The rgpA is a P. gingivalis–specific gene, one of the various virulence gene complexes of P. gingivalis, known to increase the risk of periodontal disease by inhibiting the host immune defense system and contributing to the initial inflammatory response and host oral tissue damage around the gums [14].

We obtained the sequence information (FASTA format) of the rgpA gene coding sequence region (CDS) for 22 different P. gingivalis strains (at the strain level) on the National Center for Biotechnology Information (NCBI DB) reference database (Supplementary Table 1).

For the selection of primer sequence that can detect all 22 different P. gingivalis strains, we performed the multiple sequence alignment (MSA) of each CDS information using the BioEdit 7.2v software (Fig. 1A). We selected two consistent regions confirmed through the MSA method as forward and reverse primer sequences (Table 1).

We pre-validated the suitability of the experimental application (Tm value, GC%, and potential for primer-dimer to form, etc.) about the selected primer pair using the ‘Oligo calc (http://biotools.nubic.northwetern.edu/)’ and ‘Oligo Analysis (http://www.operon.com/tools/oligo-analtsis-tool.aspx)’ open web tool. Next, we confirmed the P. gingivalis–specificity about each primer sequence and amplification region contained in each primer pair through the NCBI Nucleotide BLAST (https://blast.ncbi.nlm.nih.gov) (Fig. 1B).

Finally, we designed a specific probe in the region between each selected primer sequence (Table 1, Fig. 1A). The fluorescent reporter dye and quencher applied to the probe design are 6-FAM (6-carboxyfluorescein) and SFCQ1 (SFC probe, South Korea), which were attached to the 5' and 3' end regions on the selected probe sequence, respectively.

First, we performed PCR validation to confirm whether the pre-designed primer binds specifically to the rgpA gene CDS region on the P. gingivalis genome (Fig. 2). As a result of PCR validation, we confirmed that a correct target amplicon DNA band (approximately 110 bp) was located within samples corresponding to positive control and periodontitis groups. In contrast, amplicon DNA was not identified within samples corresponding to the negative control and healthy group.

Next, we performed qRT-PCR to confirm the relative bacterial load of the P. gingivalis within each comparison group (Table 4, Fig. 3, Supplementary Table 2). Prior to the experiment, the concentration of all double-strand DNA samples was checked through the Qubit Fluorometer 4.0 v at the Center for Bio-medical Engineering Core Facility (Dankook University, South Korea) and 1× dsDNA HS Assay kit (Thermo Fisher Scientific, Waltham, MA, USA), and then normalized to the consistent concentration (10 ng/μL). This process is essential to compare the Ct value (cycle threshold) reflecting the relative frequency of P. gingivalis on the same mDNA concentration within saliva samples of each clinical subject (healthy and periodontitis). The same experimental conditions were set by normalizing the other microbial genomic DNA concentrations (positive and negative controls) equally with the clinical comparison group. After normalizing all DNA concentrations, to confirm the relative bacterial load of P. gingivalis as various DNA concentrations range, we performed qRT-PCR by diluting each template DNA (10¹, 10⁰, 10^-1, 10^-2) through 10-fold serial dilution method (using CFX Opus 96 Real-Time PCR System, Bio-Rad, Hercules, CA, USA; The running condition of the qRT-PCR is as follows: pre-denaturation 95.0°C, 10 min; denaturation 95.0°C, 10 s; annealing 64.0°C, 15 s; elongation/flour detection 72.0°C, 15 s; and total PCR cycle was 45). As a result of qRT-PCR, we confirmed through the Ct value measurement results that the P. gingivalis rgpA gene was detected from all diluted DNA samples within the positive control and periodontitis groups. In contrast, P. gingivalis detection through confirming rgpA gene amplification could not be confirmed within the healthy and negative control groups (denoted ‘N/A’ in Table 4 and Supplementary Table 2). Although high-average Ct values (31.01_10 ng; 33.31_1 ng; 36.98_0.1 ng; 39.21_0.01 ng) were calculated from one sample in the periodontitis group (Case_04), our data show that P. gingivalis with low relative abundance within saliva sample can also be detected via qRT-PCR.

Finally, we checked the relative bacterial frequency of P. gingivalis within all samples of the healthy and periodontitis groups (four samples per each group) through the NGS-based 16S metagenome profiling analysis using the QIIEM2 microbiome analysis pipeline and compared it with the qRT-PCR result (Supplementary Table 3). The 16S microbiome analysis about eight clinical samples was performed through bacterial 16S V3-V4 amplicon sequencing using pre-extracted mDNA samples (Table 4), and the Illumina Miseq sequencer platform (300PE sequencing, Illumina, San Diego, CA, USA) was applied to the generation of raw sequencing data. We analyzed the paired-end 16S V3-V4 raw sequencing read files (i.e., fastq.gz files) using two bioinformatics pipelines: QIIME2 and DADA2 (version 1.2.0). The executing software for QIIME2 was Python software and for DADA2 was R (version 3.3.2). First, the sequencing reads were sorted using unique barcodes for each PCR product. The barcode, linker, and primer sequences were then removed from the original sequencing reads. And the removed reads were merged paired-end reads using FLASH v 1.2.11. The merged reads containing two or more ambiguous nucleotides, those with a low-quality score (average score < 20), or reads shorter than 300 bp, were filtered out. All pre-filtered sequence data quality control and feature table construction were completed using the Divisive Amplicon Denoising Algorithm 2 (DADA2) in the QIIME 2 plugin, which detects and corrects amplicon errors and filters out PhiX chimeric sequences. The pre-processed reads from each sample were used to calculate the number of Amplicon Sequence Variants (ASVs). The number of ASVs was determined by clustering the sequences from each sample using a 99% sequence identity cut-off using DADA2 pipeline-based QIIME2 software. Taxonomic relative abundance was counted with Naïve Bayes classifier using a confidence threshold of 0.7% derived from the ASVs data for each sample. The classified microbial composition based on SILVA v138 16S rRNA gene database was normalized using the value calculated from the taxonomy abundance count divided by the number of pre-processed reads for each sample. The relative bacterial frequency of P. gingivalis in each group identified through microbiome analysis was as follows: Healthy group (Normal_01: 0.00, Normal_02: 0.00, Normal_03: 0.00, and Normal_04: 0.00); Periodontitis group (Case_01: 0.00, Case_02: 0.16, Case_03: 0.16, and Case_04: 0.00). As with the qRT-PCR results, we confirmed that P. gingivalis was not detected within all samples of the healthy group. However, unlike the qRT-PCR results, we identified that the relative bacterial frequency of P. gingivalis was not calculated within two samples (Case_01 and Case_04) of the periodontitis group. These results demonstrated that qRT-PCR also sensitively captured the low abundance for P. gingivalis, which was not confirmed by the NGS-based metagenome sequencing method.