Alterations in the oral microbiome of individuals with a healthy oral environment following COVID-19 vaccination

Ethical approval for this study was granted by the Ethics Committee of the Institute of Preventive Medical Science, Health Sciences University of Hokkaido (Approval no. 2021–001). The methods used in the experiments were conducted in accordance with the Declaration of Helsinki. We enrolled 40 faculty and staff members of the Health Sciences University of Hokkaido who were scheduled to receive COMIRNATY® (Pfizer, NY, USA; BioNTech SE, Mainz, Germany). All individuals agreed to participate in this study and provided written informed consent. Individuals who were undergoing dental treatment or who had systemic conditions or smoking habits were excluded from the study. None of the individuals had contracted COVID-19. Among the participants, there were 22 males (mean ± SD years; 38.4 ± 12.3) and 18 females (32.2 ± 7.7). The number of participants in each age group was as follows: 20–29 years (n = 15); 30–39 years (n = 16); and over 40 years (n = 9). The decayed, missing, and filled teeth (DMFT) index was used to assess tooth condition according to the WHO-recommended criteria. According to the DMFT index, carious teeth received the letter D, missing teeth received the letter M, and filled or processed teeth that had not rotted received the letter F. DMFT was examined for 28 teeth, excluding the third molars. Periodontal examinations were performed using the WHO criteria [7]. The periodontal status of participants was recorded using the Community Periodontal Index (CPI). The index teeth (11, 16, 17, 26, 27, 31, 36, 37, 46, and 47) were examined. The scores adopted for CPI evaluation were as follows: 0, healthy periodontium; 1, presence of bleeding on probing; 2, presence of calculus; 3, 4–5 mm pocket; 4, ≥ 6 mm pocket.

Saliva samples were collected in conical tubes before vaccination (Before), 3 weeks after the first inoculation (After1), and 3 weeks after the second inoculation (After2). The samples were collected from the participants at their homes after they woke up. The participants were instructed to avoid gargling, eating, drinking, and brushing their teeth before sample collection (Fig. 1).

Genomic DNA was extracted from saliva samples using DNeasy Blood & Tissue Kits (Qiagen, Hilden, Germany), according to the manufacturer’s instructions. Amplicon PCR was performed to target the V3–V4 regions of the bacterial 16S ribosomal RNA (rRNA) gene [8, 9]. Sequencing libraries of the V3–V4 region were generated according to the 16S Metagenomic Sequencing Library Preparation instructions (Illumina, San Diego, CA, USA). In brief, KAPA HiFi HS ReadyMix (Nippon Genetics, Tokyo, Japan) and V3–V4 region primers (forward, 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG-3′ and reverse, 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC-3′) were used for the amplicon PCR, and KAPA HiFi HS ReadyMix and Nextera XT index kits (Illumina) were used for the index PCR. The library was diluted, mixed with PhiX (Illumina), and then sequenced using an Illumina MiSeq system with a MiSeq reagent kit v3 (600 cycles, Illumina). The sequencing depth was determined to be 14,000 reads from alpha rarefaction.

