Gut commensal Odoribacter splanchnicus attenuates hyperlipidemic periodontitis via gut-oral metabolic transmission of β-GPA Authors: Jing Xu, Ziyi Han, Qing Xue, Haoran Wang, Yutong Li, Jiyu Song, Lanzhou Li, Min Hu, Di Wang Published: 2026 ABSTRACT AbstractHyperlipidemic periodontitis (HPD) represents a prevalent comorbidity linking systemic metabolic dysregulation with local inflammation, yet the microbial mechanisms driving this gut-oral crosstalk remain elusive. Here, the comorbid state of HPD is linked to hyperlipidemia-associated gut microbiota changes, which are prominently accompanied by the depletion of Odoribacter splanchnicus in both patients and mice. Fecal microbiota transplantation demonstrates that this gut dysbiosis exacerbates periodontal destruction when local inflammation is present. Mechanistically, intragastric administration of live O. splanchnicus ameliorates HPD by remodeling the gut ecosystem and upregulating the metabolite β-guanidinopropionic acid (β-GPA). Notably, direct supplementation with β-GPA reproduces these protective effects. Furthermore, β-GPA is proposed as a systemic effector linking the gut and periodontal tissues, where its protective effect is associated with the suppression of the pro-inflammatory Toll-like receptor 4 (TLR4) signaling cascade. These findings highlight a link involving O. splanchnicus, β-GPA, and the modulation of TLR4 signaling, offering a potential microbiome-based therapeutic strategy for managing complex metabolic-inflammatory comorbidities.Subject terms: Diseases, Gastroenterology, Microbiology Introduction Periodontitis, a chronic inflammatory disease defined by the progressive destruction of tooth-supporting tissues1, impacts 45–50% of adults globally2–4. Crucially, periodontitis is fundamentally an infectious disease initiated by local pathogenic biofilms, but is mechanistically linked to systemic metabolic disorders serving as disease modifiers5,6. The co-occurrence of periodontitis and hyperlipidemia, termed hyperlipidemic periodontitis (HPD), is a complex and increasingly prevalent comorbidity defined by a bidirectional exacerbation between metabolic and inflammatory dysregulation7. While the precise molecular conduits remain elusive, clinical observations indicate that hyperlipidemia acts as an amplifier of pre-existing periodontal destruction8–10. Epidemiological evidence quantifies this risk in clinical populations, where local dental plaque is common and acts as the initiating trigger: elevated low-density lipoprotein cholesterol (LDL-C) correlates with a 14% higher likelihood of clinical attachment loss (odds ratio (OR) = 1.14), and high total cholesterol (TC) is linked to a 16% greater odds of increased pocket depth (OR = 1.16)11. The association can be even more pronounced, with high triglyceride (TG) levels reported to increase the odds of periodontitis by over 700% (OR ≈ 8.6)12. In turn, the persistent inflammation and microbial dysbiosis associated with periodontitis further disrupt systemic lipid metabolism13–15. This pathological interplay suggests that conventional local therapies may be insufficient, as they do not fully address the underlying metabolic-inflammatory dysregulation. Therefore, elucidating these crosstalk mechanisms to develop integrative therapeutic strategies is warranted. The role of the gut microbiota as a key hub connecting metabolic diseases and oral health is increasingly recognized16,17. Both hyperlipidemia and periodontitis can induce gut microbiota dysbiosis, which is typically characterized by a reduced abundance of beneficial commensals and an expansion of pro-inflammatory bacteria18,19, thereby exacerbating host metabolic disorders and inflammatory states20. Notably, non-surgical periodontal therapy in mice with comorbid hyperlipidemia and periodontitis can improve gut microbiota dysbiosis and ameliorate lipid disorders, highlighting the remote regulatory role of the gut microbiota in periodontal homeostasis21. The potential mechanisms linking gut dysbiosis to periodontitis are multifaceted. For instance, a dysbiotic gut microbiota can disrupt host metabolic pathways, altering the profile of circulating metabolites that act as signaling effectors in distant tissues, including bone22,23. This dysbiosis can also compromise intestinal barrier integrity, allowing pathogen-associated molecular patterns (PAMPs), such as lipopolysaccharide (LPS), to translocate into the bloodstream. This triggers systemic inflammation, which can remotely activate inflammatory pathways in oral tissues, fostering osteoclastogenesis and bone resorption24–26. Given that both hyperlipidemia and periodontitis disrupt the gut microbiota, modulating this “gut-oral” axis presents a potential therapeutic strategy for managing the complex metabolic-inflammatory interplay inherent to HPD. However, moving from correlation to causation and identifying specific microbial-metabolic effectors remains a challenge in the microbiome medicine field27–29. Specifically, it is unclear which key gut microbes are effective therapeutic targets to disrupt the “metabolic disorder-periodontal destruction” pathological feedback loop, and there is a lack of systematic analysis of their functional metabolites and mechanisms of action30. The treatment of periodontitis via targeted regulation of the gut microbiota is still in the exploratory stage. Key challenges include the precise identification of key functional bacteria, the development of specific intervention strategies, and the systematic elucidation of the multi-level “bacteria-metabolite-host response” mechanism of action31. Among the diverse gut commensals, Odoribacter splanchnicus, an emerging “next-generation probiotic candidate”, has garnered significant attention for its protective potential32. A reduced abundance of O. splanchnicus is closely associated with the pathogenesis of various metabolic and inflammatory disorders, including inflammatory bowel disease (IBD) and obesity33. Mechanistically, O. splanchnicus maintains intestinal homeostasis by regulating host immune responses and bolstering gut barrier function, making it a candidate for live biotherapeutic product development. For instance, its extracellular vesicles mitigate IBD pathology by inhibiting NOD-like receptor family pyrin domain-containing 3 (NLRP3) inflammasome activation and bolstering gut barrier function, while its outer membrane vesicles modulate LPS-induced cytokine responses by inhibiting interleukin-8 (IL-8) secretion and inducing interleukin-10 (IL-10)34,35. Crucially, the beneficial effects of gut microbes, such as O. splanchnicus, are often mediated by the production of functional metabolites. β-Guanidinopropionic acid (β-GPA), an endogenous creatine analog, has been shown to modulate cellular energy metabolism and inflammation by activating the AMP-activated protein kinase (AMPK) signaling pathway36–38. The established roles of these metabolites in regulating systemic metabolism and immunity underscore the therapeutic potential of targeting specific gut microbes and metabolic output to treat such complex diseases. This study highlights the critical role of gut microbiota dysbiosis in exacerbating periodontal destruction, revealing O. splanchnicus as a key protective commensal. Through a comprehensive investigation, a protective axis linking O. splanchnicus, the metabolite β-GPA, and the modulation of the periodontal Toll-like receptor 4 (TLR4) signaling pathway is identified. These findings provide a potential mechanistic framework for the crosstalk between systemic metabolic disorders and local inflammatory diseases, positioning this microbial-host interplay as a potential target for HPD intervention and providing a scientific basis for the development of next-generation probiotics. Results Gut dysbiosis in patients with HPD is associated with O. splanchnicus depletion To systematically investigate the microbiological characteristics