickening of alveolar septa due to increased collagen, and intraalveolar involvement. In contrast, PGDHi-treated mice showed fewer fibrotic lesions, which were interspersed with normal architecture, and less collagen accumulation (Fig. 3E). Pathological evaluation using a semiquantitiative morphological index of lung injury 28 , where 0 represents normal lung tissue and 5 represents high levels of inflammation and fibrosis throughout the surveyed lung lobes, demonstrated substantial lung injury in bleomycin-treated mice. PGDHi resulted in a trend towards lung injury reduction when compared to vehicle treatment (Fig. 3F). The anti-fibrotic impact of PGDHi was confirmed in mice 35 days post-intratracheal bleomycin treatment ( Supplementary Fig. 4A), wherein PGDHi significantly decreased lung hydroxyproline levels ( Supplementary Fig. 4B), and led to a trends towards preservation of lung morphology ( Supplementary Fig. 4C).

PGDHi attenuates pulmonary measurements of tissue stiffness in bleomycin-treated mice.
To determine whether PGDHi-mediated reductions in inflammation and histopathological features of fibrosis corresponded to functional improvements, we measured static lung compliance (Cst) by generating pressure-volume loops in anesthetized, tracheostomized mice before necropsy. PGDHi treatment attenuated the impact of bleomycin on baseline static compliance, inspiratory capacity, and cellular content of the bron- www.nature.com/scientificreports/ choalveolar lavage fluid, as compared to vehicle-treated counterparts (Fig. 4A). Additionally, upon methacholine challenge, PGDHi-treatment resulted in significantly increased forced oscillation technique measurements of dynamic respiratory system compliance (Crs) and decreased respiratory system elastance (Ers; Fig. 4B,C). The bleomycin-induced loss in airway responsiveness to methacholine as Newtonian airway resistance (Rn), a measure of airflow in central airways, was also significantly reversed by PGDHi such that responses did not differ from naïve mice (Fig. 4D). Lastly, the rigidity of peripheral alveoli and lung parenchyma, measured as tissue elastance (H), was significantly reduced with PGDHi as compared to vehicle treatment (Fig. 4E).

Endothelial cells, mast cells, and macrophages are PGDHi targets. Known cellular drivers of
PF pathogenesis include alveolar epithelial cells and fibroblasts 1 . To identify the cell types that PGDHi may act directly on during PF, we dissociated lung tissue from healthy mice and isolated cell populations based on surface marker expression. General hematopoietic and stromal separation on the basis of CD45 demonstrated enzyme activity in both CD45 positive and negative cells (data not shown). Stromal 15-PGDH activity likely comprises endothelial cells as the CD31+ population demonstrated moderate activity enrichment, consistent with a recent report 32 . To further delineate the identity of the 15-PGDH+ hematopoietic cells, we fractionated on the basis of CD117 to enrich mast cells, which participate in pulmonary wound healing 33 , and F4/80 to enrich alveolar macrophages, which are implicated in IPF pathogenesis 1 and have also recently been shown to express 15-PGDH 32 . Both CD117 + and F4/80 + preparations showed striking enzyme activity enrichment relative to negative counterpart cell fractions (Fig. 5A). Additionally, 15-PGDH staining colocalized with CD31, Toluidine blue, and F4/80 staining in serial sections (Fig. 5B). These data, together with the work of Bärnthaler et al. 32     www.nature.com/scientificreports/ vessels (Fig. 6), consistent with previous reports 32, 34 . The existence of PGDH+ cells in human lung tissue supports the notion that 15-PGDH inhibition may be a novel and effective therapy for IPF patients.

