Abstract Background Necroptosis, a regulated form of programmed cell death, exacerbates inflammatory responses by releasing damage-associated molecular patterns and inflammatory factors. However, the specific mechanisms underlying necroptosis in periodontitis remain largely unclear. This study integrated single-cell RNA sequencing (scRNA-seq) and bulk RNA sequencing (RNA-seq) data to identify core necroptosis-related genes (NRGs) and validated these findings using external datasets and periodontitis samples collected during our research. Methods Overlapping genes were identified through a comparative analysis of 114 NRGs sourced from GeneCards and marker genes specific to various cell types in the single-cell [40]GSE171213 periodontitis dataset. Based on these genes, cells were categorized into high- and low-necroptosis score groups. Key NRGs were identified through intersection analysis of differentially expressed genes in the high necroptosis group using the [41]GSE10334 bulk RNA-seq dataset, followed by Kyoto Encyclopedia of Genes and Genomes (KEGG)/ Gene Ontology (GO) enrichment analysis. Machine learning further identified hub genes associated with the inflammatory response in periodontitis. Consensus clustering analysis, clinical diagnostic model construction, gene set variation analysis, and gene set enrichment analysis were performed based on these hub genes. The model’s predictive performance was validated using independent datasets and periodontitis tissue samples. Results We identified 10 cell types in periodontitis tissues and observed changes in the abundance of various cell populations in affected samples. Furthermore, we selected 35 NRGs differentially expressed in specific cell populations, with neutrophils and macrophages showing higher necroptosis scores. By integrating bulk RNA-seq data, we further identified 29 key NRGs. KEGG/GO analysis indicated their enrichment in inflammatory response signaling pathways. Machine learning highlighted six hub genes (CSF3R, CSF2RB, BTG2, CXCR4, GPSM3, and SSR4), all of which were highly expressed in periodontitis tissues. Consensus clustering based on these genes divided patients with periodontitis into two subgroups with distinct expression profiles. The clinical diagnostic model constructed based on these six key genes exhibited excellent diagnostic performance. Both external independent validation sets and clinical sample tests confirmed high expression of these six key genes in periodontitis tissues. Conclusion Our study identified six hub genes (CSF3R, CSF2RB, BTG2, CXCR4, GPSM3, and SSR4) highly expressed in periodontitis tissues and positively correlated with necroptosis. These genes may serve as therapeutic targets for inflammatory diseases like periodontitis. Supplementary Information The online version contains supplementary material available at 10.1186/s12920-025-02241-1. Keywords: Periodontitis, Necroptosis, Inflammatory response, Hub genes, Therapeutic target Background Periodontitis is a chronic infectious disease initiated by dental plaque biofilm and characterized by gingival inflammation and destruction of the periodontal supporting tissues. Periodontitis is the major cause of tooth loss in adults and poses a significant health risk due to its systemic implications [[42]1]. Studies have shown that patients with periodontitis have a significantly increased risk of cardiovascular diseases, such as heart disease and stroke, possibly attributed to chronic inflammation that promoting atherosclerosis [[43]2, [44]3]. Furthermore, periodontitis is closely linked to diabetes, respiratory conditions like pneumonia, and rheumatoid arthritis, establishing a vicious cycle that intensifies the symptoms of these ailments and complicates their treatment [[45]4, [46]5]. Consequently, exploring the inflammatory regulatory mechanisms involved in periodontitis progression is pivotal for its prevention and management and may also offer fresh perspectives on mitigating the advancement of these associated diseases. Necroptosis is a form of caspase-independent cell death that is tightly regulated to maintain tissue homeostasis [[47]6, [48]7]. In contrast to uncontrolled cell necrosis, necroptosis constitutes a highly orchestrated and stringently regulated process involving intricate molecular mechanisms. Necroptosis is accompanied by organelle swelling, cell membrane rupture, and breakdown of the cytoplasm and nucleus [[49]7]. Notably, necroptosis relies on the interaction of key proteins, including receptor-interacting protein kinases (RIPK1 and RIPK3) and mixed lineage kinase domain-like pseudokinase (MLKL) rather than caspase activation [[50]8–[51]10]. Necroptosis is also associated with inflammatory diseases. During inflammation, necroptosis not only participates in cell death but also triggers or exacerbates inflammatory responses by releasing damage-associated molecular patterns (DAMPs) and inflammatory factors (such as IL-1α, IL-1β, and IL-33) [[52]11, [53]12]. This inflammatory cascade facilitates the clearance of damaged cells and defense against pathogenic infections; however, excessive or uncontrolled inflammation can result in tissue damage and disease progression. Currently, multiple studies have clearly demonstrated that necroptosis is closely associated with the onset and progression of periodontitis, and that inhibiting necroptosis can effectively slow down the progression of periodontitis. Specifically, necroptosis of gingival fibroblasts mediated by RIPK3/MLKL further exacerbates the deterioration process of periodontitis; while inhibiting the expression of RIPK3 and MLKL can, to a certain extent, alleviate the symptoms of periodontitis [[54]13, [55]14]. In addition, RIPK3/MLKL can also intensify the periodontal inflammatory response by mediating necroptosis of periodontal ligament fibroblasts [[56]15]. In terms of the pathogenic mechanism, Porphyromonas gingivalis, the major pathogen of periodontitis, can secrete gingipains that act on macrophages, thereby mediating macrophage necroptosis and ultimately exacerbating the inflammatory response [[57]16]. Overall, current research on necroptosis in periodontitis primarily focuses on gingival fibroblasts, periodontal ligament fibroblasts, and macrophages [[58]13, [59]15, [60]16]. However, the overall status of necroptosis in gingival tissues during the onset of periodontitis has not been comprehensively and thoroughly elucidated; meanwhile, the expression patterns of necroptosis-related genes (NRGs) across different cell types in periodontal diseases need further exploration; additionally, the key genes mediating necroptosis in periodontitis also urgently need to be screened and identified. In this study, we integrated single-cell RNA sequencing (scRNA-seq) and transcriptome RNA sequencing (RNA-seq) data to successfully identify core NRGs and validated these findings using external datasets and periodontitis samples collected during our research. Additionally, we investigated the specific roles of NRGs in various periodontitis subtypes. Furthermore, we explored the major cell subpopulations expressing these core genes from a single-cell perspective and analyzed their complex interactions with the immune microenvironment to provide a new perspective for understanding the role of immune regulation in the disease process. Our research findings provide valuable insights into the pathogenesis of periodontitis and lay a solid scientific foundation for developing therapeutic interventions targeting necroptosis. Methods Study design The study design is presented in Fig. [61]1. Fig. 1. [62]Fig. 1 [63]Open in a new tab Flow chart of the study Obtaining and analyzing scRNA-seq data The [64]GSE171213 single-cell dataset was downloaded from the GEO database ([65]https://www.ncbi.nlm.nih.gov/geo/), and cells originating from five patients with periodontitis and four normal tissues were analyzed using the standard protocol outlined in Seurat version 4.4.0. Initially, we implemented rigorous data quality control measures, retaining only cells with a mitochondrial gene content below 20%, erythrocyte gene content below 3%, a minimum of 300 expressed genes, and genes expressed within a range of 300–7000 in at least three cells. Subsequently, we employed the R package “harmony” (version 1.2.0) to mitigate batch effects between samples [[66]17]. Using the “Find Variable Features” function, we identified the top 2000 variably expressed genes and then scaled all genes using the “Scale Data” function. We further refined our analysis by applying the “Run PCA” function to filter and reduce the dimensionality of these highly variable genes. Batch corrections were performed using the harmony algorithm. Cells were clustered using “Find Neighbors” and “Find