of HPD, 40 participants, comprising 20 patients with HPD and 20 negative control (NC) volunteers, were recruited for a multi-omics correlation analysis (Supplementary Fig. S1A). Quantitative clinical evaluations indicated that the HPD cohort exhibited significantly elevated serum lipid profiles (TC and TG) and severe periodontal destruction (increased probing depth and clinical attachment loss), compared to the NC group (Supplementary Tables S1, S2). Analysis of the gut metagenome revealed a significant difference in microbial community composition between the two groups (Fig. 1A). At the species level, 9 gut bacteria with significantly different abundances were identified, among which O. splanchnicus exhibited a significant decrease in relative abundance in HPD patients (Fig. 1B, Supplementary Fig. S1B). Concurrently, 16S rRNA sequencing of saliva samples indicated a clear distinction in the oral microbiota composition between HPD patients and NC volunteers (Fig. 1C). Among the 14 oral bacterial genera with significantly different abundances, periodontal pathogens (Treponema, Aggregatibacter, Porphyromonas, Prevotella, Neisseria, and Tannerella) were enriched in HPD patients (Fig. 1D). Furthermore, untargeted metabolomics analysis of fecal samples identified 290 differential metabolites, with 208 significantly increased and 82 significantly decreased in HPD patients (Fig. 1E, F). O. splanchnicus was significantly negatively correlated with multiple periodontal pathogens (Fig. 1G) and was also significantly associated with various fecal metabolites, such as arginine (Fig. 1H). These clinical data suggest that HPD is associated with a notable intestinal O. splanchnicus depletion and oral pathogen enrichment. Importantly, this NC versus HPD clinical comparison was designed to identify microbial deficits characterizing the holistic comorbid state, serving as an exploratory screen to pinpoint candidate therapeutic targets for HPD. These observations pointed to a potential “gut-oral” axis link that warranted rigorous further investigation in a controlled animal model to limit human clinical confounders. Gut dysbiosis in patients with HPD is associated with O. splanchnicus depletion. A Partial least squares discriminant analysis (PLS-DA) score plot of gut microbiota composition in the HPD patients (n = 20) and the NC volunteers (n = 20). B Heatmap of gut bacteria (species level) with significantly different abundances between the two groups. The color scale indicates z-score normalized relative abundance. C Non-metric multi-dimensional scaling (NMDS) analysis score plot of oral microbiota composition between the HPD patients and NC volunteers. D Heatmap of oral bacteria (genus level) with significantly different abundances between the two groups. E Volcano plots of differential fecal metabolites in positive and F negative ion modes. Red and blue dots represent metabolites significantly upregulated and downregulated in the HPD patients, respectively (p < 0.05, VIP > 1). G Spearman correlation heatmap of differential gut microbial species and differential oral microbial genera. The color scale represents the correlation coefficient, where red indicates a positive correlation and blue indicates a negative correlation. H Spearman correlation heatmap of differential gut microbial species and the top 50 differential metabolites. A9 2-amino-1-phenylethanol, A12 DL-phenylalanine, A22 Arginine, A37 4-benzofuranethanamine, 2,3-dihydro-α-methyl-, A38 O-toluidine, A46 Benzyl alcohol, p-fluoro-α-[1-(methylamino)ethyl]-, erythro-(+/-)-, A47 3’-fluoromethcathinone. *p < 0.05, **p < 0.01. HPD manifests as exacerbated alveolar bone loss and significant depletion of intestinal O. splanchnicus To investigate the interaction between hyperlipidemia and periodontitis observed clinically, a mouse model comprising four groups was established: negative control (NC), periodontitis (PD), hyperlipidemia (HLP), and HPD (Supplementary Fig. S2A). High-fat diet (HFD) feeding successfully induced a hyperlipidemic phenotype in the HLP and HPD groups, characterized by significant weight gain, elevated blood glucose, and increased serum lipids (TC, TG, LDL-C) compared to NC mice (Supplementary Fig. S2B, C). These metabolic effects were evident as adipocyte hypertrophy in epididymal white adipose tissue (eWAT), perirenal white adipose tissue (pWAT), and inguinal white adipose tissue (iWAT). Hepatic steatosis was confirmed by Oil Red O staining and corroborated by significant increases in fat and liver indices (Supplementary Fig. S2D, E). No significant pathological changes were observed in the heart, spleen, kidney, or ileum (Supplementary Fig. S3A, B). Micro-computed tomography (Micro-CT) analysis revealed that alveolar bone loss was more severe in the HPD mice compared to the PD mice, while the HLP mice showed no significant difference from NC mice (Fig. 2A). This outcome aligns with the fundamental consensus that periodontitis requires a local mechanical or bacterial trigger; hyperlipidemia alone cannot initiate deep alveolar bone resorption but rather amplifies the destruction once local inflammation is established. Consequently, compared to the PD mice, the HPD mice exhibited an increased distance from the cementoenamel junction to the alveolar bone crest (CEJ-ABC), decreased bone mineral density (BMD), bone volume/total volume (BV/TV), trabecular thickness (Tb.Th), and trabecular number (Tb.N), and increased trabecular separation (Tb.Sp) (Fig. 2B). These findings were corroborated by histopathological analysis with hematoxylin and eosin (H&E) staining (Fig. 2A). At the molecular level, western blot analysis of alveolar bone tissue revealed that the levels of osteogenesis-related proteins, including osteoprotegerin (OPG), osteopontin (OPN), runt-related transcription factor 2 (RUNX2), Osterix, and osteocalcin (OCN), followed a decreasing trend of NC > HLP > PD > HPD, while the expression of pro-osteoclastic and inflammatory proteins such as receptor activator of nuclear factor kappa-B ligand (RANKL), inducible nitric oxide synthase (iNOS), NLRP3, interleukin-1β (IL-1β), tumor necrosis factor-α (TNF-α), and interleukin-6 (IL-6) followed an opposite trend (Fig. 2C, Supplementary Fig. S3C). HPD manifests as exacerbated alveolar bone loss and significant depletion of intestinal O.splanchnicus. A Micro-CT 3D reconstruction, 2D sagittal images of alveolar bone, and H&E staining of periodontal tissues. Scale bar = 1 mm. B Quantitative analysis of alveolar bone loss and bone microarchitecture parameters in the maxillary second molar region. C Western blot analysis showed the protein expression levels related to bone metabolism and inflammation in alveolar bone tissue. D Heatmap of the relative abundance of the top 30 oral bacterial genera in the PD and HPD mice. E LEfSe analysis of the gut microbiota in the NC and HPD mice. F Inter-group difference analysis of the relative abundance of O. splanchnicus in the NC, PD, HLP, and HPD mice. G Spearman correlation analysis heatmap of gut O. splanchnicus and differential oral genera in the NC and HPD mice. H Volcano plot of the correlation between O. splanchnicus and differential gut metabolites in the NC and HPD mice. Red dots represent positively correlated metabolites, and blue dots represent negatively correlated metabolites. Western blots for all replicates are provided in Supplementary Fig. S12. Data are presented as mean ± SD. Statistical analysis was performed using one-way ANOVA and Tukey’s multiple comparison test for (B) and (F). *p < 0.05, **p < 0.01, ***p < 0.001. The oral microbiome was significantly altered by an HFD in the setting of periodontitis. Compared to PD mice, HPD mice exhibited a dysbiotic oral microbiota, characterized by an enrichment of dysbiosis-associated genera like Streptococcus and Bacteroides. Notably, while gut-associated microbes (such as Lactobacillus and Akkermansia) were detected