P. gingivalis strain ATCC33277 was obtained from ATCC, and cultured in a brain–heart infusion (BD Bioscience, Franklin Lakes, NJ) containing 0.5% yeast extract (BD Bioscience), 10 mg/L hemin (Wako Chemicals, Osaka, Japan) and 1 mg/L 2-methyl-1,4-naphthoquinone (vitamin K3) (Tokyokasei, Tokyo, Japan) and incubated under anaerobic conditions (80% N₂, 10% CO₂, and 10% H₂) at 37°C. Bacterial suspensions were prepared in phosphate-buffered saline (PBS) without Mg²⁺/Ca²⁺, and the optical density (OD) was measured at 600 nm with a standard curve.

All animal experiments were approved by the Committee for the Care and Use of Laboratory Animals at Fudan University. C57BL/6 male mice (8 weeks old, Beijing Vital River Laboratory Animal Technology Co., Ltd., Beijing, China) were group-housed in a specific pathogen-free (SPF) controlled environment with free access to food and water under a strict 12 h light/dark cycle. Forty mice were randomized into four equal groups with ten mice each: ND (normal diet), HFD (high-fat diet), sham, and Pg.

For the obese model, mice were fed a high-fat diet (HFD, 60% fat, 20% protein, and 20% carbohydrates, Research Diets, D12492) to induce obesity for 12 weeks, while a normal diet (ND, 10% fat, 20% protein, and 70% carbohydrates, Research Diets, D12450J) was used as a control. For P. gingivalis administration, the mice were gavaged with 10⁹ CFU P. gingivalis twice a week for 6 weeks, and PBS with 2% carboxymethylcellulose was administered as a sham. Body weight was assessed in the last week, and fat mass was detected using a Minispec LF90 body composition analyzer (Bruker, Massachusetts, USA).

Mice fasting for 6 h were injected with glucose (1 g/kg) intraperitoneally. Blood glucose was measured with tail vein blood at 0 min, 15 min, 30 min, 60 min, 90 min and 120 min using a OneTouch Ultra glucometer (LifeScan, Pennsylvania, USA). The area under the curve (AUC) of the glucose level over time was calculated to evaluate the glucose tolerance ability.

The concentration of serum insulin was determined by an ELISA kit (Solarbio, Beijing, China) according to the manufacturer’s instructions. HOMA-IR in mice was then calculated using the equation (fasting glucose concentration * fasting insulin concentration)/405.

Total RNA was extracted from the subcutaneous white adipose tissue with TRIzol^® reagent (Invitrogen, California, USA). Reverse transcription was conducted with a SuperScript First-Strand cDNA Synthesis kit (Invitrogen) according to the manufacturer’s instructions. Quantitative real-time PCR (qRT–PCR) was performed with SYBR Green Master Mix (Roche Applied Science, Mannheim, Germany). Gene expression was detected with a 7500 real-time PCR system (Applied Bioscience, Foster City, CA, USA), and the thermal settings used were as follows: 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 65°C for 1 min. The primers (Shanghai Life Biotechnology Co., Ltd., Guangzhou, China) used are listed as following: TNF-α, forward primer: 5’- TGCCTATGTCTCAGCCTCTTC-3’, reverse primer: 5’- GGTCTGGGCCATAGAACTGA-3’; Cd68, forward primer: 5’-TGTCTGATCTTGCTAGGACCG-3’, reverse primer: 5’-GAGAGTAACGGCCTTTTTGTGA-3’; Adgre1, forward primer: 5’-TGACTCACCTTGTGGTCCTAA-3’, reverse primer: 5’-CTTCCCAGAATCCAGTCTTTCC-3’; Adipoq, forward primer: 5’-TGTTCCTCTTAATCCTGCCCA-3’, reverse primer: 5’-CCAACCTGCACAAGTTCCCTT-3’. Gapdh was utilized for normalization. Data were analyzed using the 2^-ΔΔCt relative expression method.

Subcutaneous white adipose tissue in all four groups and colon samples from the Pg and sham groups were harvested immediately after euthanization. Tissues were fixed in adipose tissue fixative or 10% formalin, embedded in paraffin, cut into 5 μm-thick sections and stained with hematoxylin and eosin (HE).

For immunofluorescence staining, sections were deparaffinized, rehydrated and treated with 3% H₂O₂ in methanol for 20 min to inactivate endogenous peroxidase activity. Antigen retrieval was conducted with 1% pepsin (Sigma–Aldrich, St. Louis, USA) and treated with serum albumin (BSA) for 1 h. For adipose tissues, sections were incubated with anti-F4/80 antibody (1:100, Abcam, Cambridge, MA, USA) and anti-Plin1 antibody (1:100, Abcam) overnight at 4°C, followed by three washes (5 min) in PBS. For colon tissues, sections were incubated with anti-ZO-1 antibody (1:100; Proteintech, Wuhan, China) and anti-E-cadherin antibody (1:100; Proteintech) overnight at 4°C. The secondary antibody anti-rabbit (Abcam) was applied for 1 hour at room temperature in the dark, followed by three washes (5 min) in PBS. Slides were then mounted with Vectashield hardset mounting media containing 4’,6-diamidino-2-phenylindole (DAPI). Images were obtained using an Olympus BX51 (Olympus America, Melville, NY) epifluorescence microscope, and images were captured with an Olympus DP-70 camera.