Cluster” functions, with a resolution of 0.9, yielding 18 distinct clusters. To gain deeper insights into these clusters, we used the “Find All Markers” function to identify marker genes, setting the Minpct parameter to 0.25 (indicating the expression ratio of the least differentially expressed gene [DEG]). Marker genes with a corrected p-value of less than 0.05 were selected, and cell clusters were annotated by referencing the Cell Mark 2.0 database ([67]http://117.50.127.228/CellMarker/) [[68]18]. CellMarker database compiles a comprehensive list of cell-type-specific markers across different tissues and species, providing a reliable reference for our study [[69]18]. Furthermore, we also referred to the previously reported literature to annotate each cell type. Specifically, IGHG1, IGHG2, and IGKC can be used as markers for Plasma [[70]19, [71]20], KRT6A and KRT19 can be used as markers for epithelial cells [[72]19, [73]21], VWF, PECAM1 and AQP1 can be used as markers for endothelial cells [[74]19, [75]20], MS4A1 and CD19 can be used as markers for B cells [[76]19, [77]22], FCGR3B, CXCR2 and NAMPT can be used as markers for neutrophils [[78]19], CSF1R, C1QA, and C1QB can be used as markers for macrophages [[79]23, [80]24], TPSAB1 and TPSB2 can be used as markers for Mast cells [[81]24, [82]25], COL1A1, COL1A2 and COL3A1 can be used as markers for fibroblasts [[83]19], NKG7 and GNLY can be used as markers for NK cells [[84]24], and CD3D, CD3E and TRAC can be used as markers for T cells [[85]19, [86]20]. This meticulous process led to the identification of 11 distinct cell populations, including an “others” group that was excluded from subsequent analysis. Gene set variation analysis (GSVA) GSVA is an unsupervised and nonparametric framework for assessing gene set enrichment in the transcriptome. Through rank-based transformation of gene expression profiles, GSVA converts gene-level variations into pathway-level functional activity scores, thereby elucidating the biological pathways and functions of the samples. In this study, we obtained gene set collections from the Molecular Signature Database (MsigDB) version 7.5.1 and conducted a thorough evaluation of each gene set collection using the GSVA algorithm. This enabled us to investigate the potential variations in biological functions across samples. Acquisition of necroptosis-related genes Using the GeneCards database ([87]https://auth.lifemapsc.com/), we identified a panel of 114 NRGs with relevance scores greater than 1 (Table S1). Venn diagrams were used to identify overlaps between NRGs and DEGs identified in each cell type. Expression patterns of these overlapping genes in various cells were visualized using the “pheatmap” R package (version 1.0.12), generating a heatmap for clear representation. We then compared their expression levels in healthy and periodontal disease tissues using data from the GEO dataset and conducted statistical analyses using a t-test. P-values < 0.05 were considered significant. Single-cell data necroptosis correlation score In our analysis, the AddModuleScore function belongs to the Seurat v4.4 package. Specifically, the function calculates the average expression of a predefined gene set in each cell, normalized against a background of control gene sets. By default, Seurat generates 50 control gene sets, each matched in size and expression distribution to the target gene set. This adjustment corrects for technical variations, ensuring comparable scores across cells [[88]https://satijalab.org/seurat/reference/addmodulescore]. The AddModuleScore function was used to determine the mean expression levels of NRGs for each cell in the single-cell dataset. Subsequently, the cells were categorized into high- and low-expression groups based on the median expression values obtained [[89]26–[90]28]. Kyoto encyclopedia of genes and genomes (KEGG) and gene ontology (GO) enrichment analysis The R package “clusterProfiler” (version 4.10.1) was used to perform KEGG pathway enrichment analysis on DEGs associated with necroptosis to identify significantly enriched biological processes. Additionally, GO enrichment analysis was conducted to examine the DEGs linked to necroptosis across various functional dimensions. Fisher’s exact test was used to determine the significance of each enriched term. The threshold for significance in the enrichment analysis was set at p < 0.05, and the enrichment results were further illustrated using bubble plots for enhanced visualization. Bulk RNA sequencing (RNA-seq) data analysis We employed the R package “GEOquery” (version 2.70.0) from the GEO database to retrieve the periodontitis [91]GSE10334 datasets, which served as the training set. This comprehensive dataset encompassed gene expression profiles from 183 patients with periodontitis and 63 controls. For differential gene analysis, we used the R package “limma” (version 3.58.1). To distinguish DEGs between periodontitis and healthy samples within the dataset, we set a threshold of |log[2]fold change (FC)| >0.5 and p < 0.05. Differential gene expression data was visually represented using a volcano plot. In addition, we downloaded two separate periodontitis datasets, [92]GSE173038 and [93]GSE223924, as independent validation sets. Notably, we excluded some abnormal samples from the [94]GSE173038 dataset due to irregularities. Single sample gene set enrichment analysis (ssGSEA) and gene set enrichment analysis (GSEA) ssGSEA is a widely used approach for quantifying the enrichment score of a particular gene set within individual samples. The ssGSEA score for each sample reflects the degree to which a specific gene set is systematically upregulated or downregulated. In our study, we obtained the necroptosis score for each sample from dataset [95]GSE10334 by utilizing ssGSEA from the R package “GSVA” [[96]29]. To identify the potential pathways associated with the six hub genes, we divided the periodontitis samples into high- and low-expression groups based on the average expression level of each necroptotic gene. We calculated the GSVA scores for 40 KEGG pathways and analyzed pathways showing significant differences between the high- and low-expression groups using the “limma” package. Additionally, to further explore the potential mechanisms by which key genes influence periodontitis, we performed single-gene GSEA. Gene sets with a p < 0.05 and |NES| >1 were considered significantly enriched. Immune infiltration analysis CIBERSORTx is an analytical tool for immune cell infiltration analysis [[97]30]. The immune cell abundance was estimated using a reference set containing 22 immune cell subtypes and 50 permutations. Using the LM22 characteristic gene matrix, we screened out samples with p < 0.05 to obtain the immune cell infiltration matrix. Subsequently, data with immune cell enrichment fractions greater than zero were retained, and the final results of the immune cell infiltration matrix were obtained. Spearman correlation analysis was performed to assess relationships between the infiltrating immune cells and the six hub genes. Machine learning Through intersection analysis of DEGs with high necroptosis scores from single-cell data and DEGs from periodontitis tissues, we identified 29 genes with elevated expression in periodontitis tissues. To refine this list of potential genes for periodontitis diagnosis, we employed three machine-learning techniques. The least absolute shrinkage and selection operator (LASSO) is a regression analysis technique that performs both variable selection and regularization, enhancing model interpretability and prediction accuracy. It does so by imposing a penalty on the absolute size of the regression coefficients, effectively shrinking some coefficients to zero and thus selecting a simpler, more interpretable model with fewer features. This method is particularly useful when dealing with high-dimensional datasets, as it helps prevent overfitting and improves generalization [[98]31]. LASSO was utilized to identify a subset of genes with non-zero coefficients, indicating their relevance to periodontitis diagnosis. Support vector machine (SVM) is a powerful supervised learning algorithm capable of establishing boundaries between two classes in a high-dimensional space, enabling accurate classification based on one or more feature vectors. SVM-Recursive Feature Elimination (RFE) extends this capability by recursively eliminating less important features, thereby refining the feature set and improving model performance. This method excels in scenarios where the number of features exceeds the number of samples, as it effectively reduces dimensionality while preserving predictive power [[99]32]. SVM-RFE was applied to rank and select genes based on their importance in