Distinct immune landscapes between murine and human periodontitis

Single-cell RNA sequencing reveals different immune responses between human periodontitis and murine ligature model

To gain insights into the murine ligature model,⁶ we collected gingival tissue from ligature-induced PD mice and healthy mice (Figure S1A) and performed scRNA-seq using the 10X Genomics platform. Micro-CT and histological analyses confirmed successful establishment of the ligature model, with periodontal bone loss (Figures S1B–S1D). After quality control (Figure S1E), a total of 66,873 high quality cells were retained for downstream analysis. Cell clusters were visualized using uniform manifold approximation and projection (UMAP) (Figures 1A and 1B), revealing six major cell types based on specific marker genes: fibroblasts (e.g., Col1a2, Lum), epithelial cells (e.g., Krt5, Krt6a), endothelial cells (e.g., Pecam1, Cdh5), immune cells (e.g., Ptprc, Cd52), myocytes (e.g., Acta2, Tagln), and Schwann & neuron cells (e.g., Plp1, Kcna1) (Figure 1C).^12,19,20

To compare the murine ligature model with human PD, we analyzed published scRNA-seq datasets (GEO: GSE164241), including gingival tissue from healthy individuals and patients with PD.¹² Utilizing the matching marker genes, we identified six major cell types in human PD (Figures 1D–1F). We observed a profound infiltration of immune cells in the gingival tissue of human PD (Figure 1E), reflecting exaggerated and destructive host immune response in human PD. In contrast, the immune cell populations showed a modest increase in the murine ligature model (Figure 1B), resulting distinct difference between the two species (p < 0.001). Across all samples, changes in cell proportions were consistent (Figures S1F and S1G).

Analysis of differentially expressed genes (DEGs) between PD and Ctrl groups revealed limited overlap in DEGs across human PD and murine ligature model, reflecting distinct host responses (Figure 1G). The top 5 DEGs in mice were associated with macrophage and neutrophil functions (S100a9, S100a8, Lyz2, Ccl4, and Cxcl2) (Figure 1H), whereas in humans, they were related to antibody secretion and pro-inflammatory (IGHG4, IGHG1, IGHG2, IGKC, and IL1B) (Figure 1I). Further GO analysis of DEGs in each cell type revealed that the up-regulated genes in mice PD were primarily associated with ATP metabolic processes and energy metabolism in fibroblast, endothelial cells, and myocytes; while in immune cells, up-regulated DEGs were enriched in response to wounding (Figure 1J). In contrast, in humans, the up-regulated genes were enriched in “B cell activation,” “T cell activation,” and “neutrophil activation” between the PD and Ctrl groups (Figure 1K).

Identification of immune subpopulations reveals distinct immune cell infiltration

We next focused on the immune cell populations. Ten immune clusters were identified based on classic markers in both murine ligature model and human PD (Figures 2A–2D, S2A, and S2B),^12,19,20 revealing similar immune cell subsets but with notable differences in proportions of each cell type between species. In both human and mice, T cells (including CD4⁺, CD8⁺, double-positive, double-negative αβT cells, and γδT cells) were the dominant immune population in healthy gingival tissue and decreased in both species during PD (Figures 2B and 2D). In the murine ligature model, dendritic cells (DCs) moderately decreased (24.8%–19.2%, p = 0.057), while macrophages expanded significantly (3.8%–32.9%, p = 0.001) (Figure 2B). In contrast, human PD had a smaller proportion of DCs and macrophages, with both populations decreasing in PD (DCs, from 6.5% to 3.1%, p = 0.001; macrophages, from 8.0% to 5.9%, p = 0.021) (Figure 2D). The change in macrophages (ΔPD-Ctrl) is significantly greater in mice than in humans (p < 0.0001). While B/plasma cells remained scarce in both healthy and ligature-induced murine gingival tissue (around 1.8%, compared p > 0.05) (Figure 2B), they were markedly increased in PD human tissue (from 19.5% to 36.7%, p = 0.002) (Figure 2D). The murine B/plasma Δ(PD-Ctrl) is significantly lower than that in humans (p < 0.001). Neutrophils increased both in the murine ligature model and human PD (mouse, from 3.6% to 18.0%, p < 0.001; human, from 2.4% to 10.3%, p = 0.021) (Figures 2B and 2D).

To verify that the differences are not affected by the individuals from diverse backgrounds, an East Asian population (GEO: GSE171213) was also compared to the murine dataset, and a similar consequence was observed (Figure S2C). The change in B/plasma (ΔPD-Ctrl) is significantly greater in humans (14.8% increase) than in mice (0.1% increase) (p < 0.05). And also, the change in macrophages (ΔPD-Ctrl) is significantly greater in mice (28.5% increase) than in humans (2.6% increase) (p < 0.01). There is no significant difference between human and murine neutrophil (ΔPD-Ctrl) (p > 0.05).