Intestinal permeability was determined by FITC-dextran assay. Briefly, mice fasting for 6 hours were orally administered 150 μL FITC-conjugated dextran (FITC-dextran; Sigma–Aldrich) at 80 mg/mL. Four hours later, blood was centrifuged at 1,000 x g for 30 min at 4°C for serum collection. The concentration of fluorescein was quantified at an excitation wavelength of 485 nm and an emission wavelength of 535 nm, and serially diluted samples of the FITC-dextran marker were used as standards.

Total genomic DNA was extracted from the colon contents of mice in the Pg and sham groups using a PowerSiol^® DNA isolation kit (MO BIO, Carlsbad, CA). DNA extraction was fragmented to an average size of approximately 300 bp, and the paired-end library was constructed using NEXTFLEX^® Rapid DNA-Seq (Bioo Scientific, Austin, TX, USA). Paired-end sequencing was performed on an Illumina NovaSeq/HiSeq Xten (Illumina Inc., San Diego, CA, USA) using NovaSeq Reagent Kits/HiSeq X Reagent Kits. Reads that were less than 50 bp, had a quality value < 20 or had N bases were removed. Metagenomics data were assembled using MEGAHIT, and contigs with lengths greater than or greater than 300 bp were selected as the final assembly result. Open reading frames (ORFs) from each assembled contig were predicted using MetaGene. The predicted ORFs with lengths greater than or greater than 100 bp were retrieved and translated into amino acid sequences using the NCBI translation table. All predicted genes with 95% sequence identity and 90% coverage were clustered using CD-HIT. The longest sequences from each cluster were selected as representative sequences for nonredundant gene catalog construction. The quality-controlled reads were mapped to the representative sequences with 95% identity using SOAPaligner, and gene abundance in each sample was evaluated. The KEGG annotation was conducted using BLASTP against the Kyoto Encyclopedia of Genes and Genomes database with an e-value cutoff of 1e^-5.

Untargeted metabolomics profiling of serum samples in the Pg and sham groups was performed by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). Chromatographic separation of the metabolites was performed on a Thermo UHPLC system equipped with an ACQUITY UPLC HSS T3 (100 mm × 2.1 mm i.d., 1.8 µm; Waters, Milford,USA). Mass spectrometry (MS) was performed using a Thermo UHPLC-Q Exactive Mass Spectrometer equipped with an electrospray ionization (ESI) source operating in either positive or negative ion mode. Data acquisition was performed with the Data Dependent Acquisition (DDA) mode. The detection was carried out over a mass range of 70-1050 m/z. Raw data were imported into Progenesis QI 2.3 for peak detection and alignment. The preprocessing results contained the m/z values and peak intensity. The mass spectra of these metabolic features were identified using accurate masses. MS/MS fragment spectra and isotope ratio differences with searches in internal databases and public databases. The variable importance of the projection (VIP) score generated from orthogonal partial least squares discriminate analysis was used to determine the most differentiated metabolites. Metabolites with VIP ≥ 1.0 and P value ≤ 0.05 were defined as significantly changed metabolites. A multivariate statistical analysis was performed using the R package ropls version 1.6.2.

The MetS statistical analysis was conducted with SPSS Statistics 20.0 software (IBM). Data are expressed as the mean ± SD. Differences between two groups were analyzed by Student’s t test. A two-tailed P value < 0.05 was considered statistically significant.

To investigate the effects of P. gingivalis on MetS, we gavaged C57BL/6 mice with P. gingivalis twice a week for 6 weeks, with phosphate-buffered saline (PBS) gavage as the sham group and high-fat diet (HFD)-induced obese mice as the positive control. As the results indicated, HFD increased the body weight compared with the normal diet (ND) group, while no significant difference was observed between the Pg group and the sham group ( Figure 1A ). The body composition analysis and HE staining indicated a higher proportion of fat mass and larger adipocyte cells in the HFD and Pg groups ( Figures 1B, C ). Moreover, fasting blood glucose was significantly increased in P. gingivalis-administered mice, with a similar elevation in the HFD group ( Figure 1D ). The homeostatic model assessment-insulin resistance (HOMA-IR) assessment comprising the fasting blood glucose and insulin level indicated higher insulin resistance in the Pg group accompanied by worse glucose tolerance during an intraperitoneal glucose tolerance test (IPGTT) compared with the sham group ( Figures 1E–G ).

