Arginine metabolism as a mutualistic strategy in the interaction between Porphyromonas gingivalis and Filifactor alocis

Abstract

Metabolic cross-feeding is a main driver of microbial mutualism under various stresses. Elevated arginine levels in the periodontal microenvironment reshape pathogen physiology. Here, we discovered that increased arginine levels disrupted the intracellular amino acid metabolism of the periodontal pathogen Porphyromonas gingivalis by repressing glutamate dehydrogenase activity and α-ketoglutarate production via the accumulation of NH₃. This perturbation inhibited P. gingivalis growth and was accompanied by increased expression of the PorX/PorY-associated T9SS and enhanced small-peptide acquisition. Given the frequent co-occurrence of P. gingivalis and Filifactor alocis in periodontitis, our findings suggested that F. alocis partially relieved arginine-associated growth inhibition of P. gingivalis through ArcA-dependent arginine degradation, thereby promoting its persistence under high-arginine conditions. This cooperative arginine metabolism is critical for the resilience of microbial communities and may serve as a therapeutic target.

Complex microbial communities are often shaped by cooperation. Here, the authors show that high arginine levels suppress the periodontal pathogen Porphyromonas gingivalis, while Filifactor alocis helps it persist by breaking down arginine, revealing a metabolic partnership.

Introduction

Periodontitis is a microbial dysbiosis-driven inflammatory disease that causes progressive soft tissue and bone destruction^1,2. During periodontitis, oral microbiota dysbiosis occurs with an increased abundance of periodontal pathogens such as Porphyromonas gingivalis and Filifactor alocis, among which P. gingivalis is one of the most critical pathogens. Concurrently, the microenvironment undergoes metabolic alterations with elevated levels of amino acids, iron sources and nitrate^3–6. Arginine serves as a critical modulator of microbiota-host homeostasis, and its metabolism is dysregulated in pathological conditions such as inflammatory bowel disease and microbial infections⁷. The metabolic profile of arginine and its derivatives undergoes significant changes in patients with periodontitis^8–12. Arginine concentrations are reported to increase in periodontitis, rising in whole saliva from 26.4 (12.4–34.7) mg/dL in health to 64.6 (39.4–70.8) mg/dL in chronic periodontitis¹³. Notably, some oral-care formulations contain 1.5% arginine^14,15, and certain products contain up to 8%¹⁶, generating transient local concentrations that exceed physiological levels. However, the effects of arginine on periodontal pathogens are paradoxical. For example, D-arginine has been reported to inhibit the formation of P. gingivalis biofilms and promote their dispersal¹⁷. A secreted arginine deiminase from Streptococcus intermedius has been shown to inhibit P. gingivalis biofilm development¹⁸. Consistently, clinical studies have shown that arginine-containing formulations can reduce the abundance of periodontal pathogens, including P. gingivalis¹⁵. In contrast, in vitro studies suggest that arginine-derived polyamines produced through metabolic cooperation between Streptococcus gordonii and Fusobacterium nucleatum may promote the proliferation of P. gingivalis¹⁹.

Unlike many bacteria that utilize free amino acids, P. gingivalis preferentially utilizes peptides, particularly glutamate- and aspartate-containing peptides, as primary energy and carbon sources^20,21. P. gingivalis degrades environmental proteins into oligopeptides via gingipains, and the resulting oligopeptides are imported into the periplasm by RagAB²². Periplasmic exopeptidases (DPP4/5/7/11, AOP) then hydrolyze these into di/tripeptides^23–29, which are transported across the inner membrane by SstT/Pot/Opt transporters and ultimately degraded to amino acids by cytoplasmic peptidases^30–32. The type IX secretion system (T9SS), a transmembrane complex composed of 20 core proteins and regulated by the PorX/PorY two-component signal transduction system, serves as the primary channel for P. gingivalis to export gingipains, such as Kgp and RgpA/B, to the bacterial surface or extracellular environment. The T9SS and its secreted proteases serve as critical factors for P. gingivalis predation and environmental fitness^33,34.

During periodontal infection, P. gingivalis thrives within a complex polymicrobial ecosystem and is exposed to a dynamic metabolic microenvironment^35–37. The elevated arginine-induced stress in gingival crevicular fluid and the susceptibility of P. gingivalis to arginine raise complex and important questions concerning how P. gingivalis adapts to high arginine levels and becomes one of the most abundant periodontal pathogens in the periodontal pocket. Microbial cooperative coexistence is a prevalent ecological strategy in nature. Another periodontal pathogen, F. alocis, has attracted increasing attention because of its exceptional resistance to oxidative stress and efficient amino acid utilization^38–43. F. alocis exhibits a preference for free amino acids and requires high levels of exogenous cysteine and arginine for growth^42,43. Previous studies have reported that F. alocis, similar to P. gingivalis, is frequently enriched in the subgingival microbiota of periodontitis patients. Fluorescence in situ hybridization (FISH) further revealed that these two species exhibit similar spatial clustering within diseased gingival tissues, and their in situ metabolic activities in the gingival crevice are positively correlated, collectively suggesting potential ecological interactions between them^44–47. Mechanistic investigations indicated that F. alocis promoted P. gingivalis growth via LuxS-mediated quorum sensing, and their coculture synergistically increased biofilm development and host cell invasion capabilities^48,49. On the basis of their observed association and utilization of peptides/amino acids, we hypothesize that P. gingivalis and F. alocis may establish an arginine metabolism-dependent mutualistic relationship to enhance their adaptive fitness within the periodontal pocket microenvironment.

Here, we demonstrated that elevated arginine levels suppressed P. gingivalis growth by disrupting its amino acid metabolism. Arginine negatively regulated glutamate dehydrogenase activity, consequently diminishing α-ketoglutarate production, and this metabolic disorder was accompanied by increased expression of the PorX/PorY-associated T9SS, increased Kgp activity, and enhanced small-peptide acquisition. Given the frequent co-occurrence of P. gingivalis and F. alocis in periodontitis, F. alocis partially relieved arginine-associated growth inhibition of P. gingivalis through ArcA-dependent arginine degradation, consistent with a metabolic interaction between the two species. These findings revealed the multifaceted role of arginine and its metabolism in the mutualism and resilience of microbial communities and offered potential therapeutic perspectives for periodontal disease management.

Results

Coexistence of P. gingivalis and F. alocis in periodontitis

To investigate the ecological relationship between P. gingivalis and F. alocis in oral microbial communities, we performed an integrative analysis of subgingival plaque metagenomic sequencing data derived from five independent publicly available datasets. The integrated cohort comprised a total of 102 samples, including two datasets from China (PRJNA230363 and PRJNA932553), two from the United States (PRJNA255922 and PRJNA508385), and one from Japan (PRJDB11203) (Fig. 1a). The study population consisted of 44 periodontally healthy individuals and 58 patients with chronic periodontitis. The subgingival microbiota differed between patients with periodontitis and periodontally healthy subjects (Supplementary Fig. 1a, b). Linear discriminant analysis Effect Size (LEfSe) analysis identified P. gingivalis and F. alocis as indicator species enriched in the periodontitis group (Fig. 1b). The relative abundances of P. gingivalis (RH = 0.01057, RP = 0.1171; p < 0.0001) and F. alocis (RH = 0.002129, RP = 0.008562; p < 0.0001) were both significantly higher in periodontitis samples than in healthy controls (Fig. 1c). To further explore microbial taxa ecologically associated with the keystone periodontal pathogen P. gingivalis within the subgingival microbiome, Spearman correlation coefficients were calculated between P. gingivalis and all annotated species. Figure 1d displayed all taxa exhibiting statistically significant correlations (p < 0.05), ranked from left to right in descending order of correlation strength. Species with correlation coefficients greater than 0.6 (highlighted in red) were considered to constitute a core co-occurrence network strongly associated with P. gingivalis. Among these taxa, F. alocis showed a relatively strong ecological association with P. gingivalis in the subgingival microbiome, with correlation coefficients comparable to those observed for Tannerella forsythia, Treponema denticola, and Treponema putidum. Figure 1e further illustrated this relationship, showing a positive correlation between the relative abundances of P. gingivalis and F. alocis in clinical subgingival plaque samples (Spearman’s r = 0.7636, p < 0.0001).