distinguishing periodontitis from healthy controls. By iteratively removing the least important features, we identified a core set of genes with high discriminatory ability. Random Forest is an ensemble learning method that constructs multiple decision trees and aggregates their predictions to improve accuracy, stability, and robustness. It handles non-linear relationships well, is robust to outliers and noise, and provides measures of feature importance [[100]33]. LASSO regression was executed using the “glmnet” package, SVM- RFE was implemented with the “e1071” and “caret” packages, and randomForest was applied using the “randomForest” package. Genes common to all three methods were considered hub necroptosis genes for periodontitis diagnosis. Additionally, we used the R package “corrplot” (version 0.92) to visualize the interactions between the hub necroptosis genes. To further assess the prediction model’s ability to distinguish periodontitis from healthy controls, we performed receiver operating characteristic (ROC) analysis using the R package “pROC” (version 1.18.5). Subclusters analysis with six necroptosis-related genes Using the R package “ConsensusClusterPlus“ [[101]34] and mRNA expression data of six NRGs as input, an unsupervised hierarchical clustering analysis was conducted on the 183 periodontitis samples. To ensure classification stability, we employed 80% item resampling and set the maximum evaluated k to six. Principal component analysis (PCA) was performed to validate the clustering results. GSVA was used to elucidate the functional distinctions among the necroptosis subclusters identified through prior cluster analysis. Next, a heat map was constructed to depict the differential activity of the pathways associated with the two subclusters of NRGs, while boxplots illustrated the differential expression of the six NRGs. Tissue source Ten patients with periodontitis and 10 healthy individuals were recruited from the Stomatological Hospital of Xiamen Medical College. None of the participants received medication before the procedure. Informed consent was obtained from all participants, and the study was approved by the Medical Ethics Committee of the Stomatological Hospital of Xiamen Medical College (HS20200908001). Extraction of RNA and RT-qPCR TRIzol^® reagent (Takara, Dalian, China) was used to extract total RNA from cultured cells and gingival tissue. cDNA was synthesized using the PrimeScript RT reagent kit (Takara). Quantitative real-time PCR (qPCR) was conducted using SuperReal PreMix Plus (SYBR Green, Tiangen Biotech Co., Ltd., Beijing, China) on an ABI 7500 Fast Real-Time PCR Detection system (Applied Biosystems, Foster City, CA, USA) using the primers listed in Table S2. Immunofluorescence Gingival tissue samples were initially fixed in formaldehyde (4%) for 24 h, subsequently dehydrated through a series of graded alcohols, embedded in paraffin wax, and incubated at 65 °C for 2 h to ensure proper infiltration. Sections were dewaxed in xylene and rehydrated in graded alcohols, after which they were subjected to heat-induced epitope retrieval using a citrate buffer (pH 6.0) in a pressure cooker for 10 min. After cooling, the slides were rinsed with phosphate-buffered saline. Nonspecific binding was blocked by incubating the sections with 5% normal goat serum in phosphate-buffered saline for 1 h at room temperature. Primary antibodies targeting CSF3R (1: 100, Cat#[102]C56229, Signalway Antibody LLC, Greenbelt, Maryland, USA), CSF2RB (1:50, Cat#AG1708, Beyotime, Shanghai, China), BTG2 (1:100, Cat#22339-1-AP, Proteintech, Wuhan, China), CXCR4 (1:100, Cat#11073-2-AP, Proteintech, Wuhan, China), GPSM3 (1༚100, Cat#45995, Signalway Antibody LLC, Greenbelt, Maryland, USA), and SSR4 (1:100, Cat#11655-2-AP, Proteintech, Wuhan, China) were employed. For the immunofluorescence staining, sections were initially incubated with a cocktail of primary antibodies overnight at 4 °C. After thorough washing, the sections were then incubated with a combination of species-specific secondary antibodies conjugated to different fluorophores: Alexa Fluor 488-conjugated goat anti-mouse IgG (1: 200, Cat#RGAM002, Proteintech, Wuhan, China) for BTG2 and GPSM3 (green), Alexa Fluor 568-conjugated goat anti-rabbit IgG (1:200, Cat# ab175471, Abcam, Shanghai, China) for CSF2RB and CSF3R (pink), and Alexa Fluor 647-conjugated goat anti-rabbit IgG (1:200, Cat#AS060, Abclonal, Wuhan, China) for SSR4 and CXCR4 (red). This multiplexing approach enables simultaneous visualization of multiple targets within the same tissue section. Statistical analysis All statistical analyses were performed using R software 4.3.2. Student’s t-test or Wilcoxon test was used to investigate differences between the two groups. Correlations between variables were assessed using the Pearson or Spearman parameters. Differences in means were considered statistically significant at p < 0.05, and the following statistical significance indicators were used: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Results Identification of cell types by scRNA-seq analysis We analyzed nine samples (comprising four normal and five periodontitis samples) retrieved from the scRNA-seq dataset [103]GSE171213. Through quality control, data integration, normalization, and batch-effect removal on single-cell sequencing data from patients with periodontitis and normal gingival tissues, we identified 21, 357 cells and 24, 540 genes. The integrated cells were categorized into 18 clusters using UMAP for dimensionality reduction and clustering (Fig. [104]2A). Based on the cell types present in periodontitis tissues [[105]19] and marker genes, we further classified the cells into B cells, endothelial cells, epithelial cells, fibroblasts, macrophages, mast cells, neutrophils, NK cells, plasma cells, and T cells, and a mixed group labeled “others.” For analytical clarity, we excluded the “others” group, retaining 10 distinct cell types (Fig. [106]2B). Fig. 2. [107]Fig. 2 [108]Open in a new tab Cell types in periodontitis tissues were identified by single cell analysis. A UMAP plots showing 17 clusters based on the transcriptomes of overall gene expression relationships among the 21,357 cells with an average of 24,540 genes. B UMAP distribution of cell types identified by marker genes. C Violin plots of the expression levels of the marker genes for the 10 cell types. D stacked bar plots showing the distribution of cell types in 5 periodontitis samples and 4 healthy samples. E Heatmap showing the scale normalized GSVA scores for select KEGG pathways in each annotated cell type. UMAP: uniform manifold approximation and projection; HC: healthy control; PD: periodontitis; KEGG: kyoto encyclopedia of genes and genomes Figure [109]2C presents the marker genes for each cell cluster (Table S3). A comparative analysis of cell cluster proportions across samples revealed a notable increase in endothelial cells, mast cells, and macrophages and a decrease in NK cells and T cells in periodontitis samples compared to healthy controls (Fig. [110]2D). The increased presence of endothelial cells, key components of blood vessels, suggests neovascularization or activation due to inflammatory responses in periodontitis. Activated endothelial cells exacerbate inflammation and immune cell infiltration [[111]35]. Through degranulation, mast cells release inflammatory mediators, such as histamine and leukotrienes, intensifying local inflammation [[112]36]. Macrophages, with their phagocytic and antigen-presenting abilities, clear necrotic tissues and pathogens and modulate immune responses through cytokine release [[113]37]. Alterations in cell proportions indicate the activation of inflammatory responses and immune cell imbalances in periodontitis. To investigate pathway activity changes among cell populations in periodontitis tissues, we conducted GSVA, identifying enriched gene sets for each cell type. A heat map of the 20 most significant pathways for each cell type emphasized differential pathway enrichment. Notably, in KEGG GSVA, TRYPTOPHAN_METABOLISM was markedly activated in macrophages, while BASAL_TRANSCRIPTION_FACTORS were inhibited; epithelial cells exhibited activation of BASAL_TRANSCRIPTION_FACTORS, GLYCOSPHINGOLIPID_BIOSYNTHESIS_LACTO_AND_NEOLACTO_SERIES, and GLUTATHIONE_METABOLISM; BIOSYNTHESIS_OF_UNSATURATED_FATTY_ACIDS were activated in mast cells; while FC_GAMMA_R_MEDIATED_PHAGOCYTOSIS and GLYCOSAMINOGLYCAN_BIOSYNTHESIS_CHONDROITIN_SULFATE were activated in neutrophils (Fig. [114]2E). These findings highlight the intricate changes in metabolism, biosynthesis, and immune function across different cell types in periodontitis tissues, which may be interconnected and contribute to the onset and progression of the