Subgingival microbiome in patients with healthy and ailing dental implants

Dental implants are commonly used to replace missing teeth. However, the dysbiotic polymicrobial communities of peri-implant sites are responsible for peri-implant diseases, such as peri-implant mucositis and peri-implantitis. In this study, we analyzed the microbial characteristics of oral plaque from peri-implant pockets or sulci of healthy implants (n = 10), peri-implant mucositis (n = 8) and peri-implantitis (n = 6) sites using pyrosequencing of the 16S rRNA gene. An increase in microbial diversity was observed in subgingival sites of ailing implants, compared with healthy implants. Microbial co-occurrence analysis revealed that periodontal pathogens, such as Porphyromonas gingivalis, Tannerella forsythia, and Prevotella intermedia, were clustered into modules in the peri-implant mucositis network. Putative pathogens associated with peri-implantitis were present at a moderate relative abundance in peri-implant mucositis, suggesting that peri-implant mucositis an important early transitional phase during the development of peri-implantitis. Furthermore, the relative abundance of Eubacterium was increased at peri-implantitis locations, and co-occurrence analysis revealed that Eubacterium minutum was correlated with Prevotella intermedia in peri-implantitis sites, which suggests the association of Eubacterium with peri-implantitis. This study indicates that periodontal pathogens may play important roles in the shifting of healthy implant status to peri-implant disease.


Subject recruitment
Ten individuals with healthy peri-implant sites (n = 10), eight cases with PM (n = 8), and six cases with PI (n = 6), participated in the study. All subjects were medically healthy; did not suffer from any systemic illness; were not pregnant; did not have diabetes; and had not taken antibiotics, anticoagulants, or non-steroidal anti-inflammatory drugs in the 6 5 months prior to the study. All were non-smokers. All subjects were partially edentulous patients due to severe periodontitis. They had received initial periodontal therapy (scaling and root planning) and periodontal surgery (if required). All patients commenced well-supervised maintenance care (supportive periodontal therapy) before implant treatment. Straumann Dental Implant System (Straumann, California, USA) was used, and the implants functioned for at least 1 year after the prosthesis were adopted. The project was approved by the Peking University Biomedical Ethics 10 Committee (Beijing, China). Subjects gave written informed consent with the approval of the Ethics Committee of the Peking University School and Hospital of Stomatology.

Diagnosis and sample collection
Oral examination and diagnosis were performed by one dentist, using visual, probing, and radiographic methods. 15 Intra-oral periapical radiographs were obtained using the parallel technique. Dogora imaging software was used for analysis of peri-implant bone loss by the same examiner. Average bone level on the mesial and distal aspect of each implant was accessed, using the implant-abutment junction as the reference point. Dimensional distortions and enlargements on the radiographs were adjusted. The diagnostic criteria for peri-implant diseases were in accordance with the recognized definitions of PM and PI 1 . In brief, peri-implant tissue that did not bleed on probing, which was not 20 suppurating, and for which radiography yielded no evidence of marginal bone loss, was classified as healthy. PM was diagnosed when an implant showed clinical signs of inflammation but no evidence of bone loss. PI was diagnosed based on loss of marginal bone in conjunction with inflammation of the peri-implant mucosa, as evidenced by bleeding and/or suppuration after probing. The clinical signs of inflammation for this study include bleeding on probing, increased probing depths, mucosal swelling/hyperplasia and mucosal recession 2 . Plaque samples were collected from peri-implant 25 sulci or pockets, at the maximum possible probing depth, using a sterile metal periodontal probe. Samples were suspended in 1-mL sterile tubes containing 200-μL amounts of TE buffer (20 mM Tris, 2 mM EDTA; pH = 7.4) and frozen at -80C prior to DNA isolation.
Microbial DNA extraction, 16S rRNA gene library preparation, and pyrosequencing 30 DNA from plaque samples was extracted using a TIANamp Bacteria DNA Kit (Tiangen Biotech, Beijing, China), following the manufacturer's instructions after initial treatment with lysozyme (20 mg/mL, 37C for 1 h). DNA concentrations were measured using a Qubit Fluorometer (Invitrogen, California, USA) and via qPCR. The amount of DNA per sample was 0.24-1.62 μg.
The v1-v3 hypervariable regions of bacterial 16S ribosomal RNA genes were amplified via PCR. The PCR primers were 35 27f: 5′-AGAGTTTGATCCTGGCTCAG-3′ 3 , and 534r: 5'-ATTACCGCGGCTGCTGG-3' 4 , with 10-nt barcodes tagged to the 5′-ends. PCR was performed as described in the manual of the GS FLX Amplicon DNA library preparation method (Roche, Mannheim, Germany). Briefly, genomic DNAs were used as templates. Cycling involved initial denaturation at 94C for 3 min; 30 cycles of 94C for 30 s, 57C for 45 s, and 72C for 60 s, followed by a final extension at 72C for 2 min. The libraries were pyrosequenced on a 454-GS-FLX sequencing platform (454 Life Sciences, Branford, USA) at the 40 BGI Institute (BGI Institute, Shenzhen, China).

