2.1 Patients

A total of 30 control subjects and 25 patients with AR were included in this study. More information is provided in this article's Supporting Information.

2.2 Microarray data acquisition

The GSE9150 dataset was downloaded from the Gene Expression Omnibus (GEO) database using the GEOquery package in R software.³⁷ More details can be found in the Supporting Information.

2.3 Data normalisation and identification of DEGs

With the help of the R software package limma, differentially expressed genes (DEGs) in the GSE9150 microarray were identified (adj. p < .05, |logFC| > 2).³⁸

2.4 PPI network construction and hub gene exploration

The protein–protein interaction (PPI) network was built using DEGs via STRING online tool³⁹ and visualised with Cytoscape software.⁴⁰ Moreover, Hub genes were identified using the maximal clique centrality (MCC) algorithm in the CytoHubba plugin.⁴¹

2.5 Identification of tissue/organ-specific expressed genes

An analysis of the tissue distribution of DEGs using BioGPS was conducted to identify tissue- or organ-specific DEG expression.⁴²

2.6 Cell culture

Human nasal epithelial cells (HNEPCs) were cultured in minimal essential media (MEM) supplemented with 10% foetal bovine serum (FBS) and 1% penicillin/streptomycin.⁴³ In some experiments, cells were pre-incubated with Mito-TEMPO (Selleck) for 3 h. The culture medium was replaced every 48 h.

2.7 H&E staining

Paraffin-embedded specimens were cut into 3 µm sections and stained with haematoxylin and eosin (H&E). More detailed protocol can be found in the Online Repository.

2.8 Immunohistochemical staining

Immunohistochemical (IHC) was conducted as previously described.⁴⁴

2.9 Western blotting

We followed previously established protocols for the Western blot analysis, and the antibodies utilised in this study are detailed in Table S2.^{44, 45}

2.10 Immunofluorescence staining

Immunofluorescence (IF) staining for cultured HNEPCs was performed as previously described.⁴⁵

2.11 Quantitative RT-PCR

Quantitative polymerase chain reaction (qPCR) was performed as stated elsewhere.^{44, 45}

2.12 ELISA

A commercial ELISA kit (Ruixin Biotechnology) was used to measure serum levels of total IgE and HDM-specific IgE in mice.

2.13 Transfection

HNEPCs were transfected with small interfering RNA (siRNA) targeting CMPK2 (si-CMPK2), siRNA targeting STING (si-STING), cytoplasmic mtDNA or CMPK2 expression plasmids as stated elsewhere.⁴⁵ More details can be found in the Supporting Information.

2.14 Extraction and detection of mtDNA

The mtDNA from the cytoplasm of HNEPCs was isolated utilising the Mitochondrial DNA Extraction Kit (ab65321, Abcam) in accordance with the protocol provided by the manufacturer.⁴⁶

2.15 Mitochondrial membrane potential (ΔΨm)

After treatment with HDM or Mito-TEMPO, mitochondrial membrane potential (MMP) of HNEPCs was measured using JC-1 probe (#C2003S, Beyotime).

2.16 Assessment of reactive oxygen species generation

The HNEPCs were incubated for 30 min at 37°C with DCFH-DA (10 µM, Solarbio) to detect intracellular reactive oxygen species (ROS). For mitochondrial ROS (mtROS) detection, HNEPCs were incubated with MitoSOX (5 µM, YEASEN Biotech) for 10 min at 37°C. After washing, fluorescence images were obtained by a confocal laser-scanning microscopy (Olympus).

2.17 In vivo experiments

AR was induced in mice using HDM (Greer Laboratories). The experimental protocol is depicted in Figure 4A and more information is provided in the Online Repository.

2.18 Statistical analyses

All data are presented as means and standard error from the indicated number of samples. Comparative analyses between two groups were conducted utilising a t-test, while comparisons among multiple groups were executed using a one-way analysis of variance (ANOVA) followed by Bonferroni's test. Non-normally distributed data were analysed with the Kruskal–Wallis test, and correlations were assessed using the Spearman test. All analyses were conducted with IBM SPSS 22.0, and figures were created using GraphPad Prism 7. A p value of <.05 was considered significant.

