The present study was approved by the ethics committee of Linfen Central Hospital (Ethics Approval No. 2021-42-1) and was performed in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants prior to clinical data collection and sampling.

A total of 112 participants were recruited at Linfen Central Hospital, including 31 healthy controls (HC group), 24 periodontitis patients without COPD (P group), 28 COPD patients without periodontitis(COPD group), and 29 patients with both periodontitis and COPD(P_COPD group). The diagnosis and assessment of the severity of COPD were made according to the recommendations of the Global Initiative for Chronic Obstructive Lung Disease (GOLD) committee (Vogelmeier et al., 2017). The diagnosis and assessment of the periodontitis were based on the new classification, Classification of Periodontal and Peri-implant Diseases and Conditions (Tonetti et al., 2018). Other inclusion criteria included: (1) aged ≥18 years; and (2) Periodontitis from stage II to IV, grade B. The exclusion criteria were antibiotic using before during the last three months, other systemic diseases, administration of periodontal therapy during the last three months (Cai et al., 2021). General participant demographics, including age, gender, blood routine records, pulmonary function test results and clinical treatments were collected from medical record system using a standard form.

Before sample collection, the participants were asked to rinse their mouth for removing the food residues and debris. We obtained oral samples from four first incisor teeth and four first molar teethof each participant. The first molars and incisors are the main sites of periodontal lesions, and we selected 11, 21, 31, 41, 16, 26, 36, 46 as the main sampling sites based on previous periodontal microbiology studies (Zhou et al., 2020). Clinical examination was performed before sampling to ensure that the sampling site clinical attachment loss (CAL) ≥ 3mm, probing depth (PD) ≥ 4mm and bleeding on probing. If one of these teeth was missing, the adjacent tooth was collected. After drying the target sites, gingival crevicular fluid (GCF) samples were collected with sterile absorbent paper points from gingival sulcus of each tooth. After removal of supragingival plaque, subgingival plaque (SP) samples were collected with sterile Gracey curettes from the buccal and lingual sides of each tooth. The sample of each participant was collected in the eppendorf tube. All oral specimens (subgingival plaque and gingival crevicular fluid) were stored in -80°C until DNA extraction.

Total bacterial DNA was extracted from oral samples using SteadyPure Bacterial Genomic DNA Extraction Kit(Accurate Biotechnology(Hunan)Co,Ltd,China) following the manufacturer’s instructions. Hypervariable regions (V2, V3, V4, V6-7, V8 and V9) of the 16S rRNA were amplified using two primer sets in the Ion 16STM Metagenomics Kit (ThermoFisher Scientific, UK). XP beads were used to purify the amplification products and quantified by Qubit4 (ThermoFisher Scientific, USA). Purified amplicons were ligated with barcodes and then generated for the libraries. Then the libraries were pooled in equimolar amounts on chip 530 and sequenced to single-end, 250-base-pair reads on an Ion GeneStudio S5 System (ThermoFisher Scientific, USA) based on Ion Reporter metagenomics workflow (Ion 16S mNGS). Quality filtering, trimming and dereplication of raw sequencing reads were conducted automatically on Ion Reporter metagenomics workflow, relying on default parameters. Unaligned binary data files (Binary Alignment Map, BAM) generated by the Ion Torrent PGM were uploaded to Ion Reporter and analyzed using default settings (Malczynski et al., 2021).

Quantitative variables conforming to normal distribution were presented as the mean ± SD analyzed by Student’s t test and analysis of variance (ANOVA), while Quantitative variables of non-normal distribution were presented as median and interquartile ranges (25th and 75th percentiles) and analyzed by Mann-Whitney U or Kruskal-Wallis test. Categorical variables were presented as rate or percentage, and chi-square test or Fisher test were used to analysis. The alpha diversity was evaluated using the Chao-1, Shannon, abundance-based coverage estimator (ACE) and Simpson indices, respectively. The beta diversity has been evaluated through principal coordinates analysis (PCoA) ordination of variance and compared using Bray-Curtis dissimilarity. Differential species among groups was explored with the linear discriminant analysis (LDA) effect size (LEfSe) method (Shi et al., 2021) and DESeq2 analysis (Lu et al., 2022). The microbiome phenotypes were predicted by BugBase (Ward et al., 2017). The BugBase phenotype predictions were implemented using the online web page https://bugbase.cs.umn.edu/index.html. Prediction of the abundances of functional categories was conducted using PICRUSt2 (Douglas et al., 2020). Statistics and visualization of functional data were depicted using STAMP (Chowdhry et al., 2018). P<0.05 was considered as statistically significant.