16S data processing and statistical analysis
In total, 24 samples were sequenced, and the raw data (*.sff files) generated were analyzed using (principally) the pipeline tools MOTHUR 5 and QIIME 6 . In brief, sequences were demultiplexed based on a unique barcode assigned to each 45 sample. To filter low-quality sequences, those with average quality scores ≤25 and sequence lengths <200 nt were discarded. A maximum of one barcode correction was allowed at this stage, no primer mismatch, 6 ambiguous bases were permitted. Trimmed reads were clustered into operational taxonomic units (OTUs) at a 97% similarity cutoff using the de novo OTU selection strategy. Taxonomies were assigned by the RDP classifier (version 1.27), with a confidence threshold of 0.8 7 ( Figure S1). After we obtained OTU tables and phylogenetic trees, microbial richness estimators 50 (Observed OTUs, Chao1), evenness estimators (Equitability), diversity estimators (Shannon Index, Simpson Index), and phylogenetic distances (PDs), were calculated using Perl scripts. Fixed numbers of sequences were randomly selected from each dataset to generate rarefaction curves and allow microbial diversity to be estimated. Weighted UniFrac distances were estimated within and between groups, based on the OTU tables and the phylogenetic trees 8 . Relative abundances of microbial taxa at each of the phylum, class, order, family, genus, and species levels were calculated and 55 compared. The unpaired student's t-test was used to compare alpha and beta diversities. Differences in the relative abundances of taxa in healthy implant, PM, and PI samples were analyzed using the Wilcoxon rank-sum test. Differences in prevalence were compared using Fisher's exact test. P values < 0.05 were considered to indicate statistical significance.
For each group of samples, OTUs observed in at least half of the samples were used to construct an OTU network 9,10 . We calculated the Pearson correlation coefficients (PCC) for each pair of OTUs and used the permutation test to compute the 60 statistical significance of the PCC value. Edges were set between pairs of OTUs for which the PCC was significant (P<0.01).

Quantification of bacterial loads of the Eubacterium brachy subgroup
Bacterial loads of members of the Eubacterium brachy subgroup were determined via real-time PCR using modified 65 genus-specific primers (Forward: 5'-ACACGGTCCAAACTCCTACG-3', Reverse: 5'-TTCGCRTCCCAAATTCCG-3') 11 . First, 16S rRNA genes were amplified using universal bacterial primers (27f/1492r) and the PCR products purified with the aid of a TIANquick Midi Purification Kit (Tiangen Biotech, Beijing, China). The DNA levels were adjusted to 10 ng/μL; these solutions served as templates. Each PCR reaction was performed in a volume of 20 μL, containing 10 μL Power SYBR Green PCR Master Mix (Applied Biosystems, 70 Warrington, UK), 75 nM primers, and 1 μL (10 ng) DNA template. The qPCR cycling conditions were 95C for 2 min; followed by 40 cycles of 95C for 15 s and 60C for 1 min. PCR amplicons of the Eubacterium brachy subgroup served as standards (500 pg/μL, 50 pg/μL, 5 pg/μL, 500 fg/μL, and 50 fg/μL). The presence and specificity of qPCR products were evaluated by melting curve analysis and agarose gel electrophoresis. All samples and standards were amplified in triplicate, and mean values were used in the analysis. Student's t-test was used to 75 determine the significance of differences.