Antibiotic exposure enriches streptococci carrying resistance genes in periodontitis plaque biofilms

Subjects

Patients with periodontitis and periodontally healthy subjects aged 18~65 years were recruited by a qualified periodontist from Peking University School and Hospital of Stomatology from 2018~2019. This study was approved by the Ethics Committee of Peking University Hospital of Stomatology (PKUSSIRB-201837098), and consent was obtained from all participating patients in accordance with the approved ethics protocol.

The inclusion criteria of the periodontitis patients were: no systemic disease, clinical attachment loss detected in the interproximal surfaces of more than two non-adjacent teeth, with a probing pocket depth ≥6 mm (Papapanou et al., 2018). The inclusion criteria of the healthy subjects were: no systemic or oral disease, probing depth ≤3 mm, bleeding on probing in <10% of sites, no attachment loss, and antibiotics-free within 3 months (Chapple et al., 2018). The exclusion criteria were pregnancy, smoking, systemic disease or antimicrobial therapy during the past 3 months, or periodontal therapy within 1 year. Examinations were conducted by a specialized periodontist.

Sample collection and cultivation procedures

Recruited volunteers were requested no food and water for 2 h before sampling. Subgingival plaque samples were collected from the bottom of the periodontal/gingival pocket using a sterile Gracey curette. For each periodontitis patient, plaque from four diseased sites in different quadrants with a probing pocket depth ≥6 mm and attachment loss >0 mm. For each periodontally healthy control, plaque was collected from four healthy, no probe-bleeding sites in different quadrants. The plaque (four sites) from one volunteer was pooled in individual tube containing 500 μL phosphate buffer solution (PBS, Hyclone, Logan, UT, USA), transferred to the lab with ice box and processed within 30 min after sampling. Brain Heart Infusion medium supplemented with haemin (5 mg/L) and vitamin K (1 mg/L) (BHI-HV) was used as culture broth. Antibiotics were purchased from Solarbio (Beijing, China). Samples were shaken for 30 s and diluted 1:100 in 3 mL BHI-HV with or without amoxicillin (2 μg/mL), metronidazole (16 μg/mL), clindamycin (4 μg/mL), and tetracycline (8 μg/mL). The concentration values for these antibiotics were defined based on previous periodontal microbiology studies (Feres et al., 2002; van Winkelhoff et al., 2005; van Winkelhoff et al., 2000). Since we found the MIC of metronidazole against Fusobacterium nucleatum ATCC 25586 was 16 μg/mL in another experiment, we choose 16 μg/mL as the breakpoint concentration value for metronidazole instead of 8 μg/mL that reported in the reference. All incubation was performed in an anaerobic chamber (5% H₂, 5% CO₂, and 90% N₂) at 37 °C for 48 h. For each sample, 500 μL cultures were frozen in 10% skim milk at −80 °C for further isolation experiments.

16S rRNA gene amplicon and metagenomic sequencing

Bacteria were harvested by centrifugation (8,000 rpm, 10 min), followed by DNA extraction using a QIAamp DNA Mini Kit (Qiagen, Hilden, Germany). The quality was identified by 1% agarose gel and Qubit Fluorometer (Invitrogen, Waltham, MA, USA). High-quality DNA (OD₂₆₀/OD₂₈₀ = 1.8–2.0) with concentration >20 ng/μL were used for 16S rRNA gene V3-V4 region amplification. Additional details in sequencing and data processing are provided in SI Materials and Methods. Considering the gnomic quality of samples and meanwhile the patients must be resistant to the two antibiotics, so we selected five periodontitis patients with 15 samples to conduct metagenomic shotgun sequencing (five samples from amoxicillin treatment, five samples from tetracycline treatment and five samples from non-antibiotics treatment). The raw sequencing data used in this study have been deposited in the NCBI SRA database under PRJNA952804. Additional details in sequencing, metagenomic analysis and antibiotic resistance gene identification are provided in SI Materials and Methods.

Isolation of tetracycline-sensitive and -resistant strains

To isolate tetracycline-sensitive strains, approximately 50 μL frozen stocks of bacteria cultured in medium without antibiotic were spirally spread on agar plates of BHI-HV supplemented with 5% sheep blood (BHI-HVB) and incubated at 37 °C for 4 days. Colonies of different morphologies and colors were picked, and each colony was spread on BHI-HVB agar without or with 8 μg/mL tetracycline and incubated as above. Strains that grew only on BHI-HV agar were considered tetracycline sensitive, while strains grew on BHI-HVB agar without or with 8 μg/mL tetracycline were considered tetracycline-resistant. For another tetracycline-resistant strains, 50 μL freezer stocks of bacteria cultured in tetracycline-containing medium were spirally spread on BHI-HVB agar and incubated at 37 °C for 4 days. Grown strains were all considered resistant. Individual colonies were cultured in BHI-HV medium, identified by 16S rRNA gene Sanger sequencing (27F, 1492R), and freezer stocks were prepared for storage.