Fig. 1. P. gingivalis and F. alocis are co-enriched and positively associated in the subgingival microbiota, and coinfection is associated with increased alveolar bone loss in the mouse periodontitis model.

a Overview of the integration of five independent public subgingival metagenomic datasets. b Indicator species in the subgingival microbiota of periodontally healthy subjects and patients with periodontitis identified by LEfSe analysis. c Relative abundances of Pg and Fa in subgingival plaque samples from periodontally healthy subjects and patients with periodontitis (healthy, n = 44 participants; periodontitis, n = 58 participants; two-sided Mann–Whitney U test). d Spearman correlation coefficients of bacterial taxa co-occurring with Pg in the subgingival microbiota, with Fa highlighted in orange on the x axis and taxa with correlation coefficients >0.6 shown as red bars. e Correlation between the relative abundances of Pg and Fa in subgingival plaque samples from periodontally healthy participants (n = 44 participants) and patients with periodontitis (n = 58 participants). Each point represents one independent participant (total n = 102 participants). Correlation was assessed using a two-sided Spearman rank correlation test (r = 0.7636, p < 0.0001). The solid line represents the simple linear regression fit, and the shaded area indicates the 95% confidence interval. f Representative 3D-reconstructed micro-CT images of maxillae (buccal view) from (1) the noninfected control, (2) Pg monoinfection, (3) Fa monoinfection, and (4) Pg + Fa coinfection groups (n = 10 mice). g Bone loss area (left) and CEJ-to-alveolar crest distance were mesiodistally (middle and right) measured across experimental groups (n = 10 mice; one-way ANOVA with Tukey’s multiple-comparisons test). Data in g is presented as the mean ± s.d. Only selected pairwise comparisons are shown; exact p values are above comparison bars. Source data are provided as a Source Data file. Pg P. gingivalis; Fa F. alocis.

To further evaluate the interaction between P. gingivalis and F. alocis in vivo, we employed a murine ligature-induced periodontitis model. Mice were infected with P. gingivalis, F. alocis, or both species. Compared with either monoinfection, coinfection with P. gingivalis and F. alocis resulted in greater alveolar bone loss, as reflected by a larger bone loss area and an increased cemento–enamel junction (CEJ) to alveolar crest distance (Fig. 1f, g). These findings indicated that the presence of both species enhanced periodontal bone destruction in this murine model. Two-way ANOVA revealed a significant Pg × Fa interaction effect on bone loss area, indicating that the combined effect of the two species exceeded the additive expectation of the two monoinfections. Although the CEJ to alveolar crest distance showed a similar trend, the interaction effect did not reach statistical significance for this measurement (Supplementary Fig. 1c–e).

Arginine inhibited P. gingivalis growth, which was partially alleviated by F. alocis

To assess whether amino acids influence the growth of P. gingivalis, we screened a panel of 13 amino acids under identical culture conditions. Among these, arginine, valine, ornithine, and histidine inhibited P. gingivalis growth (Fig. 2a and Supplementary Fig. 2a). We next performed a concentration-dependent analysis of arginine and quantified bacterial growth at 24 h. Arginine inhibited P. gingivalis growth in a dose-dependent manner within the range of 0.05%–0.8%. Notably, 0.2% arginine represented the highest concentration at which bacterial growth was not completely abolished and was therefore selected for subsequent analyses (Fig. 2b, c). Live/dead staining and colony-forming units (CFUs) enumeration demonstrated that 0.2% arginine reduced bacterial viability (Supplementary Fig. 2c–f). 24 h growth curves further demonstrated that 0.2% arginine significantly suppressed P. gingivalis growth dynamics (Fig. 2d), which was corroborated by FISH showing reduced bacterial biomass (Fig. 2f). In a murine ligature-induced periodontitis model, arginine attenuated P. gingivalis–associated alveolar bone loss (Fig. 2i, j). To determine whether this effect was dependent on P. gingivalis, we included additional control groups (ligature-only as Control, Control+Arg, Control+Pg, and Control+Pg+Arg). Arginine reduced bone loss in the Control+Pg+Arg group but had no significant effect in the Control+Arg group, indicating that arginine primarily modulated P. gingivalis–driven pathology rather than ligature-induced inflammation alone (Supplementary Fig. 2k–l).

Fig. 2. F. alocis partially counteracts the inhibitory effect of arginine on P. gingivalis growth.

a Effects of 13 amino acids at 0.2% (w/v) on Pg and Fa growth. Log-transformed fold changes relative to untreated controls were derived from the 0.2% conditions in Supplementary Fig. 2a, b; positive and negative values indicate growth promotion and inhibition, respectively. Orange and red bars denote Fa and Pg (n = 4). P values compare 0.2% with untreated controls and were obtained from full concentration-series analyses in Supplementary Fig. 2a, b using one-way ANOVA with Dunnett’s multiple-comparisons test. b Effect of 0–0.8% (w/v) exogenous arginine on Pg growth in monoculture or coculture with Fa (n = 3; multiple two-sided unpaired t-tests). c Fa-mediated fold change in Pg growth at 0–0.2% (w/v) arginine (n = 3; one-way ANOVA with Dunnett’s multiple-comparisons test). d, e 24 h growth curves of Pg (d) and Fa (e) in monoculture or coculture with or without 0.2% (w/v) arginine (n = 4). f FISH images (50 μm scale) of Pg (magenta) and Fa (green) in mono- and cocultures, with or without 0.2% (w/v) arginine. The experiment was repeated three times with similar results. g SEM visualization of Pg and Fa in mono- and cocultures (24 h) at magnifications (20, 10, and 5 μm scales) under 0.2% (w/v) arginine. The experiment was repeated three times with similar results. h Effect of 24 h Fa culture supernatants with or without 0.2% (w/v) arginine on the growth of Pg (n = 4; one-way ANOVA with Dunnett’s multiple-comparisons test). i Representative 3D-reconstructed micro-CT images of maxillae (buccal view) from the (1) Pg, (2) Pg + 0.4%Arg, (3) Pg + 2%Arg, (4) Pg + 0.4%Arg + Fa and (5) Pg + 2%Arg + Fa groups (n = 5 mice). j Bone loss area (left) and CEJ-to-alveolar crest distance were mesiodistally (middle and right) measured across groups (n = 5 mice; one-way ANOVA with Tukey’s multiple comparison test). Unless stated, n denotes biologically independent cultures. Data in a–h and j are mean ± s.d. (error bars, s.d.). Only selected pairwise comparisons are shown; exact p values are above comparison bars. Source data are provided as a Source Data file. Pg P. gingivalis; Fa F. alocis; Arg Arginine.

Given that arginine can be metabolized by neighboring microbes, we next investigated whether F. alocis could function as a food–web partner to modulate arginine availability. Screening of the same amino acid panel showed that P. gingivalis and F. alocis responded differently, and in many cases oppositely, to individual amino acids, with arginine and ornithine producing the most marked differences (Fig. 2a and Supplementary Fig. 2a–b). In coculture (Pg:Fa = 10:1), F. alocis partially relieved arginine-mediated growth inhibition of P. gingivalis, whereas ornithine further enhanced inhibition, suggesting a distinct mode of action (Fig. 2b and Supplementary Fig. 2g). Focusing on 0.2% arginine, time-course coculture experiments, together with FISH and scanning electron microscopy (SEM), consistently demonstrated that F. alocis partially restored P. gingivalis growth under arginine-induced stress (Fig. 2d–g). Notably, this effect was influenced by the initial coculture ratio. At a Pg:Fa ratio of 1:1, coculture further reduced P. gingivalis, whereas at a ratio of 5:1, coculture increased P. gingivalis after 24 h (Supplementary Fig. 2h). To determine whether this effect was mediated by arginine depletion, we performed supernatant transfer experiments. Supernatants from F. alocis cultured in arginine-supplemented medium partially rescued P. gingivalis growth, whereas supernatants from control cultures had no effect (Fig. 2h). Although arginine metabolism led to measurable alkalinization of the culture medium, both the inhibitory effect of arginine on P. gingivalis and the rescuing effect of F. alocis were maintained across the tested pH range, suggesting that these effects were not primarily driven by pH (Supplementary Fig. 2i, j). Consistently, in vivo, F. alocis partially attenuated the inhibitory effect of arginine on P. gingivalis–associated periodontal bone loss (Fig. 2i, j).

Arginine disrupted P. gingivalis amino acid metabolism

To elucidate the mechanism through which F. alocis affects P. gingivalis growth under arginine-induced stress, untargeted metabolomics was employed to profile changes in the levels of intracellular metabolites in P. gingivalis with or without arginine and F. alocis. Arginine treatment significantly altered the intracellular metabolic landscape of P. gingivalis (Supplementary Fig. 3a), with 98 metabolites with increased abundance and 52 metabolites with decreased abundance (Supplementary Fig. 3b, c). By classifying all the significantly altered metabolites, we found that the levels of amino acids and their metabolites—particularly small peptides—exhibited the most significant changes under both arginine treatment and F. alocis coculture (Fig. 3a, b and Supplementary Fig. 3d–e). Arginine treatment significantly increased intracellular small peptide levels (Tyr-Asn-Asp, Tyr-Glu-Val-Lys, and Asp-Gly-Asp-Lys-Asn) in P. gingivalis and significantly upregulated the expression of proteins involved in intracellular amino acid and peptide transport systems (Supplementary Fig. 3f), whereas coculture with F. alocis reduced the levels of specific peptides (Asp-Asn-Gly-Asn-Phe, Tyr-Pro-Tyr, and Lys-Glu-Asp) (Fig. 3c, d).