A total of 112 participants were enrolled in our study, and the basic characteristics of each group are listed in Table 1 . There were no differences among the groups except for gender, age and smoking. The healthy control (HC) group’s median age was significantly younger than the diseased groups. The median age of the comorbid (P_COPD) group was highest. Moreover, the healthy group had a higher proportion of female participants. Significant difference of smoking was only existed between healthy control group and periodontitis group. Gender, age and smoking status were treated as confounding factors which were corrected in the difference analysis( Figure S4A ). There were no significant differences in the GOLD grade and clinical indicators between the COPD and P_COPD groups.

All sequencing data for the four groups reached saturation at approximately 50,000 reads ( Figure S1 ). For alpha diversity, the Chao1 index in the subgingival plaque (SP) samples from the periodontitis group was significantly higher than that in the HC group ( Figure 1A , P = 0.0245). In the gingival crevicular fluid (GCF) samples, the Chao1 index of periodontitis group was significantly higher than that of the COPD group ( Figure 1C , P = 0.0068) and P_COPD group ( Figure 1C , P = 0.0063). However, no significant difference was found in the Shannon, Simpson and ACE indices among the four groups in the different sample types ( Figures 1A, C , P > 0.05). To evaluate similarities among the four groups, PCoA was based on unweighted UniFrac distances. Regardless of the SP or GCF samples, beta diversity was different in the HC and diseased groups ( Figures 1B, D ). However, in the SP samples, the bacterial compositions in the periodontitis, COPD, and P_COPD groups were indistinguishable ( Figure 1B , P>0.05). For the GCF samples, beta diversity was different between theperiodontitis and P_COPD groups ( Figure 1D ).

Eight phyla, 50 families, 98 genera were detected in the SP samples. The most abundant genera were Prevotella, Corynebacterium, Capnocytophaga, Fusobacterium, Streptococcus and Porphyromonas ( Figure 2A ). Actinomyces, Campylobacter, Capnocytophaga, Neisseria, Prevotella and Streptococcus were present in all SP samples ( Figure 2B ). Nine phyla, 57 families and 118 genera were identified in the GCF samples. The most abundant genera were Streptococcus, Prevotella, Fusobacterium, Porphyromonas, Neisseria and Capnocytophaga ( Figure 2C ). The core microbiota of the GCF samples were Actinomyces, Campylobacter, Fusobacterium, Leptotrichia, Porphyromonas, Prevotella, Streptococcus and Tannerella ( Figure 2D ).

In periodontitis group, we explored the differential taxa among stages of periodontitis. In both types of oral samples, no significant difference was found in the alpha diversity analysis and beta diversity analysis among the different stages of periodontitis. There were no difference in the taxa among stages of periodontitis ( Figure S4B, S4C , P > 0.05).

To further identify the differential taxa among these groups, LEfSe and DESeq2 analyses were conducted. According to the LEfSe analysis, in the SP samples, eight genera were predominant in the HC group, including Actinomyces, Bergeyella, Brachymonas, Cardiobacterium, Lautropia, Mannheimia, Propionibacterium and Rothia. In contrast, the abundance of Haemophilus, Filifactor, and Moraxella increased in the periodontitis group. The abundance of Atopobium and Lachnoanaerobaculum were higher in the COPD group and the abundance of Stomatobaculum, Anaeroglobus, Bifidobacterium, and Clostridium were higher in the P_COPD group. ( Figure 3A , LDA score (log10) >2, P < 0.05). According to DESeq2 analysis ( Table S1 ), there were significant differences in the oral microbiota of the three diseased groups in the SP samples but no common change among these groups. Twenty-nine genera were predominant in the periodontitis group, including Filifactor, Mogibacterium, Scardovia, Murdochiella and Odoribacter. Abiotrophia and Gemella were more abundant in the COPD group and the abundance of Cardiobacterium was higher in the P_COPD group. The abundance of Bergeyella decreased in the periodontitis and COPD groups. The abundance of Pasteurella and Propionicicella decreased in the periodontitis group. The abundance of Desulfobulbus, Soonwooa and Johnsonella decreased in the COPD group ( Figure 3C ).

