Adult female C57BL/6 mice (8–14 weeks old; Orientbio, Seongnam, Republic of Korea) were maintained at 22 ± 2°C and 50 ± 10% relative humidity on a 12-h light/dark cycle (8 a.m. – 8 p.m.), fed, and provided water ad libitum under specific pathogen-free conditions in a facility accredited by AAALAC International (# 001071). All experiments conducted in this study were approved by the Institutional Animal Care and Use Committee (approval number # 2023–0088) of Yonsei University College of Medicine. Mice were randomly divided into each group as follows: 1) The Sham group was a group with only incision; 2) the DEX group was administered dexamethasone after the deligation; 3) the Ligation group was not deligated after ligation; 4) the Deligation+PBS group was deligated two weeks after ligation and administered with PBS; and 5) the Deligation+DEX group was administered with dexamethasone after the clips that tied the duct for two weeks were removed. To confirm the side effects of dexamethasone, mice were divided into four groups: 1) DEX group administered with dexamethasone only; 2) Deligation+PBS group administered PBS after deligation; 3) Deligation+DEX group administered dexamethasone after deligation; 4) Deligation+DEX^Hi group infused with a 10-fold higher concentration of dexamethasone to assess high-dose side effects. To investigate the long-term effect of dexamethasone, the Deligation+DEX^L-T group received six administrations of the normal concentration and harvested at 16w. To serve as a long-term control, 16-week normal mice were harvested and analyzed together.

For duct ligation, mice were anesthetized via intraperitoneal injection of ketamine (50 mg/kg, Virbac) and rompun (5 mg/kg, Bayer HealthCare, Mississauga, Ontario, Canada). After fixing the mice with medical tape, povidone was spread into the front of the neck. A 1–2 cm incision was made to expose the submandibular glands (SMGs). The left excretory duct of SMG and the belonging nerve and vessels were clipped with a titanium hemostatic clip 9 (3.6 mm, # J9180, VITALITEC, Bargheim, Germany). The incision was closed using Reflex Wound Clips (7 mm, # 203–1000, CellPoint Scientific, Gaithersburg, MD, USA). After 2 weeks, the excretory duct was deligated by removing the clip, except in the Ligation group. Following an in vivo methodology (16), either dexamethasone (# D8893–1MG, Sigma-Aldrich, St. Louis, MO, USA) for the Deligation+DEX group or PBS for the Deligation+PBS group was retroductally administered through a cannula implanted in the submandibular duct orifice at 1 μg/20 μL for 10 μL/min. There were three control groups: Sham (incision only), DEX (no ligation and dexamethasone administration), and Ligation (ligation only).

To confirm the side effects of high-dose or long-term dexamethasone use, the mice were further divided into the Deligation+DEX^Hi and Deligation+DEX^L-T groups. The submandibular duct was ligated and deligated in the same manner. For the DEX^Hi group, 10 μg/20 μL dexamethasone was administered. Except for the DEX^L-T group, all mice were euthanized 2 weeks after deligation and dexamethasone administration. The DEX^L-T group received 1 μg/20 μL of dexamethasone once a week for 6 weeks and were euthanized 1 week after the final administration. As for the control group of the DEX^L-T group, normal SGs from 16-week-old mice were harvested at the same time period as the DEX^L-T group.

The mice were euthanized using a CO² chamber, and the left SMG was excised, weighed, and compared between the groups. Prior to euthanasia, mice were intraperitoneally injected with pilocarpine (5 mg/kg, # P6503; Sigma-Aldrich) to induce salivation. The lag time for salivation was measured after pilocarpine administration. 5 min after stimulation, saliva was extracted from the floor of the mouth for 5 min using a micropipette. The collected saliva was placed in 1.7 mL microtubes, and the volume was measured. Lag time was defined as the time of the first salivation following pilocarpine stimulation.

