The human mast cell line (HMC-1) was purchased from China Infrastructure of Cell Line Resources Center. Iscove’s Modified Dulbecco’s Medium (IMDM) containing 1% penicillin and streptomycin, 10% fetal bovine serum, and 2 mM L-glutamine was utilized to cultivate HMC-1 cells in an incubator with 5% CO₂ at 37°C. Hypoxic conditions were maintained at 1% O₂ concentration using a hypoxic chamber (Biospherix C21, USA). Freshly prepared 100 μM Na₂SO₃/NaHSO₃ (Sigma, USA) was used as the SO₂ donor. The working concentration of C48/80 (MCE, China) was 20 μg/ml. The rat basophilic leukemia cell MC line (RBL-2H3) was purchased from China center for culture collection and cultured in MEM medium with 10% FBS. RBL-2H3 cells were incubated overnight with 100 ng/ml anti-DNP-IgE (Sigma-Aldrich, USA) at 37°C. The following day, cells were washed and stimulated with 100 ng/ml DNP-HSA (Biosearch, USA) for 1 hour at 37°C.

HMC-1 cells were infected with lentivirus containing the small hairpin RNA (shRNA) AAT1 (Cyagen, China) to construct AAT1-knockdown MCs. Briefly, the lentivirus containing AAT1-shRNA was added when the cell density was between 60% and 70%, with a multiplicity of infection of 10 in compliance with the manufacturer’s guidelines. Fresh complete media were added after 24 h, and kept at 37°C for the entire night. For a week, puromycin (4 μg/ml) was used for screening the infected cells to get stable AAT1-shRNA-infected MCs. HMC-1 cells infected with lentivirus carrying scrambled shRNA were used as a control according to the same protocol.

SS-1 is a chemo-selective fluorescent probe for in situ detection of SO₂, kindly provided by Prof. Xiaoqi Yu and Prof. Kun Li from Sichuan University (31). Briefly, cells were incubated with SO₂ fluorescent probe for half an hour, followed by three PBS rinses and a minimum of ten minutes of fixation with 4% paraformaldehyde. After rinsing the cells as previously mentioned, the proper concentration of DAPI was applied to allow nuclear to be seen. Green fluorescence observed using a confocal microscope (Zeiss, LSM900) indicated the presence of SO₂. The signal intensity of fluorescence was measured using Image J (NIH, USA) software.

RBL-2H3 cells were collected using PBS and sonicated to obtain the cell lysate. Ear tissue was homogenized with PBS using a tissue homogenizer, centrifuged, and the supernatant was collected. AAT activity of cells and ear tissue was measured according to the instructions of AAT activity assay kit (Nanjing Jiancheng, Nanjing, China). Briefly, the substrate solution and test samples were added into a 96-well plate, and incubated at 37°C for 30 minutes. Then, 2,4-dinitrophenylhydrazine was added to each well and the plate was placed at 37°C for 20 minutes. Afterward, sodium hydroxide solution (200μl, 0.4 M) was added and gently mixed into each well. The plate was incubated at room temperature for 15 minutes. Finally, the optical density was measured at the absorption wavelength of 510 nm. Sodium pyruvate standard solution (2 μM) was used as the standard substance, and diluted according to the instructions to obtain different concentrations. The absolute absorbance values for each concentration were plotted as the y-axis, and the corresponding absorbance units were plotted as the x-axis to create a standard curve. AAT activity was calculated from the standard curve after colorimetric determination. The AAT activity of cells and ear tissue was corrected for homogenate protein concentration.

A β-N-acetylglucosaminase kit was used to evaluate the release rate of β-hexosaminidase from MCs (Yuanmu, China). After isolating the cells by centrifugation, the cell culture supernatant and MCs were collected for examination. MCs were sonicated and centrifuged to obtain cell solution. A mixture containing 10 µl of sample (cell culture supernatant or cell solution) and 50 µl of substrate buffer was incubated at 37°C for 15 minutes. Then the reaction was stopped by adding alkaline solution and the absorbance value was measured at 400 nm. Each liter of the sample was acted with the substrate at 37°C for 1 min and hydrolyzed to produce 1 μmol of p-nitrophenol as 1 unit of enzyme activity.

