Integrated Machine Learning and Bioinformatic Analyses Constructed a Network Between Mitochondrial Dysfunction and Immune Microenvironment of Periodontitis

Abstract

Periodontitis is a prevalent and persistent inflammatory condition that impacts the supporting tissues of the teeth, including the gums and bone. Recent research indicates that mitochondrial dysfunction may be involved in the onset and advancement of periodontitis. The current work sought to reveal the interaction between mitochondrial dysfunction and the immune microenvironment in periodontitis. Public data were acquired from MitoCarta 3.0, Mitomap, and GEO databases. Hub markers were screened out by five integrated machine learning algorithms and verified by laboratory experiments. Single-cell sequencing data were utilized to unravel cell-type specific expression levels of hub genes. An artificial neural network model was constructed to discriminate periodontitis from healthy controls. An unsupervised consensus clustering algorithm revealed mitochondrial dysfunction-related periodontitis subtypes. The immune and mitochondrial characteristics were calculated using CIBERSORTx and ssGSEA algorithms. Two hub mitochondria-related markers (CYP24A1 and HINT3) were identified. Single-cell sequencing data revealed that HINT3 was primarily expressed in dendritic cells, while CYP24A1 was mainly expressed in monocytes. The hub genes based artificial neural network model showed robust diagnostic performance. The unsupervised consensus clustering algorithm revealed two distinct mitochondrial phenotypes. The hub genes exhibited a strong correlation with the immune cell infiltration and mitochondrial respiratory chain complexes. The study identified two hub markers that may serve as potential targets for immunotherapy and provided a novel reference for future investigations into the function of mitochondria in periodontitis.

Keywords: Machine learning; bioinformatics; immune microenvironment; mitochondria; mitochondrial dysfunction; periodontitis.