The fixed SMG tissues were embedded in paraffin, sectioned, deparaffinized, and hydrated. For histological analysis, the sections were stained with hematoxylin and eosin (H&E) (# ab2455880, Abcam, Cambridge, UK). Periodic acid-Schiff (PAS, # ab150680, Abcam) was utilized to detect mucin secretion, while Masson’s trichrome (MTC, # ab150686, Abcam) was used to stain the collagen fibers. All staining procedures were conducted following the manufacturer’s instructions. A blinded examiner evaluated pathological changes, including inflammation (H&E), mucin production (PAS), and fibrosis (MTC). The damage score of the SMG tissues was recorded as 0–5 based on the following criteria: intact acini in image fields were scored as 0 for 90% or more, 1 for 70%–90%, 2 for 50%–70%, 3 for 30%–50%, 4 for 10%–30%, and 5 for 10% or less. The mucin area was magenta on PAS, and the fibrotic area was blue on MTC. The data were analyzed using ImageJ software (NIH, Bethesda, MD, USA).

Following the manufacturer’s protocols, total RNA was extracted using TRIZOL (# 15596018, Thermo Fisher) and reverse transcribed into cDNA using the RT Master MIX (# RRO36A, Takara, Kusatsu, Japan). Gene-specific PCR products were quantified using a SYBR reporter and measured with a QuantStudio 5 real-time PCR system (Thermo Fisher). The ΔΔCt method was used to measure gene expression after normalization to housekeeping Gapdh expression. Primer information is presented in Supplementary Table 1 .

For Immunohistochemistry, SMG tissues were cut into 5-μm-thick paraffin blocks and soaked in xylene and a series of alcohols (from 100% to 70%). The sections were subjected to antigen retrieval for 40 min in Tris-EDTA (pH 9.0). Blocking, secondary antibodies, and an avidin/biotin-based peroxidase system were utilized with the ABC-HRP Kit (# PK-6200, VECTASTAIN, Newark, CA, USA). Primary antibodies against TNF (# NBP1–19532, 1:200; NOVUS, Centennial, CO, USA), IL-1β (# NB600–633,1:200; NOVUS), F4/80 (# A18637, 1:100; Abclonal, Woburn, MA, USA), and Ly-6G (# A20861, 1:50; Abclonal) were applied. DAB Quanto (# 12623957, Epredia) was employed to expedite the peroxidase system and its reaction product turned brown.

For immunofluorescence (IF), sections were blocked with 5% normal serum for 1 h at 20–25°C and then incubated with primary antibodies overnight at 4°C. The next day, the sections were incubated with secondary antibodies for 1 h at RT. The slides were counterstained with DAPI and then mounted using a ProLong glass antifade mountant (# P36985, Invitrogen, Carlsbad, CA, USA). Detailed antibody information is presented in Supplementary Table 2 .

Six to eight images per group were randomly acquired using an Eclipse Ti–U2 inverted microscope (Nikon, Tokyo, Japan) and analyzed using NIS-Elements BR (Nikon). Data were quantified using ImageJ software. Images obtained using the same microscope were equally adjusted for thresholding, and the percentage of the total tissue area positive for a specific marker was calculated and normalized to nuclei area using ImageJ software.

Human parotid gland biopsies were acquired from individuals with benign SG tumors who provided informed consent and approval from the Institutional Review Board of Yonsei University Severance Hospital (permission number #2017–0226–001). The tissues obtained were used to generate human SG organoids as previously reported (15). Briefly, the tissues were initially sectioned with a blade and then subjected to enzymatic dissociation using collagenase type II (# 17101015, Thermo Fisher Scientific, Waltham, MA, USA) for 1 h, followed by a 10 min incubation with TrypLE Express (# 12604013, Thermo Fisher Scientific). Cells were passed through a 70-μm strainer and embedded in growth factor-reduced Matrigel (# 356231, Corning, Corning, NY, USA). The cells were then grown in GEM media containing Advanced DMEM/F12 (# 12634010, Thermo Fisher) supplemented with 10 mM HEPES (# 15630080, Thermo Fisher), 1 × GlutaMAX (# 35050061, Thermo Fisher), 0.2 μg/mL Primocin (# ant-pm, Invivogen, San Diego, CA, USA), 1× B-27 (# 17504044, Thermo Fisher), 1.25 mM N-acetyl-cysteine (# A9165, Sigma), 1% homemade RSPO1-CM, 100 ng/mL noggin (# 6057-NG, R&D, Minneapolis, MN, USA), 5 ng/mL NRG1 (# 100–03, Peprotech, Cranbury, NJ, USA), 5 ng/mL FGF2 (# 100–18B, Peprotech), 10 ng/mL FGF10 (# 100–26, Peprotech), 5 μM A83–01 (# 2939, Tocris, Abindon, UK), 10 mM niacinamide (# N0636, Sigma), 3 μM prostaglandin E2 (# 2296, Tocris), and 1 μM CHIR99021 (# 2520691, Biogems, Colinas, CA, USA). For the first 2–3 days, 10 μM Y-27632 (# 1254, Tocris) was added to the growth medium.