In vitro evaluation of resistance phenotypes and tetM fragments

An agar dilution method was used for antibiotic-susceptibility testing. Additional details are provided in SI Materials and Methods. Isolates with different tetracycline sensitivities were cultured at 37 °C overnight, and genomic DNA was extracted using a DNA extraction Kit (Qiagen, Hilden, Germany). Polymerase chain reaction (PCR) amplification was performed using the genomic DNA as the template to confirm the presence of the tetM fragment. The primer sequences are listed in Table S1.

Quantification of gene expression

Bacteria were cultivated with or without tetracycline (64 μg/mL) to the logarithmic phase respectively, and were harvested by centrifugation (12,000 rpm, 1 min), resuspended in 1 mL TRIzol reagent, heated at 95 °C for 15 min, and the remaining steps were conducted according to the manufacturer’s instructions. The quality was identified by Nanodrop 8000 and Qubit Fluorometer (Invitrogen, Waltham, MA, USA). The high-quality RNA (OD₂₆₀/OD₂₈₀ = 1.8–2.0 and concentration > 300 ng/μL) was reverse transcribed to cDNA using ReverTra Ace qPCR RT Master Mix (Toyobo, Osaka, Japan). Quantitative real-time PCR was performed using SYBR Green Master Mix (Foreverstar Biotech, Beijing, China) in an ABI 7500 system (Thermo Scientific, Waltham, MA, USA). The primer sequences are listed in Table S1. The reaction volume was 20 μL, and the conditions were initial denaturation for 10 min at 95 °C, followed by 40 cycles of 10 s at 95 °C and 1 min at 55 °C for fluorescence collection, with a final extension for 1 min at 72 °C. The tetM gene of bacteria without tetracycline treated was used for normalizing. All assays were conducted in duplicate.

Sequencing data analysis

16S rRNA gene amplicon sequencing data were processed using QIIME v1.9.1 (Caporaso et al., 2010). Metagenomic sequencing data were assembled into contigs using MEGAHIT (Li et al., 2015), and annotated against comprehensive antibiotic resistance database (CARD) 3.0.029 (Alcock et al., 2023) using DIAMOND (Buchfink, Xie & Huson, 2015). Reads coverage was evaluated by using Bowtie v10.3.0 (Langmead & Salzberg, 2012) and SAMtools (Li et al., 2009).

All gene sequences from human microbiome project (HMP) database (https://www.hmpdacc.org/HMASM/#data) were downloaded and blasted against tetM gene to screen out the target sequences. A total of 305 metagenomic sequenced samples associated with periodontitis and healthy statuses and their corresponding clinical information was obtained to assess the distribution of tetM (Table S2). Additional details in data analysis are provided in SI Materials and Methods.

Statistics

Alpha diversity and Bray-Curtis distances were calculated using the R package Vegan. Principal coordinate analysis (PCoA) was performed, with the significance was determined by the nonparametric Adonis test (P < 0.05). LEfSe was used to identify the enriched or depleted bacteria between groups. Linear discriminant analysis (LDA) score >2.0 were considered as a discriminatory signature between different antibiotic treatments. Significance comparisons between groups were conducted by t-test and the multiple comparison FDR control method (Benjamin-Hochberg).

The effect of antibiotics on oral microbiota in periodontitis

Ten patients with periodontitis and eleven periodontally healthy subjects aged 18~51 years were included. The gender (male/female) distribution was 1:1 and the age and gender were calculated with no significant difference between the two groups.

By cultivation, there were more drug-resistant samples from the periodontitis group (P-group) compared with the periodontally healthy group (H-group) after exposure to each of the four antibiotics. All P-group samples exhibited growth when exposed to different antibiotics, while all H-group samples were susceptible to amoxicillin, six were susceptible to tetracycline, two were susceptible to clindamycin, and one was susceptible to metronidazole. Totally, 85 samples (including 21 without antibiotic treatment) with bacterial growth were enrolled for 16S rRNA amplicon sequencing and strain isolation (Fig. 1A and Table S3).

As a result, in the P-group, the microbial diversity decreased significantly after exposure to amoxicillin, clindamycin, and tetracycline, but not metronidazole, compared with H-group (p = 0.038, 0.015, 0.002, and 0.148, respectively, t-test). In contrast, no significant differences were observed in the H-group (p = 0.584, 0.112, and 0.506 for metronidazole, clindamycin and tetracycline, respectively) (Fig. 1B). We next measured the inter-group community dissimilarities pre- and post- antibiotic exposure based on PCoA and pairwise Bray–Curtis distance. The results showed that the samples of H-group after exposure to antibiotics were closed to the control, while the samples of P-group were separated from their control (Figs. 1C and 1D).

Periodontitis-specific antibiotic-resistant bacteria flourish after antibiotic exposure

In view of the different response of healthy and periodontitis-derived microbiota to antibiotic exposure, we then compared bacterial taxa between these two group with and without antibiotics. Firmicutes were the most abundant phylum regardless of antibiotic treatment or not, and Bacteroidetes and Fusobacteria, on the other hand, showed periodontitis- and antibiotic-specificity and were enriched only in the P-group (Fig. 2A).