According to LEfSe analysis, in the GCF samples, the abundance of eight genera: Bergeyella, Cardiobacterium, Kingella, Lautropia. Propionibacterium, Rothia, Serratia and Staphylococcus were more abundant in the HC group. Desulfovibrio, Dorea, Filifactor, Fretibacterium, Moraxella, Pseudoramibacter and Treponema were more abundant in the P group, while the abundance of Mogibacterium increased in the COPD group. The abundance of Phocaeicola and Schwartzia was higher in the P_COPD group. ( Figure 3B , LDA score (log10) >2, P < 0.05). According to DESeq2 analysis ( Table S2 ), there were significant differences in the oral microbiota of the three diseased groups, but no common changes were observed among these groups. Fifteen genera were predominant in the periodontitis group: Pasteurella, Phocaeicola, Pseudoramibacter, Pseudoramibacter, Pyramidobacter, Scardovia, Schwartzia, Shuttleworthia, Slackia, Stomatobaculum and Synergistes. The abundance of Corynebacterium, Bacillus, Ottowia and Neisseria decreased in the periodontitis group. The abundance of Serratia decreased in the COPD group. The abundance of Kingella, Alloprevotella and Dialister decreased in the P_COPD group ( Figure 3D ).

The observed links between the respiratory microbial community and disease prompted us to examine the interactions between the taxa and their clinical features. The taxa analyzed were the top 20 genera regarding abundance. For SP samples, the relative abundance of Streptococcus was negatively correlated with neutrophil, white blood cell and monocyte counts. The relative abundance of Rothia was negatively correlated with lymphocyte counts. The relative abundance of Leptotrichia and Campylobacter were positively correlated with the basophil counts, and the relative abundance of Aggregatibacter and Neisseria were negatively correlated with the basophil counts ( Figure 3E ). In the GCF samples, the relative abundance of Rothia, Streptococcus, and Haemophilus was positively correlated with the lymphocyte percentages. The relative abundance of Streptococcus and Rothia was negatively correlated with the neutrophil percentages, and the relative abundance of Veillonella was positively correlated with neutrophil percentages. The relative abundance of Treponema was positively correlated with neutrophil counts. In addition, the relative abundance of Leptotrichia was negatively correlated with the eosinophil percentages ( Figure 3F ).

We analyzed the predicted phenotypes based on taxonomic classification using BugBase. In different sample types, the relative abundance of aerobic bacteria in the HC group was higher than that in the diseased groups ( Figure S2A , S2F ). In comparison, the relative abundance of anaerobic bacteria was lower in the HC group than that in the diseased groups ( Figure S2B , S2G ). The ability to form biofilms in the HC group was greater than that in the diseased groups ( Figure S2C , S2H ). The potential pathogenicity in the HC group was lower than that in the diseased groups in the SP samples ( Figure S2D ). In the GCF samples, the potential pathogenicity in the P group was lower than that in the other groups ( Figure S2I ). In addition, in the SP samples, the relative abundance of gram-positive bacteria in the HC group was higher than that in the other diseased groups, whereas gram-negative bacteria showed the opposite trend ( Figures S2E , S2J ).

Through PICRUSt2, putative biological functions of the microbiota of the four groups were illustrated. No significant differences were observed between the GCF samples. As shown in Figure 4A , the periodontitis group exhibited significantly enriched metabolism of cofactors and vitamins (thiamine metabolism, nicotinate and nicotinamide metabolism), translation, protein families: genetic information processing (translation factors), amino acid related enzymes, and carbon fixation in photosynthetic organisms. The COPD group showed significantly enriched protein families: genetic information processing (transfer RNA biogenesis, ribosome, mitochondrial biogenesis, DNA replication proteins, translation factors), translation (ribosome, aminoacyl-tRNA biosynthesis, RNA transport), replication and repair (homologous recombination, mismatch repair, DNA replication), protein families: metabolism (peptidases and inhibitors, amino acid related enzymes, peptidoglycan biosynthesis and degradation proteins), glycan biosynthesis and metabolism(peptidoglycan biosynthesis, other glycan degradation, other types of O-glycan biosynthesis and mannose type O-glycan biosynthesis) ( Figure 4B ). The P_COPD group showed significantly enriched protein families: genetic information processing (DNA repair and recombination proteins, transfer RNA biogenesis, ribosome, chromosome and associated proteins), protein families: metabolism (amino acid related enzymes, peptidases and inhibitors, peptidoglycan biosynthesis and degradation proteins), translation (ribosome, aminoacyl-tRNA biosynthesis, RNA transport), glycan biosynthesis and metabolism(peptidoglycan biosynthesis, other glycan degradation, lipopolysaccharide biosynthesis), metabolism of cofactors and vitamins(lipoic acid metabolism, porphyrin and chlorophyll metabolism, riboflavin metabolism, thiamine metabolism) and other functions ( Figure 4C ).

The HC group showed significantly enriched signal transduction (two-component system), lipid metabolism (biosynthesis of unsaturated fatty acids), and metabolism of other amino acids (glutathione metabolism, phosphonate and phosphinate metabolism), compared to the other three groups ( Figure S3 ).