Fig. 3. The P. gingivalis PorX/PorY two-component signal transduction system is involved in the T9SS-dependent predation in the presence of arginine and F. alocis.

a Class I metabolite scatterplot; b bubble plot of differentially abundant amino acids/metabolites and glycerophospholipids; c, d top 20 VIP-ranked amino acids and metabolites for Pg + 0.2%Arg versus Pg (a–c) or Pg + 0.2%Arg + Fa versus Pg + 0.2%Arg (d). Differential metabolites were defined by OPLS-DA VIP > 1 and FDR-corrected p < 0.05 from multiple two-sided unpaired t tests. e Hemagglutination of Pg, Pg + Fa, Pg + 0.2%Arg and Pg + 0.2%Arg + Fa groups; left scale, bacterial dilution factors. The experiment was independently repeated twice with similar results. f, g Initial Kgp reaction velocity (V₀) in Pg with or without 0.2% Arg (f) and Pg + Fa with or without 0.2% Arg (g) (n = 3; two-sided unpaired t tests). h, i Kgp and T9SS gene expression with or without 0.2% Arg (h) and with or without Fa under 0.2% Arg (i) (n = 4). j Schematic of PorX/PorY-mediated T9SS regulation. k Hemagglutination of ΔporX, ΔporX + Fa, ΔporX + 0.2%Arg and ΔporX + 0.2%Arg + Fa groups. The experiment was independently repeated twice with similar results. l, m, Initial Kgp reaction velocity (V₀) in ΔporX with or without 0.2% Arg (l) and ΔporX + Fa with or without 0.2% Arg (m) (n = 4; two-sided unpaired t tests). n Arginine-induced T9SS transcriptional responses in Pg WT, ΔporX, and ΔporY strains (n = 4; multiple two-sided unpaired t tests). o Representative 3D-reconstructed micro-CT images of maxillae from Pg WT, Pg WT + 2%Arg, Pg WT + 2%Arg + Fa, ΔporX, ΔporX + 2%Arg and ΔporX + 2%Arg + Fa groups (n = 5 mice). p Bone loss area (n = 5 mice; one-way ANOVA with Tukey’s multiple comparison test). q Arginine-attenuated alveolar bone resorption in Pg WT and ΔporX infections (n = 5 mice; two-sided unpaired t test). Unless stated, n denotes biologically independent cultures. Data in h, i, n, p and q are mean ± s.d. (error bars, s.d.). Data in f, g, l, and m are floating min-to-max plots (center lines, medians; limits, minima and maxima). Only selected pairwise comparisons are shown; exact p values are above comparison bars. Source data are provided as a Source Data file. Pg P. gingivalis; Fa F. alocis; Arg Arginine; WT wild type; ΔporX porX deletion mutant; ΔporY porY deletion mutant. Amino acids in peptide sequences are denoted using standard three-letter codes.

The increased abundance of intracellular small peptides in P. gingivalis suggested that arginine perturbed amino acid metabolism and might enhance proteolytic activity mediated by the Kgp/Rgp system, which could facilitate protein degradation and nutrient acquisition. Indeed, arginine increased the hemagglutination ability and Kgp protease activity of P. gingivalis, whereas these effects were reversed by F. alocis (Fig. 3e–g; Supplementary Fig. 3g–i); however, the Rgp protease activity did not significantly change (Supplementary Fig. 3j–m), suggesting that arginine preferentially enhanced Kgp-associated activity in P. gingivalis. We then screened the relevant genes that might affect the activity of the Kgp enzyme and examined their expression at 4, 12, and 24 h after arginine treatment or coculture with F. alocis. The expression of the Kgp-encoding gene decreased at 4 h but then gradually increased, and the expression of the relevant genes encoding the T9SS (porL) and the two-component sensor system porX/Y, which regulate the T9SS, also significantly increased in the early stage (Fig. 3h). Interestingly, F. alocis significantly reversed the upregulation of porX/Y at both 4 h and 12 h (Fig. 3i), supporting the involvement of PorX/PorY in the T9SS-associated response of P. gingivalis to arginine (Fig. 3j). The porX and porY deletion mutants of P. gingivalis were subsequently constructed. In comparison with the wild type (Fig. 3e), ΔporY showed a slight decrease in hemagglutination function with or without arginine and F. alocis (Supplementary Fig. 4c), but hemagglutination was nearly abolished in ΔporX (Fig. 3k). Kgp activity was reduced in ΔporY and was further diminished in ΔporX. Arginine failed to enhance Kgp activity in both ΔporX and ΔporY mutants (Fig. 3l, m; Supplementary Fig. 4a, b, d–g). To further assess transcriptional responses, we examined the expression of T9SS-related genes in P. gingivalis wild type (WT), ΔporX, and ΔporY following arginine stimulation. Arginine increased the expression of T9SS-associated genes in the wild-type strain, whereas this induction was markedly reduced in ΔporX and partially attenuated in ΔporY (Fig. 3n). These results suggested that arginine-associated regulation of the T9SS involved the PorX/PorY regulatory system, with PorX playing a predominant role and PorY making a partial contribution.

To evaluate the contribution of PorX in vivo, a mouse periodontitis model was established with six groups: the Pg WT, Pg WT + 2%Arg, Pg WT + 2%Arg + Fa, ΔporX, ΔporX + 2%Arg, and ΔporX + 2%Arg + Fa groups. The degree of periodontal destruction caused by ΔporX was less than that caused by Pg WT. Specifically, compared with ΔporX, Pg WT led to more severe reductions in mesiodistal alveolar bone height and greater loss of alveolar bone area (Fig. 3o–p; Supplementary Fig. 4h–k). Arginine reduced alveolar bone resorption induced by both Pg WT and ΔporX, with a stronger effect observed in the ΔporX background (Fig. 3p, q). Moreover, F. alocis reversed the protective effects of arginine on both Pg WT and ΔporX (Fig. 3o, p and Supplementary Fig. 4h–k). Although our data showed that arginine modulated the expression of porX/porY and that the arginine-dependent increase in Kgp activity required both regulators, direct biochemical evidence for arginine–dependent activation of PorY or PorY–mediated phosphorylation of PorX was not obtained in our study.

F. alocis restored P. gingivalis glutamate metabolism

To further elucidate the mechanism through which arginine alters the P. gingivalis nutrient acquisition system to increase small peptide uptake, we hypothesized that arginine disrupts amino acid metabolism in P. gingivalis since P. gingivalis preferentially catabolizes peptides, particularly glutamate/aspartate-rich ones, as its primary carbon and energy sources compared to some free amino acids. According to the results of the metabolome analysis, the intracellular levels of proline, 4-hydroxyproline, glycocyamine, glutamate, aspartate, and ornithine were significantly elevated in P. gingivalis, whereas the levels of citrulline, 1-pyrroline-4-hydroxy-2-carboxylate, and α-ketoglutarate were significantly reduced (Fig. 4a), indicating that the addition of arginine affected the metabolism of glutamate, which is among the preferred amino acids for P. gingivalis growth. Arginine can be metabolized via the urea cycle pathway to produce ornithine and NH₃, while glutamate is catabolized by glutamate dehydrogenase into α-ketoglutarate and NH₃ (Fig. 4b). Since NH₃ is a common byproduct of both arginine and glutamate metabolic pathways, we hypothesized that elevated arginine levels forced P. gingivalis to produce more ornithine and NH₃ and that the accumulated NH₃ inhibited glutamate dehydrogenase activity to reduce α-ketoglutarate production. As expected, under nutrient-rich conditions, exogenous addition of NH₃ inhibited the growth of P. gingivalis and similarly suppressed the activity of glutamate dehydrogenase (Supplementary Fig. 5a–c). Arginine significantly increased NH₃ concentrations in culture supernatants (Fig. 4c) but decreased intracellular α-ketoglutarate levels (Fig. 4d) and glutamate dehydrogenase activity (Fig. 4e). Notably, α-ketoglutarate supplementation partially alleviated arginine-mediated growth inhibition and restored P. gingivalis growth (Fig. 4f). To further validate whether F. alocis could restore glutamate dehydrogenase activity and glutamate metabolism to induce α-ketoglutarate production, a Transwell system of P. gingivalis-F. alocis coculture was constructed (Fig. 4g). The addition of F. alocis significantly decreased the NH₃ concentration in the culture supernatant (Fig. 4h) and increased the intracellular α-ketoglutarate concentration (Fig. 4i) accompanied by the restoration of glutamate dehydrogenase activity (Fig. 4j). These results suggested that arginine perturbed glutamate metabolism in P. gingivalis, leading to metabolic imbalance associated with depletion of key intermediates such as α-ketoglutarate. This metabolic disturbance was accompanied by changes in the PorX/PorY-regulated T9SS and increased Kgp activity. Consistently, supplementation with α-ketoglutarate reversed the arginine-associated increase in porX, supporting a link between PorX/PorY and metabolic perturbation related to glutamate pathway disruption (Supplementary Fig. 5d). In contrast, acute starvation in PBS did not reproduce this response, as both porX and porY were downregulated under these conditions (Supplementary Fig. 5e). Moreover, deletion of kgp, which impairs peptide acquisition and alters nitrogen availability, increased porX and porY expression (Supplementary Fig. 5f). Together, these findings were more consistent with an adaptive response to intracellular metabolic imbalance linked to glutamate/α-ketoglutarate disruption than with a direct effect of arginine or ammonia.