Discussion
In this study, we identified the presence of abundant 15-PGDH+ cells in murine pulmonary tissue. Compared to bone marrow and colon; tissues in which we have previously demonstrated enhanced regeneration and wound healing via pharmacologic 15-PGDH inhibition, lung tissue shows greater expression and enzyme activity. As PGE2 is well known to antagonize TGFβ-mediated fibrotic signaling, we tested the capacity for 15-PGDH inhibition to protect against PF pathology in mice exposed to bleomycin. Our data show that PGDHi strikingly attenuates early inflammation at day 7, which correlated with reduced weight loss and mortality. We find that continuous PGDHi treatment reduces the occurrence of fibrotic lesions and interstitial remodeling in the lung at day 35. Moreover, PGDHi attenuated lung dysfunction, as measured by changes in compliance, resistance, and tissue rigidity. Our data identify pulmonary endothelial cells, macrophages, and mast cells as the direct cellular targets of PGDHi, thus confirming and extending findings published in a recent report 32 . Lastly, the presence of PGDH+ cells in human pulmonary tissue identifies 15-PGDH as a potential novel therapeutic target in human IPF. These data extend previous work demonstrating the capacity for PGE2 or specific agonism of EP2/EP4 receptors to attenuate TGFβ-mediated pro-fibrotic signaling in pulmonary fibroblasts 19,[23][24][25] , and provide a welltolerated and efficacious alternative to the delivery of long-acting prostaglandin analogs 18 . Our dual effect on both acute inflammation and chronic fibrosis, as well as our finding that pulmonary endothelial cells, mast cells, and macrophages are highly 15-PGDH active, suggest multiple discrete mechanisms contribute to PGDHi-mediated protection. These results also indicate that systemic 15-PGDH inhibition may exert a more potent protective effect than targeted PGE2 delivery to endothelial cells 21 or alveolar macrophages 22 during IPF. Early in disease, PGDHi may improve alveolar epithelial wound repair, in a manner similar to that which has been demonstrated in the colonic epithelium 26 . Indeed, the concept that PGDHi may protect in part by improving alveolar stem and progenitor cell function is consistent with recent reports that senescent epithelial cells and fibroblasts drive IPF pathology 35 . Later in disease, and of particular clinical relevance, PGDHi reduces ECM protein expression and fibrotic remodeling in the lung, consistent with the impact of PGE2 administration or specific EP2/EP4 agonism, on human pulmonary fibroblasts from IPF patients 19 . Notably, PGDHi was associated with fewer fibrotic foci, preservation of alveolar architecture, and striking improvements to the dynamic compliance, resistance, and parenchymal rigidity of lungs upon forced oscillation analysis.
During the preparation of this manuscript, corroborating data was published by Bärnthaler et al. 32 . Both studies agree in the ability of PGDHi to increase local PGE2 levels, to prevent the accumulation of collagen, and to improve lung function in murine PF. Our work additionally demonstrates robust PGDH enzymatic activity in pulmonary tissue as compared to additional organs in which PGDHi has therapeutic efficacy 26 . Moreover, we demonstrate PGDHi-mediated amelioration of acute inflammatory pathology, including local diminution of Il1b and Il6, and systemic reductions in CXCL1, and TNFα, which likely contribute to the subsequent reductions in fibrosis and improvements in pulmonary function. These results are consistent with the known anti-inflammatory actions of PGE2 36 . Importantly, Bärnthaler et al. demonstrate that PGDHi initiated during fibrotic stage disease still confers protection 32 . Future studies to test the durability of the therapeutic effect after discontinuing PGDHi treatment are warranted.
PGDHi provides a novel and well-tolerated therapeutic approach to IPF, which is increasing in prevalence worldwide, particularly among aging populations. In the United States prevalence is estimated at 4.0 per 100,000 persons aged 18-34, and 227.2 per 100,000 persons among those > 75 years 37 . A history of cigarette smoking is the strongest risk factor, with additional risks including exposure to stone, wood, and organic dusts, and gastroesophageal reflux, which may contribute to lung injury via microaspiration 38,39 . While recent work has yielded greater insight into the pathogenesis of IPF and related progressive fibrosis that occurs in interstitial pneumonias of unknown cause, treatment options remain only partially effective. FDA approved therapies include Nintedanib, an inhibitor of nonreceptor and receptor tyrosine kinases, and Pirfenidone a small molecule of unknown mechanism. Neither are curative but demonstrate an ability to slow disease progression and can be combined with palliative care approaches to improve quality of life 38,39 . Patients deemed critical may be eligible for lung transplantation, which provides a 40-50% 5-year survival rate 40 . With a clear need for improved therapies for IPF and other interstitial lung diseases, our work demonstrates PGDHi holds significant promise either as a standalone therapy to prevent myofibroblast differentiation and reduce collagen deposition, or as a combination strategy with the currently approved therapies and palliative care.