To validate these findings, we performed flow cytometry analysis (Figure S2D) and observed an increase in CD45⁺ immune cells in the murine PD group (Figure S2E). B/plasma cells (CD79A⁺) were slightly increased (Figure 2E), while macrophages (CD11B⁺F4/80⁺) (Figure 2F) and neutrophils (CD11B⁺LY6G⁺) (Figure S2F) significantly increased. Immunofluorescence staining also validated these observations, showing that B/plasma cells and macrophages were the most prominent differences between the two species, while neutrophil infiltration remained relatively consistent (Figures 2G and 2H).

Moreover, neutrophil subgroups were identified based on established markers (Table S3).²¹ Significant differences between human and mice were observed in: elevated Neutro b1 subgroup (related to myeloid leukocyte migration and chemotaxis) and decreased Neutro c subgroup (associated with ribonucleoprotein complex biogenesis and cytoplasmic translation) in human PD, while no comparable changes were noted in mice (Figure S2G). Notably, overall neutrophil proportions and trends of cellular differentiation pseudotime revealed relatively consistent results between human and mice (Figures S2H and S2I).

B cells in the murine ligature model are fewer and primarily in an early inactive state

To further investigate the B/plasma cell compartments, we integrated murine and human datasets by FindIntegrationAnchors. Through SingleR and manual annotation (Table S1), we identified 8 distinct sub-clusters, including naive B cells, activated B cells, transitional B cells, conventional memory B cells, CD11c⁺CD27⁻memory B cells, atypical memory B cells, regulatory B cells (Breg), and plasma cells (Figures 3A and 3B). Through pseudotime ordering of integrated B/plasma cells (Figure 3C), we observed that the distribution of eight B/plasma subsets was consistent with the trajectory of differentiation (Figure 3D). In the murine model, B/plasma cells were predominantly located at the early stages of the pseudotime trajectory (Figure 3E), with a composition primarily of naive B cells (Figure 3A). In the murine PD group, naive B cells showed an increased trend, but plasma cells remained scarce (Figure 3A). In human PD, the increased plasma cells were predominantly located at the late stages of the pseudotime trajectory (Figure 3E). Compared to the murine model, the human B/plasma Δ(PD-Ctrl) is significantly higher in the plasma cell subset (p = 0.001) (Figure 3A). The trends of plasma cells (CD138⁺), naive B cells (IgD⁺), and conventional memory B cells (CD27⁺) were verified by immunofluorescence staining (Figure 3F).

Gene expression analysis further highlighted these differences. Murine B/plasma cells expressed genes associated with early differentiation (e.g., Lyn, Cd24, Ms4a1), whereas human cells demonstrated higher expression of late-stage, antibody-related genes such as MZB1, IGKC, and JCHAIN. Despite this distinction, both murine and human PD groups showed a general trend of later-stage gene expression compared to their respective Ctrl groups (Figure 3G).

Interestingly, transcription factor (TF) analysis revealed species-specific changes. In human PD, BTG1, a key TF that inhibits hyperactive B cell proliferation,²² was significantly downregulated (p < 0.001), while XBP1, a hallmark TF that promotes B cells transforming into antibody-secreting phenotype,²³ was significantly upregulated (p < 0.001) (Figure 3H). These transcription factors showed no significant changes in the murine ligature model (Btg1, p = 0.173; Xbp1, p = 0.780), underscoring key differences in B/plasma cell regulation between species (Figure 3I).

Macrophages in the murine ligature model show elevated differentiation

We performed cross-species integration of monocytes/macrophages between the murine ligature model and human PD samples using FindIntegrationAnchors. Through combined SingleR prediction and manual annotation (Table S2), we identified nine distinct subpopulations: classical monocytes, intermediate monocytes, non-classical monocytes, two M1-like macrophage subsets (Macro 1 and 2), and four M2-like macrophage subsets (Micro 3, 4, 5, and 6) (Figures 4A–4D). The trends of monocytes (classical, CD14⁺CCR2⁺; non-classical, CD14⁺CX3CR1⁺), and macrophages (CD206⁺ for M2-like; iNOS⁺ for M1-like) were verified by immunofluorescence staining (Figures 4E and 4F). Pseudotime ordering revealed their differentiation trajectory (Figures 4G–4J).

In mice, macrophages from the Ctrl group were primarily located at the early stages of pseudotime, and shifted toward the later stages in the PD group (Figure 4I). In the murine PD group, classical monocytes markedly increased (p = 0.002), while non-classical monocytes were significantly reduced (p = 0.002) (Figures 4B and 4E). This likely indicated that: 1) concurrent recruitment and M1-polarization of classical monocytes during acute inflammation; 2) exhaustion of non-classical monocyte reserves to M2-like cells. Compared to humans, the murine Δ(PD-Ctrl) is significantly higher in M1-like Micro 1 (pro-inflammatory, p = 0.032) and M2-like Micro 3 (tissue repair, p < 0.001) (Figures 4B and 4F), reflecting the acute inflammatory features associated with mechanical injury.