Fig. 4. Excessive arginine inhibits P. gingivalis growth by disrupting its amino acid metabolism.

a Heatmap of differentially abundant amino acids and downstream metabolites in the Pg and Pg + 0.2%Arg groups (n = 6). Metabolite abundances were normalized and scaled by row as z scores. Red and blue indicate relatively higher and lower abundance, respectively. b Glutamate and arginine metabolic pathways of Pg according to the KEGG database. c Ammonia concentrations of culture supernatants with/without exogenous arginine treatment (n = 3; two-sided unpaired t test). d Intracellular α-ketoglutarate concentrations in Pg with/without exogenous arginine treatment (n = 3; two-sided unpaired t test). e Glutamate dehydrogenase activity in Pg with/without exogenous arginine treatment (n = 3; two-sided unpaired t test). f Effect of α-ketoglutarate supplementation on Pg growth in 0.2% arginine-containing medium (n = 3; two-sided unpaired t test). g Schematic diagram of Pg and Fa coculture in a Transwell system. Created in BioRender. Fangjie, Z. (https://BioRender.com/ncr6fen). h Ammonia concentrations of culture supernatants of Pg and Fa coculture in Transwell plates (n = 3; two-sided unpaired t test). i Intracellular α-ketoglutarate concentration in Pg during Transwell coculture with Fa (n = 3; two-sided unpaired t test). j Glutamate dehydrogenase activity in Pg during Transwell coculture with Fa (n = 3; two-sided unpaired t-test). Unless stated, n denotes biologically independent cultures. Data in c–f, h–j are floating min-to-max plots (center lines, medians; limits, minima and maxima). Only selected pairwise comparisons are shown; exact p values are above comparison bars. Source data are provided as a Source Data file. Pg P. gingivalis; Fa F. alocis; Arg Arginine; AKG α-ketoglutarate.

ArcA-dependent arginine metabolism in F. alocis alleviated arginine-associated stress in P. gingivalis

To further investigate how F. alocis affected the impact of arginine on P. gingivalis, we initially compared the arginine-consuming capacity of P. gingivalis, F. alocis and their coculture in the presence of 10 mM exogenous arginine. P. gingivalis showed very weak arginine consumption, but F. alocis showed strong consumption, while their coculture showed even greater arginine consumption (Fig. 5a). The utilization of arginine by F. alocis predominantly occurs via the following two pathways: converting arginine to citrulline through ArcA (HMPREF0389_01584) or generating agmatine through SpeA (HMPREF0389_00102) (Fig. 5b). In P. gingivalis and F. alocis cocultures, the expression of arcA was significantly upregulated, but there was no obvious change in the expression of speA (Fig. 5c). Then, ΔarcA and ΔspeA were constructed (Supplementary Fig. 6a, b). In comparison with that of the Fa WT, the arginine metabolic capabilities of both ΔarcA and ΔspeA decreased, and the arginine consumption of ΔarcA almost completely decreased (Fig. 5d). These findings indicated that ArcA represented the predominant arginine catabolic pathway in F. alocis under these conditions, whereas SpeA contributed to a minor or auxiliary route (Supplementary Fig. 6c, d). P. gingivalis alone or with F. alocis WT/ΔarcA/ΔspeA was subsequently cultured with different concentrations of arginine. The Fa WT significantly restored P. gingivalis growth, whereas ΔarcA and ΔspeA significantly reduced this capacity, and no such effects were observed in arginine-free medium (Fig. 5e). Compared with the Fa WT, the ΔarcA mutant failed to decrease extracellular NH₃ levels or restore intracellular α-ketoglutarate levels, as it did not normalize P. gingivalis glutamate dehydrogenase activity (Fig. 5f–h), whereas the activity of ΔspeA was similar to that of the Fa WT (Fig. 5f–h), which is consistent with its unchanged expression in the coculture (Fig. 5c). We further used a murine periodontitis model to validate the effects of F. alocis arcA and speA in vivo. The groups were the Pg, Pg + 2%Arg, and Pg + 2%Arg with Fa WT, ΔarcA, and ΔspeA groups. Arginine significantly improved the symptoms of P. gingivalis infection, whereas Fa WT synergistically exacerbated alveolar bone loss, increasing both the bone resorption area and the CEJ-to-alveolar crest distance (Fig. 5i, j). This synergistic effect was abolished in the ΔarcA group, which exhibited less bone loss than the WT coinfection group (WT group: 0.71 ± 0.03 mm²; ΔarcA group: 0.55 ± 0.03 mm²) (Fig. 5i, j). In the ΔspeA-coinfected group, only a slight decrease in bone destruction was detected (0.69 ± 0.09 mm² in the ΔspeA group; p = 0.3941) (Fig. 5i, j). These findings indicated that ArcA-dependent arginine metabolism in F. alocis contributed to depletion of extracellular arginine and alleviation of arginine-associated metabolic stress in P. gingivalis, thereby supporting its growth and pathogenicity under arginine-rich conditions.

Fig. 5. ArcA-mediated arginine consumption by F. alocis relieves P. gingivalis growth suppression, aggravating periodontal disease.

a Arginine consumption in culture supernatants of Pg and Fa monocultures and cocultures (n = 3; one-way ANOVA with Dunnett’s multiple-comparisons test). b Schematic diagram of the Fa arginine metabolic pathway from the KEGG database. c Expression of the arcA and speA genes in Fa under monoculture and Pg coculture conditions (n = 3; multiple two-sided unpaired t-tests). d Arginine consumption in culture supernatants of the Fa WT, ΔarcA, and ΔspeA strains (n = 4; one-way ANOVA with Dunnett’s multiple-comparisons test). e Growth of Pg in monoculture and coculture with Fa WT, ΔarcA, and ΔspeA strains under 0.2% (left), 0.1% (middle), and 0% (right) arginine conditions (n = 4; one-way ANOVA with Tukey’s multiple comparison test). f Ammonia concentrations of the supernatants of cocultures with the Fa WT, ΔarcA, and ΔspeA strains in Transwell plates (n = 3; one-way ANOVA with Dunnett’s multiple-comparisons test). g Intracellular α-ketoglutarate concentration in Pg during Transwell coculture with the Fa WT, ΔarcA, and ΔspeA strains (n = 3; one-way ANOVA with Dunnett’s multiple-comparisons test). h Glutamate dehydrogenase activity in Pg during Transwell coculture with the Fa WT, ΔarcA, and ΔspeA strains (n = 3; one-way ANOVA with Dunnett’s multiple-comparisons test). i Representative 3D-reconstructed micro-CT images of maxillae (buccal view) from (1) Pg, (2) Pg + 2%Arg, (3) Pg + 2%Arg + Fa WT, (4) Pg + 2%Arg + ΔarcA, and (5) Pg + 2%Arg + ΔspeA. j Bone loss area (left) and CEJ-to-alveolar crest distance were mesiodistally (middle and right) measured across experimental groups (n = 5 mice; one-way ANOVA with Tukey’s multiple comparison test). Unless stated, n denotes biologically independent cultures. Data in a, c–e and j are mean ± s.d. (error bars, s.d.). Data in f–h are floating min-to-max plots (center lines, medians; limits, minima and maxima). Only selected pairwise comparisons are shown; exact p values are above comparison bars. Source data are provided as a Source Data file. Pg P. gingivalis; Fa F. alocis; Arg Arginine; WT wild type; ΔarcA arcA deletion mutant; ΔspeA, speA deletion mutant.