in the oral cavity, their presence is likely confounded by murine coprophagy rather than representing true oral colonization (Fig. 2D). Linear discriminant analysis effect size (LEfSe) analysis identified O. splanchnicus as a key bacterium enriched in the NC mice (Fig. 2E). To evaluate the relative contributions of systemic and local factors to this microbial shift, a four-group relative abundance analysis was performed (Fig. 2F). Specifically, while the PD group showed a modest, non-significant downward trend relative to the NC group, the HLP group induced a significant depletion of this commensal (p < 0.05). Crucially, the superimposition of these conditions in the HPD group exhibited the most profound depletion (p < 0.01). This indicates an additive dysbiotic effect, suggesting that the loss of O. splanchnicus is primarily associated with the systemic metabolic disorder but reaches its maximum severity under the cumulative burden of the comorbid state. Depletion of O. splanchnicus is a prominent feature of the hyperlipidemia-associated dysbiosis that culminates in the HPD state, consistent with our clinical observations. Furthermore, untargeted metabolomics analysis confirmed that the superimposition of hyperlipidemia generates a distinct metabolic alteration independent of the background periodontal inflammation (Supplementary Fig. S4A). Specifically, differential analysis between the PD and HPD groups revealed alterations in numerous specific metabolites, further highlighting the metabolic signature associated with hyperlipidemia (Supplementary Fig. S4B). To capture the ecological shifts characterizing this holistic comorbidity, broad multi-omics network analyses on NC versus HPD were maintained as an exploratory assessment. Since the hyperlipidemic state appears to be the predominant factor underlying these shifts, this comparison reflects the overarching microbial-metabolic differences of the comorbid state rather than a uniquely HPD-specific crosstalk. A correlation analysis comparing the NC and HPD mice revealed that the abundance of gut O. splanchnicus was negatively associated with the dysbiosis-associated genus Streptococcus (Fig. 2G). This association extended to the metabolic level, as its abundance was significantly correlated with 14 of the 298 identified differential gut metabolites (Fig. 2H). Hyperlipidemia-associated gut dysbiosis in HPD contributes to the exacerbation of periodontal destruction To determine whether the hyperlipidemia-associated gut dysbiosis in HPD is a mere byproduct or a contributing factor to periodontal destruction, a fecal microbiota transplantation (FMT) model was established to investigate its functional contribution in HPD (Supplementary Fig. S5A). Mice receiving microbiota from HPD donors (HPD-FMT) exhibited more severe periodontal tissue destruction compared to those receiving microbiota from PD donors (PD-FMT) (Fig. 3A). In HPD-FMT mice, the CEJ-ABC distance and Tb.Sp were significantly increased, while BMD, BV/TV, Tb.Th, and Tb.N were decreased (Fig. 3B). Histological observations were consistent with the quantitative imaging results (Fig. 3A). Recipient mice showed no significant differences in body weight or blood glucose levels (Supplementary Fig. S5B, C). At the molecular level, the alveolar bone of HPD-FMT mice showed significantly downregulated expression of OPG, RUNX2, and OCN, and upregulated expression of RANKL, NLRP3, and IL-6 (Fig. 3C, Supplementary Fig. S5D). This was confirmed by immunofluorescence, which showed reduced RUNX2 and enhanced RANKL and NLRP3 signals (Fig. 3D, Supplementary Fig. S5E). Consistent with these findings, tartrate-resistant acid phosphatase (TRAP) staining revealed greater osteoclast infiltration in the HPD-FMT mice (Fig. 3E). Hyperlipidemia-associated gut dysbiosis in HPD contributes to the exacerbation of periodontal destruction. A Micro-CT 3D reconstruction, 2D sagittal images, and H&E staining. Scale bar = 1 mm. B Quantitative analysis of alveolar bone loss and bone microarchitecture parameters. C Western blot analysis of osteogenesis-related proteins and pro-bone resorption/inflammatory proteins in alveolar bone tissue. D Representative immunofluorescence staining (RUNX2, green; RANKL, NLRP3, red; DAPI, blue) of alveolar bone. Scale bar: 20 μm. E TRAP staining (red indicates multinucleated osteoclasts) of alveolar bone sections. Scale bar: 100 μm. F Oral microbiota genus-level hierarchical clustering analysis. G LEfSe analysis of differential bacterial genera in the oral microbiota. H Heatmap of differential bacterial species in the gut microbiota. Western blots for all replicates are provided in Supplementary Fig. S13. Data are presented as mean ± SD. Statistical analysis was performed using a two-tailed unpaired t-test. **p < 0.01, ***p < 0.001. Significant differences were observed in oral microbiota composition (Supplementary Fig. S5F, G), with the PD-FMT mice being predominantly colonized by the early-colonizing genus Streptococcus (Fig. 3F). LEfSe analysis showed an elevated relative abundance of the genus Prevotella in the HPD-FMT mice, consistent with the pathogenic signature observed in the HPD patients (Fig. 3G). In the gut, the microbial structure also differed significantly (Supplementary Fig. S5H, I). Analysis of the top 150 most abundant bacteria revealed that HPD-FMT mice exhibited a significant reduction in several known beneficial bacteria and an increase in potentially pathogenic bacteria like Helicobacter ganmani. Notably, the abundance of O. splanchnicus was lower in the HPD-FMT mice (Fig. 3H). Furthermore, liver Oil Red O staining confirmed the transmission of the metabolic phenotype; HPD-FMT mice showed more extensive hepatic lipid accumulation than PD-FMT mice (Supplementary Fig. S6A, B), while other organs remained unaffected (Supplementary Fig. S6C, D). These results suggest that the hyperlipidemia-associated gut microbiota contributes to the exacerbation of periodontal destruction and transmits metabolic disturbances, highlighting the therapeutic potential of targeting key beneficial microbes such as O. splanchnicus. Live O. splanchnicus protects against HPD by remodeling gut metabolism to upregulate β-GPA To investigate the therapeutic potential of O. splanchnicus, an intervention model was established (Supplementary Fig. S7A). Compared to the HPD mice, the live O. splanchnicus-treated (OS) mice exhibited a markedly reduced CEJ-ABC distance, improved bone microarchitecture parameters, and less inflammatory infiltration (Fig. 4A, B). Although the heat-inactivated O. splanchnicus-treated (OS-I) mice exhibited slight radiological improvements in alveolar bone compared to the HPD group, the therapeutic effect was more pronounced in the OS group than in the OS-I group. The oral microbial community structure of the OS group clustered closer to that of the NC group in ordination space, whereas the improvement in the OS-I group was limited (Fig. 4C). Live O. splanchnicus protects against HPD by remodeling gut metabolism to upregulate β-GPA. A Representative Micro-CT 3D reconstructions, 2D sagittal plane images of maxillae, and H&E staining of periodontal tissue. Scale bar = 1 mm. B Quantitative analysis of alveolar bone loss and bone microarchitecture parameters. C NMDS analysis of oral microbial communities based on Bray-Curtis distance. D Western blot analysis of osteogenesis-related proteins and osteoclast/inflammation-related proteins in alveolar bone tissue. E Immunofluorescence staining of RUNX2, Osterix (green), RANKL, NLRP3, and IL-1β (red) in alveolar bone tissue. Nuclei were counterstained with DAPI (blue). Scale bars = 20 μm. F TRAP staining to label osteoclasts. Scale bars = 100 μm. G Volcano plot of metabolites in positive and negative ion modes. H Correlation network of differential gut metabolites and microbes. Red lines represent positive correlations, and green lines represent negative correlations. The