Venn analysis is an analysis of different datasets used to categorize the intersecting data. Three datasets were used in this study: vascular smooth muscular cell (VSMC) sulfenylome dataset (30), positive regulation of MC activation Gene Ontology dataset (GO: 0033005), and negative regulation of MC activation Gene Ontology dataset (GO: 0033004). In this study, the number of shared genes was statistically analyzed using the sulfenylome and positively regulated/negatively regulated MC activation datasets to show the similarity and overlap of gene compositions between the different datasets.

The sulfenylation of Gal-9 in HMC-1 cells and human purified Gal-9 protein (MCE, China) were quantified using BSA (30). Briefly, HMC-1 cells were lysed in non-denatured lysis buffer (Applygen, China) supplemented with 5 mM DAz-2 for 20 minutes. The supernatant was gathered by a centrifugation at 4°C at a speed of 16000 g for 5 minutes, and then incubated at 37°C with gentle shaking for 2.5 h. Subsequently, 250 μM p-biotin (Cayman, USA) was added, and the mixture was incubated for two hours at 37°C. UltraLink™ Immobiolized NeutrAvidin ™ (Thermo Fisher Science, USA) was added at a 1:10 volume ratio and incubated at 4°C for 4 h. The beads were washed 5 times with PBS and centrifuged to enrich sulfenylated protein. Finally, non-denatured loading buffer was added and heated for ten minutes at 100°C. Western blot was conducted to quantify the level of Gal-9 sulfenylation.

Human purified Gal-9 protein was divided into three groups: control, SO₂ and SO₂+Dithiothreitol (DTT) groups. Each sample used 0.2 μg of protein. The protein was incubated with SO₂ and DTT at 37°C for 2 h. The working concentration of SO₂ and DTT were both 100 μM. After the incubation, the sulfenylation of Gal-9 was detected according to the same protocol described in the cell experiments.

From the UniProt database (https://www.uniprot.org/), the Gal-9 protein sequence for each species was obtained. The protein IDs were listed as follows: human (O00182), mouse (O08573), rat (P97840), bovine (Q3MHZ8), Camelus dromedarius (A0A5N4D4M7), and pig (Q9XSM9). BioEdit software was used for the homology analysis of the above sequences.

Human purified Gal-9 protein was dissolved in PBS (10 μg in 30 μl). Then 100 μM SO₂ and 10 mM dimethyl ketone were added, shaken gently, and incubated for 2 hours at 37°C. The protein samples were loaded into a non-reducing loading buffer and separated using 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The coomassie bright blue R-250 dye was applied to the protein gel, then enzymatically cut. LC-MS/MS was utilized for the extraction and identification of the peptide mixture. The pFind 3 software, developed by Professor He Simin and the members of his group at the Institute of Computing Technology, Chinese Academy of Sciences, was used to analyze the mass spectra (32).

HMC-1 and RBL-2H3 cells were transfected with plasmids using the SF 4D-NucleofectorTM X Solution Kit (Lonza, Germany). Electrotransformation mix (100 μl) containing 1 μg of plasmid was prepared in compliance with the manufacturer’s guidelines. The pcDNA3.1 vectors for Gal-9-WT-His, Gal-9-C74S-His and Gal-9-C312S-His were constructed (General, Anhui, China). Briefly, the cells underwent transfection and were left to stand at room temperature for ten minutes. Then cells were gently resuspended, aspirated into culture wells, and gently mixed with a pre-warmed medium. After six hours, the medium was replaced with a new one.