To further investigate whether arginine catabolism by F. alocis generated metabolites that promoted P. gingivalis growth, we performed untargeted metabolomic analysis of F. alocis culture supernatants after 24 h growth in the presence or absence of 0.2% arginine. Compared with supernatants from cultures without arginine supplementation, the five most significantly increased metabolites in the arginine-treated group were 1,2-dioleoyl-sn-glycerol, pimelic acid, guanidinosuccinic acid, butyric acid, and dodecanoic acid (Supplementary Fig. 7a–b). However, none of these candidates showed a detectable growth-promoting effect on P. gingivalis in our in vitro validation assays (Supplementary Fig. 7c–g). These findings suggested that the growth-promoting effect of F. alocis was unlikely to be explained by a single dominant secreted metabolite under our experimental conditions. This interpretation was further supported by targeted amino acid metabolomics, which showed that F. alocis preferentially utilized arginine, citrulline, and ornithine in the culture supernatant (Supplementary Fig. 8a–d), consistent with the view that arginine depletion was a major mechanism by which F. alocis supported P. gingivalis growth under high-arginine conditions.

Discussion

Periodontitis is accompanied by pronounced shifts in the subgingival microbiota and oral metabolites, including increased levels of arginine and its derivatives, which can influence host–microbe and microbe–microbe interactions^50–52. Our study revealed that excess arginine disrupted amino acid metabolism in P. gingivalis, reduced bacterial growth, and attenuated pathogenicity. Mechanistically, arginine catabolism was associated with ornithine and NH₃ accumulation, and elevated NH₃ coincided with reduced glutamate dehydrogenase activity and impaired conversion of glutamate to α-ketoglutarate. Amino acid metabolic dysregulation was associated with increased expression of the PorX/PorY-regulated T9SS response, leading to elevated Kgp protease activity and enhanced acquisition of exogenous small peptides to compensate for intracellular amino acid imbalance. In parallel, F. alocis, a periodontal pathobiont frequently detected together with P. gingivalis, preferentially utilized cysteine and arginine and hydrolyzed arginine through the ArcA pathway, thereby relieving arginine-associated stress in P. gingivalis and modulating its growth and pathogenesis. This arginine metabolism-based mutualistic strategy provides a potential ecological basis for their co-enrichment in periodontitis (Fig. 6).

Fig. 6. Schematic diagram of the mechanism underlying the arginine-mediated interaction between P. gingivalis and F. alocis.

Elevated arginine levels are accompanied by NH₃ accumulation, which represses glutamate dehydrogenase activity and α-ketoglutarate production in P. gingivalis, thereby disrupting intracellular amino acid metabolism and inhibiting bacterial growth. This metabolic perturbation is accompanied by increased expression of the PorX/PorY-associated type IX secretion system and enhanced acquisition of extracellular small peptides. In parallel, F. alocis degrades arginine through the ArcA-dependent pathway, partially relieving arginine-associated growth inhibition and promoting the persistence of P. gingivalis under high-arginine conditions. This cooperative arginine metabolism may contribute to the resilience of periodontal microbial communities.

Elevated arginine concentrations have been reported in oral fluids of patients with periodontitis^11–13, and arginine is widely incorporated into oral formulations^14–16. These observations motivated our investigation of the role of arginine in the periodontal microenvironment. The mechanism underlying the effect of arginine on periodontal pathogens remains controversial. Arginine-containing formulations inhibit P. gingivalis growth by targeting both bacterial viability and key virulence mechanisms^15,53. However, arginine may also function as a nitrogen source to support periodontal pathogen growth¹⁹. Our findings offered additional insights: bacterial amino acid/polypeptide utilization is tightly regulated by environmental factors⁵⁴, such as substrate preferences²¹, transport competition^31,55,56, and metabolic feedback⁵⁷. We demonstrated that excess free arginine disrupted P. gingivalis’ intracellular amino acid‒polypeptide equilibrium, leading to a compensatory increase in T9SS-mediated virulence functions that facilitate acquisition of preferred amino acid species. Secretion systems, including types IV, VI, and IX, serve as bacterial weapons for spatial and nutrient competition by facilitating the transport of various toxins. This strategy allows energy-efficient nutrient acquisition in competitive environments^52,58. Our data placed the PorX/PorY–T9SS response in the context of nutrient limitation. Excess arginine disrupted amino acid metabolism in P. gingivalis, particularly along the glutamate/α-ketoglutarate axis, and this metabolic imbalance was associated with increased porX and porY expression, elevated T9SS-associated transcription, and enhanced Kgp activity. Rescue experiments further supported a link between this response and amino acid starvation. Although our genetic data supported the involvement of the PorX/PorY system under these conditions, direct biochemical evidence for arginine-dependent activation of PorY or PorY-mediated phosphorylation of PorX was not obtained in this study. Thus, the PorX/PorY system likely functions as part of a broader adaptive program that enables P. gingivalis to respond to metabolic stress and maintain virulence output under nutritionally challenging conditions.

The concentrations of arginine used in our in vitro experiments (0.0125–0.2%) exceed the micromolar levels reported in gingival crevicular fluid¹¹, but overlap with, or moderately exceed, the millimolar range measured in whole saliva¹³. In addition, commercially available oral formulations can contain 1.5% arginine^14,15, and some products contain up to 8%¹⁶, generating transient local exposures well above endogenous levels. Therefore, our experimental design models a spectrum spanning from physiologically relevant salivary exposure to pharmacological concentrations achievable during topical oral-care interventions. Likewise, the 0.4% and 2% in vivo regimens should be regarded as pharmacological exposures used to probe therapeutic potential rather than as physiological surrogates. Accordingly, the residual bone loss observed in vivo under high-arginine conditions should not be interpreted as evidence that P. gingivalis is unaffected by arginine, but rather as being consistent with partial suppression of the P. gingivalis-dependent pathogenic component in a complex ligature-induced model.

Microbial communities utilize metabolic cross-feeding as an essential ecological strategy to mitigate environmental stress^59–61. Enterococci modulate the intestinal metabolic milieu by regulating ornithine and arginine levels, thereby altering the metabolic programming of Clostridium difficile⁵¹. While the arginine deiminase system (ADS) in oral bacteria (e.g., Streptococcus) has been well characterized in cariogenic contexts⁶², its role in periodontal ecosystems remains less defined. P. gingivalis PPAD primarily catalyzes the citrullination of peptidyl-arginine residues rather than free arginine molecules^63,64. This process is functionally linked to gingipains and represents a distinct arginine-related virulence pathway in P. gingivalis⁶⁵. Although ADS is not unique to F. alocis, this organism is consistently enriched in periodontal lesions and frequently co-detected with P. gingivalis, making it a clinically relevant model partner for examining ADS mediated metabolic interactions in periodontal ecosystems. Our study revealed that elevated free arginine concentrations inhibited the growth of P. gingivalis, whereas F. alocis-mediated arginine metabolism alleviated this inhibitory effect. These findings suggested that ADS-positive bacteria can buffer arginine-induced metabolic stress in P. gingivalis. While other oral commensals possessing ADS may potentially perform similar functions, F. alocis represents a periodontal pathogen-associated partner in which this metabolic interaction is ecologically relevant. Rather than indicating a species-exclusive mutualism, our results supported a context-dependent metabolic cooperation that may contribute to the co-persistence of ADS-positive organisms and P. gingivalis within periodontal pockets.

Several limitations should be considered. The clinical association between P. gingivalis and F. alocis was supported by public metagenomic datasets, but the subgingival datasets lacked sufficiently complete clinical metadata to determine whether their co-occurrence is directly linked to greater periodontal destruction in patients. The interaction between these organisms is also likely to be abundance-dependent. In our model, F. alocis relieved arginine-associated stress in P. gingivalis, but this effect may not be maintained across all community structures, particularly when F. alocis becomes dominant. In addition, the mechanistic analyses were performed mainly in the P. gingivalis W83 background, and strain-level variation in metabolic behavior may limit the generality of these observations. Although untargeted metabolomics and targeted amino acid profiling were performed on F. alocis supernatants, no single metabolite was identified as a dominant growth-promoting factor for P. gingivalis in this system, consistent with the interpretation that the major effect of F. alocis is the removal of excess arginine. Further validation of additional candidate metabolites will be needed. More broadly, the nutritional biology of P. gingivalis remains incompletely resolved, including its relative use of polypeptides versus free amino acids, substrate switching under nutrient fluctuation, and the extent to which nutrient state is coupled to virulence regulation. These issues, together with the contribution of other arginine-metabolizing organisms in polymicrobial communities, warrant further investigation.

In summary, our work elucidated the arginine-mediated symbiotic interactions between P. gingivalis and F. alocis and their effects on periodontitis. Our results highlighted the pivotal role of metabolic programming in regulating pathogen virulence and microbial crosstalk in microbial communities. The critical nodes identified in these metabolic pathways or virulence regulatory networks may serve as potential targets for developing novel therapeutics by reshaping the microbial dysbiosis in periodontitis patients.

