Immunological link between periodontitis and type 2 diabetes deciphered by single-cell RNA analysis

Abstract

Background: Type 2 diabetes mellitus (DM) is a complex metabolic disorder that causes various complications, including periodontitis (PD). Although a bidirectional relationship has been reported between DM and PD, their immunological relationship remains poorly understood. Therefore, this study aimed to compare the immune response in patients with PD alone and in those with both PD and DM (PDDM) to expand our knowledge of the complicated connection between PD and DM.

Methods: Peripheral blood mononuclear cells were collected from 11 healthy controls, 10 patients with PD without DM, and six patients with PDDM, followed by analysis using single-cell RNA sequencing. The differences among groups were then compared based on intracellular and intercellular perspectives.

Results: Compared to the healthy state, classical monocytes exhibited the highest degree of transcriptional change, with elevated levels of pro-inflammatory cytokines in both PD and PDDM. DM diminished the effector function of CD8+ T and natural killer (NK) cells as well as completely modified the differentiation direction of these cells. Interestingly, a prominent pathway, RESISTIN, which is known to increase insulin resistance and susceptibility to diabetes, was found to be activated under both PD and PDDM conditions. In particular, CAP1+ classical monocytes from patients with PD and PDDM showed elevated nuclear factor kappa B-inducing kinase activity.

Conclusions: Overall, this study elucidates how the presence of DM contributes to the deterioration of T/NK cell immunity and the immunological basis connecting PD to DM.

Keywords: chronic inflammation; insulin resistance; periodontitis; single-cell RNA sequencing; type 2 diabetes mellitus.

FIGURE 1

Peripheral blood mononuclear cells (PBMCs) and clinical profiles of healthy donors and of patients with periodontitis (PD) and both PD and diabetes mellitus (DM) (PDDM). (A) Clinical information of each group. Body mass index (BMI), haemoglobin A1c (HbA1c), C‐reactive protein (CRP) and erythrocyte sedimentation rate (ESR) were recorded. Error bars represent mean ± standard deviation (SD). All differences with p‐values <.05 are marked: Wilcoxon rank‐sum test. ^* p < .05, ^** p < .01, ^*** p < .0005, ns: not significant. (B) Uniform manifold approximation and projection (UMAP) projections of integrated PBMCs. The cell compartments consist of naïve T cells, effector and memory T cells, regulatory T cells (Treg), natural killer (NK) cells, naïve B cells, switched memory B cells, classical monocytes (classic mono), intermediate monocytes (intermediate mono), non‐classical monocytes (nonclassic mono), gamma delta T cells (gdT), mucosal‐associated invariant T cells (MAIT), myeloid dendritic cells (mDC), plasmacytoid DCs (pDC), red blood cells (RBC) and platelets. (C) Bar plot of the immune cell fraction for each individual. (D) Bar plot showing the number of immune cells. Healthy, grey; PD, yellow; and PDDM, blue. (E) Boxplot of the ratio between lymphocyte‐lineage immune cells and monocytes. The left panel shows the ratio of lymphocytes, including B lymphocytes, T lymphocytes, and NK cells to monocytes. The middle panel shows the ratio of T lymphocytes to monocytes and the right panel shows the ratio of B lymphocytes to monocytes. The horizontal line in the boxplot indicates the average ratio for each group.

FIGURE 2

Immunological shifts of monocytes and CD4 effector cells in periodontitis (PD) and both PD and diabetes mellitus (DM) (PDDM). (A) Number of differentially expressed genes (DEGs) in each cellular compartment. Bar plots display the number of DEGs when comparing patients with PD to healthy controls (left) and patients with PDDM to healthy controls (right). (B) The top 20 Gene Ontology (GO) terms of DEGs in classical monocytes. Functions of upregulated genes in patients with PD (left) and PDDM (right) compared with those in the healthy controls are shown. (C) Expression levels of pro‐inflammatory and anti‐inflammatory cytokines and chemokines in monocytes. (D) Expression of human leukocyte antigen‐DR (HLA‐DR). Patients with HLA‐DRA, HLA‐DRB5 or HLA‐DRB1 mutations were included in this study. The dotted plot in the upper panel shows the expression levels of HLA‐DR genes and the ridge plot below shows the distribution of HLA‐DR gene expression. The significance of the differences was computed using the Wilcoxon rank‐sum test. ^* p < .05, ^** p < .01, ^*** p < .001, ^**** p < .0001, ns: not significant. (E) The top 20 GO terms of DEGs in CD4+ effector cells. Functions of downregulated genes in PD (left) and PDDM (right) compared to those in the healthy controls are shown. (F) Violin plot showing the CD4+ effector cell activity score regulating alpha–beta T cell activation. The expression levels of genes listed in GO:0046634 (regulation of alpha–beta T cell activation) were measured.