Following the lysis of the cells in RIPA buffer, the protein concentrations were determined using a BCA kit (Beyotime, China). Proteins in equal amounts were separated using 12% SDS-PAGE and then transferred onto nitrocellulose membranes. These membranes were then blocked using a blocking solution that contained 5% milk powder. The primary antibodies were incubated with the following dilutions: anti-AAT1 (1:1000; Abcam, USA), anti-β-tubulin (1:2000; Zsbio, China), anti-Galectin-9 (1:1000; Proteintech, China), and anti-His (1:1000; Zsbio, China). The corresponding secondary antibodies were used subsequently. The bands incubated with chemiluminescent detection reagents were analyzed by a FluorChem M MultiFluor system (Protein Simple, USA).

Mice were purchased from Si Pei Fu Biotechnology Co., Ltd (Beijing, China) and housed in the Experimental Animal Center of Peking University First Hospital. Mice were housed in a specific pathogen-free and viral antibody-free animal facility in accordance with the guidelines established by the Institutional Committee on Animal Use and Care (IACUC) and Laboratory Animal Resource Center (LARC).

For IgE-mediated passive cutaneous anaphylaxis (PCA) model, the animal study was approved by the Animal Research Ethics Committee of Peking University First Hospital (Ethics No.: J2024039). Briefly, the mouse PCA model was constructed by an intradermal injection with 500 ng of anti-DNP-IgE in 20 μl PBS in the left ear and an intravenously challenge with DNP-HSA (50 μg in saline containing 1% Evans blue) after 24h in the eight-week-old female BALB/c mice. Thirty minutes after the challenge, skin areas were photographed and the thickness of the ears was measured, after which the mice were euthanized. Evans blue dye was extracted by incubating the skin tissues in 300 μl of formamide for 24 h at 63°C, after which absorbance was measured on a spectrophotometer at 620 nm. Mice in IgE+SO₂ group were intraperitoneally injected three days with SO₂ donor (Na₂SO₃:68.04 mg/kg; NaHSO₃:18.72 mg/kg body weight) before IgE sensitization (33). Mice in control group received three intraperitoneal injections with the same amount of physiological saline and an intradermal injection with 20 μl of PBS. SO₂ donor was freshly dissolved in saline.

The animal study about chronic hypoxia/SU5416 (SuHx)-stimulated mouse model was approved by the Animal Research Ethics Committee of Peking University First Hospital (Ethics No.: J2023008). C57BL/6 mice were exposed to normoxia (ambient air, 21% O₂) or hypoxia (10% O₂) in a ventilated chamber (Ox-100, TOW, Shanghai, China) for 3 weeks. The mice in the SuHx and SuHx+SO₂ group received a single weekly subcutaneous injection of the VEGFR2 antagonist SU5416 at 20mg/kg, while the normoxia group received a vehicle injection (34). SO₂ donors (Na₂SO₃:68.04 mg/kg; NaHSO₃:18.72 mg/kg body weight) were administered intraperitoneally daily.

SO₂ levels in ear tissues were measured by HPLC-FD (Agilent, Palo Alto, CA, USA) as described in the previous study (30). In brief, ear tissue was homogenized with PBS using a tissue lyser, centrifuged, and the supernatant was collected. The SO₂ in the sample was reduced to sulfhydryl compounds by sodium borohydride, and then labeled with monobromobimane to generate fluorescent derivative. Perchloric acid was used to deproteinize. After centrifugation, the supernatant was neutralized by Tris-HCl (pH 3.0) and prepared for HPLC-FD analysis. The fluorescent derivatives were separated by chromatographic column and detected at excitation/emission wavelength of 392/479 nm.

Mouse tissues were fixed with 10% formalin and embedded in paraffin. The fixed tissues were cut into 4 µm sections. Sections were dewaxed and incubated with toluidine blue dye solution (Solarbio, China) for 10 min. Excess dye was then washed away and the background color was separated with 95% alcohol for 5min. The quantification of mast cells and their degranulation status relied on the identification of abnormal morphology and the presence of extracellular cytoplasmic granules under a light microscope using 400× magnification (Olympus, Japan) (35). The number of degranulated mast cells was counted in two fields per section.