thickness of the line represents the absolute value of Spearman’s correlation coefficient. I Spearman’s correlation heatmap of gut microbiota and metabolites. A1 1-palmitoyl-sn-glycero-3-phosphocholine, A2 1-stearoyl-sn-glycero-3-phosphocholine (LPC(18:0)), A3 1-stearoyl-2-hydroxy-sn-glycero-3-phosphocholine, A4 1-palmitoyl-2-hydroxy-sn-glycero-3-phosphoethanolamine, A5 1-stearoyl-sn-glycero-3-phosphocholine, A6 1-(1Z-hexadecenyl)-sn-glycero-3-phosphocholine, A7 1-pentadecanoyl-sn-glycero-3-phosphocholine, A8 1-heptadecanoyl-sn-glycero-3-phosphocholine, A9 1-myristoyl-sn-glycero-3-phosphocholine. Western blots for all replicates are provided in Supplementary Fig. S14. Statistical analysis was performed using one-way ANOVA followed by Tukey’s post-hoc multiple comparisons test. Data are presented as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001. ns not significant. Importantly, western blot analysis revealed that live O. splanchnicus significantly upregulated osteogenic proteins (OPG, OPN, RUNX2, Osterix) and downregulated osteoclastogenic (RANKL) and pro-inflammatory markers (iNOS, NLRP3, IL-1β, TNF-α, IL-6) in alveolar bone. In contrast, the heat-inactivated O. splanchnicus failed to significantly rescue these osteogenic and inflammatory markers. This suggests that the therapeutic effect on periodontal tissue is dependent on the live metabolic activity of the bacteria (Fig. 4D, Supplementary Fig. S7B). Immunofluorescence staining detected increased expression of RUNX2 and Osterix, decreased expression of RANKL, NLRP3, and IL-1β (Fig. 4E, Supplementary Fig. S7C), coupled with a marked reduction in osteoclasts (Fig. 4F). The therapeutic benefits also extended systemically. Regarding body weight, the administration of live O. splanchnicus blunted the rate of weight gain (evaluated as percent change from the pre-intervention baseline), exhibiting a more gradual upward trend compared to the HPD group (Supplementary Fig. S7D). In addition, it ameliorated dyslipidemia, decreased adiposity, and attenuated hepatic steatosis, without inducing adverse effects (Supplementary Figs. S7E, S8A–D). Untargeted metabolomics of intestinal contents identified 522 differential metabolites between the OS and HPD groups (212 upregulated and 310 downregulated) (Fig. 4G). Notably, β-GPA was significantly elevated in the OS mice (fold change = 3.49, p < 0.05), and it was a central node in the metabolite-microbe correlation network (Fig. 4H). A correlation analysis between differential gut bacteria and the key metabolites upregulated by live O. splanchnicus treatment confirmed that β-GPA was significantly correlated with multiple microbial species (Fig. 4I). The elevation of β-GPA following live O. splanchnicus administration, coupled with the bacterium’s therapeutic effects, suggests that β-GPA may be a functional metabolite mediating these protective effects. β-GPA supplementation alleviates HPD phenotypes To directly assess the therapeutic potential of β-GPA, the metabolite was administered to HPD mice (Supplementary Fig. S9A). β-GPA significantly alleviated alveolar bone loss, improved bone microarchitecture, and suppressed periodontal inflammation (Fig. 5A, B). Consistent with these structural improvements, the expression of osteogenic proteins (OPG, OPN, RUNX2, Osterix) was upregulated, while osteoclast-related (RANKL, cathepsin K (CTSK)) and pro-inflammatory (iNOS, NLRP3, IL-1β, TNF-α, IL-6) markers were downregulated in the alveolar bone (Fig. 5C, D, Supplementary Fig. S9B). Immunofluorescence staining revealed an enhanced expression of RUNX2 and Osterix proteins and decreased expression of RANKL, CTSK, NLRP3, and IL-1β proteins, culminating in a marked reduction in osteoclast numbers (Fig. 5E, F, Supplementary Fig. S9C). β-GPA supplementation alleviates HPD phenotypes. A Representative Micro-CT 3D reconstruction, 2D sagittal images, and H&E staining of periodontal tissue. Scale bar = 1 mm. B Quantitative analysis of alveolar bone loss and bone microarchitecture parameters. Western blot of C osteogenesis/osteoclast-related proteins and D pro-inflammatory-related proteins in alveolar bone tissue. E Representative immunofluorescence staining images (RUNX2, Osterix, green; RANKL, CTSK, NLRP3, IL-1β, red; DAPI, blue). Scale bar = 20 μm. F TRAP staining images (osteoclasts, red) of alveolar bone tissue. Scale bar = 100 μm. G Composition of the oral microbiota at the phylum level. H LEfSe analysis of the gut microbiota. Western blots for all replicates are provided in Supplementary Fig. S15. Data are presented as mean ± SD. Statistical tests were performed using one-way ANOVA with Tukey’s post-hoc multiple comparisons test. **p < 0.01, ***p < 0.001. The impact of β-GPA on host microbial communities was also examined. In the oral cavity, β-GPA treatment remodeled the microbiome by altering its structure and alpha diversity (Supplementary Fig. S9D, E) and reducing the abundance of the phylum Proteobacteria, lowering the Proteobacteria/Firmicutes (P/F) ratio (Fig. 5G). In the gut, the microbial composition of the β-GPA group was distinct from the HPD group (Supplementary Fig. S9F). LEfSe analysis indicated that β-GPA treatment enriched for several health-associated species, while the HPD group was enriched with the inflammation-associated Parasutterella (Fig. 5H). Treatment with β-GPA mitigated the metabolic dysregulation induced by an HFD, leading to reduced body weight, hyperglycemia, and dyslipidemia, alongside decreased adiposity and attenuated hepatic steatosis (Supplementary Figs. S9G, H, S10A, B). Micro-CT fat scans showed that the fat volume to total volume ratio was reduced in the β-GPA group (Supplementary Fig. S10C). Importantly, these therapeutic effects were achieved without inducing adverse events in major organs (Supplementary Fig. S10D, E). Since exogenous β-GPA ameliorated both periodontal and systemic health, the specific molecular pathway through which it exerts its protective action in the periodontal tissue was subsequently investigated. β-GPA ameliorates HPD and is associated with the modulation of the TLR4–MyD88 signaling pathway Proteomic analysis of periodontal tissues identified 251 differentially expressed proteins (DEPs) between the β-GPA and HPD groups, of which 143 were upregulated, and 108 were downregulated (Fig. 6A). Gene Ontology (GO) analysis showed that these DEPs were primarily enriched in biological processes related to immune response, inflammatory signaling, and bone metabolism (Fig. 6B). Among these DEPs, TLR4 was identified as one of the most significantly downregulated core proteins. Protein-protein interaction (PPI) network analysis based on the STRING database revealed that TLR4 is a key regulatory hub in the network (Fig. 6C). Validation experiments confirmed that β-GPA treatment was associated with a significant reduction in the protein levels of TLR4 and its downstream effectors, myeloid differentiation primary response 88 (MyD88), and TNF receptor-associated factor 6 (TRAF6). This was accompanied by markedly suppressed phosphorylation of IκB kinase α/β (IKKα/β) and nuclear factor-κB (NF-κB) p65, while the phosphorylation of the pro-osteogenic Smad1/5/9 was significantly increased (Fig. 6D, Supplementary Fig. S11A). In the gingival tissues of HPD mice, TLR4 and MyD88 were highly expressed, whereas β-GPA treatment significantly reduced the expression and localization of these molecules (Fig. 6E, Supplementary Fig. S11B). These results indicate that β-GPA, a key metabolite upregulated by O. splanchnicus intervention, effectively alleviates HPD-associated bone loss. Notably, this protective effect is closely associated with the downregulation of the TLR4-MyD88 signaling pathway in periodontal tissues, thereby reducing local inflammation and shifting the bone remodeling balance towards osteogenesis. β-GPA ameliorates HPD and is associated with the modulation of the TLR4–MyD88 signaling pathway. A