Methods

Ethics statement

All animal experiments were conducted in accordance with all relevant ethical regulations and were approved by the Ethics Committee of West China Hospital of Stomatology, Sichuan University (approval no. WCHSIRB-D-2019-183). The human microbiome datasets analyzed in this study were obtained from publicly available, de-identified databases. No new human participants were recruited, and no identifiable private information was accessed; therefore, no additional ethics approval or informed consent was required for this secondary analysis.

Metagenomic data processing, species annotation and bioinformatic analysis

Publicly available subgingival plaque shotgun metagenomic datasets were retrieved from the NCBI Sequence Read Archive (SRA) and BioProject databases through December 2025 using the terms “periodontitis”, “periodontal disease”, “subgingival plaque”, “subgingival microbiome”, “metagenome”, and “shotgun sequencing”. Datasets were included if they met the following criteria: (1) human subgingival plaque samples; (2) explicit classification of subjects as periodontally healthy or periodontitis; (3) shotgun metagenomic sequencing; and (4) accessible raw sequencing data with associated sample metadata. Duplicate or ineligible studies were excluded after manual screening. Sample information is provided in Supplementary Data 1. Five public datasets were included, comprising 102 samples in total, including 44 periodontally healthy controls and 58 patients with chronic periodontitis, from China (PRJNA230363 and PRJNA932553), the United States (PRJNA255922 and PRJNA508385), and Japan (PRJDB11203). Raw reads were quality-filtered using fastp (v0.23.4) with default parameters, and host-derived reads were removed using KneadData (v0.12.0) against the human reference genome GRCh37. Taxonomic profiling was performed using Kraken2 (v2.1.2) against the Standard Library PlusPF-16 database, followed by abundance refinement with Bracken (v2.8). Species-level profiles were further filtered against the Human Oral Microbiome Database (HOMD) to retain oral taxa and standardize species annotation. Principal coordinate analysis (PCoA) was performed based on species-level relative abundance profiles. The relative abundances of the top 20 taxa in each group were visualized using stacked bar plots. LEfSe analysis was performed to identify differentially enriched taxa between groups using the default α value of 0.05 for the Kruskal–Wallis test and a logarithmic Linear Discriminant Analysis (LDA) score threshold of 3.0. PCoA, species relative abundance plots, and LEfSe analysis were performed using the OECloud platform (https://cloud.oebiotech.com/#/home). Microbial co-occurrence was assessed by Spearman’s rank correlation analysis using R v4.4.1 based on species-level relative abundance profiles. Correlations with p < 0.05 were retained for downstream interpretation and visualization.

Microbial strains and growth conditions

P. gingivalis W83 and F. alocis ATCC35896 were sourced from the American Type Culture Collection. All the bacterial strains and culture media used in this study are listed in Supplementary Tables 1 and 2. P. gingivalis W83 and its isogenic mutants were grown in BHIV broth. F. alocis ATCC 35896 and its isogenic mutants were grown in BF broth. At 37 °C, all bacterial strains were grown anaerobically utilizing an atmosphere of 80% N₂, 10% CO₂, and 10% H₂. Single colonies of isogenic mutants were selected from blood agar plates (for P. gingivalis) and BF plates (for F. alocis) supplemented with 5 μg/mL erythromycin.

Coculture assays

For coculture experiments, all bacterial strains were first cultured anaerobically overnight and then subcultured in BB broth containing arginine or ornithine at the indicated concentrations (0–0.8%, as required). P. gingivalis and F. alocis were mixed at a 10:1 ratio to approximate their relative abundance in subgingival plaque, in which the median P. gingivalis: F. alocis ratio was 12 (Supplementary Data 2). Additional coculture assays were performed at 5:1 and 1:1 ratios. The inoculum was standardized to 1 × 10⁶ CFUs/ml P. gingivalis (including its mutants) and 1 × 10⁵ CFUs/ml F. alocis (including its mutants). Growth dynamics were assessed through bacterial quantification of samples harvested at 3-h intervals over 24 h. Additionally, coculture samples were collected after 24 h for bacterial quantification in concentration-gradient experiments, as well as for SEM and FISH. For gene expression analysis, RNA was extracted at 4, 12 and 24 h post-inoculation.

Effects of pH on arginine response and coculture assays

To determine whether environmental pH influences the effect of arginine and the interaction between P. gingivalis and F. alocis, monoculture and coculture experiments were performed at initial pH 6.5, 7.0, 7.5 and 8.0. Overnight cultures were subcultured into medium containing 0.2% arginine under the indicated pH conditions. P. gingivalis and F. alocis were cultured alone or together and incubated anaerobically at 37 °C for 24 h. Bacterial biomass of each species was then quantified by qPCR as described in the section Quantification of bacterial biomass.

Quantification of bacterial biomass

Owing to the lack of selective media suitable for distinguishing P. gingivalis and F. alocis, coupled with their slow growth on solid agar and high susceptibility to environmental interference, quantitative polymerase chain reaction (qPCR) was employed for reliable quantification. The quantified bacterial suspensions were centrifuged and washed twice with PBS to obtain bacterial pellets. The extraction procedure commenced with the incubation of bacterial pellets with 50 mg/mL lysozyme (J&K Chemicals, Beijing, CHN) to lyse cell walls. The genomic DNA was subsequently purified using the TIANamp Bacteria DNA Kit (TIANGEN, Beijing, CHN). Species-specific primers targeting the 16S rRNA gene were used for qPCR quantification, with SYBR Green fluorescence detection. Supplementary Table 3 lists the sequences and details of all primers used in this study. Cycle threshold (Ct) values were converted to bacterial counts on the basis of standard curves generated from serial dilutions of known bacterial concentrations. Standard curves correlating optical density at 600 nm (OD600), Ct values, and CFUs for both species were constructed following the methodology described by Hwang⁶⁶.

Bacterial viability assay

The short-term effect of arginine on P. gingivalis viability was evaluated by CFUs enumeration and live/dead staining. P. gingivalis was cultured anaerobically for 4 h in the presence of 0% or 0.2% exogenous arginine, collected by centrifugation, washed with PBS, and resuspended for subsequent analyses. For CFU enumeration, suspensions adjusted to an OD₆₀₀ of ~0.1 were diluted 10⁴ fold, and 50 μL was plated onto blood agar and incubated anaerobically at 37 °C for 3 d before colony counting. Survival ratio (%) was calculated relative to the untreated control. For live/dead staining, suspensions adjusted to an OD₆₀₀ of ~0.1 were diluted fivefold and stained with the DMAO/PI bacterial viability kit (Beyotime) according to the manufacturer’s instructions. After 30 min, 10 μL of stained suspension was applied to a glass slide and imaged using a fluorescence microscope (DM8, Leica, Germany).

Scanning electron microscopy

The morphological characteristics of the P. gingivalis and F. alocis coculture system were analyzed by SEM after 24 h of incubation. Bacterial cells were pelleted by centrifugation and processed for SEM as follows: aliquots were washed twice with PBS, followed by a single wash with 4% (w/v) sucrose solution. Dehydration of the samples was achieved using a graded ethanol series, with each concentration step lasting 10 min. Processed samples were mounted on carbon-coated conductive tape, subjected to critical point drying, and sputter-coated with a 10 nm gold‒palladium layer under vacuum. SEM images were acquired using an Inspect S50 scanning electron microscope (FEI, Hillsboro, OR, USA) at optimized magnifications.

Treatment of P. gingivalis with spent media

F. alocis was cultured in BB broth supplemented with 0.2% arginine (designated medium A+ ) or without arginine supplementation (designated medium A− ) at 37 °C for 24 h under anaerobic conditions (80% N₂, 10% CO₂ and 10% H₂). Culture supernatants (spent media A+ and A− ) were collected by centrifugation followed by filtration. Spent medium A+ was then serially twofold diluted with fresh medium A+ in 96-well plates to generate final spent-medium proportions of 50%, 25%, 12.5%, 6.25% and 3.125%, after which P. gingivalis was inoculated into each well. In parallel, spent medium A− was serially twofold diluted with fresh medium A− to generate the corresponding dilution series, followed by inoculation of P. gingivalis. After 24 h of incubation, bacterial growth was assessed by measuring optical density in the 96-well plates.