HE staining was conducted following standard procedures. Briefly, tissue sections were immersed in hematoxylin solution, followed by rinsing in PBS to eliminate excess stain. Subsequently, the sections were differentiated in 1% acid alcohol for 1 s, and rinsed in water for 1 min. Finally, sections were sequentially stained with eosin dye, dehydrated in gradient ethanol and xylene, and mounted with resinene.

The sections were soaked in xylene to dewax, and then rehydrated in gradient ethanol. Subsequently, the sections were stained with the Verhoeff staining solution (a mixture of alcohol hematoxylin, ferric chloride, and iodine solution) for 30 min. Following this, the sections were stained with Van Gieson’s solution (a mixture of saturated picric acid and acidic magenta in a volume ratio of 9:1) for 1–3 minutes, followed by a quick water wash. Finally, the sections were sequentially immersed in anhydrous ethanol and xylene for 1–5 minutes, and sealed with neutral resin. Images were acquired and analyzed using Leica Q550 CW. Total vessel area was defined as the area within the lamina elastica externa, and lumen area was defined as the area within the lamina elastica interna. Medial area is defined as the area between the lamina elastica externa and lamina elastica interna. The media area was expressed as a percentage of the total external area of the vessel (36, 37). The vessel median thickness was measured at four points at 12, 3, 6 and 9 o’clock on each vessel section, and the average value was calculated (38).

For statistical analysis, Graphpad Prism 8 and SPSS17.0 were both utilized, and the results were shown as mean ± SD. GraphPad Prism 8 software was used to generate graphs. The two-tailed Student’s t-test was performed for the comparison between two groups, and one-way ANOVA followed by Bonferroni post-doc analysis was used to compare the difference among multiple groups. A significance level of P < 0.05 was applied.

MC degranulation activation is a crucial initiation and development process of the immune inflammatory response. The β-hexosaminidase release is a landmark event of MC degranulation. It has been confirmed that AAT1 catalyzes the generation of endogenous SO₂ in MCs (26). In the present study, we verified the effect of endogenous SO₂ on MC degranulation under physiological condition by employing AAT1 knockdown and SO₂ rescue. The data showed that AAT activity and SO₂ content were decreased in AAT1-knockdowned HMC-1 compared with scrambled cells, while the supplementation of SO₂ donor restored the SO₂ content in cells of sh-AAT1 group ( Figure 1A , Supplementary Figure 1A ). Simultaneously, the release of β-hexosaminidase in HMC-1 cells was significantly increased after AAT1 knockdown compared with that in the scramble group, which was reversed by SO₂ donor ( Figure 1B ). However, in the MCs transfected with scramble lentivirus, SO₂ donor increased the SO₂ content in the cells but did not affect MC degranulation ( Supplementary Figure 2 ). The above findings suggested that AAT1-derived endogenous SO₂ might be an important stabilizer by suppressing MC degranulation under physiological condition.

Subsequently, we explored the probable mechanism underlying endogenous SO₂-controlled MC degranulation. As shown in Figure 1B , a thiol-reducing agent DTT significantly abolished the SO₂-suppressed β-hexosaminidase release, suggesting that a thiol-dependent chemical modification might be involved in the regulation of SO₂-controlled MC degranulation. Furthermore, we screened the candidate target proteins by SO₂ from a SO₂-mediated sulfenylome dataset as previously described (30). Huang et al. found that 1137 thiol groups of cysteine residue on 658 proteins were responsive to SO₂ in VSMC (30). Based on the GO dataset, 16 genes were related to negative regulation for MC activation and 75 genes were related to positive regulation. Venn analysis was performed on the SO₂ sulfenylome dataset using positively and negatively regulated MC activation datasets. The only candidate gene, LGALS9, encoding the protein Gal-9, was found to regulate MC activation, and the thiol group of the Gal-9 protein could be modified by SO₂-mediated sulfenylation ( Figure 1C ).