3.1 Identification of the haematologic/immune system-specific hub genes associated with allergy

First, the datasets GSE9150, which included primary nasal epithelial cells exposed to HDM from five monotypic HDM-allergic samples and five healthy control samples, were selected for identification of the DEGs between these two groups. Boxplots of the normalised gene expression profiles (Figure S1A), along with principal components analysis (PCA) and Uniform Manifold Approximation and Projection (UMAP), indicated significantly different expression profiles between allergic samples and normal controls (Figure S1B,C). A total of 482 DEGs were identified, with 250 up-regulated and 232 down-regulated genes (Figure 1A,B). Among these, 83 are tissue-/organ-specific genes as identified by BioGPS (Table S6). The PPI network of the DEGs was constructed utilising the STRING tool and subsequently visualised with Cytoscape (Figure S1D). Subsequently, the top 20 hub genes were determined employing the CytoHubba MCC algorithm, as illustrated in Figure S1D and detailed in Table S7. To investigate the role of HDM pathogenic hub genes, we conducted an intersection analysis between the top 20 hub genes and those specifically expressed in the haematologic/immune system. This analysis led to the identification and validation of four hub genes specific to the haematologic/immune system: RRM1, RSAD2, IFI44 and CMPK2 (Figures 1C and S1F). These tissue-specific hub genes may play a key role in the pathogenesis of AR, and CMPK2, with the most significant difference, was selected for further study.

3.2 Expression of CMPK2 and NLRP3 inflammasome in both nasal mucosa and HNEPCs of AR

Next, Western blotting and immunohistochemistry were used to assess CMPK2 protein levels in nasal tissues from control subjects and AR patients. CMPK2 was detected in nasal epithelial and inflammatory cells in the lamina propria, with its protein levels being significantly higher in individuals with AR than in healthy nasal tissues (Figure 1D,E). Moreover, we explored the expression of CMPK2 in HDM-stimulated airway epithelial cells. We found that HDM increased CMPK2 protein expression in a dose- and time-dependent manner in HNEPCs (Figure 1F,G). Similar results were obtained using BEAS-2B cells, which indicates that CMPK2 may be involved in the pathogenesis of AR (Figure S1G,H). After optimisation, HNEPCs were treated with 50 µg/mL HDM allergen for 24 h to establish AR-HNEPCs for subsequent study.

In light of previous research indicating that CMPK2 plays an important role in activating NLRP3, we propose that HDMs may enhance NLRP3 activation via CMPK2 up-regulation, thereby advancing the progression of AR, thus promoting the progression of AR.^{47, 48} We initially investigated NLRP3 inflammasome activation in AR patients. IHC staining revealed significantly higher expression of NLRP3, caspase-1, ASC and IL-1β in AR nasal tissues compared to controls (Figure 2A). Similar up-regulation protein trends were observed by Western blot (Figure 3A,B). Notably, correlation analysis revealed that CMPK2 staining intensity was positively correlated with the staining intensities of NLRP3, caspase-1, ASC and IL-1β in AR (Figure 2B–E).

3.3 CMPK2 is an upstream regulator of the NLRP3 inflammasome

To investigate the potential critical role of CMPK2 in the activation of the NLRP3 inflammasome in AR, we employed siRNA to down-regulate CMPK2 expression in HNEPCs. The efficacy of the knockdown was subsequently validated through quantitative reverse transcription polymerase chain reaction (qRT-PCR) and Western blot analysis (Figure S1I,J). Our data showed that knockdown of CMPK2 significantly reduced IL-1β mRNA expression, but had little effect on the expression of IL-6, IL-8, CSF2, periostin and CCL5 (Figure S1K). Moreover, the qRT-PCR results showed that NLRP3 inflammasome components, including NLRP3 and CASP1, were significantly decreased after CMPK2 knockdown (Figure 3C). Consistently, the NLRP3 inflammasome activation was mitigated by knocking down CMPK2, as shown by reduced cleavage of CASP1 and IL-1β in response to HDM stimulation (Figures 3E and S2A,B). Similar results were observed by IF staining (Figure 3D). Moreover, we transfected HNEPCs with a CMPK2 expression plasmid (transfection efficiencies shown in Figure S2C), and CMPK2 overexpression significantly increased C-IL-1β expression (Figure 3F). This indicates that CMPK2 is essential for NLRP3 inflammasome activation in response to HDM stimuli.