FIGURE 3

The cytotoxicity, exhaustion and activity scores of CD8 T and NK cells. (A–C) Cytotoxicity and exhaustion of (A) CD8+ effector T, (B) CD16‐high NK and (C) CD56‐high NK cells separated by group. The significance of strength between groups was compared using the Wilcoxon rank‐sum test. ^* p < .05, ^** p < .01, ^*** p < .001, ^**** p < .0001, ns: not significant. (D) Activity of CD16‐high and CD56‐high NK cells. Figure descriptions are similar to those of Figure 2A–C. (E) Violin plot of cytotoxicity and exhaustion scores according to clusters, which are identified as CD8+ effector T cells in cell‐type annotation step. The bar plot located above the violin plot shows the proportion of the clusters in each individual group. (F) The violin plot of cytotoxicity and exhaustion scores according to clusters, which are identified as CD16‐high NK cells (clusters 29, 6, 26, 7, 2 and 30) and CD56‐high NK cells (cluster 32) in cell‐type annotation step. The description of the figure is similar to that of Figure 2E.

FIGURE 4

Trajectory analysis of CD8+ T cells and NK cells reveals separated diffusion paths depending on the health conditions. (A) Inferred pseudotime of the CD8+ T‐cell lineage overlaid on uniform manifold approximation and projection (UMAP). CD8+ T cells are coloured according to their pseudotime, with yellow indicating early differentiation and dark blue representing late differentiation. (B) Expression levels of SELL and CCR7, early‐stage markers of CD8+ T cells, ordered by pseudotime. (C) The UMAP was separated by the healthy group. The red text depicts the terminal states detected in the CD8+ T cell lineage pathway. (D) Percentage of cells in each terminal cluster of CD8+ T cells by patient group. (E) Heatmap representing differentially expressed genes for each terminal cluster of CD8+ T cells. The scaled expression levels of extracted genes correspond to the colours depicted in the colour bar. (F) Inferred pseudotime of the NK cell lineage overlaid on the UMAP. NK cells are coloured according to their pseudotime, with yellow indicating early differentiation and dark blue representing late differentiation. (G) Expression levels of diffusion stage markers in NK cells ordered by pseudotime. IL7R is a marker of NK cell precursors, and IL2RB is a marker of NK cell maturation during differentiation. (H) The UMAP was separated by health group. The red text depicts the terminal states detected in the NK cell lineage pathway. (I) The percentages of cells by patient groups for each terminal cluster of NK cells. (J) Heatmap showing differentially expressed genes in each terminal cluster of NK cells. The scaled expression levels of the extracted genes correspond to the colours depicted in the colour bar.

FIGURE 5

The TNF pathway is strengthened under PD and PDDM conditions whereas the GRN pathway is enhanced under PDDM conditions. (A) Circle plot of TNF signalling in PD and PDDM. The cell–cell crosstalk of TNF signalling in PD is shown in the left panel, and that of PDDM is shown in the right panel. The arrow starts from the signal sender and ends directly at signal receiver. The colour of the arrow matches that of the sender cell. The strength of the crosstalk is correlated with the edge width. (B) Violin plot of TNF expression in monocyte subpopulations. Expression levels between groups were compared using the Wilcoxon rank‐sum test. ^* p < .05, ^** p < .01, ^*** p < .001, ^**** p < .0001, ns: not significant. (C) Percentage of TNF+ non‐classical monocytes in the control, PD and PDDM groups. (D) Circle plot of GRN signalling. The cell–cell crosstalk of TNF signalling in healthy, PD and PDDM groups is drawn in the left, middle and right, respectively. (E) GRN expression and percentage of GRN+ cells in the control, PD and PDDM groups. Violin and bar plots in the left panel show the expression level of GRN in pDCs and the proportion of GRN‐expressing pDCs, whereas the right panel shows the GRN expression and proportion of GRN‐expressing classical monocytes. (F) SORT1 expression and percentage of SORT1+ cells in the control, PD and PDDM groups. Violin and bar plots in the left panel show the expression level of SORT1 in mDCs and the proportion of SORT1‐expressing mDCs, whereas the right panel shows SORT1 expression and the proportion of SORT1‐expressing intermediate monocytes.

