Construction of a single-cell resolution spatial transcriptomics atlas of human gingivae

To construct a single-cell resolution spatial transcriptomics atlas, gingivae were collected from eight patients with periodontitis, four with periodontitis (P), four with smoking-associated periodontitis (SP), and healthy gingivae (HG) were collected from four healthy volunteers. Formalin-fixed paraffin-embedded (FFPE) tissue blocks were generated from these samples (P = 4, SP = 4, HG = 4). Sections of the FFPE tissue blocks were processed for Visium HD-based analysis to investigate the effects of smoking on the periodontal tissue microenvironment.

The Visium HD platform allows for high-resolution spatial gene expression analysis using single-cell resolution probes targeting the entire transcriptome. Visium HD slides are structured with a reference frame of 8 × 8 mm, containing a capture area of 6.5 × 6.5 mm. Each capture region comprises approximately 11 million 2 × 2 µm squares arranged in a continuous array of unique barcode oligonucleotides. These 2 µm squares are directly adjacent to each other, forming a continuous and gap-free capture region (Fig. 1a). The high-resolution data generated by Visium HD were used to perform unsupervised clustering, identifying distinct cell subpopulations and mapping those populations onto the morphological features of the gingival tissue (Fig. 1b).

Fig. 1

Construction of a single-cell resolution spatial transcriptomic atlas of human gingivae. a Schematic diagram of the single-cell resolution spatial transcriptomic analysis. b Image of H&E-stained gingival tissue and cell clustering of Visium HD data. c Identification of 7 cell types in the gingival tissues of the healthy gingivae (HG), periodontitis (P) and smoking-associated periodontitis (SP) groups. d UMAP plots showing single-cell clustering of the gingival tissue, displaying 7 major cell types. e Dot plot showing the expression of specific marker genes in 7 major cell types. The colour of the dot represents the expression level of the marker genes in each cell. High expression levels are shown in brown, and low expression levels are shown in blue. f GO analysis depicting the functions of the marker genes for each cell type

The cells were categorized into four main groups—epithelial cells, fibroblasts, endothelial cells, and immune cells—through integration and clustering techniques. Subsequent single-cell transcriptomic analysis combined with spatial transcriptomics allowed further classification of immune cells into T cells, plasma cells, macrophages, and neutrophils, which were then projected onto the gingival tissue morphology (Fig. 1c). Cluster analysis was conducted, and the cell populations were visualized on a uniform manifold approximation and projection (UMAP) plot (Fig. 1d and Supplementary Fig. 1) using data from 250 000 cells after quality control.

Marker gene expression levels were assessed for each cell group and depicted in dot plots using R software to identify specific marker genes for each cell type: epithelial cells (KRT4 and KRT6B), fibroblasts (COL1A1 and COL12A1), endothelial cells (VWF and AQP1), plasma cells (IGHG1 and IGKC), macrophages (CD68 and LYZ), neutrophils (FCGR3B and MPO), and T cells (TRAC and CD3D) (Fig. 1e). A pathway enrichment analysis was conducted on the marker genes within each cell group. Marker genes of epithelial cells were associated with structural components of the skin epidermis and cell–cell adhesion mediator activity; fibroblast marker genes were linked to extracellular matrix and collagen binding; endothelial cell marker genes were related to growth factor binding and platelet-derived growth factor binding; plasma cell marker genes were related to antigen binding and immunoglobulin receptor binding; macrophage marker genes were associated with MHC class II protein complex binding and cysteine-type endopeptidase activity; neutrophil marker genes were related to chemokine activity and actin monomer binding; and T cell marker genes were linked to immune receptor activity and T cell receptor binding(Fig. 1f). These results were consistent with the typical functional pathways of these cell subpopulations. The single-cell resolution spatial transcriptomics atlas of human gingivae obtained here provides the basis for further analyses of cellular changes under smoking conditions.