Of interest, this work has additional therapeutic implications for disease states also characterized by fibrotic lesions. Primary myelofibrosis (PMF) is a rare bone marrow disorder characterized by abnormal blood cell production, extensive bone marrow scarring, activation of extramedullary hematopoiesis to the spleen, and a high incidence of leukemic transformation 41 . Median overall survival is only six years using current standards of care ruxolitinib, hydroxyurea, or bone marrow transplantation [42][43][44] . PGDHi may hold significant promise for PMF therapy, as inhibition of the TGFβ signaling axis has demonstrated efficacy in preclinical PMF models 45,46 . TGFβ transformation of fibroblasts plays a significant role in additional fibrotic diseases including hepatic, skin, and post-radiation induced fibrosis of multiple organs 47 , as well. We propose that PGDHi-mediated disruption of myofibroblast differentiation will have therapeutic benefit in a number of these models and provide novel treatment strategies for these poor prognosis conditions.

Methods
Reagents. Bleomycin sulfate was purchased from Cayman Chemical and was dissolved in 0.9% sodium chloride to a concentration of 16 mg/mL. 80 mg bleomycin per kg mouse body weight was administered as a single dose at day 0 by retro-orbital injection under isoflurane anesthesia. For intratracheal instillation, 2 mg bleomycin per kg mouse body weight was administered to anesthetized, intubated mice by slowly instilling into Semiquantitative morphological evaluation. The degree of pulmonary inflammation and fibrosis in day 35 lung samples was semiquantitatively evaluated on Masson's Trichrome-stained sections of the whole lung, in toto. Briefly, we employed a previously reported semiquantitative morphological scoring system 28  Real time pcR. On Day 0 and 7 post bleomycin induction, animals were harvested via CO2 inhalation followed by cervical dislocation and whole lungs excised and homogenized in RLT buffer. For comparison of bone marrow, colon, and lung, hindlimb marrow was flushed and the distal colon was dissected, prior to homogenization in RLT buffer. RNA isolation was performed using the RNAEasy Kit (Qiagen) and real-time PCR performed using commercial primers purchased from Applied Biosystems: IL2 (Mm00434256_m1), IL1B(Mm00434228_ m1), IL6 (Mm00446190_m1), TNF (Mm00443258_m1), CXCL13 (Mm04214185_s1). Values were tabulated graphically with error bars corresponding to standard error of the means and compared using 2-tailed t-tests.
Multiplex ELISA. Peripheral blood from mice 7 days post-bleomycin exposure was collected into Microtainer serum-separator tubes (Becton-Dickinson) by submandibular cheek puncture. Whole blood was allowed to clot at room temperature and then spun at 6,000×g for 3 min to separate serum. Serum was removed and stored at -80 prior to analyzing with the V-PLEX ProInflammatory Panel 1 Mouse Kit (Meso Scale Diagnostics).
Lung mechanics. Mice were weighed and anesthetized with an intraperitoneal injection of ketamine/xylazine and placed in a supine position. Once mice were nonresponsive to toe pinch, midline tracheostomy was performed to insert a blunt tip cannula. Mice were then paralyzed with pancuronium bromide and mechanically ventilated by the flexiVent system (SCIREQ, FX2 module) with continuous ECG monitoring. The maximum baseline static compliance was measured using computer software slope-fit of a quasi-static pressure-volume loop (Salazar-Knowles equation). Dynamic compliance (Crs), elastance (Ers), Newtonian airway resistance (Rn), and tissue elastance (H) were calculated by forced oscillation technique (Snapshot-150 and Quick Prime-3). Following two recruitment breaths (30 cm H 2 O for 3 s), increasing doses of methacholine (0-200 mg/mL; Sigma-Aldrich) were administered over 10 s via an in-line ultrasonic nebulizer (AeroNeb, SCIREQ). After each methacholine dose, 6 dynamic measurements were made and the mean values were analyzed. Subsequently, the bronchoalveolar lavage fluid was collected by instilling 0.5 mL of PBS into the lung with a 10 s dwell time.
Cell separation. Briefly, lungs were minced and enzymatically-digested using previously described methods 48 to generate a single cell suspension. Cells were then isolated by surface marker expression using Miltenyi microbead kits and LS column separation. 15-PGDH enzymatic activity was measured in lysed cell fractions, or in homogenized whole organs, as previously reported 26 . Activity was then normalized to input pro-