To investigate the effect of dexamethasone, the organoid groups were divided into four: untreated (None), dexamethasone-only administration (DEX), inflammatory substances (I.S) administration, and I.S. administration after dexamethasone (DEX+I.S). To induce inflammatory responses in the organoids, dexamethasone (1 μM) was added to the media on day 4, followed by agitation to ensure thorough mixing. After 2 h, 10 μg/mL LPS (# L3021, Sigma) and 10 ng/mL TNF (# 300–01A, Peprotech) were added and mixed well, and each group of organoids was harvested 72 h later. To assess the potential side effects of high concentration or prolonged treatment, the high-concentration group (DEX^Hi+I.S) received 10 μM dexamethasone, representing a concentration 10 times higher than the conventional dexamethasone dosage (1 μM). The long-term treatment group (DEX^L-T+I.S) was harvested on day 12 and compared to DEX+I.S, which was removed after 3 days, and DEX^L-T+I.S, which received 1 μM dexamethasone continuously until day 12.

Mouse BMDMs were extracted from the tibia and femur of 8-week-old female C57BL/6 mice. Bone marrow cells were collected and cultured on a suspension plate in DMEM (# L0103–500, Biowest, France) supplemented with 10% FBS (# 16000044, Thermo Fisher), 1% P/S (# 15140122, Thermo Fisher), and 20 ng/mL murine M-CSF (# 315–02, Peprotech). The medium was replaced every 2–3 days.

Media from each of the four SG organoid groups (None, DEX, I.S, and DEX+I.S), as well as the inflamed organoids, were harvested and stored at -20°C. We labeled the media in the None, DEX, I.S, and DEX+I.S group media as CM1, CM2, CM3, and CM4, respectively. The group that did not treat CM was labeled Ctrl. Seven days after starting the mouse BMDM culture, media mixed with BMDM culture media and organoid CM at a 1:1 ratio were added. For the polarization to M1 macrophages, LPS and IFNγ (# 575302, Biolegend) were added to the media, while IL4 (# 214–14, Peprotech) and IL13 (# 210–13, Peprotech) were employed to polarize M2 macrophages. The mouse BMDMs were harvested after 24 hours of stimulation for RNA extraction.

All data were analyzed using GraphPad Prism (GraphPad Software ver. 10, San Diego, CA, USA). We used the Shapiro-Wilk test to check the normality of the data. If the data displayed a normal distribution, we employed t-tests for two groups and one-way ANOVA for more than three groups. Tukey’s post hoc test was conducted to compare values and determine statistical significance. In some data that did not demonstrate a normal distribution, the Kruskal-Wallis H test, followed by Dunn’s test, was performed for multiple group comparisons. We considered * p < 0.05, ** p < 0.01, and *** p < 0.001 as the significance levels. Further details are provided in the figure legends.

First, we investigated the therapeutic effects of dexamethasone on obstructive sialadenitis using in vivo duct ligation and deligation models. To induce severe obstruction, we ligated the nerves and vessels along the submandibular duct based on our previous protocol (15). Duct ligation was performed at week 0, deligation was conducted 2 weeks after ligation, and SMGs were extracted 2 weeks after retroductal administration of either dexamethasone or PBS following deligation ( Figure 1A ). Ligation damage induced SG atrophy in the Ligation group; however, the SGs in the Deligation+DEX group returned to a size similar to that in the Sham group ( Figure 1B ). The body weight slowly increased in the duct-ligated groups (Ligation, Deligation+PBS, and +DEX) and tended to recover after deligation. Only the Deligation+DEX group showed a more significant recovery in body weight than the Ligation group ( Figure 1C ).