There were remarkable inter-group variations in the bacterial genera resistant to each antibiotic based on LDA of the effect size. After exposure to amoxicillin, Prevotellaceae, Flavobacteriia, and Leptotrichiaceae of Bacteroidetes were enriched. Firmicutes, Aerococcaceae, Carnobacteriaceae, Streptococcaceae, and Lactobacillales bacteria were related to tetracycline resistance, which indicated that there was heterogeneity in antibiotic resistance among phylum members (Fig. 2B). Then, the relative abundance of genera in groups with and without antibiotic treatment was calculated. The prevalence of Veillonella, Prevotella, Granulicatella, Haemophilus genus were abundant in all the antibiotic-exposed periodontitis groups. Among the healthy subjects, only the genera Virungella, Granulopsis and Klebsiella had high relative abundance after exposure to clindamycin and tetracycline, while the genera Granulopsis, Actinomyces and Klebsiella had high abundance after exposure to metronidazole (Fig. 2C). Streptococcus were abundant in both groups, regardless of antibiotic exposure or not, which was the main taxon of tetracycline resistance in most samples except P8 (Fig. S1). Based on the 16S rRNA sequencing results, the diversity and relative abundance of genus was significantly decreased after amoxicillin and tetracycline exposed, which indicated amoxicillin and tetracycline were the most sensitive antibiotic in clinical therapy.

Antibiotic exposure drives heterogeneity of antibiotic resistance genes carried by periodontal bacteria

Then we further investigated the causes of heterogeneity in antibiotic resistance in the oral microbiome. A total of 15 samples that met the requirements for metagenomic sequencing (P1, P3, P7, P8, and P10 with or without amoxicillin- and tetracycline-exposure in P-group) were used to analyze the effects of amoxicillin- and tetracycline-exposure on antibiotic resistance genes (ARGs) generation to uncover the causes of resistance differences. We assembled metagenomic reads into 2.1 × 10⁶ contigs under the length cutoff of 500 bp. After gene prediction and ARGs annotation, 741 contigs were annotated with 149 ARGs among all the five samples (Fig. 3). Then we calculated the abundance of ARGs belong to each gene based on reads mapping coverage. We found that compared with control, amoxicillin-exposed samples were virtually unchanged, and four genes in tetracycline-exposed samples were significantly enriched (t-test, adjusted p < 0.1) (Fig. 3). Two of them were tetracycline-resistance gene tetM (Table S4). Accordingly, we next focus on tetM mediates tetracycline resistance.

tetM gene in Streptococcus spp. accounts for the tetracycline resistance

In order to measure the location of tetM-mediated tetracycline resistance in the body parts and which bacteria in humans are more likely to carry the gene, we retrieved tetM gene sequences from the HMP database and tracked their bacterial sources and the distribution of body parts. As a result, tetM can be detected in multiple body sites. The oral cavity was its main reservoir, especially for supragingival and tongue dorsum. Conversely, the detection rate was very low in stools. In addition, tetM was found to be widely distributed in various genera. For each body site, tetM carriage was restricted to a few genera. In oral cavity, tetM was mainly carried by Streptococcus (Fig. 4A).

Considering that Streptococcus genus harboring tetM might be resistant to tetracycline, enabling them to proliferate in the presence of tetracycline in oral microbiota, we isolated Streptococcus strains with different tetracycline-resistance phenotypic characteristics (Fig. 4B). Resistant strains were isolated mainly from tetracycline-exposure samples in P-group (Table S5). Resistant and sensitive strains of the same species co-existed in several samples (e.g., P1S3 vs. P1m8; P7S2 vs. P7m8), indicating strain-level differences in resistance. As expected, all isolates lacking tetM were sensitive to tetracycline, while those with tetM were resistant, except one SI intermedius strain (H11-2) isolated from sample H11, which contains tetM but was sensitive (Table S5 and Fig. S2).

To verify whether the tetM gene is also periodontitis-specific, 16 sequences were recruited from our isolates, which were Streptococcus genus, and found that tetM from periodontitis and healthy sources were priority clustered separately. While tetM from the same strain were not showed homologous (Fig. 5A). We also analyzed the relative abundance of tetM in 305 metagenomic data of periodontally healthy subjects and periodontitis patients. Regardless of the sample type (saliva or plaque), the amount of tetM carried by healthy population was relatively stable, while the distribution of tetM among periodontitis patients varied greatly, with some patients carried more tetM genes (Fig. 5B).

To avoid inter-species discrepancy, we used isolates within the same species for comparison to test whether the periodontitis-derived tetM mediate stronger tetracycline-resistance. A total of 22 isolates belonging to five Streptococcus species were selected for MIC test. All tetracycline-resistant SI sanginosus isolates had the same MIC (32 μg/mL). While within the same species, SI oralis sub. tigurinus, SI salivarius, and SI intermedius isolates had varied MICs (ranged from 8 to 64 μg/mL) (Table S6), and the MICs of periodontitis-derived strains were higher or equal to that of healthy-derived strains. Finally, we evaluated the expression of tetM in strains with higher tetracycline exposure (64 μg/mL). The results showed that the expression of tetM at the transcriptional level in different strains was consistent with the resistant phenotypes of the strains, and strains derived from periodontitis expressed higher or equal to that of healthy-derived (Fig. 5C).