First, we performed reverse transcription-quantitative polymerase chain reaction (RT-qPCR) to examine the mRNA expression levels of mPGES-1 (encoded by PTGES) and PGE₂ receptors (EP1–4, encoded by PTGER1–4) in the nasal polyps of ECRS or non-ECRS patients. Patients with ECRS had significantly higher JESREC scores and peripheral blood eosinophil percentages (%) than patients with non-ECRS ( Figure 1A ). The nasal polyps isolated from patients with ECRS had significantly lower PTGES expression than those isolated from patients with non-ECRS, whereas the expression of EP receptors did not differ significantly between ECRS and non-ECRS patients ( Figure 1B ). These results raised the possibility that decreased PGE₂ production would be involved in increased eosinophil infiltration into the nasal polyps of ECRS patients.

To address the role of PGE₂ in eosinophil infiltration into the nasal tissue, we took advantages of Ptges ^–/– mice. Following intranasal administration of Aspergillus oryzae protease (AP) and ovalbumin (OVA) to littermate Ptges ^+/+ (WT) and Ptges ^–/– (KO) mice 3 times/week for 6 weeks, eosinophil accumulation was observed beneath the mucosal epithelium in both groups, which was more prominent in KO mice than in WT mice ( Figure 2A ). Neither epithelial tissue inflammation nor submucosal eosinophil infiltration was evident in phosphate-buffered saline (PBS)-treated control groups. Strikingly, epithelial hypertrophy, with massive eosinophil infiltration under the mucosa, were observed in the respiratory epithelium ( Figure 2B ) and maxillary sinus ( Figure 2C ) of AP+OVA-challenged Ptges KO, but not WT, mice. Periodic acid-Schiff (PAS) staining of the nasal cavity revealed that mucus production was greatly increased in KO mice compared to WT mice ( Figure 2D ). The nasal respiratory epithelium of KO mice had approximately twice as many sub-mucosal eosinophils as did WT mice in regions I (septum) and III (maxillary concha), and to a lesser extent in region II (dorsal concha) ( Figure 2E ), with the mucous layer in region I of KO mice being significantly thicker than that of WT mice ( Figure 2F ). Thus, Ptges deficiency promotes eosinophil recruitment into the nasal mucosa with formation of epithelial hypertrophy, which had been barely or only mildly observed in previously reported eosinophilic sinusitis models using WT C57BL/6 or BALB/c mice ( Table 1 ) (4–7). Because of the severe eosinophilic inflammation with epithelial hyperplasia that extended from the respiratory epithelium (septum and maxillary concha) to the maxillary sinus, we regarded this condition as an ECRS-like model.

We next examined the expression levels of various cytokines and chemokines in this model by qPCR. Among the proinflammatory cytokines Il1b, Il6 and Tnf ( Figure 3A ), the Th2 cytokines Il4, Il5, and Il13 ( Figure 3B ), the type-2 epithelial cytokines Il25, Il33 and Tslp ( Figure 3C ), and the eosinophil-attracting chemokines Ccl11 and Ccl24 ( Figure 3D ), the expression levels of Il1b, Il4, Il5, Il13, Il25, Tslp and Ccl11 were significantly higher in Ptges KO mice than in WT mice at 2 weeks after AP+OVA challenge. Thus, the increased eosinophil accumulation in nasal mucosal tissues of KO mice may rely on the increased expression of these cytokines and chemokines involved in nasal mucosal inflammation and type-2 immunity at an early stage (2 weeks after AP+OVA challenge) of the sinusitis model.

Based on previous studies demonstrating that cys-LTs were overproduced in aspirin-induced asthma and that PGD₂, thromboxane (TX) A₂, LTB₄, and cys-LTs promoted, while PGE₂ inhibited, pulmonary eosinophilic inflammation (17), we analyzed the expression of eicosanoid-biosynthetic enzymes and receptors in the AP+OVA-induced ECRS-like model. Among the phospholipase Pla2g4a (cPLA₂α) ( Figure 4A ), the cyclooxygenases Ptgs1 (COX-1) and Ptgs2 (COX-2) ( Figure 4B ), the PGE₂ receptors Ptger1–4 (EP1–4) ( Figure 4C ), and the LT-biosynthetic enzymes and receptors Alox5 (5-LOX), Alox15 (15-LOX), Lta4h (LTA₄ hydrolase = LTB₄ synthase), Ltb4r1 (BLT₁, an LTB₄ receptor) and Ltc4s (LTC₄ synthase) ( Figure 4D ), the expression levels of Ptgs1, Ptgs2, Alox5, Alox15, Lta4h and Ltc4s were significantly higher in Ptges KO mice than in WT mice at 2 weeks after AP+OVA challenge. Ptges was constantly expressed in WT mice, but not in KO mice as expected, over 6 weeks ( Figure 4B ). Thus, corroborating the changes in the expression of several cytokines and chemokines ( Figure 3 ), that of several eicosanoid-biosynthetic enzymes and receptors was elevated at an early stage (2 weeks after AP+OVA challenge) of the sinusitis model in Ptges KO mice.