Based on the grouping shown in Figure 1B , we verified whether endogenous SO₂ sulfenylated the Gal-9 protein in HMC-1 cells. BSA results showed that the sulfenylation of Gal-9 protein in the MCs of the sh-AAT1 group was substantially lower than that of the scramble group. Compared with the sh-AAT1 group, Gal-9 sulfenylation increased after SO₂ supplementation, while the sulfenylation of Gal-9 by SO₂ was significantly decreased when the cells treated with DTT supplementation ( Figure 1D ). These results indicated that HMC-1 cell-derived SO₂ modified the thiol group of the Gal-9 protein by sulfenylation.

Furthermore, an enhanced sulfenylation of Gal-9 in HMC-1 cells of SO₂ group was shown compared with the control group, which was reversed by the treatment of DTT ( Figure 1E ). Subsequently, the SO₂-induced Gal-9 sulfenylation was verified in the cell-free experiment. The BSA results showed that the SO₂ treatment upregulated the sulfenylation of Gal-9 in the purified human Gal-9 protein, while DTT supplementation blocked SO₂-induced Gal-9 sulphenylation ( Figure 1F ). These findings suggested that Gal-9 was sulfenylated by SO₂ in cellular and cell-free experiments.

To further investigate the exact sulfenlyation site on the Gal-9, a bioinformatics analysis, LC-MS/MS, and site-directed mutation were used. Firstly, as showed in the Figure 2A , human Gal-9 has six cysteine residues: Cys74, Cys102, Cys169, Cys259, Cys312, and Cys316. Homologous conservation analysis of cysteine sites was performed on the Gal-9 sequences of different species, and it was found that the cysteine residues 74 and 312 in humans, mice, rats, cows, pigs, and other species were highly conserved ( Figure 2A ).

Secondly, LC-MS/MS results showed that sulfenylation of purified Gal-9 protein by SO₂ occurs at the residue of cysteine 74 ( Figure 2B ), which was in accordance with the finding demonstrated in the SO₂-mediated sulfenylome in the previous study. Next, we constructed Gal-9-wild type (WT), Gal-9-C74S (mutation of cysteine-74 to serine), and Gal-9-C312S (mutation of cysteine-312 to serine) mutant plasmids. HMC-1 cells were transfected using the electrotransfer method. Here, Gal-9-C312S mutant plasmid was used as negative control. The results of BSA detection suggested that SO₂ treatment promoted the sulfenylation of Gal-9 in cells transfected with Gal-9-WT and Gal-9-C312S mutant plasmids. However, SO₂-induced Gal-9 sulfenylation was not found in cells transfected with Gal-9-C74S mutant plasmid ( Figure 2C ). These above findings suggested that sulfenylation of Gal-9 by SO₂ occurred at the 74^th cysteine residue.

Considering IgE is a classical stimulator of MC degranulation, we further investigated the effect of SO₂ on MC degranulation in the IgE-challenged MCs and mouse models. The rat basophilic leukemia cells, RBL-2H3, are a tumor analogue of mast cells with a high expression of FcϵRI on the cell surface, and can be activated by the IgE-antigen complex (39, 40). Our results showed a significant reduction in SO₂ content, AAT1 protein expression, and AAT activity within IgE-insulted RBL-2H3 cells compared with the control group ( Figures 3A–C ). Concurrently, MC degranulation was increased in IgE-treated cells compared with IgE-untreated cells ( Figure 3D ). Also, the supplementation of SO₂ donor restored the SO₂ content and reduced MC degranulation in IgE-stimulated cells ( Figures 3A, D ). Those in vitro findings suggested that IgE stimulation downregulated the endogenous SO₂/AAT1 pathway in MC, as well as the deficiency of endogenous SO₂ might participate in IgE-induced MC degranulation.

Furthermore, we constructed a mouse IgE-mediated PCA model to investigate the effect of SO₂ on the mast cell-dependent inflammation in vivo ( Figure 4A ). Local anti-DNP-IgE sensitization and systemic administration of DNP-HSA together with Evans blue triggered a remarkable dye extravasation and ear swelling, but reduced AAT activity and SO₂ content in the IgE-sensitized ear compared with the control group ( Figure 4 ). As observed in the in vitro experiments, the pretreatment of SO₂ donor recovered the SO₂ content in ear tissue, and subsequently prevented the dye extravasation and ear swelling in the IgE-challenged mouse ( Figure 4 ), further reinforcing that SO₂ suppressed IgE-related MC activation and inflammation.