To determine whether the administration of dexamethasone facilitated the recovery of function after deligation, we measured SG weight, saliva volume, and lag time to salivation. The Ligation group had a significantly lower SG weight than the Sham group (23%). However, the Deligation+PBS group showed a recovery in gland weight of up to 47%, and the Deligation+DEX group showed a recovery of up to 70% compared with the Sham group ( Figure 1D ). Saliva volume was measured for 5 min, and only the Deligation+DEX group showed an increase in saliva volume compared to the Ligation group. No significant changes were observed in the Deligation+PBS group ( Figure 1E ). There was considerable lag time in the Ligation group, indicating a lack of salivation, which significantly recovered after deligation and dexamethasone treatment ( Figure 1F ).

To examine the structure of SG, H&E and PAS staining was performed. The duct-ligated groups showed dilated ducts, whereas deligation and dexamethasone administration reduced the diameter of the dilated ducts ( Figure 1G , black circles). PAS staining revealed a decrease in the Ligation group and an increase in mucin production in the Deligation+DEX groups ( Figure 1H , arrows). Moreover, the Deligation+DEX group also exhibited the most substantial recovery from SG damage, as evidenced by the morphological improvement in the acinar and ductal structures ( Figures 1G, I ). The magenta area, as assessed by PAS staining, demonstrated an increased mucin production ratio in the Deligation+PBS group. However, dexamethasone treatment resulted in a more significant restoration of mucin production compared to the Deligation+PBS group ( Figures 1H, J ). Gene expression analysis revealed that Krt5 (basal cell marker) and Krt7 (luminal cell marker) were upregulated by ligation and downregulation by dexamethasone treatment. Conversely, Aqp5 (pro-acinar marker) and Bhlha15 (mature acinar marker) were downregulated following injury but restored by dexamethasone treatment. Similarly, Smgc (adult murine submandibular specific mucin marker) exhibited a pattern consistent with the acinar markers, while Bpifa2 (pro-acinar cell marker) followed a trend similar to the ductal markers ( Figure 1K ). These results suggest that administering dexamethasone retroductally following the relief of SG obstruction promotes both microscopic and functional recovery from SG damage.

Following ligation and deligation, we investigated alterations in SG epithelial and stromal cell markers, focusing on the SGs with dexamethasone treatment. We focused on the changes in salivary acinar, ductal, and myoepithelial cells. The count of KRT5⁺ basal cells and KRT7⁺ luminal cells elevated post-injury ( Figures 2A, D ). Dexamethasone administration significantly reduced the KRT7-positive areas, suggesting an improvement in luminal duct dilation. The number of AQP5⁺ pro-acinar cells and BHLHA15⁺ mature acinar cells was dramatically reduced by ligation injury, with barely any visible BHLHA15⁺ cells ( Figure 2B ). However, partial recovery occurred with PBS administration post-deligation, whereas dexamethasone treatment led to almost complete regeneration of AQP5-and BHLHA15⁺ cells ( Figures 2B, E ). KRT14⁺ basal progenitor cells decreased following injury but exhibited a greater increase with dexamethasone than in the Sham condition. However, ACTA2+ myoepithelial cells did not show any difference between groups ( Figures 2C, F ). Our findings suggest that duct ligation mainly affects mucin-secreting acinar and ductal cells rather than myoepithelial cells. Moreover, dexamethasone appears to facilitate the regeneration of salivary epithelial cells following duct ligation.

Furthermore, we observed that PECAM1⁺ endothelial cells were present around the duct and acinar in the Sham group. However, their expression was reduced following ligation-induced damage. Dexamethasone treatment resulted in an increased number of PECAM1⁺ cells along the duct ( Figures 2G, I ). In addition, dexamethasone administration significantly increased the number of TUBB3⁺ neurons ( Figures 2H, J ). Among the control groups, there was a slight increase in PECAM1⁺ and TUBB3⁺ areas in the DEX group, although the difference was not significant ( Figures 2I, J ). These results suggest that dexamethasone may stimulate SG stromal cells during the repair of damaged SG epithelial cells.