Volcano plot of DEPs in periodontal tissues from HPD versus β-GPA groups. Red dots represent upregulated proteins, and blue dots represent downregulated proteins (screening threshold: Fold Change > 1.5, p < 0.05). B GO enrichment analysis of DEPs. C The PPI network was constructed using the STRING database. D Western blot validation of protein expression levels in alveolar bone tissue. E Immunofluorescence staining of TLR4 and MyD88 (red) in gingival tissue, with DAPI (blue) for nuclear counterstaining. Scale bar = 20 μm. Western blots for all replicates are provided in Supplementary Fig. S16. Discussion This study proposes a protective axis involving the gut commensal O. splanchnicus and its functionally associated metabolite, β-GPA. By integrating clinical data with animal model validation, we show that HPD is tightly linked to the depletion of intestinal O. splanchnicus compared to healthy controls, and that transplantation of HPD microbiota can transmit the exacerbated disease phenotype. Intragastric supplementation with live O. splanchnicus alleviates periodontal destruction and metabolic disturbances by remodeling the gut microbial ecosystem to upregulate intestinal β-GPA. As the downstream effector molecule, β-GPA subsequently reaches periodontal tissues, where it exerts protective effects closely linked to the suppression of the TLR4-MyD88 signaling pathway, ultimately mitigating inflammatory bone resorption. These results provide a potential mechanistic rationale for modulating the gut microbiome to manage the interplay between systemic metabolic disorders and local inflammatory diseases. Our results support the concept that hyperlipidemia is not an independent trigger for alveolar bone resorption but rather a systemic risk modulator that exacerbates pre-existing periodontal tissue destruction, aligning with previous mechanisms demonstrating that dysregulated cholesterol homeostasis amplifies periodontal inflammation39. Crucially, our FMT results suggest that hyperlipidemia-associated gut dysbiosis may act as an upstream mediator linking this systemic metabolic disorder to distal inflammation in the alveolar bone40. As explicitly shown in our model, transferring HPD-derived microbiota alone failed to initiate disease; however, it exacerbated alveolar bone loss once the local ligature trigger was applied. Therefore, the gut microbiota is not a mere passive bystander but an active conduit that transmits the metabolic-inflammatory burden of this comorbid state41,42. These findings suggest that systemic metabolic disorders prime a susceptible host immune microenvironment via gut remodeling, thereby lowering the threshold for destructive inflammation at distant sites like the periodontium. Ultimately, this highlights the potential value of combining metabolic intervention with local therapy for HPD patients and lays the foundation for elucidating the specific regulatory role of key species. O. splanchnicus, known for maintaining gut homeostasis and mitigating conditions like inflammatory bowel disease43,44, extends its protective benefits to oral health in our HPD model. While the strictly anaerobic nature of O. splanchnicus raises valid concerns about its survival during gavage due to gastric acidity and transient oxygen exposure, previous studies indicate that its bile tolerance and production of outer membrane vesicles confer considerable resilience to gastrointestinal stressors45,46. Furthermore, genomic analyses indicate that many gut-derived anaerobes, including Odoribacter species, encode oxidative stress defense mechanisms, such as thioredoxin and other detoxifying enzymes. These systems likely provide the necessary transient aerotolerance during upper gastrointestinal transit32,47. Upon reaching the anoxic lower gut, its metabolic capacity to utilize host-derived and dietary glycans facilitates its survival and integration into the resident microbial ecosystem48. This colonization provided a basis for evaluating whether bacterial viability is required for therapeutic efficacy. Intriguingly, we observed a divergence in outcomes between live and heat-inactivated O. splanchnicus treatments. While the OS-I group still showed partial improvements in hepatic steatosis, it was insufficient to restore alveolar bone. This partial local benefit aligns well with the “postbiotics” concept49,50, where residual structural components or pre-synthesized active substances within the inactivated bacteria can still exert mild regional effects, a phenomenon also observed with pasteurized Akkermansia muciniphila51,52. However, because these residual components cannot be continuously synthesized, their limited quantity fails to sustain the necessary systemic signaling required for distant tissue repair. These findings suggest that periodontal protection requires the continuous presence of live, metabolically active O. splanchnicus to dynamically reshape the gut ecosystem and sustainably modulate circulating effectors53. This continuous metabolic activity, combined with ecological remodeling, underpins its therapeutic efficacy. Specifically, our findings suggest β-GPA as a systemically circulating effector upregulated by live O. splanchnicus. β-GPA is known for its multi-faceted benefits, including improving metabolic disorders and reducing inflammation54,55. Our work suggests its role as a potential gut-oral signaling molecule. Physicochemically, β-GPA is a highly water-soluble endogenous creatine analog with good oral bioavailability. It is well-established in the pharmacokinetic literature that, following oral or intragastric delivery, β-GPA is rapidly and efficiently absorbed into the systemic circulation via specialized intestinal transport systems56–58. Once in circulation, it accumulates in distal target tissues, exerting anti-inflammatory and metabolic-regulating effects59,60. Anatomically, the heavily vascularized periodontal tissues, combined with the highly permeable state induced by HPD-associated systemic inflammation, provide a conduit for circulating β-GPA. This allows it to extravasate into the local microenvironment and gingival crevicular fluid, where it may interact with resident cells to modulate downstream inflammatory signaling. Mechanistically, within these resident cells, β-GPA has been reported to function primarily as a creatine analog to activate the AMPK signaling pathway, which in turn improves metabolic homeostasis and bone metabolism36,37,61. Furthermore, its ability to activate the nuclear factor erythroid 2-related factor 2 (Nrf2)/antioxidant response element (ARE) pathway confers additional anti-inflammatory effects62,63. The therapeutic significance of β-GPA is underscored by its supplementation, which favorably remodels both gut and oral microbial communities, suggesting a beneficial feedback loop. Clinical data from HPD patients revealed that the depletion of gut O. splanchnicus was significantly correlated with an accumulation of fecal arginine, a key precursor for β-GPA synthesis64,65. Building on this finding, we demonstrated that O. splanchnicus administration restored intestinal β-GPA levels. Genomic analysis reveals that while O. splanchnicus lacks the enzymes for direct β-GPA synthesis, its genome does encode arginine metabolic pathways33. This genetic profile suggests that it likely functions not as a direct producer but as a key metabolic regulator of the arginine-to-β-GPA conversion within the gut ecosystem. Its presence likely repairs this dysfunctional metabolic link by modulating the overall metabolic network of the gut microbiota. Such indirect regulation is likely multifactorial, potentially involving the production of essential substrates for other β-GPA-producing microbes, the alteration of the gut environment (e.g., pH, redox potential), or the modulation of substrate availability by outcompeting pathogens for arginine32. Following β-GPA intervention, TLR4 was identified as