We also induced a mouse model of pulmonary vascular remodeling using chronic hypoxia combined with SU5416 to disclose the probable meaning of SO₂ in non-IgE-stimulated MC degranulation pathological model in vivo ( Figure 5A ). A decrease in SO₂ content and AAT activity, along with an increase in the number of degranulated mast cells were found in lung tissues of SuHx mice compare with those of normoxic mice ( Figures 5B–D ). Simultaneously, HE and EVG staining results indicated that compared with the normoxia group, the mice in the SuHx group exhibited an increased media thickness and area of pulmonary arterioles ( Figures 5E, F ). Furthermore, after exogenous supplementation of SO₂, the number of degranulated mast cells, the media thickness and area of the arterial were significantly reduced in the lung tissues of SuHx mice ( Figures 5D–F ). The above results suggested a potential association among the decreased SO₂, the activation of MCs, and the pulmonary vascular remodeling in the hypoxic, a non-IgE stimulation, mouse pathological model.

To demonstrate if SO₂-mediated Gal-9 sulfenylation resulted in the inhibition of IgE-stimulated MC degranulation by SO₂, we compared the effects of SO₂ on the IgE-challenged RBL-2H3 cells transfected with Gal-9-WT, Gal-9-C74S mutant, and Gal-9-C312S mutant plasmids. Figure 6A showed that SO₂ supplement restored the intracellular SO₂ level in IgE-treated and increased SO₂ content in IgE-untreated cells. In all 3 types of RBL-2H3 cells transfected with Gal-9 WT and mutant plasmids, IgE stimulation induced cell degranulation ( Figure 6B ). However, SO₂ supplement only suppressed IgE-stimulated cell degranulation in the cells transfected with Gal-9-WT and Gal-9-C312S mutant plasmids, but not in those transfected with Gal-9-C74S mutant plasmid ( Figure 6B ). The above findings implied that SO₂-sulfenylated Gal-9 at Cys74 was necessary to suppress IgE-activated MC degranulation.

To further clarify the significance of SO₂-mediated Gal-9 sulfenylation in the inhibition of non-IgE-stimulated MC degranulation by SO₂, we firstly examined the changes of AAT activity under hypoxic stimulation. The results showed that hypoxia decreased AAT activity in MC compared with normoxic cells ( Supplementary Figure 1B ). Next, we compared the effects of SO₂ on the MCs transfected with Gal-9-WT and two mutant plasmids under hypoxic stimulation. As shown in Figure 7A , SO₂ supplement increased the intracellular SO₂ level in both hypoxic and normoxic cells. In all 3 types of HMC-1 transfected with Gal-9 WT and mutant plasmids, hypoxia induced cell degranulation ( Figure 7B ). However, SO₂ supplement only suppressed hypoxia-induced cell degranulation in the cells transfected with Gal-9-WT and Gal-9-C312S mutant plasmids, but not in those transfected with Gal-9-C74S mutant plasmid ( Figure 7B ). Moreover, C48/80 is a compound that induces MC degranulation in a non-IgE-dependent manner (41, 42). Like the hypoxia and IgE stimulation, C48/80 also inhibited AAT activity and SO₂ content in HMC-1 cells ( Supplementary Figure 1C , Figure 8A ), while SO₂ supplement restored the intracellular SO₂ level in the C48/80-induced cells ( Figure 8A ). Concurrently, only in the cells transfected with Gal-9-C74S mutant plasmids, SO₂ could not prevent the C48/80-induced MC degranulation ( Figure 8B ). The above results suggested that SO₂-sulfenylated Gal-9 at Cys74 also participated in the suppressive effect of SO₂ on the non-IgE-stimulated pathophysiological MC degranulation.