Conclusion
Gene biomarkers of periodontitis are highly predictive. The pro-inflammatory role of F. nucelatum is not related to butyrate synthesis.
Methods
Sixteen individuals were studied, from which six were diagnosed with chronic periodontitis. Illumina sequencing of the total messenger RNA (mRNA) yielded ~42 million reads per sample. A total of 324 human oral taxon phylotypes and 366,055 open reading frames from the HOMD database reference genomes were detected.
Results
The transcriptionally active community shifted from Bacilli and Actinobacteria in health to Bacteroidia, Deltaproteobacteria, Spirochaetes and Synergistetes in periodontitis. Clusters of orthologous groups (COGs) related to carbohydrate transport and catabolism dominated in health, whereas protein degradation and amino acid catabolism dominated in disease. The LEfSe, random forest and support vector machine methods were applied to the 2,000 most highly expressed genes and discovered the three best functional biomarkers, namely haem binding protein HmuY from Porphyromonas gingivalis, flagellar filament core protein FlaB3 from Treponema denticola, and repeat protein of unknown function from Filifactor alocis. They predicted the diagnosis correctly for 14 from 16 individuals, and when applied to an independent study misclassified one out of six subjects only. Prevotella nigrescens shifted from commensalism to virulence by upregulating the expression of metalloproteases and the haem transporter. Expression of genes for the synthesis of the cytotoxic short-chain fatty acid butyrate was observed by Fusobacterium nucleatum under all conditions. Four additional species contributed to butyrate synthesis in periodontitis and they used an additional pathway.