Rhythmic-phasic c-di-AMP signaling: a new paradigm for bacterial growth synchronization governed by a pseudo-phosphodiesterase and modifiable diadenylate cyclase

Abstract

Background

Cyclic di-adenosine monophosphate (c-di-AMP) is an essential bacterial second messenger that regulates bacterial physiology, survival, and pathogenesis. While current models describe c-di-AMP regulation as a linear synthesis-degradation paradigm through opposing activities of diadenylate cyclases (DACs) and c-di-AMP-specific phosphodiesterases (PDEs), they fail to capture the temporal complexity of c-di-AMP signaling dynamics in coordination with discrete growth-phase transitions, metabolic state, and redox homeostasis. In Porphyromonas gingivalis, a key Gram-negative proteolytic oral pathobiont relevant to oral and systemic disease, measurable canonical c-di-AMP degradation is absent, suggesting that c-di-AMP signaling operates through an atypical regulatory architecture mediated by the atypical phosphodiesterase PDE_Pg. Here, we identify rhythmic-phasic c-di-AMP signaling scheme in which PDE_Pg plays a critical role by coordinating with bacterial growth phase, metabolism, and cell cycle progression.

Methods

We employed an integrated experimental and structural modeling framework that combined bacterial genetics, growth-phase-resolved nucleotide profiling, whole membrane fraction biochemical assays, untargeted metabolomics, transcriptomics, proteomics, post-translational modification profiling, ultrastructural transmission electron microscopy, structural modeling, and molecular dynamics simulations to dissect the pseudoenzymatic-controlled c-di-AMP regulatory module in P. gingivalis.

Results

This study demonstrates that growth phase synchronization in P. gingivalis requires rhythmic-phasic c-di-AMP signaling, characterized by coordinated depletion-replenishment episodes of intracellular c-di-AMP during growth-phase transitions. This regulatory network is orchestrated by two key components: the atypical HD-domain PDE homolog, PDE_Pg, and the conserved serine residue (Ser-226) within the DAC_Pg SEE loop. PDE_Pg maintains c-di-AMP homeostasis through a noncanonical regulatory mechanism rather than through measurable PDE-mediated hydrolysis. Disruption of PDE_Pg profoundly remodels bacterial physiology, resulting in altered c-di-AMP dynamics, impaired peptide-dependent metabolic adaptation, disrupted ATP homeostasis, and widespread post-translational remodeling, including condition-specific phosphorylation of the c-di-AMP cyclase DAC_Pg at the conserved Ser226 residue. Functional genetic analyses demonstrate that Ser226 is required for proper coordination of rhythmic-phasic c-di-AMP signaling and metabolic adaptation, while pantothenate-mediated rescue further supports the contribution of metabolic homeostasis to these processes. Along with structural modeling and molecular dynamics simulations, these findings support a model in which noncanonical regulatory protein (pseudoenzymatic function) PDE_Pg coordinates rhythmic-phasic c-di-AMP signaling and physiological adaptation in P. gingivalis.

Conclusion

Together, our findings establish a new paradigm in bacterial second messenger signaling, showing that pseudoenzymatic scaffolds coordinate rhythmic c-di-AMP dynamics that synchronize nutrient acquisition, redox metabolism, and cell-cycle progression, with broad implications for microbial physiology and pathogenesis.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12964-026-03108-3.

Background

Cyclic di-adenosine monophosphate (c-di-AMP) has emerged as a pivotal bacterial second messenger with broad regulatory functions across diverse pathogens. Initially characterized in Firmicutes, c-di-AMP is now recognized as an essential regulator that coordinates a wide range of cellular processes, including cell wall homeostasis, central metabolism, DNA integrity, osmotic adaptation, antibiotic resistance, virulence, and pathogenesis [1–11]. Its pleiotropic roles make c-di-AMP unique among nucleotide messengers, and both its deficiency and excess are detrimental to bacterial growth and survival. Tight regulation of intracellular c-di-AMP concentration is therefore a universal requirement for bacterial fitness, survival, and pathogenesis.

Biochemically, c-di-AMP is synthesized by diadenylate cyclases (DACs) from ATP and typically degraded by c-di-AMP-specific phosphodiesterases (PDEs), producing 5′-pApA and eventually AMP. In well-studied Gram-positive model organisms such as Listeria monocytogenes, Bacillus subtilis, and Staphylococcus aureus, a balanced interplay between DAC and PDE activities ensures proper c-di-AMP turnover and signaling [2, 12]. Disruption of this balance leads to defects in DNA replication, envelope stability, osmoadaptation, and host-pathogen interactions, underscoring the centrality of c-di-AMP homeostasis to bacterial physiology and pathogenesis [13, 14]. Although the mechanisms of c-di-AMP synthesis and degradation are well characterized in Gram-positive species, the regulation of c-di-AMP in Gram-negative pathogens remains far less explored, even as growing evidence of its importance in numerous physiological and adaptive processes. Our previous work addressed this gap in Porphyromonas gingivalis (Bacteroidota), a keystone anaerobic proteolytic model pathobiont implicated in chronic periodontitis and multiple systemic diseases [6, 10]. As a strictly proteolytic anaerobe, P. gingivalis relies heavily on extracellular protein degradation and peptide import to fuel metabolism and maintain redox balance processes that must be tightly coordinated with cell-envelope remodeling and bioenergetic/redox-coupling mechanisms, likely necessitating dynamic and temporal regulation of c-di-AMP to synchronize its growth and survival. However, the classical linear synthesis-degradation paradigm does not fully explain how c-di-AMP dynamics are temporally coordinated with discrete growth-phase transitions, nutrient acquisition, as well as redox-linked metabolic states in this pathobiont.