a key downregulated core protein in periodontal tissue. The TLR4 signaling pathway is a well-established critical driver of inflammatory bone destruction66, and plays a key role in periodontal pathogenesis67, typically activated by bacterial LPS68,69. While β-GPA’s direct interaction with TLR4 is yet to be established, its activation of AMPK provides a well-documented indirect regulatory route. Activated AMPK is known to phosphorylate and inhibit key downstream components of the TLR4 pathway, such as TRAF6, thereby suppressing NF-κB activation70. The suppression of the TLR4-MyD88 pathway appears to have a dual effect on periodontal tissue homeostasis: it simultaneously suppresses inflammation and shifts the bone remodeling balance toward anabolism. This anabolic shift is primarily driven by the attenuation of osteoclastogenesis, as TLR4 activation on myeloid precursors is a trigger for their differentiation into bone-resorbing osteoclasts, a process heavily reliant on NF-κB-mediated RANKL signaling71. Our findings of a reduced RANKL/OPG ratio and fewer osteoclasts align with this observation. Concurrently, this axis promotes osteogenesis. Chronic TLR4 signaling is known to antagonize bone morphogenetic protein (BMP) signaling by inhibiting the phosphorylation of Smad1/5/9, thereby impeding osteoblast differentiation72,73. Consistent with the attenuation of TLR4 signaling, β-GPA treatment was associated with enhanced osteogenic signaling, as evidenced by increased p-Smad1/5/9 levels and the upregulation of the transcription factors RUNX2 and Osterix. This proposed mechanism illustrates how the O. splanchnicus-β-GPA axis promotes a shift from a catabolic, inflammatory state to an anabolic, regenerative microenvironment within the periodontium, thereby tilting the balance from bone resorption to formation. Several limitations of this study warrant consideration. First, our findings indicate that O. splanchnicus depletion is primarily a hyperlipidemia-associated systemic shift that reaches peak severity in HPD, rather than a uniquely “HPD-specific” phenomenon. Consequently, the absence of a hyperlipidemia-only donor group in the FMT experiment limits our ability to fully isolate the specific contribution of the comorbid state. While the HPD-derived microbiota exacerbated bone loss compared with the PD-derived microbiota, we cannot rule out the possibility that this exacerbating effect is strongly influenced by the donors’ underlying hyperlipidemic microbial background. Second, while our data demonstrate an association between β-GPA administration and the downregulation of periodontal TLR4-MyD88 signaling, our study lacks direct functional validation. Future investigations utilizing isotope-labeled tracing (13 C-β-GPA) to track tissue distribution, or employing genetic knockout models (e.g., TLR4−/−mice) and specific pharmacological antagonists, are needed to further establish direct receptor-level antagonism and causal pathway dependency. Additionally, expanding the clinical cohort size and quantifying the precise pharmacokinetics of β-GPA will help translate these findings into clinical practice. In conclusion, this study demonstrates that the gut commensal O. splanchnicus alleviates HPD by remodeling the gut microbiome to upregulate the key functional metabolite β-GPA. This metabolite mitigates inflammatory bone loss and is associated with the downregulation of periodontal TLR4 signaling. In the broader context of gut-oral crosstalk, building upon previous findings that highlight local epithelial barrier repair, the present study provides an additional perspective. Elucidating these diverse microbial-host interactions paves the way for developing multi-strain probiotic consortia as a potential therapeutic strategy. Ultimately, these findings underscore the potential of microbiome-based interventions in addressing complex metabolic comorbidities such as HPD. Methods Human study and recruitment The human study protocol was approved by the Institutional Review Board of the Hospital of Stomatology, Jilin University (Approval No. SJDKQ2024-003), and all participants provided written informed consent. We recruited 20 patients with HPD and 20 healthy volunteers, aged 25–60 years. It is important to note that the clinical cohort analyzed in this study represents an expanded dataset from an ongoing, large-scale clinical project investigating the gut-oral axis in metabolic diseases. An earlier subset of this continuous cohort was recently reported to identify the barrier-protective role of Limosilactobacillus mucosae74. Because gut dysbiosis in HPD represents a complex ecological shift, the overarching clinical objective of this continuous recruitment is to uncover a consortium of parallel therapeutic targets systematically. Specifically, the current cohort includes 21 participants from our previously reported clinical cohort (NC: 8, HPD: 13) and 19 newly enrolled, completely independent patients (NC: 12, HPD: 7)74. Furthermore, all multi-omics re-evaluations and downstream targeted mechanism discoveries presented in this study were conducted independently. The inclusion criteria for patients included abnormal blood lipid levels (TC > 5.70 mmol/L or TG > 1.70 mmol/L) and a diagnosis of periodontitis, as classified by the 2017 guidelines of the American Academy of Periodontology/European Federation of Periodontology (AAP/EFP). The NC group had normal blood lipid levels and no periodontal disease. Exclusion criteria included using antibiotics or probiotics within the last 3 months, pregnancy, severe systemic diseases, or immunodeficiency. Saliva samples were collected from participants for 16S rRNA gene sequencing, while fecal samples were collected for metagenomic sequencing and untargeted metabolomics. The participants’ information is presented in Supplementary Table S1. Animal models All animal experiments were conducted in strict adherence to the international guidelines for the use and management of laboratory animals. These procedures received approval from the Animal Experimental Ethics Committee of Jilin University (Approval Nos. SY202310022, SY202409017, and SY202403027). Eight-week-old specific-pathogen-free (SPF) male C57BL/6J mice (purchased from Changzhou Cavens Experimental Animal Co., Ltd.) were used in this study. All mice were housed in individually ventilated cages (IVCs) with a 12 h light/dark cycle and had free access to food and water. To establish the HPD model, mice were randomly divided into four groups (n = 6 per group): NC, PD, HLP, and HPD. Hyperlipidemia was induced by feeding mice in the HLP and HPD groups an HFD (Research Diets, D12492; 60% kcal from fat, 20% from protein, 20% from carbohydrate) for 20 weeks. Mice in the NC and PD groups were fed a standard chow diet (Research Diets, D12450J; 10% kcal from fat, 20% from protein, 70% from carbohydrate). At week 19, a periodontitis model was established in the PD and HPD groups using the silk ligature method. Specifically, under isoflurane anesthesia, a 5-0 sterile silk ligature was tied around the cervix of the maxillary left second molar. After maintaining the ligature for one week, the mice were euthanized for analysis. For the FMT experiment, fresh feces were collected from donor mice (PD and HPD groups) and prepared as a suspension (50 mg feces/200 μL sterile phosphate-buffered saline (PBS)) in an anaerobic chamber. After thorough vortexing, the suspension was centrifuged at 4 °C (800 rpm, 3 min) to remove large particles, mixed with glycerol, and stored at −80 °C. To establish a pseudo-germ-free model, SPF C57BL/6J recipient mice were treated with an antibiotic cocktail (vancomycin (0.5 g/L), neomycin (1 g/L), metronidazole (1 g/L), and ampicillin (1 g/L)) in their drinking water for 4 weeks. Following depletion of their endogenous microbiota, recipient mice were randomly assigned to two groups (n = 6 per group): the PD-FMT group, which received fecal