3.4 CMPK2 deficiency impairs NLRP3 inflammasome activation in HDM-challenged AR mice

Subsequently, we aimed to investigate whether a deficiency in CMPK2 mitigates the progression of AR in vivo. According to the experimental schedule depicted in Figure 4A, a mice model of AR induced by HDM was established. In alignment with the findings derived from human samples and HNEPCs, the protein expression levels of CMPK2 and the NLRP3 inflammasome-related proteins (NLRP3, ASC, C-CASP1, C-IL-1β) were significantly elevated in the nasal mucosa of mice with HDM-induced AR when compared to those of the control group (Figure 4B,D). In order to further elucidate the role of CMPK2 in activating the NLRP3 inflammasome in vivo, we used CMPK2^−/− mice. As shown in Figure S3, CMPK2 was not expressed in the nasal mucosa of CMPK2-deficient mice. The H&E staining demonstrated an increased number of eosinophils in the nasal mucosa of HDM-challenged wild-type (WT) mice, whereas a reduced number was observed in CMPK2^−/− mice (Figure 4C). Furthermore, the number of periodic acid schiff (PAS)-stained goblet cells in HDM-challenged mice was significantly elevated compared to control mice, and CMPK2 knockout attenuated goblet cell hyperplasia (Figure 4C). In addition, HDM-sensitised mice had markedly increased frequency of sneezing and nose scratching compared with the control group, but CMPK2 knockout relieved these allergic symptoms (Figure 4E,F). Similarly, the elevation in serum total IgE and HDM-specific IgE levels in WT mice after HDM challenge was ameliorated in CMPK2 knockout mice (Figure 4G,H). Notably, a significant reduction in NLRP3 inflammasome activation occurred in the nasal mucosa of HDM-challenged AR mice with CMPK2 deficiency (Figure 3I). Furthermore, the IHC staining results exhibited a similar trend of attenuated inflammasome activation in CMPK2-deficient mice (see Figure S4A). These findings indicate that CMPK2 plays a role in the pathogenesis of AR by facilitating the activation of the NLRP3 inflammasome.

3.5 mtROS played a key role in HDM-induced activation of the NLRP3 inflammasome in HNEPCs

We conducted a further investigation into the mechanisms underlying the activation of the NLRP3 inflammasome induced by HDM in HNEPCs. We first investigated the relationship between HDM and mitochondrial dysfunction, which may lead to mtDNA leakage into the cytosol, thereby activating the NLRP3 inflammasome.⁴⁹ After HDM treatment, the green fluorescence representing intracellular ROS increased significantly (Figure S5A). Subsequently, our findings revealed an elevation in mtROS in HDM-treated HNEPC, as demonstrated by IF staining (Figure S5B). Furthermore, HDM exposure resulted in a decrease in mitochondrial membrane potential, as illustrated by a reduction in JC-1 aggregates, and an increase in JC-1 monomers (Figure S5C). To further substantiate the contributory role of mtROS, Mito-TEMPO, a specific mtROS scavenger, was employed. Our study showed that Mito-TEMPO (50 µmol/L) pretreatment significantly reduced HDM-induced mtROS accumulation and restored mitochondrial membrane potential (Figure 5A–D). Furthermore, Western blot analysis revealed that Mito-TEMPO effectively reversed the up-regulation of CMPK2 and the activation of the NLRP3 inflammasome in AR (Figure 5E–J).

Previous studies indicate that mtROS can oxidise mitochondrial DNA (ox-mtDNA), leading to its release into the cytoplasm, where it activates the inflammasome by binding to NLRP3.^50-52 We then used a fluorescent antibody to label 8-OHdG, a marker of DNA oxidation. Consistent with previous findings, the oxidised DNA was increased in HNEPCs after HDM treatment and co-localised with cytosolic NLRP3, which can promote the activation of NLRP3 inflammasome (Figure 5K,L). However, pretreatment with Mito-TEMPO markedly reversed this phenomenon (Figure 5K,L). These data indicated that Mito-TEMPO treatment dramatically alleviated mitochondrial oxidative stress and mitochondrial dysfunction, which may serve as a potential target for AR therapy.