disease. Scoring cell types based on the 35 expressed NRGs To explore the significance of necroptosis in periodontitis, we conducted an intersection analysis between 114 NRGs retrieved from the GeneCards database and marker genes specific to various cell populations. This identified 35 overlapping differentially expressed NRGs (Fig. [115]3A, Table S4). Further examination of the expression patterns of these 35 NRGs across cell clusters revealed pronounced differential expression in neutrophils. Specifically, HSPA5, JUN, HMGB1, and PTGES3 exhibited low expression levels, while MYD88, NFE2L2, SLC25A37, TLR4, PELI1, and FLOT2 were highly expressed in neutrophils (Fig. [116]3B). ZBP1, known for its role in triggering necroptosis by recognizing and binding to Z-DNA/Z-RNA [[117]38], was highly expressed in plasma cells, indicating its potential involvement in necroptosis in these cells. Fig. 3. [118]Fig. 3 [119]Open in a new tab Scoring cell types based on the expression 35 NRGs. A Venn diagram of NRGs and marker genes. B Heatmap showing the expression of 35 NRGs in each annotated cell type. C Bubble map showing KEGG analysis results of 35 NRGs. D Bubble map showing GO analysis results of 35 NRGs. E UMAP showing the score of the cell activity based on the 35 overlapping NRGs. Yellow indicates high necroptosis activity, while black indicates low necroptosis activity. F Violin plots showing the necroptosis score for each cell type. G Stacked bar plots showing the abundance of each cell type in the high and low necroptosis groups. NRGs: necroptosis-related genes; UMAP: uniform manifold approximation and projection Subsequently, we used GO/KEGG enrichment analysis to elucidate the potential biological relevance of these 35 NRGs in periodontitis. GO analysis revealed that these genes primarily regulate innate immunity and the NF-κB signaling pathway, critical for mediating inflammatory responses (Fig. [120]3C). KEGG analysis demonstrated that NRGs influence necroptosis, apoptosis, and multiple inflammatory signaling pathways, including the NF-κB, NLR, TNF, and TLR signaling pathways (Fig. [121]3D). These findings suggest a potential association between the 35 NRGs and periodontitis onset and progression. Considering the pivotal role of necroptosis in the progression of periodontitis, we assigned a necroptosis score to each cell based on the expression of these 35 NRGs, visualized using UMAP distribution (Fig. [122]3E). Our analysis revealed higher necroptosis scores in neutrophils and macrophages, whereas T cells exhibited lower scores (Fig. [123]3F). Notably, all cell types within the periodontitis group exhibited relatively high necroptosis scores (Fig. S1A). Stratifying cells into high- and low-necroptosis groups based on the median scores revealed a marked increase in the abundance of neutrophils and macrophages, accompanied by a decrease in T cells, in the high-necroptosis group (Fig. [124]3G). Furthermore, GSEA enrichment analysis based on differentially expressed genes between high- and low-necroptosis groups revealed significantly elevated activity of necroptosis pathways and inflammation-related pathways in the high-necroptosis group (Fig. S1B). Screening of key necroptosis-related genes through combined bulk RNA-seq analysis To identify pivotal genes involved in periodontitis, we conducted a combined bulk RNA-seq analysis, which yielded 522 DEGs (Fig. [125]4A). To further refine our focus on pinpointing key NRGs involved in periodontitis progression, we utilized the Findmarks function and obtained 522 DEGs from the high-necroptosis group in our single-cell analysis. By comparing these genes with the 408 DEGs derived from the bulk RNA-seq data, we identified 29 overlapping genes differentially expressed in both datasets (Fig. [126]4B, Table S5). Analysis of their expression patterns in periodontitis tissues revealed that most of these genes were upregulated (Fig. [127]4C-D). Fig. 4. [128]Fig. 4 [129]Open in a new tab Screening of key necroptosis-related genes through combined bulk RNA-seq analysis. A Volcano plots showing differentially expressed genes in the periodontitis-related dataset [130]GSE10334. Red represents up-regulated genes and pale blue represents down-regulated genes. B Venn diagram showing the overlap between differentially expressed genes in the hyper-necroptosis group and differentially expressed genes in periodontitis tissue. C Heatmap showing the expression of 29 overlapping genes in periodontitis tissues compared to healthy controls. D Boxplot showing the expression of 29 overlapping genes in periodontitis tissues compared to healthy controls. E Bar plot showing KEGG analysis results of 29 NRGs. F Bubble map showing GO analysis results of 29 NRGs. Adj.P: adjusted p value; Not sig: not significant; NRGs: necroptosis-related genes; DEGs: differentially expressed genes. ***p < 0.001 GO analysis indicated that these 29 overlapping DEGs were significantly associated with processes such as leukocyte migration, leukocyte chemotaxis, neutrophil migration, granulocyte migration, regulation of granulocyte differentiation, and secretory granule membrane function, which could potentially be involved in periodontitis progression (Fig. [131]4E). KEGG analysis further revealed that these NRGs were mainly enriched in pathways, including cytokine-cytokine receptor interactions, TNF signaling, JAK-STAT signaling, and prolactin signaling (Fig. [132]4F). Leukocyte migration and chemotaxis are crucial steps in the inflammatory response, involving the directed movement of cells towards inflammatory sites in response to inflammatory signals such as chemokines. Neutrophils and granulocytes are key effector cells in acute inflammatory responses that migrate to sites of inflammation and participate in pathogen clearance and damaged cell removal through phagocytosis [[133]39]. These DEGs may influence key molecules involved in immune cell function, such as cytokine receptors and signal transduction molecules, via the TNF, JAK-STAT, and prolactin signaling pathways, which could be related to periodontitis onset and progression. Consequently, the 29 overlapping DEGs were selected for further analysis. Screening of hub genes by machine learning and analysis of their association with immune infiltration To further identify the key genes involved in periodontitis progression, we conducted a screening process using machine-learning techniques, specifically LASSO regression, Random Forest, and SVM-RFE algorithms. Through LASSO regression analysis, we identified 15 feature genes associated with NRGs (Fig. [134]5A-B). Similarly, the SVM-RFE algorithm (Fig. [135]5C) and Random Forest algorithm also identified 15 feature genes (Fig. [136]5D). Subsequently, we intersected the feature genes obtained from these three algorithms to derive a final list of six overlapping genes, namely CSF3R, CSF2RB, BTG2, CXCR4, GPSM3, and SSR4 (Fig. [137]5E, Table S6), which were considered hub genes for further research. To predict periodontitis, we analyzed the ROC curves of these six feature genes. CXCR4 had the highest area under the curve (AUC) value (0.894), while the AUC values for CSF3R, CSF2RB, BTG2, GPSM3, and SSR4 were 0.87, 0.888, 0.826, 0.882, and 0.874, respectively (Fig. [138]5F). These results indicate that all six genes have potential diagnostic value. Fig. 5. [139]Fig. 5 [140]Open in a new tab Screening of hub genes by machine learning and analysis of their association with immune infiltration. A The coefficients of 29 overlapping genes over different values of the penalty parameter. B LASSO coefficient values of the 29 overlapping genes. The vertical dashed lines are the optimal log(λ) values. C DEGs profiles based on SVM-RFE algorithm. D DEGs profiles based on random forest algorithm. E Venn diagram showing the key overlapping genes obtained by the above three algorithms. F ROC curve of the 6 overlapping NRGs in periodontitis diagnosis. G Boxplot showing the expression of MLKL in periodontitis tissues compared to healthy controls. H Box plot showing the necroptosis scores in periodontitis samples and healthy control samples from the [141]GSE223924. I Mantel test correlation heatmap displaying the correlation between MLKL expression, necroptosis score and the expression of 6 hub genes. J Boxplot showing the difference of immune cell subtypes validated by CIBERSORT algorithm in periodontitis tissues compared to healthy controls. K Heatmap showing the relationship between the 6 overlapping NRGs and immune cell subtypes. RMSE: root mean squared error; AUC: area under the curve.