suspensions from PD donors, and the HPD-FMT group, which received suspensions from HPD donors. The mice were orally gavaged three times a week for 8 weeks. At the 7th week of gavage, periodontitis was induced in all recipient mice using the ligature method described above. For the O. splanchnicus intervention experiment, mice were assigned to four experimental groups (n = 6 per group): an NC group on a normal diet, and three groups on an HFD, designated as the HPD group, the OS group, and the OS-I group. Starting from week 21, the three HFD-fed groups received an intragastric gavage (using a specialized ball-tipped gavage needle delivered directly into the stomach to minimize local oral exposure) of either sterile PBS, live O. splanchnicus (1 × 10⁹ colony-forming units (CFU)/mL), or heat-inactivated O. splanchnicus (1 × 10⁹ CFU/mL) at a dose of 200 μL per mouse, three times weekly for 8 weeks. Periodontitis was induced at the seventh week of gavage. For the β-GPA intervention experiment, mice were randomly allocated to three groups (n = 6 per group): an NC group on a normal diet, an HPD group on a high-fat diet, and a β-GPA treatment group on an HFD. Starting from week 21, mice were administered either sterile PBS (HPD group) or β-GPA (100 mg/kg; β-GPA group) via intragastric gavage, explicitly bypassing the oral cavity to minimize local topical effects on the gingiva, with each mouse receiving 200 μL of the solution three times weekly for 8 consecutive weeks. A periodontitis model was established in the HPD and β-GPA groups at week 27. β-GPA was purchased from Sigma-Aldrich Corporation (USA) (catalog number: G6878, CAS Number: 353-09-3) and dissolved in sterile PBS. At the end of the respective experimental periods, mice underwent a standardized terminal procedure for sample collection. Specifically, mice were anesthetized using 5% isoflurane in oxygen via an induction chamber. Once deep anesthesia and the loss of the pedal withdrawal reflex were confirmed, terminal blood was collected via the retro-orbital sinus. Mice were then immediately euthanized by cervical dislocation while still fully anesthetized. This procedure, performed in strict accordance with the AVMA Guidelines for the Euthanasia of Animals, was selected to minimize animal pain and distress while preventing hypoxic or chemical artifacts from confounding downstream multi-omics and biochemical analyses. Bacterial culture Odoribacter splanchnicus (BNCC359789) was sourced from the China National Collection for Bioprocesses and Cell Cultures (BNCC, Beijing, China). A single colony was isolated, streaked for purity on a fresh Columbia Blood Agar (CBA) plate, and incubated at 37 °C for 4–5 days in an anaerobic chamber (85% N₂, 10% H₂, 5% CO₂). A single resulting colony was then inoculated into modified Chopped Meat Medium (CMM) for liquid expansion. Bacteria were harvested during the logarithmic growth phase (Optical Density at 600 nm (OD₆₀₀) ≈ 0.8–1.0, typically after 24–36 h of incubation). The bacterial cells were collected by centrifugation (3000 rpm, 10 min, 4 °C) and washed with sterile PBS to remove medium residue. The final bacterial pellet was resuspended in sterile PBS, and the concentration was adjusted to 1 × 10⁹ CFU/mL using a combination of turbidimetry and plate counting. To prepare the heat-inactivated bacteria, a live bacterial suspension was incubated at 70 °C for 30 min, followed by rapid cooling to 4 °C. Successful inactivation was confirmed by plating the suspension on CBA and observing no growth after anaerobic incubation. Multi-omics analysis For 16S rRNA gene sequencing and bioinformatic analysis, mouse saliva was collected under sterile conditions and immediately frozen at −80 °C. Total DNA was extracted from samples using a commercial DNA isolation kit. The V3-V4 hypervariable region of the 16S rRNA gene was amplified by polymerase chain reaction (PCR) using universal primers (341F: CCTACGGGNGGCWGCAG; 805R: GACTACHVGGGTATCTAATCC). Purified amplicons were used to construct sequencing libraries and were sequenced on an Illumina NovaSeq 6000 platform using a paired-end strategy. The raw sequencing data were processed and analyzed using the QIIME2 (v2022.2) platform. The DADA2 plugin was used for sequence denoising, merging, and chimera removal to generate amplicon sequence variants (ASVs). ASVs were taxonomically classified against the SILVA database (v138). Alpha diversity (e.g., Shannon and Observed species indices) and beta diversity (Principal coordinates analysis (PCoA) based on Bray-Curtis distances) were calculated in QIIME2. Permutational multivariate analysis of variance (PERMANOVA) was used to assess significant differences in microbial community structure between groups. LEfSe was employed to identify microbial biomarkers with significantly different abundances between groups. For metagenomic sequencing analysis, cecal contents were aseptically collected from mice, and total DNA was extracted as described above. Metagenomic libraries were constructed through fragmentation, end-repair, A-tailing, and adapter ligation, followed by sequencing on an Illumina NovaSeq platform. The KneadData tool was used to control the quality of raw data and remove host-derived sequences. MetaPhlAn3 was used for taxonomic profiling, and HUMAnN3 was used for functional pathway abundance analysis with reference to the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. For untargeted metabolomics analysis, ~20 mg of intestinal content was extracted with a methanol-water solution (4:1, v/v). After centrifugation, the supernatant was analyzed using an ultra-high-performance liquid chromatography-quadrupole time-of-flight mass spectrometry (UHPLC-Q-TOF-MS) system in both positive and negative ionization modes. Raw data were processed for peak picking, alignment, and correction using the XCMS and CAMERA packages. Metabolites were identified by searching the online Human Metabolome Database (HMDB), KEGG, and LIPID MAPS databases. Orthogonal projections to latent structures discriminant analysis (OPLS-DA) models were used to screen for differential metabolites, using the criteria of Variable Importance in Projection (VIP) > 1, p < 0.05, and fold change >1.5. Proteomics analysis Gingival and jawbone tissues were frozen in liquid nitrogen and mechanically homogenized. Total protein was extracted using RIPA lysis buffer containing protease and phosphatase inhibitors. Proteomic analysis was performed using Tandem Mass Tag (TMT) labeling coupled with liquid chromatography-tandem mass spectrometry (LC-MS/MS). Proteins were identified by searching the UniProt mouse protein database with Mascot software (v2.6). The criteria for differentially expressed proteins were a fold change >1.5 and a p-value < 0.05. The Database for Annotation, Visualization and Integrated Discovery (DAVID) online tool and KEGG database were used for Gene Ontology (GO) functional annotation and KEGG pathway enrichment analyses of the differentially expressed proteins. Micro-computed tomography analysis For in vivo body composition analysis, mice were anesthetized with isoflurane in O₂ (5% for induction, 1.5% for maintenance) and positioned prone on the scanner bed. Whole-body scans, excluding the head and tail, were acquired using a NEMO® Micro-CT system (PINGSENG Healthcare (Kunshan) Inc., China). The scanning parameters were set to 65 kV, 190 μA, with an isotropic voxel size of 80 μm. Following the scan, 2D projection images were reconstructed into 3D tomograms using the proprietary software (Recon, PINGSENG Healthcare (Kunshan) Inc., China). Adipose tissue was identified and segmented by applying a specific grayscale threshold. The total fat and tissue volumes were calculated based on the segmented 3D models. The Fat Volume/Total Volume was then determined for each mouse. For ex vivo maxillary bone microarchitectural analysis, after euthanasia, maxillae were fixed in 4% paraformaldehyde