Smoking-induced damage in gingival epithelial cells

Exposure to smoking-related toxins has a direct effect on the integrity of the oral mucosal epithelial barrier, increasing vulnerability to injury. Gingival epithelial cells were classified into four distinct subgroups, and their correlation with tissue morphology was analysed (Fig. 2a). Utilizing clustering data, distinct marker genes were identified for each epithelial subgroup within the gingival tissue morphology: Cornulin (CRNN) for Epi-1, Keratin 6B (KRT6B) for Epi-2, Ly-6 Domain Containing 1 (LY6D) for Epi-3, and Keratin 5 (KRT5) for Epi-4 (Fig. 2b). These genes are essential for maintaining epithelial function and integrity. For instance, CRNN is a crucial epithelial cell protein predominantly found in tissues like the skin, oral mucosa, and gastrointestinal tract mucosa, where it has protective, barrier, and repair functions.18,19KRT6B, a member of the keratin family, forms network structures with other keratin proteins to uphold cellular mechanical strength and stability, which are particularly vital for barrier function and resilience against external stress.20,21LY6D, belonging to the Ly-6 superfamily, encodes a membrane-bound protein involved in cell signalling, emphasizing its role in preserving epithelial homoeostasis.22,23KRT5, another keratin family member, is mainly expressed in basal layer cells, where it plays a crucial role in providing structural stability and ensuring epithelial integrity under conditions of friction and stretching.24,25 Furthermore, immunofluorescence staining was conducted to validate the spatial localization of these molecular markers in the tissue, confirming consistency with the sequencing results (Supplementary Fig. 2). This validation reinforces the accuracy and biological significance of the single-cell RNA sequencing data and further elucidates the specific expression patterns and functional attributes of these markers in distinct epithelial subgroups.

Fig. 2

Smoking-induced damage in gingival epithelial cells. a Spatial map of the four subgroups of epithelial cells in the Visium HD date. b Spatial map of marker gene expression in the four cell subgroups. c Dot plots showing upregulated pathways in epithelial cells from the SP group. d Visium HD data analysis revealing the expression of KRT1, DSG1, PI3, and GPX2 in the gingival tissue; scale bar = 200 µm. e RT–qPCR analysis of the KRT1, DSG1, PI3, and GPX2 expression levels in epithelial cells treated with LPS and nicotine (n = 3). *P < 0.05

To investigate the alterations in the gene expression of epithelial cells induced by smoking, we identified DEGs in the gingival epithelial cells of the P and SP groups. Gene Ontology (GO) pathway enrichment analysis revealed that, compared with chronic periodontitis, smoking upregulated the expression of genes associated with pathways such as Staphylococcus aureus infection, oxidative stress, the humoral immune response, and keratinized membrane formation in the gingival epithelium (Fig. 2c). Key genes such as Keratin 1 (KRT1), Desmoglein 1 (DSG 1), Phosphoinositide 3-kinase (PI3), and Glutathione Peroxidase 2 (GPX2) are crucial genes in the above pathways. These genes are essential for maintaining the structure and function of epithelial cells, connecting epithelial cells, and participating in the stress response and inflammation signalling, as well as the response to oxidative stress. To confirm the spatial expression differences of these pathways in the gingival epithelium, we further mapped these genes in both chronic periodontitis tissues and smoking-associated periodontitis tissues (Fig. 2d), revealing statistically significant differences (Supplementary Fig. 3). Additionally, the in vitro stimulation of gingival epithelial cells with 100 nm LPS mimicked the inflammatory microenvironment, and the combination of LPS and nicotine (100 nm) was used to simulate the effect of smoking on the gingival epithelium. The PCR results confirmed that the changes in the expression of KRT1, DSG1, PI3, and GPX2 were correlated with the spatial expression patterns observed in the tissue samples (Fig. 2e). The above results indicate that smoking increases the risk of periodontitis by disrupting the structure and function of the gingival epithelium through its impact on the expression and function of these key genes.

Smoking-induced fibroblast alterations and the dysregulation of fibroblast‒epithelial cell communication

Gingival fibroblasts play a critical role in maintaining the integrity and function of the gingival epithelium. By conducting GO enrichment analysis, we determined the relevant biological functions of the differentially expressed genes in fibroblasts between the SP and P groups. We specifically focused on statistically significant GO terms that are associated with pathways involving inflammation, immune response, and tissue healing. Our results revealed that individuals in the smoking group, as opposed to healthy controls, presented upregulated expression of genes linked to ageing, intrinsic apoptotic signalling, and mitotic processes (Fig. 3a). In contrast to those associated with chronic periodontitis, genes associated with cell chemotaxis, leucocyte proliferation, and ageing were enriched in the smoking group (Fig. 3b). The subsequent clustering of fibroblasts into eight subgroups allowed mapping of their spatial distribution onto the gingival tissue morphology (Fig. 3c and Supplementary Fig. 4). Statistical assessments of fibroblast subgroup proportions revealed significantly higher Fib-8 expression in the smoking-related periodontitis group than in both the periodontitis group and the normal group (Fig. 3d).