Treatment of P. gingivalis and F. alocis with hit metabolites

P. gingivalis and F. alocis were cultured in BB broth under anaerobic conditions (37 °C, 24 h, 80% N₂, 10% CO₂, and 10% H₂) supplemented with one of the following test metabolites (Macklin, Shanghai, CHN) at specified concentrations: arginine, glutamine, leucine, valine, ornithine, isoleucine, methionine, histidine, lysine, serine, threonine, aspartate, asparagine, citrulline, ammonium chloride, 1,2-dioleoyl-sn-glycerol, pimelic acid, guanidinosuccinic acid, butyric acid, and dodecanoic acid. Growth monitoring was performed by assessing the optical density at 600 nm with a spectrophotometer.

Fluorescence in situ hybridization

To visualize the biofilm architecture and composition, FISH was employed with species-specific probes targeting P. gingivalis and F. alocis (Supplementary Table 4), following a previously described method⁴⁶. The resulting biofilm images were subsequently acquired and processed on a Nikon A1+ confocal laser scanning microscope (Nikon, Tokyo, JP).

pH measurement

Culture pH was measured in P. gingivalis and F. alocis monocultures and cocultures with or without 0.2% arginine using a pH meter (FE28, Mettler Toledo, Switzerland) at 3 h intervals for 24 h.

Untargeted metabolomics

Untargeted metabolomics was performed in two analyses. Intracellular metabolomic profiling included 36 samples from six groups: P. gingivalis, P. gingivalis with 0.2% arginine, F. alocis, F. alocis with 0.2% arginine, P. gingivalis–F. alocis coculture and coculture with 0.2% arginine, with six independent samples per group. Extracellular metabolomic profiling included uninoculated medium and F. alocis culture supernatants prepared in the presence or absence of 0.2% arginine, with four independent samples per group. For intracellular metabolite profiling, P. gingivalis and F. alocis were cultured anaerobically in BHIV medium with or without arginine at 37 °C for 4 h. Cells were harvested by centrifugation, washed with PBS, and homogenized at 30 Hz for 20 s. Metabolites were extracted with 400 μL methanol:water (4:1, v/v) containing internal standards. Samples underwent three freeze–thaw cycles and were centrifuged at 12000 rpm for 10 min at 4 °C. The supernatants were incubated at −20 °C for 30 min, centrifuged again, and transferred for LC–MS analysis. For metabolomic analysis of F. alocis culture supernatants, media collected before and after F. alocis cultivation in the presence or absence of exogenous 0.2% arginine were prepared as described in the section Treatment of P. gingivalis with spent media. 50 μL of medium or culture supernatant was extracted with 150 μL of 20% acetonitrile in methanol containing internal standards, followed by centrifugation and protein precipitation.

Samples were analyzed using a Vanquish UHPLC system coupled to a Q Exactive HF-X mass spectrometer (Thermo Fisher Scientific). Chromatographic separation was performed on a Waters ACQUITY Premier HSS T3 column (1.8 μm, 2.1 × 100 mm) at 40 °C and a flow rate of 0.4 mL/min, using 0.1% formic acid in water and 0.1% formic acid in acetonitrile as mobile phases. Data were acquired separately in positive- and negative-ion electrospray modes using full-scan MS and data-dependent MS/MS acquisition. Detailed chromatographic gradients and MS acquisition parameters are provided in Supplementary Table 5. Pooled quality-control samples, prepared by combining aliquots of all samples, were analyzed after every ten study samples. Raw data were converted to mzML format using ProteoWizard v3.0.2 and processed using XCMS v4.8.0 for peak detection, alignment and retention-time correction. Features with more than 50% missing values were removed, and the remaining missing values were imputed using K-nearest-neighbor or minimum-value-based imputation. Peak areas were corrected using support vector regression. Metabolites were annotated using an in-house spectral library, integrated public databases and metDNA v1.23. Features with an annotation score >0.5 and a coefficient of variation <0.30 in pooled QC samples were retained. Principal coordinates analysis (PCoA) was performed using the Metware Cloud platform (https://cloud.metware.cn/#/tools/tool-list) to assess overall differences in metabolic profiles among groups. Orthogonal partial least squares–discriminant analysis (OPLS-DA) was performed using MetaboAnalystR v1.0.1 after log₂ transformation and scaling, and model robustness was evaluated using 200 permutation tests. Metabolites with a variable importance in projection (VIP) score >1 and a two-sided unpaired t-test p value < 0.05 were considered differentially abundant. After differential metabolite identification, Class I metabolite scatter plots, bubble plots, lollipop plots of the top 20 metabolites ranked by VIP score, and heatmaps were also generated using the Metware Cloud platform (https://cloud.metware.cn/#/tools/tool-list)^67–70. Representative extracted ion chromatograms of key metabolites are shown in Supplementary Fig. 9a, b. Comprehensive lists of all metabolites identified in the two analyses are provided in Supplementary Datas 3 and 4. All source data underlying these analyses and plots are provided in the Source Data file.

Targeted amino acid metabolomics

Targeted analysis included 16 samples divided into four groups: uninoculated medium without 0.2% arginine, uninoculated medium supplemented with 0.2% arginine, F. alocis culture supernatant prepared without 0.2% arginine, and F. alocis culture supernatant prepared with 0.2% arginine, with four independent samples per group. A 50-μL aliquot of each sample was extracted with 250 μL of 20% acetonitrile in methanol, centrifuged at 12000 rpm for 10 min at 4 °C, incubated at −20 °C for 30 min, and centrifuged again before analysis.

A panel of 94 amino acids and related metabolites was analyzed using an ExionLC AD UPLC system coupled to a QTRAP 6500+ mass spectrometer equipped with an electrospray-ionization source (SCIEX). Separation was performed on an ACQUITY BEH Amide column (1.7 μm, 2.1 × 100 mm), and metabolites were detected in positive- and negative-ion multiple-reaction-monitoring modes. Metabolites were identified by comparison with authentic standards and quantified using external standard curves. Detailed chromatographic conditions, source parameters, MRM transitions and calibration information are provided in Supplementary Table 6. Differentially abundant metabolites were identified on the basis of fold change, with metabolites showing a fold change of ≥2 or ≤0.5 considered increased or decreased, respectively. Heatmaps were generated using the Metware Cloud platform (https://cloud.metware.cn/#/tools/tool-list)^71,72. Representative MS/MS spectra of key metabolites are shown in Supplementary Fig. 9c. A comprehensive list of all metabolites identified in the analyze is provided in Supplementary Data 5. All source data underlying these analyses and plots are provided in the Source Data file.

Hemagglutination assay

Hemagglutination was evaluated following an established methodology⁷³. P. gingivalis W83 strains and derived mutants treated with arginine or cocultured with F. alocis were pelleted through centrifugation at 6000 × g for 10 min. The bacterial pellets were washed twice with PBS to remove residual medium and arginine, and then resuspended in PBS to an OD₆₀₀ of 2.0. Parallel preparations involved washing sheep red blood cells twice with PBS and reconstituting them to 1% (v/v) in PBS. In round-bottomed 96-well plates, 100 μL bacterial samples were subjected to serial twofold PBS dilutions. These were combined with equal volumes of 1% erythrocyte suspension and maintained at ambient temperature for 3 h. Visual inspection was used to determine the endpoint titer, which was recorded as the maximal dilution demonstrating unambiguous hemagglutination.

Detection of Kgp/Rgp gingipain activity

The proteolytic activities of Rgp and Kgp were measured using BAPNA (4 mM) and ALNA (4 mM) as substrates in an activation buffer containing 0.1 M Tris, 5 mM CaCl₂, 9 mM L-cysteine, and 0.2 M NaCl. These assays were performed to examine the protease activities of P. gingivalis WT, ΔporX, and ΔporY strains following arginine treatment or coculture with F. alocis. Bacterial cultures (OD₆₀₀ = 1.5) were centrifuged at 6000 × g for 10 min. The bacterial pellets were subjected to three washes with PBS and subsequently adjusted to OD₆₀₀ = 1.0, after which a reaction mixture was prepared by combining 16 μL of either the suspension or supernatant with 184 μL of activation buffer. Subsequently, 100 μL of substrate solution was added at a 1:1 ratio. The reaction times varied by protease (Rgp: 5, 10, and 15 min; Kgp: 30, 60, and 90 min), and the measurement of absorbance at 405 nm was performed with a Bio-Rad microplate reader (CA, USA). Gingipain activity was calculated as the initial reaction velocity (V₀), expressed as mOD/min, from the linear portion of the absorbance–time curve.