*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 Next, we analyzed the expression changes of MLKL, the executor of necroptosis [[142]8], in periodontitis tissues and found that MLKL expression was significantly upregulated. These tissues also showed higher necroptosis scores (Fig. [143]5G-H). Additionally, we constructed a Mantel test correlation heatmap using the expression levels of MLKL, necroptosis scores, and the six hub genes. Both MLKL expression and necroptosis scores showed a tendency of positive correlation with the expression levels of these six hub genes, suggesting a possible association with necroptosis (Fig. [144]5I). Furthermore, we analyzed the infiltration of immune cells in periodontitis tissues and found a significant increase in plasma cell infiltration, with the six feature genes positively correlated with plasma cell infiltration. However, these genes were negatively correlated with the infiltration of memory B cells, activated and resting dendritic cells, M1 and M2 macrophages, resting mast cells, follicular helper T cells, and regulatory T cells (Fig. [145]5J-K). Overall, these six hub genes were associated with the degree of immune cell infiltration, indicating their important roles in the immune microenvironment of periodontitis tissues. Expression analysis of hub genes at the single-cell level To identify the cellular populations influenced by the six hub genes, we observed their expression patterns across various cell populations at the single-cell level. The results revealed that most hub genes were predominantly expressed in neutrophils. Specifically, CSF3R, CSF2RB, BTG2, and GPSM3 showed the highest expression levels in neutrophils. These six hub genes were also expressed in macrophages (Fig. [146]6A-B). This observation might offer a possible explanation for the relatively the higher necroptosis scores of neutrophils and macrophages. In contrast, CXCR4 was most abundantly expressed in T cells, while SSR4 exhibited the highest expression in plasma cells (Fig. [147]6B). These findings suggest that the six hub genes may have an impact on necroptosis in multiple cell types within periodontitis tissues, influencing the onset and progression of the disease. Fig. 6. [148]Fig. 6 [149]Open in a new tab Expression analysis of hub genes at the single-cell level. A UMAP showing the expression abundance of the six hub genes across various cells. B Bubble plot displaying the expression abundance of the six hub genes in different cells. UMAP: uniform manifold approximation and projection Consensus clustering of periodontitis based on the expression of six hub genes Based on the assessment of the six differential hub genes, we employed consensus clustering to investigate the patterns of necroptosis alterations in periodontitis. This analysis revealed distinct differences between two sample clusters when k = 2, leading to the establishment of two distinct clusters (Fig. [150]7A-C). As illustrated in the PCA plot, genes within these two subgroups exhibited different expression patterns (Fig. [151]7D). Furthermore, we compared the differences in the expression of the necroptosis executor MLKL and the six hub genes between the two subgroups and found that they were all highly expressed in Group 2 (Fig. [152]7E). This suggests that patients in Group 2 displayed the high necroptosis and may have more severe periodontitis. Fig. 7. [153]Fig. 7 [154]Open in a new tab Consensus clustering periodontitis based on expression of 6 hub genes. A Heatmap of consensus clustering analysis based on the 6 hub genes. B CDF curves showing consensus distributions from k = 2 to k = 6. C Area fraction under the CDF curve for k = 2–6. The horizontal axis indicats the number of categories (k), while the vertical axis indicats the relative changes in the area under the CDF curves. D PCA diagram showing the distribution of different subclusters. E Boxplot showing the expression of MLKL and 6 hub genes in Group 2 compared to Group 1. F Heatmap showing the scale normalized GSVA scores for select KEGG pathways in 2 subclusters. G Boxplot showing the difference of immune cell subtypes validated by CIBERSORT algorithm in Group 2 compared to Group 1. H Heatmap showing the correlation between infiltrated immune cells. I Heatmap showing the relationship between the 6 hub genes and immune cell subtypes in 2 subclusters. CDF: cumulative distribution function; Dim: dimension. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 Subsequently, we analyzed the differentially expressed pathways between the two subgroups using the GSVA function and found that the TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY, closely related to inflammation, was significantly more active in Group 2. Additionally, immune-related pathways such as PRIMARY_IMMUNODEFICIENCY and LEUKOCYTE_TRANSENDOTHELIAL_MIGRATION appeared to have increased activity in Group 2. Conversely, pathways like LINOLEIC_ACID_METABOLISM, METABOLISM_OF_XENOBIOTICS_BY_CYTOCHROME_P450, and HISTIDINE_METABOLISM were downregulated (Fig. [155]7F). By comparing the infiltration of immune cells between the two subgroups, we observed that plasma cells accounted for a large proportion of all immune cells and were significantly more infiltrated in Group 2, negatively correlating with the infiltration of most other immune cells. Moreover, the infiltration levels of B memory cells, monocytes, M1 and M2 macrophages, resting dendritic cells, and resting mast cells were significantly reduced in Group 2, with a negative correlation to plasma cell infiltration (Fig. [156]7G-H). Furthermore, correlation analysis between the hub genes and the degree of immune cell infiltration revealed that all six hub genes were positively correlated with plasma cell infiltration but negatively correlated with resting dendritic cells, M2 macrophages, resting mast cells, and follicular helper T cells infiltration (Fig. [157]7I). These results indicate distinct immune microenvironments between the two subgroups, and the six hub genes may be associated with immune cell infiltration. Construction of a diagnostic nomogram To explore the clinical application of NRGs, a clinical diagnostic model for periodontitis was constructed using the “rms” package in R, based on the expression levels of the six hub genes (CSF3R, CSF2RB, BTG2, CXCR4, GPSM3, and SSR4) in the training set ([158]GSE10334) (Fig. [159]8A). Subsequently, the diagnostic performance of the model was evaluated using calibration and ROC curve analyses. In the training set, the calibration curve demonstrated a minimal difference between the actual and predicted risks of periodontitis, highlighting the practicality of the model for predicting periodontitis (Fig. [160]8B). Furthermore, the ROC curve showed an AUC of approximately 0.92, suggesting a relatively high diagnostic performance (Fig. [161]8C). The diagnostic efficacy of the model was validated using independent datasets ([162]GSE173078 and [163]GSE223924). Notably, the AUC results from the calibration and ROC curves of these datasets were consistent with those of the training dataset (Fig. [164]8D–G). Therefore, the clinical diagnostic model based on the six hub genes may have a certain diagnostic performance and could potentially provide an effective reference for the prediction and treatment of periodontitis. Fig. 8. [165]Fig. 8 [166]Open in a new tab Construction of a diagnostic nomogram. A diagnostic nomogram was constructed using six hub genes. B The calibration curve evaluated the predictive ability of the nomogram model in the training set [167]GSE10334. C The clinical discriminative ability of the model was evaluated using an ROC curve in the training set [168]GSE10334. D-G The calibration curve and ROC curve of the model were evaluated using an independent validation set [169]GSE173078 and [170]GSE223924. PD: periodontitis; AUC: area under the curve; ROC: receiver operating characteristic Enrichment of signaling pathways for the six hub genes We analyzed the signaling pathways enriched by the six characteristic genes to investigate their potential molecular mechanisms influencing periodontitis progression. GSVA indicated that CSF3R, when highly expressed, primarily activates the CHEMOKINE_SIGNALING_PATHWAY, GLYCINE_SERINE_AND_THREONINE_METABOLISM, JAK_STAT_SIGNALING_PATHWAY, and TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY (Fig. [171]9A). CSF2RB, when highly expressed, mainly activates the CHEMOKINE_SIGNALING_PATHWAY, N_GLYCAN_BIOSYNTHESIS, TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY, JAK_STAT_SIGNALING_PATHWAY, APOPTOSIS, and NOD_LIKE_RECEPTOR_SIGNALING_PATHWAY (Fig. [172]9B). BTG2, when highly expressed, primarily activates LEUKOCYTE_TRANSENDOTHELIAL_MIGRATION, N_GLYCAN_BIOSYNTHESIS, and MAPK_SIGNALING_PATHWAY (Fig. [173]9C). CXCR4, when highly expressed, predominantly activates LEUKOCYTE_TRANSENDOTHELIAL_MIGRATION, N_GLYCAN_BIOSYNTHESIS, CHEMOKINE_SIGNALING_PATHWAY, and JAK_STAT_SIGNALING_PATHWAY (Fig. [174]9D). GPSM3, when highly expressed, primarily activates the CHEMOKINE_SIGNALING_PATHWAY, HEMATOPOIETIC_CELL_LINEAGE, JAK_STAT_SIGNALING_PATHWAY, and TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY (Fig. [175]9E). SSR4, when highly expressed, mainly activates the HEMATOPOIETIC_CELL_LINEAGE, CHEMOKINE_SIGNALING_PATHWAY, GLYCINE_SERINE_AND_THREONINE_METABOLISM, JAK_STAT_SIGNALING_PATHWAY, and TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY (Fig. [176]8F). Fig. 9. [177]Fig. 9 [178]Open in a new tab GSVA analysis of pathways activated upon up- and down-regulation of CSF3R, CSF2RB, BTG2, CXCR4, GPSM3, SSR4. A GSVA analysis of CSF3R; B GSVA analysis of CSF2RB; C GSVA analysis of BTG2; D GSVA analysis of CXCR4; E GSVA analysis of GPSM3; F GSVA analysis of SSR4. These panels show differences in pathway activities scored by GSVA between high and low expression levels (from top to bottom) of specific genes. The GSVA scores, plotted on the X-axis, are sorted in descending order based on their rank, reflecting the significant pathway enrichment levels of the gene set. The blue and green colors indicate significantly enriched pathways. GSVA: gene set variation analysis; LExp: low expression; HExp: high expression GSEA indicated that CSF3R is enriched in NF-κB signaling pathway, chemokine signaling pathway, cell cycle, and tyrosine metabolism (Fig. [179]10A); CSF2RB is enriched in toll-like receptor signaling pathway, TNF signaling pathway, beta-alanine metabolism, and tight junction (Fig. [180]10B); BTG2 is enriched in leukocyte transendothelial migration, apoptosis, steroid biosynthesis, and fatty acid metabolism (Fig. [181]10C); CXCR4 is enriched in the chemokine signaling pathway, TNF signaling pathway, fatty acid degradation, and histidine metabolism (Fig. [182]10D); GPSM3 is enriched in the chemokine signaling pathway, hematopoietic cell lineage, fatty acid degradation, and cell cycle (Fig. [183]10E); SSR4 is enriched in cytokine-cytokine receptor interaction, N-glycan biosynthesis, peroxisome, and histidine metabolism (Fig. [184]10F). These findings suggest that these hub genes may have an impact on periodontitis progression by potentially regulating tyrosine, beta-alanine, histidine, and fatty acid metabolism inflammatory responses mediated by the NF-κB, TLR, and TNF signaling pathways, as well as cell cycle. Fig. 10. [185]Fig. 10 [186]Open in a new tab GSEA analysis of KEGG Gene Sets. A GSEA analysis of CSF3R-up and CSF3R-down. B GSEA analysis of CSF2RB-up and CSF2RB-down. C GSEA analysis of BTG2-up and BTG2-down. D GSEA analysis of CXCR4-up and CXCR4-down. E GSEA analysis of GPSM3-up and GPSM3-down. F GSEA analysis of SSR4-up and SSR4-down. NES: normalized enrichment score Validation of external data sets of hub genes To further validate the differential expression of these six necroptosis-related hub genes in periodontitis tissues, we detected their differential expression in the independent validation sets [187]GSE173038 and [188]GSE223924 and found that the expression of all six hub genes was significantly upregulated in periodontitis tissues. Furthermore, the periodontitis group exhibited higher necroptosis scores (Fig. [189]11A-D). Fig. 11. [190]Fig. 11 [191]Open in a new tab Validation of external data sets of hub genes. A Boxplot showing the differential expression of six hub genes in periodontitis samples and healthy control samples from the independent validation set [192]GSE173038. B Boxplot showing the necroptosis scores in periodontitis samples and healthy control samples from the independent validation set [193]GSE173038. C Boxplot showing the differential expression of six hub genes in periodontitis samples and healthy control samples from the independent validation set [194]GSE223924. D Boxplot showing the necroptosis scores in periodontitis samples and healthy control samples from the independent validation set [195]GSE223924. E-J ROC curves of the six hub genes in the independent validation set [196]GSE173028. AUC: area under the curve. *p < 0.05; **p < 0.01; ***p < 0.001 Next, we investigated the diagnostic value of these six hub genes in periodontitis and normal tissue samples. In the training dataset, all NRGs demonstrated high diagnostic value, as indicated by the AUCs, with all AUCs greater than 0.8 (Additional file 2: Fig. S2A). CSF3R (AUC: 0.87), CSF2RB (AUC: 0.888), BTG2 (AUC: 0.826), CXCR4 (AUC: 0.894), GPSM3 (AUC: 0.882), and SSR4 (AUC: 0.874). This trend was confirmed in the independent validation set [197]GSE110169, where CASP3 had an AUC of 0.944, followed by CSF2RB (AUC: 0.917), BTG2 (AUC: 0.819), CXCR4 (AUC: 0.847), GPSM3 (AUC: 0.917), and SSR4 (AUC: 0.931) (Fig. [198]11E-J). Results from the independent dataset [199]GSE223924 further supported these findings (Additional File 2: Fig. S2B). These results suggest that these six hub genes have great potential for periodontitis diagnosis and treatment. Validation of hub genes in clinical samples To explore the potential clinical applications of the six hub genes, we conducted a comparative analysis of their expression levels in the gingival tissues of patients with periodontitis and healthy controls. Total RNA was extracted from the gingival tissues of 10 patients with periodontitis and 10 healthy controls. RT-qPCR analysis clearly demonstrated that all selected NRGs were significantly upregulated in periodontitis tissues compared to healthy controls (Fig. [200]12A-F). Fig. 12. [201]Fig. 12 [202]Open in a new tab Validation of hub genes in clinical samples. A-F The mRNA expression of CSF3R, CSF2RB, BTG2, CXCR4, GPSM3 and SSR4 in periodontitis human gingival tissue (n = 10) and normal gingival epithelial tissue (n = 10) were detected by RT-qPCR. G-H The expression of CSF3R, CSF2RB, BTG2, CXCR4, GPSM3, and SSR4 in gingival tissues from periodontitis patients and healthy controls were detected by Multiplex immunofluorescence. scale bar 500 μm. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 To further enhance the reliability of these findings, multiplex immunofluorescence was used. The results similarly indicated that these six hub genes exhibited high expression patterns in the gingival tissues of patients with periodontitis (Fig. [203]12G-H). By combining the RT-qPCR and multiplex immunofluorescence results, we confirmed that the mRNA and protein expression levels of these six hub genes in clinical samples of periodontitis were highly consistent with the results previously obtained from the training dataset and independent validation set. These findings strongly suggest that these six hub genes represent potential therapeutic targets in periodontitis progression. Discussion Necroptosis, a programmed cell death mode distinct from traditional apoptosis, is characterized by cell membrane rupture and the release of inflammatory factors, a process that significantly activates the body’s immune response [[204]7]. Previous studies that primarily focused on necroptosis in gingival fibroblasts, periodontal ligament fibroblasts, and macrophages, lacking a comprehensive depiction of the necroptosis landscape in gingival tissue affected by periodontitis. This study innovatively combines single-cell sequencing with large-scale RNA sequencing analysis techniques to provide a comprehensive and systematic portrayal of the complex situation of cellular necroptosis in periodontitis. In this study, 114 NRGs were meticulously selected from the GeneCards database and subjected to in-depth cross-analysis with specific marker genes of various cell populations. Through this rigorous analytical process, 35 overlapping and differentially expressed NRGs were successfully identified. These 35 NRGs are primarily enriched in innate immune responses and signaling pathways related to inflammatory responses, including the NF-κB signaling pathway, NLR signaling pathway, TNF signaling pathway. This suggests that these NRGs may remodel the periodontal immune microenvironment and drive tissue inflammatory responses by mediating necroptosis. Further analysis revealed that these NRGs were differentially expressed in neutrophils, including TLR4 and MYD88, which are closely associated with necroptosis. Studies have shown that TLR4/MyD88 may be involved in the transduction of necroptotic signals in astrocytes [[205]40]. Additionally, the activation of the TLR4/MyD88/NF-κB pathway induces necroptosis in L8824 cells [[206]41]. These findings suggest that the TLR4/MyD88 pathway in neutrophils is activated in periodontitis tissues and induces necroptosis in neutrophils. Therefore, targeted intervention of the TLR4/MyD88 pathway may be a new strategy for inhibiting neutrophil necroptosis and subsequently inhibiting periodontitis progression. However, this hypothesis must be validated