Fig. 3

Smoking-induced fibroblast alterations and the dysregulation of fibroblast‒epithelial cell communication. a Bar plots showing enriched GO terms of DEGs (SP group vs. HG group) in fibroblasts. b Bar plots showing enriched GO terms of DEGs (SP group vs. P group) in fibroblasts. c Spatial map of the eight subgroups of fibroblasts in the Visium HD date. d Proportions of fibroblasts subgroups in each group. e GO analysis of highly expressed genes in Fib-8 (SP group vs. P group). f Ligand–receptor interactions between fibroblasts and epithelial cells and Visium HD data showing the cell communication signals of COL3A1-ADGRG1 and EPGN-EGFR. g Ligand–receptor interactions between epithelial cells and fibroblasts and Visium HD data showing the cell communication signals of AGRN-PTPRS and WNT5A-SFRP2. *P < 0.05

Smoking was found to upregulate the expression of Fib-8 and genes associated with wound healing (e.g., IGFBP2, COMP), chemotaxis (e.g., CXCL14, HSPB1, S100A14), and ageing (e.g., S100A8, S100A9) pathways (Fig. 3e). Analyses of epithelial-fibroblast interactions using CellChat revealed that fibroblasts release ligands that interact with epithelial cell receptors, including COL3A1-ADGRG1, EPGN-EGFR, NECTIN1-NECTIN4, WNT5A-SFRP1, and SEMA4D-PLXNB1. Specifically, the EPGN-EGFR signalling pathway may promote excessive epithelial proliferation, undifferentiated migration, and barrier disruption, ultimately compromising epithelial barrier function (Fig. 3f). An analysis of cell communication between epithelial cells and fibroblasts revealed an increase in signals such as AGRN-PTPRS, IGFBPS-TMEM219, and WNT5A-SFRP2 (Fig. 3g). In smoking-associated chronic inflammatory environments, the upregulation of PTPRS and SFRP2 on fibroblasts aligns with previous studies demonstrating increased expression of these markers in fibroblasts with a pro-inflammatory phenotype.26,27 The accumulation of pro-inflammatory fibroblasts may drive abnormal epithelial cell proliferation,28 compromise epithelial barrier stability, and facilitate the infiltration of bacteria or inflammatory factors, thereby exacerbating the inflammatory response.29,30 Signals such as AGRN-PTPRS and WNT5A-SFRP2 may lead to fibroblast dysfunction or abnormal proliferation, thereby exacerbating damage to periodontal tissues.27,31,32,33,34,35,36 These results suggest that smoking upregulates the expression of genes related to ageing, chemotaxis, and wound healing in gingival fibroblasts and that the dysregulation of epithelial‒fibroblast interactions potentially affects the integrity and function of periodontal tissues.

Smoking induces immune microenvironment disruption by promoting macrophage dysfunction

Smoking disrupts the local immune microenvironment in periodontal tissues. To investigate its impact on each subgroup, immune cells in gingival tissue were categorized into four subpopulations: plasma cells, macrophages, T cells, and neutrophils (Fig. 4a). Analyses revealed a significant increase in immune cell numbers in smoking-related periodontitis tissues (40.6% of total cells), with plasma cells exhibiting the most notable increase from 18.7% in chronic periodontitis to 27.8% in smoking-related periodontitis. The percentage of macrophages also increased from 3% in chronic periodontitis to 5.6% in smoking-related periodontitis (Fig. 4b).

Fig. 4

Smoking induces an immune microenvironment disruption by promoting macrophage dysfunction. a Map showing the infiltration of the four immune cell subgroups into the gingiva. b Proportions of each immune cell subgroup in each group. c CellChat analysis revealing outgoing signalling patterns of each immune cell subgroup in each group. d GO analysis of highly expressed genes in macrophages (SP group vs. P group). e GO analysis of downregulated genes in macrophages (SP group vs. P group). f Spatial analysis revealing the spatial relationship of macrophages and endothelial cells. g Visium HD data showing the spatial map of macrophages and endothelial cells

The CellChat analysis of signalling patterns indicated that macrophages in the smoking group significantly upregulated the expression of genes in proinflammation-related signalling pathways, such as VEGF, WNT, CD40, and IL2 (Fig. 4c). Differential gene analysis of macrophages revealed that, compared with those in the chronic periodontitis group, the expression of genes related to the “response to wounding,” “leucocyte-mediated immunity,” “regulation of apoptotic signalling,” “response to bacteria,” and “inflammatory response” pathways was upregulated in the smoking group (Fig. 4d). The downregulated genes were associated with the following pathways: “interleukin-4 and interleukin-13 signalling,” “positive regulation of cell migration,” and “regulation of insulin-like growth factor transport” (Fig. 4e).

