Immune landscape of periodontitis unveils alterations of infiltrating immunocytes and molecular networks-aggregating into an interactive web-tool for periodontitis related immune analysis and visualization

Abstract

Background: Immunity reaction plays an essential role in periodontitis progress and we aim to investigate the underlying regulatory network of immune reactions in periodontitis.

Methods: CIBERSORT was used to estimate immunocyte fractions in different clinical statuses. Logistic regression was used to assess the immunocyte weight in periodontitis. Immune-related periodontitis subtypes were identified by the Nonnegative Matrix Factorization algorithm. Gene-set enrichment analysis and Gene-set variation analysis were conducted to analyze pathway activities. Immunocytes related gene modules were identified by Weighted gene co-expression network analysis.

Results: Altered immunocytes in healthy versus periodontitis, aggressive versus chronic, male versus female and age were identified. Immunocytes enriched in periodontitis were calculated, and their correlation was also explored. Two distinct immune-related periodontitis subtypes were identified and one is characterized by B cell reactions and the other is IL-6 cytokine reactions. 463 statistically significant correlations between 22 immunocytes and pathways were revealed. Immunocytes and clinical phenotypes matched their gene modules, and their functions were annotated. Last, an easy-to-use and user-friendly interactive web-tool were developed for periodontitis related immune analysis and visualization ( https://118.24.100.193:3838/tool-PIA/ ).

Conclusions: This study systematically investigated periodontitis immune atlas and caught a glimpse of the underlying mechanism of periodontitis from gene-pathway-immunocyte networks, which can not only inspire researchers but also help them in periodontitis related immune researches.

Keywords: Bioinformatics; Gene; Immune; Immunocyte; Pathway; Periodontitis; Web-tool.

Fig. 1

Infiltrating immunocytes differences between healthy and periodontitis samples. a The relative fraction of immunocytes identified by the CIBERSORT algorithm, which estimates relative subsets of 22 types of immunocyte from known RNA transcripts. The relative distributions of these 22 immunocytes were presented by bar-plots concerning different disease status (all samples for the left and average for the right). b Compositional differences of 22 immunocytes between healthy and periodontitis presented by violin-plot (blue means healthy and red means periodontitis). c The volcano-plot demonstrates the fold changes of 22 immunocytes in periodontitis compared with healthy. d Principal component analysis (PCA) of 22 infiltrating immunocytes between healthy and periodontitis. The two first principal components (PC1, PC2) which explain the most of the variables are plotted. e Correlation matrix of 22 immunocytes proportions. The top right is correlations of 22 immunocytes in all samples and the left bottom is correlations of 22 immunocytes in periodontitis samples

Fig. 2

Periodontitis associations of immunocytes. a Forest-plot demonstrates associations between different immunocyte subsets and periodontitis by univariate logistic regression. b Forest-plot demonstrates the independent associations between periodontitis-related immunocytes and periodontitis by multivariate logistic regression

Fig. 3

Identification of immune subtypes of periodontitis by consensus clustering. a The consensus matrix was obtained from 200 random runs of the Brunet et al.'s algorithm. Values range from 0 to 1. Columns and rows were ordered by hierarchical clustering based on the euclidean distance with average linkage. b Heatmap of the metagene matrix. Each row corresponds to a gene. The most metagene-specific genes were selected using Kim and Park's scoring and filtering methods. This resulted in the selection of 116 genes. Rows were scaled to sum to one and ordered by hierarchical clustering based on the euclidean distance and average linkage. c The protein–protein interaction network of 116 metagenes. The blue is immune subtype cluster-1 and red is subtype cluster-2. Only the genes with interaction with others are presented

Fig. 4

Functions and biological characteristics of immune subtypes. a The gene ontology enrichment analysis for the metagenes in two immune subtypes concerning their biological processes. GO categories are grouped according to functional. b, c The KEGG pathway enrichment analysis for the metagenes in two immune subtypes respectively (b for cluster-1 and c for cluster-2)

Fig. 5

Correlations between immune subtypes and immunocytes. a Compositional differences of 22 immunocytes between immune subtype cluster 1 and cluster 2 which was presented by violin-plot (blue means cluster-1 and red means cluster-2). b Comparison of immune-related pathways’ activity between the two immune subtypes. c Comparing periodontitis subtypes, genders, and ages between immune subtype cluster 1 and cluster 2. The heatmap illustrates the association of different clinical characters with cluster 1 and cluster 2 patients. Statistical significance was performed by the χ² test

Fig. 6

Transcriptome differences between healthy and periodontitis. a The protein–protein interaction network of significantly differentially expressed genes between healthy and periodontitis. The node size means the absolute value of logFC, the node with yellow frame means the hub genes with a high connective degree in the network, the red nodes mean the genes up-regulated in periodontitis and blue nodes mean the genes down-regulated in periodontitis. b–e Gene-set enrichment analysis from four aspects (including immunologic signatures, hallmark gene sets, GO biological processes and canonical pathways) revealed the underlying biomolecular differences between healthy and periodontitis

Fig. 7

Correlations between 22 immunocytes fractions and activity of 50 important biological hallmark-related pathways. a The number of significant pathways is correlated with individual immunocyte. The upper panel is for positively correlated pathways, and the bottom panel is for negatively correlated pathways. b Network diagram demonstrating the correlation between immunocytes and pathways. Red represents a positive correlation, and blue represents a negative correlation. The size of the nodes corresponds to the number of links, and the thickness of the line represents the correlation coefficient

Fig. 8

Identification of immunocytes and clinical characteristics related to gene modules. a The sample clustering was based on the expression data of all samples. The top 25% of variation genes were used for the analysis by WGCNA and outlier samples were excluded. b Gene dendrogram obtained by average linkage hierarchical clustering. The color row underneath the dendrogram shows the module assignment determined by the Dynamic Tree Cut, in which 27 modules were identified. c Heatmap of the correlation between module eigengenes and the immunocyte fractions

Fig. 9

Plasma cells related genes. a A scatterplot of gene significance (GS) for plasma cell fraction versus module membership (MM) in the turquoise module. GS and MM exhibit a very significant correlation, implying that hub genes of the turquoise module also tend to be highly correlated with plasma cell fraction. b GO enrichment analysis concerning biological processes revealed the functions of hub genes in the turquoise module. c The top ten hub genes in the turquoise module and their GS and MM were presented by bar-pot. d, e The most plasma cell fraction correlated with two genes was presented by a scatter plot

Fig. 10

Overview of Periodontitis Immune Atlas app