in subsequent experimental studies. Further intersection analysis was conducted between the 114 NRGs and marker genes of various cell populations in single-cell data, leading to the identification of 35 key genes. Based on these 35 intersecting genes, necroptosis scores were calculated for each cell population in the single-cell dataset [207]GSE171213. It was found that neutrophils and macrophages exhibited high necroptosis scores, with neutrophils having the highest score and the largest number. We also found that the proportions of neutrophils and macrophages are significantly increased in the high necroptosis score group, highlighting their central role in necroptosis. Neutrophils, as core effector cells of the innate immune system and the most abundant leukocytes in the circulatory system, play a crucial role in defending against bacterial and fungal pathogens [[208]42]. However, their potent antibacterial efficacy can become detrimental to host tissues, and improper regulation may trigger self-mediated diseases [[209]43]. Although the necroptosis of neutrophils has not been a focal point of research, existing studies have shown that inhibiting this process can effectively curb the progression of various inflammatory diseases. For example, Tim-3-knockout macrophages recruit and induce necroptosis of neutrophils, subsequently disrupting the intestinal mucosal barrier by releasing DAMPs and triggering colitis. Pharmacological inhibition of RIP1 or RIP3 alleviates this pathological process [[210]44]. In Staphylococcus aureus pneumonia, PSMα-induced necroptosis is a key factor in lung pathological changes, and the RIP1 inhibitor Nec reduces lung tissue damage by inhibiting necroptosis of neutrophils, thereby improving survival in infected mice [[211]45]. Macrophages, major producers of inflammatory factors, play a pivotal role in the initiation and progression of inflammation. Extensive research indicates that necroptosis in macrophages promotes the release of inflammatory factors, triggering or exacerbating inflammatory responses; conversely, inhibiting this process effectively controls inflammation spread. RIPK3 kinase activity is essential for TREM-1 activation-induced macrophage necroptosis, and the RIPK3 inhibitor GSK872 significantly inhibits this process and reduces the secretion of TNF-α and IL-1β by macrophages [[212]46]. Furthermore, the PTRF-IL33-ZBP1 signaling pathway-mediated necroptosis of macrophages contributes to HDM-induced airway inflammation. As a key regulator of necroptosis, RIPK1 mutation (K45A) impairs kinase function, inhibiting macrophage necroptosis and inflammatory response activation. Mice carrying this mutation exhibit enhanced resistance and higher survival rates following lipopolysaccharide treatment [[213]47]. These studies underscore the significance of inhibiting neutrophil and macrophage necroptosis in the treatment of inflammatory diseases, suggesting that inhibiting necroptosis in these two cell types may effectively suppress the malignant progression of periodontitis. To further explore the relationship between necroptosis and periodontitis, we analyzed bulk RNA-seq data and identified MLKL, the executor of necroptosis, as highly expressed in periodontitis tissues. Using ssGSEA, we further confirmed elevated necroptosis scores in these tissues, supported by external validation sets. These findings further confirm the close association between necroptosis and periodontitis occurrence and development. To identify NRGs associated with periodontitis more accurately, we employed machine learning methods and successfully identified six key genes: CSF3R, CSF2RB, BTG2, CXCR4, GPSM3, and SSR4. All six genes were highly expressed in periodontitis tissues, and their expression levels were significantly and positively correlated with MLKL expression and necroptosis scores. We also experimentally validated the high expression patterns of these six hub genes in periodontitis tissues. Further analysis of the expression levels of these six genes in various cell populations in periodontitis tissues revealed that CSF3R, CSF2RB, BTG2, and GPSM3 were particularly abundant in neutrophils, with all six genes also expressed in macrophages. These findings partially explain why neutrophils and macrophages exhibit high necroptosis scores in periodontitis. Unfortunately, the associations between the six hub genes we have identified and necroptosis have not been reported in the literature, and their potential connections also remain unclear. CSF3R, a tyrosine kinase receptor, binds granulocyte colony-stimulating factor (G-CSF), promoting the proliferation and differentiation of granulocytes in the bone marrow, thereby effectively increasing granulocyte counts in peripheral blood [[214]48]. In addition to regulating granulopoiesis, CSF3R is actively involved in granulocyte activation. During infections or injuries, G-CSF levels rise sharply, and upon binding to CSF3R, granulocytes are activated and rapidly migrate to inflammation sites to perform essential functions such as phagocytosis and sterilization [[215]49, [216]50]. Granulocytes, as the core cells of the inflammatory response, eliminate pathogens, damaged cells, and tissue debris and finely regulate the extent and scope of the inflammatory response by releasing inflammatory mediators and chemokines. Given the central role of CSF3R in granulopoiesis and activation, it is closely associated with the initiation and progression of inflammation. Dysfunction of CSF3R can affect normal granulocyte generation and activation, leading to abnormal inflammatory responses or immunodeficiency. Neutrophils, as important granulocyte subsets, express high levels of CSF3R on their cell membrane surfaces, making them highly responsive to elevated G-CSF levels in periodontal tissues. The CSF3R structure includes an extracellular ligand-binding domain, a transmembrane region, and an intracellular tyrosine kinase domain. When G-CSF binds to CSF3R, it triggers receptor dimerization and autophosphorylation, subsequently activating downstream signal transduction pathways [[217]48]. GSVA enrichment analysis showed significant enrichment of the CHEMOKINE_SIGNALING_PATHWAY, JAK_STAT_SIGNALING_PATHWAY, and TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY in the CSF3R high-expression group. Both the JAK/STAT [[218]51] and TLR signaling pathways are closely related to necroptosis, suggesting that CSF3R may enhance the transmission of these two signaling pathways, inducing neutrophils necroptosis and exacerbating periodontitis severity. CSF2RB, also known as colony-stimulating factor 2 receptor β, is a multi-colony stimulating factor receptor that serves as a common β subunit for cytokines like GM-CSF, IL-3, and IL-5, mediating signal transduction processes [[219]52]. Upon GM-CSF binding, CSF2RB activates downstream signaling pathways like JAK/STAT and MAPK, precisely regulating the expression of inflammation-related genes and the activation status of inflammatory cells. Aberrant CSF2RB expression or dysfunction is closely associated with various inflammatory diseases. For instance, in Alzheimer’s disease, high CSF2RB expression is observed in peripheral blood monocytes from both patients and rats injected with Aβ in the brain [[220]53], which may be directly related to the inflammatory response in the brain. CSF2RB is also involved in the pathogenesis of inflammatory diseases such as glomerulonephritis [[221]54]. GSVA further revealed significant enrichment of the TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY, JAK_STAT_SIGNALING_PATHWAY, and NOD_LIKE_RECEPTOR_SIGNALING_PATHWAY in the CSF2RB high-expression group, suggesting that CSF2RB induces necroptosis by enhancing the transmission of these signaling pathways. The protein encoded by the BTG2 gene belongs to the BTG/Tob family, which share structural similarities and exhibit antiproliferative properties, participating in the regulation of the G1/S cell cycle transition [[222]55]. Although there is currently no direct evidence that BTG2 directly participates in or regulates the inflammatory process, our study is the first to validate the high expression of BTG2 in gingival tissues of patients with periodontitis, preliminarily revealing a potential connection between BTG2 and inflammation. Although GSVA showed that BTG2 is associated with leukocyte migration and MAPK signaling pathway transmission, further experimental validation is needed to confirm its influence on necroptosis via the MAPK pathway. GPSM3, encoding G-protein signaling modulator 3, plays a pivotal role in intracellular signaling networks. Studies have shown that GPSM3 directly binds to NLRP3, inhibiting the secretion of the inflammatory cytokine IL-1β triggered by