Next, we performed lipidomics analysis of the nasal mucosal tissue at 0 and 2 weeks after AP+OVA administration. Consistent with the increased expression of both COX isoforms, PGE₂ was increased approximately three times at 2 weeks after AP+OVA treatment in WT mice, whereas it was barely detected in Ptges KO mice as expected ( Figures 4E, F ). Furthermore, other COX metabolites such as PGD₂ and TXB₂ (a stable end product of TXA₂), as well as LOX metabolites such as LTC₄, LTD₄ and 12-HETE, were increased more markedly in KO mice than in WT mice at 2 weeks ( Figures 4E, F ), suggesting the shunting of AA toward other prostanoids and LOX metabolites in the absence of mPGES-1-driven PGE₂ synthesis. Thus, in the context of lipid mediator signaling, the increased eosinophil infiltration in Ptges KO mice at 2 weeks after AP+OVA administration may rely, at least in part, on the increased production of pro-allergic COX (e.g., PGD₂ and TXA₂) and LOX (e.g., LTB₄ and cys-LTs) metabolites in association with the decreased production of anti-allergic PGE₂. In addition, the decrease of lipoxins (LXA₄ and LXB₄; AA-derived specialized pro-resolving mediators (18)) at 2 weeks in KO mice relative to WT mice ( Figures 4E, F ) might also contribute to the exacerbation of nasal inflammation caused by Ptges deficiency. Although various EPA ( Supplementary Figures 2A, B ) and DHA ( Supplementary Figures 2C, D ) metabolites, including the pro-resolving mediators resolvins and protectin (18), were also elevated at 2 weeks after AP+OVA challenge, their levels did not significantly differ between KO and WT mice. Overall, these findings indicate that the increased production of pro-allergic cytokines, chemokines, and lipid mediators occurred during an early phase of the sinusitis pathology in Ptges KO mice.

Given that the aspirin-induced, asthma-like lung response in Ptges KO mice is alleviated by treatment with AE1–259-01, an EP2 agonist (12), it is possible that activation of PGE₂–EP2 signaling may also improve excessive eosinophilic inflammation in the nasal sinuses. Therefore, we investigated whether several EP agonists could prevent the ECRS-like pathology observed in Ptges KO mice. To this end, AP+OVA, along with either a PGE₂ analog (16,16-dimethyl-PGE₂; dm-PGE₂), an EP2 agonist (butaprost), an EP1/3 agonist (sulprostone), or an EP4 agonist (CAY10598), were administered intranasally into KO and WT mice 3 times/week for 6 weeks, and eosinophil infiltration into the nasal mucosa was evaluated at 6 weeks ( Figure 5A ). Significantly fewer eosinophils were found in the nasal mucosal regions I, II, and III of the groups treated with dm-PGE₂, butaprost, or CAY 10598 than in those of the group without treatment in KO mice and even WT mice ( Figure 5B ). In contrast, sulprostone failed to suppress eosinophil infiltration. These findings suggest that the ECRS-like pathology observed in Ptges KO mice is alleviated by intranasal treatment with agonists for EP2 or EP4, both of which are coupled with Gs-dependent cAMP signaling (19).

EYBELIS ophthalmic solution with 0.002% omidenepag isopropyl (OI), an eyedrop drug used to treat glaucoma, is a clinically approved EP2 agonist in humans (20). Therefore, we next investigated whether OI treatment could prevent the ECRS-like pathology caused by AP+OVA challenge in Ptges KO mice. OI was intranasally administered into Ptges KO and WT mice along with AP+OVA 3 times/week for 6 weeks. Thereafter, we analyzed eosinophil infiltration into the nasal mucosa ( Figure 6A ). Our results showed that the group treated with AP+OVA in the presence of OI had significantly fewer eosinophils in mucosal regions I–III than did the group treated with AP+OVA alone ( Figure 6B ). Moreover, in KO mice, the increased expression of Il4, Il13, Il25, and Tslp at 2 weeks after treatment with AP+OVA was markedly suppressed by OI ( Figure 6C ). Thus, drug repositioning of OI to human ECRS is expected to lower the risk of undesirable side-effects and significantly shorten the time required for new drug development.