To analyze how dexamethasone ameliorates salivary inflammation and promotes recovery, we measured the expression of TNF and IL-1β using Immunohistochemistry. These pro-inflammatory cytokines increase under inflammatory conditions (17). The expression levels of TNF and IL-1β were significantly increased by salivary duct ligation ( Figures 3A, B, G, H ). We examined whether the recruitment of macrophages resulted from cytokine chemotaxis. F4/80⁺ macrophages exhibited a 13% increase in the Ligation group compared to the Sham group, while the Deligation+DEX group exhibited a significant reduction to 32% of the macrophage percentage, approaching normal levels ( Figures 3C, I ). However, no significant changes were observed in the LY6G⁺ neutrophil population. ( Figures 3D, J ).

We next explored the changes in macrophages that showed significant change ( Figures 3C, I ). Myofibroblasts, major contributors to fibrosis, become activated through autocrine and paracrine signaling from macrophages (18). Macrophages can trigger fibroblast activation and pro-inflammatory activity during tissue repair (19). MTC staining indicated a statistically significant reduction in the percentage of fibrotic cells with dexamethasone treatment ( Figures 3E , K ). As the increase in ACTA2 expression was not significant ( Figures 2C, F ), we conducted IF staining for PDGFRB, a marker of salivary stromal fibroblasts, to explore the involvement of other stromal cell subtypes in ligation-induced fibrosis. The results revealed a significant increase in the number of PDGFRB⁺ cells post-ligation, followed by a significant decline during recovery, consistent with the MTC findings ( Figures 3F, L ). This finding suggests that macrophages, recruited via chemoattraction during ductal ligation, and fibrosis, induced by ligation were relieved by dexamethasone.

Our in vivo experimental design aimed to assess the impact of dexamethasone over 2 weeks to ensure significant damage and subsequent recovery. Because of the short half-life of dexamethasone (36–54 h), we developed a 3-dimensional organoid culture system to efficiently investigate the underlying mechanism (20). After 3 days of growth, the organoids were treated with LPS (10 µg/mL) and TNF (10 ng/mL) to induce inflammation, while dexamethasone (1 µM) was administered to assess its anti-inflammatory effects. Treatment with inflammatory substances resulted in increased KRT5 and KRT7 expression and decreased expression of the acinar markers AQP5 ( Figures 4A, B ). However, ACTA2 did not exhibit inflammation-dependent changes ( Figure 4C ). In the inflamed organoid model, the mRNA levels of ductal, acinar, and myoepithelial cell markers in the DEX+I.S group fully recovered to the expression levels of the None group, closely matching the results of the in vivo experiments ( Figures 4A–C ). Additionally, we analyzed the cytokine genes to study the immunoregulatory mechanisms of SG epithelial cells. For cytokines, we used the inflammatory cytokines TNF and IL6 as markers. For chemokines, we used CCL2, which attracts monocytes/macrophages in inflammatory situations; CXCL5, which recruits leukocytes, especially neutrophils; and CXCL12, a chemokine secreted in the presence of inflammatory stimuli such as LPS, TNF, or IL1. The results demonstrated a significant increase in immune-related gene expression in epithelial cells due to inflammation, including elevated levels of TNF and IL6 ( Figure 4D ) and CCL2, CXCL5, and CXCL12 ( Figure 4E ), which were subsequently restored to normal levels with dexamethasone treatment. Furthermore, IF staining demonstrated that the inflammatory substances reduced the expression of AQP5 and increased KRT7 levels, while treatment with dexamethasone reversed these effects, similar to the None group ( Figures 4F, G ). Thus, KRT5 levels in IF were increased in response to inflammation and decreased by treating with dexamethasone, whereas ACTA2 showed no significant difference ( Figures 4H, I ). These results are in agreement with the in vivo IF results in Figures 2A–F , demonstrating that the investigations of dexamethasone through organoid inflammation modeling can be representative of in vivo biological changes.