Utilizing Squidpy’s co-occurrence score, we analysed the Visium dataset and identified significant spatial interactions among distinct cell clusters. In particular, macrophages displayed closer spatial proximity to endothelial cells than to other cell types, such as neutrophils, T cells, and plasma cells (Fig. 4f). Furthermore, in the comparison of distances between endothelial cells and macrophages across the three groups, the SP group exhibited the shortest proximity between these cell types (Supplementary Fig. 5). Spatial mapping of macrophages and endothelial cells in the gingiva section illustrated that macrophages were predominantly situated around endothelial cells, particularly in the smoking group (Fig. 4g). The intimate spatial association between macrophages and endothelial cells implies a potentially important role of their interaction in the progression of SP. Collectively, these findings indicate the significant impact of smoking on perturbing the immune microenvironment in periodontal tissues through the modulation of macrophage function, augmentation of inflammatory signalling, and alteration of gene expression, consequently exacerbating tissue damage.

Endothelial cell inflammation and its interaction with macrophages potentially aggravate SP

We first focused on functional changes in endothelial cells in the gingival tissue of SG to investigate the role of the interaction between endothelial cells and macrophages in exacerbating periodontal inflammation. PECAM, a marker of endothelial cells, showed increased expression in the SP group compared to the P group (Fig. 5a). We then examined gene expression in endothelial cells across the HG, P, and SG groups. Compared to HG individuals, SP patients exhibited a significant upregulation of genes associated with DNA damage, intrinsic apoptotic signalling, vascular morphogenesis, and acute inflammatory responses to antigen stimulation (Fig. 5b). Moreover, the TF‒gene network analysis identified downstream gene targets of key transcription factors such as NFKB1, RELA, and STAT3, suggesting their involvement in the observed differential gene expression in the endothelial cells of the smoking group (Supplementary Fig. 6a).

Fig. 5

Endothelial cell damage in the gingiva is an important factor contributing to the immune microenvironment disruption. a Visium HD data analysis revealing the relative expression levels of PECAM (marker gene of endothelial cells). b Bar plots showing enriched GO terms of DEGs (SP group vs. HG group) in endothelial cells. c Bar plots showing enriched GO terms of DEGs (SP group vs. P group) in endothelial cells. d Alluvial plot showing the interaction of endothelial cells–immune cells. e Visium HD data analysis revealing the relative expression levels of CXCL12 and CXCR4 in the gingival tissue. f Violin plot showing the relative expression levels of CXCL12 in each cell subgroup. g Visium HD data analysis revealing the colocalization of PECAM + cells and CXCL12+ cells. h Visium HD data analysis revealing the colocalization of LYZ+ cells and CXCR4+ cells

Differences in gene expression in endothelial cells between the SP group and P group were analysed. GO analysis revealed that smoking-associated periodontitis upregulated the expression of genes involved in DNA damage regulation, apoptosis signalling, macrophage proliferation, and responses to DNA damage and bacterial stimulation (Fig. 5c). The activation of DNA damage and apoptosis pathways in endothelial cells could impair their function and compromise vascular endothelial integrity,37 fostering an inflammatory milieu and augmenting the recruitment and activation of immune cells, particularly macrophages, thereby exacerbating periodontal tissue degradation.38 Additionally, compromised endothelial cell responses to bacterial stimulation may intensify local inflammation, thereby aggravating periodontitis in smokers.39 An analysis of transcription factors revealed the participation of inflammatory-related factors, such as NFKB1, SP1, RELA, STAT3, and HIF1A, in the endothelial cells of the SP (Supplementary Fig. 6b). Therefore, endothelial cell dysfunction likely exacerbates periodontal inflammation in the context of smoking.