3.6 CMPK2 facilitates NLRP3 inflammasome activation through mtDNA

Recently, it was found that mtDNA is essential for NLRP3 inflammasome activation and that this process is dependent on the mitochondrial nucleotide monophosphate kinase CMPK2, which is a critical enzyme responsible for mtDNA synthesis.²⁶ Our findings revealed that stimulation with HDM significantly increased the mtDNA copy number in HNEPCs compared to the control group, whereas knockdown of CMPK2 resulted in a substantial reduction in mtDNA copy number (Figure S6A,B). In addition, our study showed that mtDNA in the cytoplasm of HNEPCs was significantly increased after HDM treatment (Figure S6C–E), and knockdown of CMPK2 significantly attenuated this effect (Figure 6A,B).

To elucidate the role of cytosolic mtDNA in the activation of the NLRP3 inflammasome triggered by HDM, we employed EtBr to inhibit mtDNA replication.⁵³ After the application of EtBr, the mtDNA copy number of HNEPCs was reduced by approximately 60% (Figure 6C). As expected, the Western blotting results showed that NLRP3 inflammasome-related protein (NLRP3, C-CASP1 and C-IL-1β) levels were significantly up-regulated in HNEPCs treated with HDM compared with the levels in the untreated controls. However, such elevation of the NLRP3 inflammasome was inhibited by EtBr pretreatment (Figure 6D–G). Furthermore, immunofluorescent staining results showed a similar trend to the Western blot assay results (Figure 6H). The findings indicate that mtDNA serves as a crucial mediator in the activation of the NLRP3 inflammasome by CMPK2, potentially involving the cGAS-STING signalling pathways.

3.7 Enhanced cGAS-STING pathway in HDM-induced AR

Studies have indicated that the cGAS-STING pathway is crucial for activating the NLRP3 inflammasome. Our experiments showed that HDM stimulation increased cGAS protein expression and STING phosphorylation both in vivo (Figure 7A) and in vitro (Figure 7B), indicating activation of the cGAS-STING pathway in the AR model. Furthermore, the IHC staining results indicated that STING expression was elevated in patients with AR relative to the control group (Figure 7C). Additionally, there was a positive correlation between the intensity of CMPK2 in IHC staining and STING expression in AR patients (Figure 7D). However, the absence of CMPK2 abrogated STING activation in the AR model (Figures 7E,F and S7A). Taken together, these results suggest that the cGAS-STING pathway is activated in HDM-induced AR and is regulated by CMPK2.

3.8 CMPK2 promotes NLRP3 inflammasome activation via the cGAS/STING pathway

To further explore whether cGAS-STING participates in the activation of NLRP3 inflammasome in HNEPCs, STING siRNA and H151 (a selective STING antagonist) were used to inhibit cGAS-STING signalling. The knockdown efficiencies of the siRNAs were validated in HNEPCs by Western blot and qRT-PCR (Figure S7B,C). We found that si-STING treatment or pharmacological inhibition of STING with H-151 inhibited HDM-induced activation of NLRP3 inflammasome in HNEPCs (Figure 7G,H). To verify whether cytosolic mtDNA activates NLRP3 via the cGAS-STING pathway, we isolated mtDNA and transfected it into HNEPCs. Western blot analysis demonstrated that the transfection of cytoplasmic mtDNA significantly activated the STING signalling pathway and subsequently induced the activation of the NLRP3 inflammasome. This activation was notably inhibited upon STING knockdown (Figure S8A,B). Furthermore, pretreatment of Mito-TEMPO can also reduce the phosphorylation level of STING (Figure S9A).

Next, STING knockout mice were used to determine the role of STING signalling in the activation of NLRP3 inflammasome in murine AR model. The knockout efficiency of STING was assessed by Western blotting (Figure S7D). STING deficiency reduced eosinophilic infiltration and goblet cell hyperplasia in the nasal mucosa of AR mice compared to the WT group (Figure 8A). Not surprisingly, STING deficiency significantly inhibited the activation of NLRP3 inflammasome in the nasal mucosa of HDM-challenged AR mice (Figure 8B–H). The IHC staining analysis consistently yielded results analogous to those obtained through Western blot analysis (Figure 8I). Collectively, these findings indicated that inhibition of STING could alleviate NLRP3 inflammasome activation and disease progression in HDM-induced AR.