RNA isolation and RT‒qPCR

Total RNA was extracted from F. alocis/P. gingivalis cultures grown with/without arginine or from P. gingivalis-F. alocis cocultures using TRIzol Reagent (Invitrogen, CA, USA) in accordance with the manufacturer’s protocols. Complementary DNA (cDNA) was synthesized following the manufacturer’s protocol, employing the HiScript III RT SuperMix for qPCR from Vazyme (Nanjing, China). The upstream and downstream primers, template cDNA, and 2 × Taq Pro Universal SYBR qPCR Master Mix (Vazyme, Nanjing, CHN) were subsequently blended as per the requirements stipulated in the instructions, and the real-time PCR procedure was executed on a LightCycler 480 II (Roche, Basel, CH). Using the 16S rRNA genes of F. alocis and P. gingivalis as internal references, the expression of genes related to arginine metabolism in F. alocis (arcA, speA, amid, carb and aspa), proteases (kgp), and the T9SS (porX, porY, sigP, porL, porU, porK, porP, porN, porM) of P. gingivalis was evaluated. Target gene sequences were retrieved from the National Center for Biotechnology Information (NCBI) database. Primers were designed using NCBI Primer-BLAST, and their specificity was assessed by BLAST analysis. All primers were synthesized by Beijing Tsingke Biotech Co., Ltd. (Beijing, CHN). The sequences and details of all primers used in this study are provided in Supplementary Table 3.

Isogenic mutant construction

Isogenic mutants of P. gingivalis and F. alocis were generated using the long PCR-based fusion method as previously reported^39,74. The upstream and downstream flanking regions (~800 bp each) of the target genes were amplified by PCR using chromosomal DNA from P. gingivalis and F. alocis as the template. The primers used to construct the mutations are listed in Supplementary Table 3. To generate different ermF cassette variants, PCR was carried out using the Pg-ermF-F/Pg-ermF-R or Fa-ermF-F/Fa-ermF-R primer pairs to amplify the cassette with or without its native promoter and terminator. Fuse the upstream, ermF and downstream fragments. The purified fusion fragments were subsequently electroporated into P. gingivalis- and F. alocis-competent cells. Transformants were initially selected based on their ability to grow in the presence of erythromycin (5 μg/mL). Positive clones were then confirmed by sequencing.

Assays for ammonia, α-ketoglutarate, and glutamate dehydrogenase activity

Following treatment with arginine or coculture with F. alocis, ΔarcA, or ΔspeA in a Transwell system, P. gingivalis cells were collected by centrifugation, washed, and resuspended. A portion of the supernatant was retained for ammonia quantification. Bacterial suspensions from all experimental groups were normalized to an OD₆₀₀ of 1 before cell lysis. For α-ketoglutarate and glutamate dehydrogenase (GDH) activity assays, lysis was performed using kit-recommended solvents. Ammonia, α-ketoglutarate, and GDH activity were subsequently quantified with the following commercial kits according to the manufacturer’s protocols: an Ammonia Assay Kit (Sigma‒Aldrich, MO, USA) for ammonia; an Amplex Red α-Ketoglutarate Assay Kit (Beyotime, Shanghai, CHN) for α-ketoglutarate; and a Glutamate Dehydrogenase Activity Assay Kit (Elabscience, Wuhan, CHN) for GDH activity.

Assessment of arginine metabolic capacity

P. gingivalis and F. alocis were cultured individually or in coculture for 24 h in BB broth supplemented with 10 mM exogenous L-arginine. After centrifugation, the supernatant was collected and filtered. The arginine concentration in the supernatant was measured using an arginine detection kit (Solarbio, Beijing, CHN) according to the manufacturer’s instructions. The baseline arginine level was determined using blank medium, and the Δarginine concentration was calculated by subtracting the arginine concentration of the experimental groups from this baseline. The arginine metabolic capacity of F. alocis WT and its mutant strains was assessed using the same methodology.

Mouse experiments

The animal experiments conducted in this research were approved by the Ethics Committee of West China Hospital of Stomatology, Sichuan University (approval no. WCHSIRB-D-2019-183). A ligature-induced periodontitis (LIP) mouse model was used. The inoculated exogenous microorganisms, including P. gingivalis and F. alocis as well as their mutant strains, served as the experimental variables. Six-week-old male C57BL/6 mice were obtained and maintained under specific pathogen-free (SPF) conditions at the Experimental Animal Center of West China School of Basic Medical Sciences. Only male mice were included, consistent with previously published ligature-induced periodontitis models using male C57BL/6 mice⁷⁵. The study focused on the effects of bacterial inoculation and arginine treatment and was not designed or statistically powered to assess sex-dependent effects. Therefore, sex was not included as an experimental variable and no sex-stratified analyses were performed. The housing environment was maintained at an ambient temperature of 22 ± 2 °C and a relative humidity of 50–60%, with a 12 h light/12 h dark cycle. Mice had ad libitum access to autoclaved standard chow and water. All mice were sourced from Dossy Experimental Animals Co., Ltd. (Chengdu, CHN). After one week of environmental adaptation, the LIP model was constructed. Before all the experimental manipulations were performed, the mice were anesthetized using isoflurane. 6–0 ligatures (Gold Medical Co., Ltd., Shanghai, CHN) were tied around the cervical region of the maxillary second molar. Inoculation with the infectious bacterial suspension was initiated one day after the operation. The strains that were cultivated to the logarithmic growth phase were prepared as single or mixed infection bacterial suspensions in 2% carboxymethyl cellulose (CMC) so that the final bacterial concentrations were 1 × 10⁹ CFUs/mL P. gingivalis and 1 × 10⁸ CFUs/mL F. alocis, respectively. During the inoculation procedure, a sterile cotton swab was used to collect an equal volume of the inoculum suspension (50 μL per side of the ligature thread), which was gently applied around the ligature. Each ligature site was treated three times per session, with inoculations performed at two-day intervals. Arginine was administered via drinking water. After 4 weeks, the mice were euthanized by gradual displacement with CO₂ (20–30% chamber volume/min). The maxillae were dissected and fixed in 4% paraformaldehyde for 48 h. The samples were examined using a micro-CT scanner (SCANCO Medical AG, Bassersdorf, CH) at a voltage of 70 kVp and a current of 200 μA. Three-dimensional reconstruction analysis was performed on the scanned data. The distance from the buccal mesiodistal alveolar crest of the second molar to the cementoenamel junction (CEJ) was measured on the reconstructed three-dimensional image using Photoshop (Adobe, CA, USA). The area of alveolar bone resorption was also quantified on the reconstructed three-dimensional images. The upper limit was defined by the line connecting the cementoenamel junctions of the three molars, the lower limit was the superior margin of the alveolar bone, and the mesial aspect of the third molar and the distal aspect of the first molar served as the left and right boundaries, respectively, to delineate the range.

Statistics and reproducibility

Sample sizes were selected based on those commonly used in previous comparable microbiological and murine periodontitis studies. No statistical method was used to predetermine sample size. For the metagenomic analysis, the sample size was determined by publicly available datasets that met the predefined inclusion and exclusion criteria. Exact sample sizes and the definitions of biological replicates are provided in the corresponding Methods sections and figure legends. No data were excluded from the analyses. Mice were randomly allocated to the experimental groups before treatment. Randomization was not applicable to the in vitro experiments. The investigators were not blinded to allocation during experiments and outcome assessment. Data are presented as mean ± s.d. from at least three biologically independent replicates. Statistical analyses were performed using GraphPad Prism 9. Comparisons between two groups were performed using two-sided t tests, whereas comparisons among multiple groups were performed using one-way ANOVA followed by Tukey’s or Dunnett’s multiple-comparisons test, as appropriate. Multiple two-sided unpaired t tests were performed with false discovery rate (FDR) correction. Exact p values and full details of the statistical tests used are provided in the corresponding figure legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

F.Z. and Y.W. contributed equally in the design and conduct of experiments and wrote the paper. F.Z., Y.W., W.N. and X.W. conducted experiments. Y.L. and X.P. provided the methodology. B.R. and J.L. designed, conducted and analyzed experiments. All authors edited the manuscript.

Peer review

Peer review information

Nature Communications thanks Jan Potempa, Yasuo Takeuchi and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This research was supported by the National Natural Science Foundation of China (81991501, 82170949, 32470205 and 82401116), Sichuan Province Science and Technology Support Program (2023YFS0048), Sichuan Medical Association Research Project (S20250069), Natural Science Foundation of Sichuan Province (2024NSFSC0546, 2026NSFSC1046), Fund of the State Key Laboratory of Oral Diseases (SKLOD-2025RD008).

Data availability

The subgingival plaque metagenomic sequencing data used in this study are publicly available from the Sequence Read Archive (SRA) under BioProject accessions: PRJNA230363, PRJNA932553, PRJNA255922, PRJNA508385, PRJDB11203. The raw metabolomics data generated in this study have been deposited in the MetaboLights repository under accession codes: MTBLS14827, MTBLS14833, MTBLS14834. Source data are provided with this paper.

Code availability

No custom code was developed or used to analyse data in this manuscript.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77383-6.