Metagenomic sequencing data were analyzed using the software package Quantitative Insights into Microbial Ecology 2 (QIIME2 v2020.2) against the 16S rRNA gene sequences that were assigned to the 16S rDNA database (Greengenes v13.8) [8, 9]. Analysis of the amplicon sequence data employed the DADA2 pipeline. Alpha diversity was estimated using the observed OTUs, Faith’s phylogenetic diversity (PD), and Shannon index. Statistical significance was set at p < 0.05. Beta diversity was evaluated based on UniFrac distances representing the fraction of the branch length of the phylogenetic tree shared between the groups. Three-dimensional principal coordinate analysis (PCoA) was used to generate UniFrac scatterplots to visually compare microbial compositions across groups. The differences in bacterial communities among the Before, After1, and After2 groups were analyzed using the unweighted and weighted UniFrac distance metric. Permutational multivariate analysis of variance (PERMANOVA) was used with the unweighted and weighted UniFrac distance matrix to determine significant differences in microbial communities between the different groups. Statistical significance was set at p < 0.05. All results are presented as uncorrected p-values. Significant differences in microbial taxon abundance among the Before, After1, and After2 groups were analyzed using the analysis of composition of microbiomes (ANCOM) in QIIME2. The final significance was expressed as the empirical distribution of W. Feature volatility analysis was performed to determine the linear mixed-effects (LME) model results for bacterial taxonomy. Several bacterial taxa were selected for subsequent analysis, depending on their importance, which were analyzed using the LME model to detect important bacterial taxonomies [10]. The phylogenetic investigation of communities by reconstruction of unobserved states (PICRUSt2) software was used to predict functional abundances based only on marker gene sequences. The statistical analysis of metagenomic profiles (STAMP) software package was used to analyze the PICRUSt2 results. Welch's t-test was used for two-group comparisons, and the Welch's inverted confidence interval method was used to calculate the confidence interval. The Benjamini–Hochberg false discovery rate method was used to calculate the adjusted p-value (p < 0.05).

The mean DMFT values were D (mean ± SD teeth; 0.4 ± 1.0), M (0.2 ± 0.5), and F (5.2 ± 4.0). The average DMFT index before and after vaccination did not change for any participant (Fig. 2). To understand the actual condition of periodontal disease, CPI was used. A comparison of CPI before and after vaccination revealed little change in CPI and no change in the number of participants with code 4 (Fig. 2).

All 120 samples were sequenced using MiSeq, and 7,073,846 total sequences were amplified from the Before, After1, and After2 groups, ranging from a minimum of 14,347 to a maximum of 175,709 sequences per sample, with a mean of 52,150 sequences per sample. The sequence data generated and analyzed in this study were deposited in the DNA Data Bank of Japan (DDBJ) under the BioProject accession number PRJDB12809. The following summarizes our data: PSUD ID; PSUB 016463; and PRJDB12809.

To evaluate the mean diversity of species in different sites or habitats within a local scale in each participant, alpha diversity was analyzed. The observed OTUs were significantly higher in the After1 and After2 groups than in the Before group (p = 0.040 and p = 0.007, respectively; Fig. 3a). No significant differences were observed in the Faith’s PD and the Shannon index among the Before, After1, and After2 groups (Fig. 3b, c).

To evaluate the diversity differences among the Before, After1, and After2 groups, PCoA of UniFrac distance was performed. PCoA plots demonstrated clustering among the groups. The weighted UniFrac distance metric differed significantly between the Before and After2 groups based on PERMANOVA (p = 0.020; Fig. 3d). The unweighted UniFrac distance metric differed significantly between the Before and After2 groups and between the After1 and After2 groups based on PERMANOVA (p = 0.032; Fig. 3e).

QIIME2 detected a total of 142 different bacterial genera in the Before, After1, and After2 groups. The most abundant genus among all samples was Prevotella, followed by Neisseria, Veillonella, and Streptococcus (Fig. 3f). At the genus level, the ANCOM test revealed Bacteroides as a differentially abundant genus among the Before, After1, and After2 groups (W = 138); the genus was present at a lower proportion in the After1 and After2 groups than in the Before group (Table 1). Feature volatility and LME analyses were performed on the highly important bacterial genera identified by the volatility plot. Four bacterial genera were identified as being significantly differentially abundant among the Before, After1, and After2 groups (p < 0.05). Vaccination increased the abundance of the genera Lachnoanaerobaculum, Moryella, Parvimonas, and Peptostreptococcus (Table 2).

In all, 60 MetaCyc pathways were significantly altered in the After2 group compared with those in the Before group. Twenty-eight MetaCyc pathways, including GLYCOGENSYNTH-PWY and PWY-6470, were increased, whereas 32 pathways, including PANTO-PWY and PWY-5659, were decreased with COVID-19 vaccination (Fig. 4). Additionally, the analysis identified the highest number of pathways in the After2 samples compared with those obtained in the Before samples.