FIGURE 6

The RESISTIN pathway is intensified in both PD and PDDM, sequentially. (A) Circle plot of RESISTIN signalling. The RESISTIN network in the healthy, PD and PDDM groups are drawn in the order: left, middle and right, respectively. (B) Percentage of ligand‐expressing cells in the RESISTIN pathway in the control, PD and PDDM groups. The bar plots represent the ratio of RETN+ cells in mDCs, and in non‐classical and intermediate monocytes, which play a role in signalling, in that order. (C) Percentage of cells expressing the receptors of the RESISTIN pathway in the control, PD and PDDM groups. Each panel shows the proportion of TLR4+ and CAP1+ cells among mDCs and classical monocytes. (D) Differential functions between CAP1+ and CAP1– classical monocytes. The upper panel depicts the augmented biological functions of CAP1+ classical monocytes in the PD group, whereas the lower panel shows the same functions in the PDDM group. (E) The activity score of NF‐κB‐inducing kinase and the relative expression of genes associated with this activity (GO:0007250) were categorised according to each group. The score was calculated at the single‐cell level, and the expression levels between the two groups were compared using the Wilcoxon rank‐sum test. ^* p < .05, ^** p < .01, ^*** p < .001, ^**** p < .0001, ns: not significant.

FIGURE 7

Comparison with publicly available datasets GSE165816 and GSE164241. (A) Summary of the origins of public data samples. (B) Uniform manifold approximation and projection (UMAP) of GSE165816. CD8+ effector T, CD16‐high NK and CD56‐high NK cells; classical, intermediate and non‐classical monocytes; and mDCs were identified for comparison with prior observations. (C) UMAP of GSE164241. Classical monocytes and mDCs were identified for comparison with prior observations. (D–F) Cytotoxicity and exhaustion score of (D) CD8+ effector T cells, (E) CD16‐high NK cells and (F) CD56 high‐NK cells derived from GSE165816. Significant differences between groups were determined using the Wilcoxon rank‐sum test. ^* p < .05, ^** p < .01, ^*** p < .001, ^**** p < .0001, ns: not significant. (G) Proportion of TNF+ non‐classical monocytes derived from GSE165816. (H) Proportions of cells expressing the ligands and receptors involved in the GRN pathway (GSE165816). We examined GRN+ classical monocytes, SORT1+ mDCs and SORT1+ intermediate monocytes. (I) Proportion of cells expressing the ligands and receptors involved in the RESISTIN pathway (GSE165816). RETN+ mDCs, RETN+ non‐classical monocytes, RETN+ intermediate monocytes and CAP1+ classical monocytes were examined. (J) Heatmap of gene expression involved in the NF‐κB‐inducing kinase activity within classical monocytes (GSE165816). The cells were separated into the following four groups based on CAP1 expression and health conditions: CAP1– healthy, CAP1– diabetes mellitus (DM), CAP1+ healthy and CAP1+ DM. (K) Proportion of TNF+ macrophages derived from GSE164241. (L) Proportion of cells expressing the ligands and receptors involved in the RESISTIN pathway (GSE164241). RETN+ mDCs, RETN+ macrophages and CAP1+ macrophages were investigated. (M) Heatmap of gene expression involved in the NF‐κB‐inducing kinase activity within macrophages (GSE164241). Macrophages were separated into the following four clusters based on CAP1 expression and health conditions: CAP1– healthy, CAP1– periodontitis (PD), CAP1+ healthy and CAP1+ PD.