To assess the impact of cytokines and chemokines secreted by epithelial cells on macrophages, we exposed macrophages to an organoid CM to evaluate macrophage polarization. M1 macrophage markers with pro-inflammatory roles include the cytokines Tnf, Il6, Il1b, and Cd86, which are involved in antigen presentation (21). Upon induction with M1 macrophages, there was a significant increase in M1 markers in the M1+CM3 group, which consisted of an organoid CM treated with inflammatory substances. After the induction of inflammation, the M1+CM4 group treated with dexamethasone showed a significant decrease in both markers ( Figure 4J ). For M2 macrophage markers, we used Ym1, which plays a role in macrophage activation; Il10, an immunosuppressive cytokine; Cd163, a scavenger receptor for monocytes/macrophages; and Klf4, which regulates M2 polarization (21–24). In CM2, the medium of organoids treated with dexamethasone, all M2 markers were upregulated. In the M2+CM3 group, most showed significantly lower expression compared to the M2+CM1 group. The M2+CM4 group also showed an increase in M2 marker levels, similar to those in the M2+CM2 group ( Figure 4K ). Inflammatory organoids accurately replicated epithelial conditions in vivo, thereby facilitating mechanistic studies. Furthermore, we demonstrated that cytokines and chemokines released by epithelial cells strongly regulated macrophage polarization, and that epithelial cells treated with dexamethasone downregulated M1 macrophages and upregulated M2 macrophages.

Next, we assessed the side effects of dexamethasone when used at high doses or in the long term. These experiments involved the addition of two new groups: DEX^hi and DEX^L-T. The high-dose group (DEX^Hi) received a 10-fold higher concentration (10 μg/20 μL) than the DEX group, and the long-term group (DEX^L-T) received dexamethasone once a week for a total of six doses, with each dose consisting of 1 μg/20 μL. Figure 5A shows the experimental design used to test the side effects of dexamethasone in vivo. High-dose administration of dexamethasone resulted in significantly lower gland weights ( Figure 5B ). Saliva volume was assessed to verify the effect of dexamethasone, and no significant differences were found in the Deligation+DEX^Hi ( Figure 5C ). The lag time tended to increase again in the Deligation+DEX^Hi group compared to the Deligation+DEX group, but statistical significance is not observed due to the variation in values ( Figure 5D ). Gene expression analysis revealed that the genes restored to the expression pattern of the DEX group by dexamethasone treatment deteriorated in the Deligation+DEX^Hi group. Specifically, Krt7 increased, and Aqp5 and Bhlha15 decreased. Krt5 and Acta2 showed no change ( Figure 5E ). In the Deligation+DEX^L-T group, there was no difference in SG weight compared to the 16w Normal group, but there was a significant reduction in saliva volume and a significant increment in lag time ( Supplementary Figures S1A–C ). Gene expression analysis of the Deligation+DEX^L-T group showed a significant increase in Krt7 and Acta2 and a statistically significant decrease in Aqp5 and Bhlha15 but no change in Krt5 ( Supplementary Figure S1D ). In the Deligation+DEX group, both KRT7 and AQP5 exhibited recovery patterns similar to those observed in the DEX group. However, in the side effects group, elevated KRT7 levels were observed owing to duct dilation. Notably, the KRT7 layer showed thickening, and the presence of acinar loss around the ducts, as indicated by yellow arrows ( Figure 5F ). As indicated in the quantification graph, a significant decrease in AQP5 was observed in the DEX^Hi and the DEX^L-T group compared to the Deligation+DEX group, while KRT7, which was decreased in the Deligation+DEX group, was rebounded by long-term usage of dexamethasone ( Figure 5G ). This suggests the occurrence of acino-ductal metaplasia in the high-dose and long-term groups.

We performed the SG organoid experiments to confirm the adverse effects of dexamethasone. The high-dose group was treated with dexamethasone (10 µM), and the long-term group, which was treated with dexamethasone for 9 days, showed markedly increased expression of KRT5 and KRT7 and significantly reduced expression of AQP5 ( Figures 5H, I ). In conclusion, high-dose, long-term administration of dexamethasone caused acino-ductal transitions. Therefore, prescribing dexamethasone for an extended period may cause adverse effects during recovery from SG dysfunction.