Among patients who received surgery at the Department of Otolaryngology and Head and Neck Surgery at the University of Tokyo Hospital from November 1, 2020 to December 31, 2022, we chose five patients, with ECRS who relapsed within 6 months of surgery, and five patients without ECRS, who showed improvement after surgical treatment. Patients with a JESREC score (34) of ≥11 and a nasal polyp tissue eosinophil count (400× magnification) of ≥70 were diagnosed as ECRS, and the other patients were diagnosed as non-ECRS. Analyses were performed on surgically excised nasal polyps from nasal sinuses of the patients. Oral corticosteroids were given to all ECRS patients for 1 week before surgery. Before collecting any samples, each patient provided a written informed consent. This study was authorized by the Ethics Review Board of the University of Tokyo Hospital (2020214NI).

Ptges ^–/– mice on the C57BL/6 background were described previously (35). C57BL/6 mice were purchased from Japan SLC. Animals were housed in a specific pathogen-free room with a humidity of 50 ± 10%, a temperature of 23 ± 1°C, and a 12-h light/dark cycle (light: 8:00 to 20:00, dark: 20:00 to 8:00). The mice were provided ad libitum access to CE-2 feed (Japan SLC) and ultrafiltered water. All animal experiments were conducted in accordance with the University of Tokyo Animal Experiment Regulations (P17–032).

Human ethmoid sinus, which is prone to tissue changes due to ECRS, corresponds to a part of the nasal turbinate and septum in mice. Therefore, previous reports have analyzed the respiratory epithelium of the nasal turbinates and septum as a model of sinusitis (4–7). We modified a previously described mouse model of eosinophilic sinusitis (4) by intranasally administering AP+OVA to the mice. In a protocol of 2-week treatment, 9-week-old male mice were treated with nasal spray containing 2 units of AP (Sigma-Aldrich) and 75 μg of grade-V OVA (Sigma-Aldrich) suspended in 20 μl of PBS (–) three times/week for 2 weeks. In a protocol of 6-week treatment, the same doses of AP and OVA in PBS (–) were intranasally administered to 5-week-old mice three times/week for 6 weeks. Thereafter, 0.3 mg/kg of the dm-PGE₂ (Cayman Chemical), butaprost (Cayman Chemical), sulprostone (Abcam), or CAY 10598 (Cayman Chemical) was mixed with the AP+OVA suspension and administered intranasally as above. As required for experiments, mice were intranasally challenged with AP+ OVA in the presence of OI in EYBELIS ophthalmic solution 0.002% (Ube Industries and Santen Pharmaceutical).

After deep anesthesia, right atrial appendages of mice were incised. Thereafter, 10 ml of saline and then 10 ml of 10% neutral-buffered formalin solution (Fujifilm-Wako) were administered into the left ventricle for perfusion fixation. The heads of the mice were removed, immersed in 10% neutral-buffered formalin solution, and left at 24°C for 24 h. Thereafter, the solution was replaced with 10% EDTA-2Na solution (pH 7.0; Muto Pure Chemicals) and shaken at 24°C at 100 rpm. Finally, a neutral demineralization treatment was performed for 14 days. After decalcification, histology of the respiratory epithelium of the nasal cavity was analyzed after sectioning the head coronally, which allowed the horizontal parts of the incisor roots to be observed. Each tissue sample was placed in a Unicassette (Sakura), immersed in 70% ethanol (EtOH) for dehydration purposes, and placed in a Tissue-Tek VIP 5 Jr. tissue processor (Sakura). Thereafter, the tissue samples were immersed in paraffin, heated to 60°C using a Tissue-Tek TEC (Sakura), cooled to 4°C, and embedded. The embedded tissue was placed in a litratorome REM-710 (Yamato) and sliced to a thickness of 4 μm. To prepare the tissue sections, they were immersed in a 45°C hot bath, allowed to adhere to the slides, and then dried overnight at 45°C in a Slide Warmer (Sakura).