and scanned using a NEMO® Micro-CT system (50 kV, 500 μA, 9 μm isotropic voxel size). Following 3D reconstruction using Recon software, alveolar bone loss was quantified by measuring the CEJ-ABC distance. A standardized region of interest (ROI) at the maxillary second molar furcation was defined for microarchitectural analysis. Within this ROI, key parameters including BMD, BV/TV, Tb.Th, Tb.N, and Tb.Sp were calculated. Histological analysis At the end of the experiment, following the terminal procedure and euthanasia, the heart, liver, spleen, kidneys, ileum, four types of adipose tissue (iWAT, eWAT, pWAT, and brown adipose tissue (BAT)), and maxillae were quickly collected. The heart, spleen, kidneys, ileum, and adipose tissue were fixed in 4% paraformaldehyde, then processed for paraffin embedding and sectioned at 5 μm. The maxillae were fixed and decalcified in 10% ethylenediaminetetraacetic acid (EDTA) for 2 weeks before undergoing the same paraffin processing and sectioning. Liver tissue was divided into two parts: one was fixed in 4% paraformaldehyde for paraffin embedding, and the other was snap-frozen in liquid nitrogen for cryosectioning. All paraffin sections were stained with H&E to observe basic tissue morphology. Frozen liver sections (10 μm thick) were stained with Oil Red O to assess lipid deposition, then observed and imaged under a microscope. Maxillary sections were stained for TRAP to label osteoclasts, following the manufacturer’s instructions (Sigma-Aldrich, Cat. No. N2125). Biochemical analysis Fasting blood glucose levels were measured using a portable glucose meter (Sannuo Biosensing Co., Ltd., China) via a minor tail vein prick. For serum biochemical profiling, terminal blood samples were collected via the retro-orbital sinus under deep isoflurane anesthesia, as described in the euthanasia protocol. The collected blood was allowed to clot at room temperature for 30 min, and then centrifuged at 3000 rpm for 10 min at 4°C (twice) to separate the serum. Serum TC, TG, LDL-C, and high-density lipoprotein cholesterol (HDL-C) were measured using commercial kits (Nanjing Jiancheng Bioengineering Institute, China, Cat. Nos. A111-1-1, A110-1-1, A113-1-1, and A112-1-1, respectively). Western blot analysis Total protein from alveolar bone tissues was extracted and quantified using the bicinchoninic acid (BCA) method. Equal amounts of protein were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidene fluoride (PVDF) membrane. The membranes were incubated overnight at 4 °C with primary antibodies, followed by a 1 h incubation at room temperature with secondary antibodies. Bands were detected using enhanced chemiluminescence (ECL) and quantified using ImageJ software. The expression levels of the following proteins were measured: NLRP3, IL-6, IL-1β, TNF-α, CTSK, RUNX2, Osterix, OPG, OPN, OCN, RANKL, iNOS, TLR4, MyD88, TRAF6, IKKα/β, p-IKKα/β, NF-κB p65, p-NF-κB p65, Smad1/5/9, and p-Smad1/5/9. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the loading control. The specific information of the antibodies can be found in Supplementary Table S3. The expression of each protein was assessed using samples from three independent biological replicates. Representative blots are shown, and all experiments were repeated independently three times with similar results. Immunofluorescence staining For immunofluorescence staining, maxillary sections underwent antigen retrieval and were then incubated overnight at 4 °C with primary antibodies (Supplementary Table S4). Subsequently, sections were incubated with corresponding fluorescent secondary antibodies for 1 h at room temperature and counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Images were captured, and fluorescence intensity was quantified using ImageJ software (NIH, USA) by averaging three non-overlapping fields of view per sample. Each staining experiment was performed in triplicate. Statistics and reproducibility Data were analyzed using SPSS software (version 25.0; IBM, NY, USA). Quantitative data are presented as the mean ± standard deviation (SD). For normally distributed continuous variables, comparisons between two independent groups were performed using a two-tailed unpaired Student’s t test, while comparisons among multiple groups were conducted using a one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc multiple comparisons test. Non-normally distributed data were analyzed with the Mann–Whitney U test for two groups, or the Kruskal–Wallis test for multiple groups. Spearman correlation was used for correlation analysis. Microbiome data were statistically analyzed in R using packages such as vegan and phyloseq. A p-value < 0.05 was considered statistically significant. Supplementary information Supplementary material Acknowledgements We gratefully acknowledge all patients and healthy volunteers for their participation in this study, as well as the essential contributions of the laboratory animals that made this research possible. This work was supported by grants from the National Natural Science Foundation of China (82470987), the Science and Technology Project of Jilin Province Department of Finance (jcsz2025678-2), and the “Medicine+X” Interdisciplinary Innovation Project (2024JBGS07). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Author contributions J.X.: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Visualization, Writing - original draft. Z.H.: Investigation, Methodology, Validation. Q.X.: Investigation, Methodology, Validation. H.W.: Investigation, Methodology. Y.L.: Methodology, Project administration. J.S.: Investigation, Methodology. L.L.: Conceptualization, Supervision, Validation, Writing - review & editing. M.H.: Conceptualization, Funding acquisition, Project administration, Supervision, Validation, Writing - review & editing. D.W.: Conceptualization, Funding acquisition, Project administration, Supervision, Validation, Writing - review & editing. Data availability The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive in the National Genomics Data Center, China National Center for Bioinformation / Beijing Institute of Genomics, Chinese Academy of Sciences, under the accession numbers CRA031242, CRA031295, CRA031308, and CRA031309. These datasets are publicly accessible at https://ngdc.cncb.ac.cn/gsa. The metabolomics data have been deposited to MetaboLights repository with the study identifiers MTBLS13125 and MTBLS13135. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (https://proteomecentral.proteomexchange.org) via the iProX partner repository with the dataset identifier PXD069376. All other relevant data supporting the findings of this study are available within the article and its supplementary information files. Code availability No custom code was used in this study. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Lanzhou Li, Email: lilanzhou@jlau.edu.cn. Min Hu, Email: humin@jlu.edu.cn. Di Wang, Email: jluwangdi@jlu.edu.cn. Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41522-026-01033-4. References Associated Data Supplementary Materials Supplementary material Data Availability Statement The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive in the National Genomics Data Center, China National Center for Bioinformation / Beijing Institute of Genomics, Chinese Academy of Sciences, under the accession numbers CRA031242, CRA031295, CRA031308, and CRA031309. These datasets are publicly accessible at https://ngdc.cncb.ac.cn/gsa. The metabolomics data have been deposited to MetaboLights repository with the study identifiers MTBLS13125 and MTBLS13135. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (https://proteomecentral.proteomexchange.org) via the iProX partner repository with the dataset identifier PXD069376. All other relevant data supporting the findings of this study are available within the article and its supplementary information files. No custom code was used in this study.