NLRP3 in bone marrow-derived macrophages [[223]56]. GPSM3-knockout mice exhibited milder symptoms in collagen antibody-induced arthritis models, and monocytes from these mice showed suppressed expression of pro-inflammatory chemokine receptors and cytokines [[224]57]. These findings highlight the unique anti-inflammatory role of GPSM3. However, research has also found that neutrophils lacking GPSM3 have significantly reduced migration ability to primary inflammation sites [[225]58]. Our single-cell analysis indicated that GPSM3 was highly expressed in neutrophils in periodontal tissue and confirmed its overall high expression in periodontal tissue. GPSM3 is enriched in the CHEMOKINE_SIGNALING_PATHWAY, HEMATOPOIETIC_CELL_LINEAGE, JAK_STAT_SIGNALING_PATHWAY, and TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY, suggesting that in periodontal tissue, GPSM3 may play a pro-inflammatory role by promoting neutrophil chemotaxis to the inflammatory site and possibly inducing necroptosis of neutrophils via the JAK/STAT and TLR signaling pathways, exacerbating the inflammatory response. CXCR4, a G-protein-coupled receptor, plays a complex dual role in periodontitis by binding to the specific ligand CXCL12 (also known as SDF-1, stromal cell-derived factor-1). CXCL12 expression is upregulated in periodontal tissues, attracting periodontal ligament stem cells expressing CXCR4 to inflammation sites to aid in the repair and regeneration of periodontal tissues [[226]59, [227]60]. Conversely, the binding of CXCR4 to CXCL12 can also guide neutrophils to migrate directionally along the CXCL12 concentration gradient, exacerbating inflammatory damage in periodontal tissues [[228]61]. This dual role of CXCR4 appears to be closely related to the cell type in which it is expressed. Currently, CXCR4 inhibitors have demonstrated significant therapeutic effects in the treatment of various inflammatory diseases, including skin inflammation [[229]62] and neuroinflammation [[230]63], suggesting their potential therapeutic efficacy in periodontitis as well. Our analysis showed that CXCR4 is mainly expressed in immune cells like T cells, NK cells, macrophages, and B cells, with altered immune cell abundance in periodontal tissues, suggesting that CXCR4 is intricately linked to the remodeling of the immune microenvironment in periodontal tissues. SSR4, a member of the TRAP-delta family, is the delta subunit of the transporter-associated protein complex involved in protein transport across the endoplasmic reticulum (ER) membrane. Endoplasmic reticulum stress is an important trigger for inflammatory responses, and SSR4 knockout can effectively attenuate the transcriptional signature of endoplasmic reticulum stress [[231]64]. However, endoplasmic reticulum stress may also lead to the accumulation of unfolded or misfolded proteins, further activating inflammatory signaling pathways, including necroptosis. ER stress leads to the release of calcium ions into mitochondria and the cytoplasm. Elevated Ca^2+ levels can cause mitochondrial dysfunction and excessive production of reactive oxygen species (ROS), thereby promoting RIPK3-dependent necroptosis [[232]65]. Furthermore, ER stress-mediated mitochondrial calcium overload and the subsequent generation of ROS can also promote the opening of the mitochondrial permeability transition pore (mPTP), thereby inducing necroptosis in cardiomyocytes [[233]66]. Given the close association of SSR4 with ER stress, it is suggested that its high expression pattern in periodontitis tissue may drive the occurrence of necroptosis by mediating ER stress. Necroptosis is frequently regulated by tissues or the microenvironment. Although this study, through single-cell analysis, found that neutrophils and macrophages exhibited high necroptosis scores, these findings have not yet been validated. Subsequent validation using multi-immunofluorescence and flow cytometry is necessary. Additionally, although we have identified six hub genes and preliminarily verified the upregulation of CSF3R, CSF2RB, BTG2, CXCR4, GPSM3, and SSR4 in periodontal tissues, the sample size remains relatively limited. Future studies should expand the sample size to enhance the stability and reliability of the results. In future research, we also need to directly isolate individual cell populations from periodontal tissues to verify whether these hub genes exhibit differential expression in specific cell populations. Subsequently, we can further overexpress or knock down these hub genes in specific cell populations to investigate whether they can influence necroptosis in those populations, thereby mediating periodontal inflammatory responses. Moreover, we can use small-molecule inhibitors of the corresponding hub genes in vivo or generate conditional knockout mice for these genes to explore the therapeutic potential of targeting these hub genes in periodontitis. These efforts will enable a more comprehensive assessment of their roles in cell necroptosis and the pathogenesis of periodontitis. Conclusions This study systematically unveiled the necroptosis landscape in periodontitis and identified six key NRGs. These six hub genes are all highly expressed in periodontitis tissues and exhibit a positive correlation with necroptosis. Targeting these genes may be a novel therapeutic strategy for managing periodontitis. Supplementary Information [234]12920_2025_2241_MOESM1_ESM.xlsx^ (528.5KB, xlsx) Additional file 1: Table S1. Primers for amplification of indicated genes. Table S2. 114 necroptosis-related genes from the GeneCards database. Table S3. Marker genes for various cell populations. Table S4. Intersection genes between the 114 necroptosis-related genes and various cell populations. Table S5. Intersection of differentially expressed genes in the high necroptosis score group and differentially expressed genes in the GSE10334 dataset. Table S6. Intersection of genes selected by LASSO, SVM-RFE, and RF algorithms. Table S7. The GEO dataset used in this study. [235]12920_2025_2241_MOESM2_ESM.docx^ (1.2MB, docx) Additional file 2: Fig. S1. Analysis of necroptosis score in the periodontitis group and analysis of pathway activity changes in the group with high necroptosis score. Fig. S2. ROC curves of the six hub genes in the training set and independent validation set. Table 1. Demographic characteristics of the study population and clinical characteristics of the sampling sites for mRNA level analysis. Abbreviations AUC area under the curve GO gene ontology GSEA gene set enrichment analysis GSVA gene set variation analysis G-CSF granulocyte colony-stimulating factor KEGG kyoto encyclopedia of genes and genomes LASSO least absolute shrinkage and selection operator MLKL mixed lineage kinase domain-like pseudokinase MsigDB molecular signature database qPCR quantitative real-time PCR ROC receiver operating characteristic ssGSEA single sample gene set enrichment analysis scRNA-seq single-cell RNA sequencing SVM support vector machine RNA-seq transcriptome RNA sequencing DAMPs damage-associated molecular patterns NRGs necroptosis-related genes GEO gene expression omnibus DEGs differentially expressed genes RF RandomForest PCA principal component analysis Authors’ contributions Conceptualization, M. X., F.L., S. S., and X. D.; methodology, F. Z., Y. G.,Y.L.,R.H.,and X. D.; formal analysis, M. X., F. Z., Y. G.Y.L,, and R.H.; investigation, M. X., F. Z, Y.G.,Y.L., and R.H.; writing—original draft preparation, M. X., F. Z. and X. D.; writing—review and editing, M. X., Q. W. and X. D. Funding This research was supported by National Natural Science Foundation of China (No 82270988, 82071128 and 82301047), Xiamen Natural Science Foundation Project (NO. 3502Z20227291). Data availability The data that support the findings of our study were derived from the following resources available in the public domain: Gene Expression Omnibus at [236]https://www.ncbi.nlm.nih.gov/geo/ ([237]GSE171213, [238]GSE10334, [239]GSE173038 and [240]GSE223924) and the GeneCards: The Human Gene Database at [241]https://www.genecards.org/. Declarations Ethics approval and consent to participate The clinical operation involved in this paper was told to the patients and got the informed consent from the patients and totally adhered to the Declaration of Helsinki. The present study was approved by the Ethics Committee of Stomatological Hospital of Xiamen Medical College, and all authors and patients have read and agreed to the published version of the manuscript. Consent for publication Not Applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Feixiang Zhu and Mingyan Xu contributed equally to this work. Contributor Information Qianju Wu, Email: wuqianju@sjtu.edu.cn. Xiaoling Deng, Email: xiaolingdeng@xmu.edu.cn. References