Eosinophils were stained with Sirius Red as described (36). Each sample preparation was transferred to a staining vat. Deparaffinization was performed twice with Xylene for 5 min, 99% EtOH for 2 min, 95% EtOH for 2 min, and 70% EtOH for 2 min. Each preparation was removed from the basket and the area around the tissue section was blocked with a water-repellent pen (Dako). Thereafter, 100 μl of Sirius Red staining solution (Muto pure chemicals) was added dropwise, and the mixture was incubated at 24°C for 1 h in a humid chamber. Subsequently, each specimen was washed with running water for 5 min, with hematoxylin for 2 min, and again with running water for 5 min. The samples were dehydrated and cleared by incubating them with 70% EtOH for 1 min, 95% EtOH for 1 min, 99% EtOH for 1 min, and xylene for 1 min twice, after which they were mounted with soft mount (Fujifilm-Wako). Tissue photos were taken using an all-in-one fluorescence microscope (BZ-X710, Keyence). Among the immune cells that infiltrated into the nasal mucosa (200-μm wide) at three locations, i.e., I: nasal septum, II: dorsal concha, and III: maxillary concha, eosinophils stained with Sirius Red were counted.

Reagents and chemicals required for quantitative RT-qPCR were purchased from Thermo Fisher Scientific. Mice were euthanized by cervical dislocation and then decapitated, the head was sectioned sagittally from the midline, and the mucous membrane of the respiratory epithelial region of the nasal cavity was collected. TRIzol reagent (Thermo Fisher Scientific, 500 μl) was then added to the samples and homogenized with a bead homogenizer (Precellys, Bertin Instruments). Thereafter, total RNA was extracted and reverse-transcribed to cDNA using a High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). qPCR was conducted with a StepOnePlus Real-Time PCR System (Thermo Fisher Scientific) using TaqMan Gene Expression Master Mix and pre-designed primer probes (TaqMan gene Expression Assay) listed in Table 2 , with Gapdh (mouse) or GAPDH (human) as an internal control to normalize the expression levels of individual genes.

Sample extraction methods using solid-phase extraction have been previously described (37, 38). Mouse mucous membrane of the respiratory epithelial region of the nasal cavity was collected as described above. The sample was immediately frozen at -80°C, pulverized with a multi-beads shocker (YASUI KIKAI). 500 μl of methanol (MeOH) was added to the samples, sonicated for 10 min, and left at -30°C overnight. The samples were then centrifuged at 15000 x g for 5 min at 4°C, and the supernatant was collected. A 3 cc Oasis HLB cartridge (Waters) was used to extract polyunsaturated fatty acid metabolites. After adding 4,500 μl ultrapure water to the supernatant, the pH was adjusted to 3 with HCl and passed through the equilibrated columns. Thereafter, the column was washed with ultrapure water and hexane, and eluted with methyl formate. Finally, it was dried under nitrogen gas, dissolved in 50 μl MeOH, and used for analysis. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) based lipidomics was performed on an Exion LC™ Series UHPLC coupled with a Quadrupole linear ion trap hybrid mass spectrometer (QTRAP 6500⁺) System (AB Sciex).

The extracted lipids applied to a C18 column (2.1 mm i.d. x 150 mm length, 1.7 µm particle, Phenomenex, Inc.) were separated by gradient elution with mobile phase A (water containing 0.1% acetic acid) and mobile phase B (acetonitrile:methanol, 4:1, v/v) at flow rate 0.2 mL/min at 45°C. The MS/MS analysis was performed in negative mode, and the lipids were identified by multiple reaction monitoring (MRM) transition and retention times and quantified based on the peak area of the MRM transition. The calibration curve was obtained with an authentic standard for each compound. d4-labeled PGE₂ (Cayman Chemicals) and d5-labeled EPA (Cayman Chemicals) were added to each sample as internal standards.

Data are expressed as mean ± standard error of the mean (SEM), and differences between two groups were determined by Mann-Whitney U test. Differences between three or more groups were determined by two-way analysis of variance (ANOVA) with Tukey’s multiple-comparison test. Statistically significant differences were defined as follows: *p < 0.05, **p < 0.01, ***p < 0.001, or ****p < 0.0001. GraphPad Prism 9 (GraphPad) was used for statistical analysis.

Schematic diagrams were created for this study using BioRender (BioRender.com).