Adipose inflammation plays a detrimental role in obesity-related metabolic dysbiosis. Thus, we determined the expression of inflammatory factors in adipose tissue with P. gingivalis administration. qRT–PCR indicated the expression of TNF-αwas increased in the HFD and Pg groups( Figure 2A ). Adipose tissue macrophages play vital roles in obesity-induced inflammation. Cd68 and Adgre1 (adhesion G protein-coupled receptor E1), which act as homologs of F4/80, are important markers of macrophages. qRT-PCR demonstrated elevated expression of Cd68 and Adgre1 in the HFD and Pg groups, with remarkably higher expression of Cd68 with P. gingivalis administration ( Figures 2B, C ). However, the expression of the anti-inflammatory factor Adipoq was decreased in the HFD and Pg groups ( Figure 2D ). Moreover, immunofluorescence against F4/80 demonstrated higher macrophage infiltration in the adipose tissue of P. gingivalis-gavaged mice, similar to the HFD group ( Figures 2E, F ).

Increasing evidence has indicated the role of gut microbiota in the progression of metabolic disorder; thus, we analyzed the composition, abundance and function of gut microbiota with fecal samples in P. gingivalis-administered mice via metagenome sequencing. The principal coordinate analysis (PCoA) of Bray–Curtis distances revealed significant differences in the composition and abundance of gut microbiota between the Pg and sham groups ( Figure 3A ). At the phylum level, the proportion of Firmicutes was remarkably lower with P. gingivalis administration, while Bacteroidetes showed an elevated tendency ( Figure 3B ). The linear discriminant analysis effect size (LEfSe) based on an LDA score ≥ 3.0 demonstrated that the gut microbiota of mice in the Pg group was distinguished from that in the sham group by the family, phylum, and genus ( Figure 3C ). At the genus level, the proportion of unclassified Lachnospiraceae was decreased in P. gingivalis-administered mice, while the proportions of unclassified Muribaculaceae, Akkermansia, Prevotella and Porphyromonadaceae were increased ( Figures 4A, B ). Moreover, KEGG annotation showed that mice treated with P. gingivalis exhibited a definitely separate enrichment in microbial function and pathways ( Figure 4C ). For example, Biosynthesis of secondary metabolites”, “Metabolic pathways”, and “Biosynthesis of amino acids” were enriched with P. gingivalis administration.

Intestinal barrier function is closely associated with gut microbiota (Dabke et al., 2019); thus, we further investigated gut permeability with P. gingivalis administration. HE staining of the intestinal epithelia indicated a higher infiltration of inflammatory cells, such as neutrophils, with P. gingivalis administration than in the sham group ( Figure 5A ). The in vivo barrier function assay showed that administration of P. gingivalis increased gut permeability compared with the sham group ( Figure 5B ). Moreover, we determined the expression of junctional proteins located in the gut. As the results indicated, administration of P. gingivalis significantly decreased the expression of ZO-1, a major component of tight junction proteins ( Figures 5C, D ), and E-cadherin, a major component of adhesion molecules ( Figures 5E, F ).

Microbial metabolites are key actors in host-microbiota crosstalk, and we further analyzed the serum samples of P. gingivalis- and sham-administered mice with untargeted metabolomics. The principal component analysis (PCA) based on nuclear magnetic resonance demonstrated a divergent separation of the serum metabolic profile between P. gingivalis- and sham-administered mice ( Figure 6A ). A total of 568 metabolites were identified in both groups, and 80 metabolites were upregulated while 76 metabolites were downregulated according to a variable value set at a VIP > 1 and a P value < 0.05 in the Wilcoxon rank-sum test ( Table S1 ). The pathway classification showed a total of 88 metabolites enriched in metabolic pathways, including lipid metabolism and amino acid metabolism ( Figure 6B ). The heatmap analysis identified 39 metabolites related with metabolic pathways, with 18 metabolites that were upregulated and 21 metabolites that were downregulated with P. gingivalis administration ( Figure 6C ). The analysis of the metabolic pathways demonstrated 10 of the most relevant pathways associated with the metabolites in the Pg and sham groups, including tryptophan metabolism, protein digestion and absorption, and choline metabolism ( Figure 6D ).

The above results demonstrated that P. gingivalis administration induces metabolic disorders, gut microbiota dysbiosis and serum metabolome alterations. As gut microbiota interact with the host to produce metabolites, acting as intermediates or end-products of microbial metabolism, we further performed a correlation analysis involving the above 39 metabolites with all of the bacterial genera ( Figure 7 ). Porphyromonadaceae, to which P. gingivalis belongs, showed a wide correlation with most metabolites, such as 5-hydroxyindoleacetic acid, indole-3-acetaldehyde (IAAld), P-salicylic acid, phosphatiylcholine (PC), and creatine (r = 0.61, 0.72, 0.79, and 0.69, respectively, Table S2 ). Akkermansia has been shown to be correlated with almost all of the metabolites. PC, as the precursor of trimethylamine (TMA) and TMAO, has been demonstrated to be positively correlated with Prevotella and Porphyromonadace and negatively correlated with Lachnospiraceae and Clostridiales. These results indicate that serum metabolome alteration and gut microbiota dysbiosis are closely related to P. gingivalis administration.