Furthermore, our analysis of cell communication between endothelial and immune cells in individuals with smoking-related periodontitis conditions reveals that endothelial cell-derived CXCL12 influences macrophages by binding to CXCR4. The expression level of CXCL12 was higher than that of other endothelial cell ligands binding to macrophage receptors (Fig. 5d). A comparison of CXCL12 expression in endothelial cells between the P and SP groups confirmed the significant upregulation of CXCL12 in endothelial cells of the SP group (Fig. 5e). Additionally, endothelial cells were identified as the primary source of CXCL12 (Fig. 5f). Spatial transcriptomic visualization further supported the activation of CXCL12-CXCR4 signalling between endothelial cells and macrophages in the SG group, showing spatial proximity between endothelial cells with high CXCL12 expression (Fig. 5g) and macrophages with high CXCR4 expression (Fig. 5h). Taken together, endothelial cell inflammation and its interaction with macrophages may exacerbate SP.

Targeting Endothelial CXCL12 Promotes Macrophage Polarization to an Anti-Inflammatory Phenotype, Alleviating Periodontal Inflammation and Bone Resorption

Given the higher CXCL12 expression in endothelial cells from the SP group compared to the P group and the proximity of endothelial cells to macrophages, we investigated CXCL12-regulated genes in macrophages. The prediction using the STRING database suggests that CXCL12 may upregulate proinflammatory genes, such as CXCL14, S100A9, and S100A8. Consistent with this result, our spatial transcriptomic data showed upregulated expression of CXCL14, APOE, C1QC, C1QA, S100A9, and S100A8 in macrophages of the SG group compared to the P group (Fig. 6a).

Fig. 6

Targeting endothelial CXCL12 promotes macrophage polarization to an anti-inflammatory phenotype, alleviating periodontal inflammation and bone resorption. a Associations between endothelial CXCL12 signalling and macrophage inflammation-related genes were predicted using the STRING database. b RT–qPCR analysis of the CXCL12 expression levels in endothelial cells treated with AAV-shRNA-CXCL12. c, d Flow cytometry analysis showing the expression of iNOS and CD206 in macrophages treated with different endothelial cell culture media. e, f Multicolour IF staining showing the expression of CXCL12 and the endothelial cell marker CD31. The nuclei were stained with DAPI. Bar = 50 μm. g, h Three-dimensional reconstructions of the maxilla in each group generated by micro-CT scans; the black arrows indicate the distance between cement enamel junction and alveolar bone crest (CEJ-ABC), scale bar = 500 μm. i, j Images of the HE-stained periodontium from each group showing bone resorption; the black arrows indicate the distance between CEJ-ABC, scale bar = 50 μm. k, l IF staining showing the expression of TNF-α. Bar = 50 μm. *P < 0.05

Furthermore, we explored the impact of endothelial cell-derived CXCL12 on macrophage functions. Tie1 is an endothelial cell-specific promoter; we constructed AAV-Tie1-shRNA-CXCL12 (AAV-shCXCL12) to drive the expression of an shRNA via the Tie1 promoter and specifically inhibit the expression of CXCL12 in endothelial cells. We stimulated endothelial cells with nicotine while simultaneously introducing AAV-shCXCL12 and control AAV (AAV-con), and the RT-qPCR analysis revealed that the expression of CXCL12 in endothelial cells was significantly downregulated after the addition of AAV-shCXCL12 (Fig. 6b). The collected endothelial cell culture medium was then added to bone marrow-derived macrophages (BMDMs). The medium from endothelial cells with suppressed CXCL12 expression reduced the proinflammatory shift in macrophages, with the downregulation of the proinflammatory-related gene CD86 and upregulation of the anti-inflammatory-related gene CD206 (Fig. 6c, d).

Furthermore, in a mouse model of smoking-induced periodontal inflammation simulated by ligature placement and nicotine injection, immunofluorescence staining revealed that treatment with AAV-shCXCL12 targeting endothelial cells led to a marked downregulation of CXCL12 expression in endothelial cells (Fig. 6e, f). Micro-CT analysis revealed that treatment with AAV-shCXCL12 alleviated the damage to the alveolar bone caused by local nicotine injection (Fig. 6g, h). HE staining revealed that alveolar bone destruction in mice was significantly relieved following AAV-shCXCL12 treatment (Fig. 6i, j). Moreover, the immunofluorescence results revealed a substantial reduction in the expression of inflammatory cytokines, such as TNF-α, within the periodontal tissues (Fig. 6k, l). The interaction between macrophages and endothelial cells, which is mediated by CXCL12 signalling, exacerbates smoking-induced periodontal inflammation, and targeting CXCL12 in endothelial cells with AAV-shCXCL12 can reduce proinflammatory macrophage activation and alleviate alveolar bone destruction in a mouse model.