In contrast, human macrophages displayed a more even distribution across pseudotime in both the Ctrl and PD groups (Figure 4J). Unlike the murine ligature model, the human PD group did not show a marked increase in classical monocytes or M1-like macrophages (p > 0.05), but a significant increase in non-classical monocytes (p = 0.046) (Figures 4B, 4E, and 4F). Compared to mice, the human Δ(PD-Ctrl) is significantly lower in Micro 4 (metabolic tissue protection, p = 0.039), and higher in Micro 5 (lysosome function, p = 0.002) (Figure 4B), suggesting differences in the macrophage responses and functional dynamics between the two species.

Additionally, the cell density of different groups and representative genes gradually changed with pseudotime (Figures 4K and L). In the murine Ctrl group, the cell density distribution was primarily concentrated in the early-stage monocytes, whereas the PD group exhibited a broader distribution across all stages, with upregulated genes associated with acute immune responses (e.g., Sell, Ereg, C1qb, and C2). In contrast, the cell density distribution in human Ctrl and PD groups differed from that in mice, with genes gradually expressed along pseudotime being linked to energy metabolism (e.g., ATP5F1E, NDUFS7), antioxidation (e.g., SELENOK, ME1), and protein homeostasis (e.g., TENT5A, CREG1). These findings suggested distinct transcriptional profiles existed in PD between the two species.

Application of Porphyromonas gingivalis induces moderate changes in the immune response of the murine ligature model

Given the limitations of the simple ligature (Lig) model in mimicking human PD, we investigated whether combining ligature with Porphyromonas gingivalis (Pg) application (LigPg) could better recapitulate the immune response of PD.²⁴ PCR validation confirmed Pg colonization in the LigPg samples (Figure S3A). Furthermore, LigPg mice exhibited elevated serum TNF-α and IL-6 levels, reflecting a systemic inflammatory impact (Figure S3B). LigPg group showed a similar level of local inflammation and osteoclastic bone resorption to that of the Lig group (Figures S3C and S3D).

After quality control (Figure S3E), single-cell RNA sequencing of Lig and LigPg murine gingival tissue yielded a total of 20,122 high quality cells (Figure 5A). Compared with the Lig group, Pg application did not induce an expansion of the immune cells (p > 0.05) (Figure 5B). And 9,045 immune cells were then classified into subgroups for further analysis (Figure 5C). Notably, Pg application did not significantly increase the proportions of B/plasma cells or macrophages (p > 0.05), and neutrophils showed an elevated trend with no statistical significance (p > 0.05) (Figure 5D). Flow cytometry analysis demonstrated no significant changes in B/plasma cell or macrophage populations (Figures 5E, 5F, and S3F–S3H). Consistent with these findings, immunofluorescence staining revealed comparable CD19⁺ and CD68⁺ signals between the LigPg and Lig groups (Figure 5G). These results suggest that Pg exposure exerts minimal impact on mimicking human PD immunity in this model.

When comparing DEGs between Lig vs. Ctrl and LigPg vs. Ctrl separately, most DEGs showed overlap between the Lig&Ctrl and LigPg&Ctrl comparisons (Figure 5H). And comparing LigPg with Lig group, most genes showed no significant differences, especially B/plasma cell (up-regulated DEGs, 0.09%; down-regulated DEGs, none). DC, macrophages, and neutrophils displayed higher percentage of DEGs (up-regulated, about 0.49%–2.68%; down-regulated, about 0.14%–0.79%) (Figure 5I). Up-regulated DEGs were mainly linked to inflammatory responses and cellular stress (e.g., Mt4, Perp, Lagls7, and Ly6d), while down-regulated DEGs were mostly associated with mitochondrial metabolism (e.g., mt-Nd2, mt-Nd4 l, and mt-Nd5).

Limitations of the study

In our study, a transcriptional-level comparison was made between murine 9-day-ligature model and human PD, and proposed it may not be fully suitable for exploring more complex immune responses, especially in B/plasma cells and macrophages. However, our study merely highlights differences at the immune cellular transcription level, yet the ligature model remains a classic system for alveolar bone resorption and acute inflammation, requiring further validation with advancing technologies. Beyond this, potential factors for further improvement including time, age, microbiota, and species-specific biology might be considered and explored in the future studies to identify more appropriate animal models that accurately recapitulate the immune responses and microbial dynamics observed in human PD.

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Quan Yuan (yuanquan@scu.edu.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

Key resources table

Experimental model and study participant details

Method details

Quantification and statistical analysis

To evaluate cross-species proportion changes, the differences between PD and control groups Δ(PD-Ctrl) in both human and murine samples were calculated for comparative analysis. For murine PD vs. Ctrl, human PD vs. Ctrl, and human Δ(PD-Ctrl) vs. murine Δ(PD-Ctrl), statistical analysis was performed as follows: normality was assessed by the Shapiro–Wilk test, and homogeneity of variance was assessed with Levene’s test (α = 0.05); data with a normal distribution were analyzed using an unpaired Welch’s t-test, while non-normally distributed data were analyzed using the Mann–Whitney U test. Flow cytometry data underwent unpaired t-tests following normality and variance confirmation. Differential gene expression analysis employed Seurat’s Wilcoxon rank sum tests. Statistical significance was defined as P<0.05 for all analyses.