We identified and characterized the core components of the c-di-AMP regulatory system in P. gingivalis, including PGN_0523 (DAC_Pg, the diadenylate cyclase), PGN_0521 (PDE_Pg, an atypical phosphodiesterase), and PGN_1486 (a CdaR-like regulator) [10]. These studies established that c-di-AMP is indispensable for P. gingivalis growth and influences lipopolysaccharide composition, cell envelope integrity, and virulence. Notably, we showed that PDE_Pg does not exhibit measurable hydrolytic activity toward c-di-AMP despite its sequence similarity to canonical PgpH/HD-type PDEs. Instead, PDE_Pg is required to maintain optimal intracellular c-di-AMP levels and appears to positively regulate both c-di-AMP synthesis and ATP turnover rather than mediating c-di-AMP degradation [10]. Accordingly, deletion of pde_Pg resulted in abnormally elevated ATP accumulation, reduced c-di-AMP levels, and a pronounced delay in the lag-to-log transition, demonstrating that P. gingivalis employs a fundamentally different strategy to regulate this signaling system [10]. Complementation of the Δpde_Pg mutant restored intracellular c-di-AMP levels and biofilm growth to wild-type levels, demonstrating the essential role of PDE_Pg in maintaining c-di-AMP homeostasis and growth [6, 10]. Together, these findings reveal that PDE_Pg function and c-di-AMP signaling in P. gingivalis diverge substantially from canonical models and are governed by distinct and synthesis-centered regulatory architecture.

Building on these findings, the present study investigates how the atypical PDE_Pg regulates c-di-AMP signaling and growth-phase progression. We first employed comprehensive metabolic, transcriptomic, and proteomic analyses to generate a global view of cellular alterations in the Δpde_Pg mutant compared with the wild type (WT). This was followed by targeted approaches, including site-specific mutagenesis, structural modeling, molecular docking, and dynamics simulations, along with complementary biochemical and microbiological assays, to dissect the mechanistic basis of PDE_Pg function. These analyses demonstrate that P. gingivalis undergoes recurrent phasic cycles of c-di-AMP depletion and replenishment during growth-phase transitions, a process that strictly depends on the presence of PDE_Pg. The Δpde_Pg mutant fails to exhibit this rhythmic-phasic c-di-AMP signaling and instead displays extensive rewiring of nutrient acquisition pathways, impaired peptide uptake, and altered redox-coupling mechanisms. We further discovered that Ser-226 within DAC_Pg is post-translationally modifiable and functions as an additional regulatory hotspot controlling c-di-AMP dynamics in this anaerobe.

Collectively, these findings establish a new paradigm in bacterial cyclic dinucleotide signaling, nutrient acquisition, redox control, and growth-phase transitions. By delineating the molecular interplay between the atypical PDE_Pg and DAC_Pg in P. gingivalis, our study uncovers, for the first time, a synthesis-centered, temporal signaling regulatory mechanism orchestrated by the PDE_Pg as a pseudoenzymatic temporal coordinator, coordinating cell-cycle progression and revealing potential therapeutic vulnerabilities in this pathobiont, and potentially other Bacteroidota.

Methods

Bacterial strains, culture conditions, and chemicals

P. gingivalis wild-type strain 381 and the mutant 381 Δpde_Pg (ΔPGN_052) were used in this study. Construction of the Δpde_Pg mutant and restoration of intracellular c-di-AMP levels and biofilm growth were performed in our previous study [6, 10]. For each independent experiment, both the WT and Δpde_Pg strains were freshly revived from the genetically validated frozen stocks and cultured under identical experimental conditions with minimal in vitro passaging prior to analysis. Frozen stocks of both WT and mutant strains were revived on BAPHK agar plates consisting of BD™ Bacto™ Tryptic Soy Broth without dextrose (TSB) supplemented with 5 µg/mL hemin (H), 1 µg/mL menadione (Vitamin K), and 5% defibrinated sheep blood (Northeast Laboratory Services, USA). Plates were incubated anaerobically for 3–4 days at 37 °C in a Coy Lab Products chamber (5% H₂, 10% CO₂, 85% N₂) until colonies reached a brownish pigmentation, which was then used as an inoculum. All subsequent procedures were conducted inside the anaerobic chamber. For liquid cultures and growth rate assay, a half-loop of colony was inoculated into 10 mL of pre-reduced TSBYHK (TSB supplemented with 0.5% yeast extract, 1 µg/mL H, and 1 µg/mL K). Cultures were incubated statically overnight at 37 °C, and turbidity was confirmed the following day. Cells were harvested by transferring 1.8 mL aliquots into four 2 mL microcentrifuge tubes, centrifuged at ~ 6000 rpm for 1 min, and washed once with assay medium (BHIYHK: Brain Heart Infusion supplemented with yeast extract, H, and K). The washed pellets were combined and resuspended in 1 mL of fresh BHIYHK to generate a concentrated cell stock. For growth assays, ~ 32 mL of cell suspension at OD₆₀₀ ≈ 0.2 was prepared by diluting 400 µL of the concentrated cell stock into 35 mL of assay medium in a 50 mL Falcon tube. Optical density at 600 nm (OD₆₀₀) readings were taken directly in test tubes to ensure accuracy, with cuvette-based measurements used to confirm consistency. Final inocula were distributed into test tubes containing 4 mL assay medium per condition. Three replicates were used for each condition. Cultures were then incubated at 37 °C under anaerobic conditions, and OD₆₀₀ readings and sample collection were performed at designated time points over 51 h. Solvents used for the analysis of nucleotide second messengers were HPLC/MS grade.

Quantification of c-di-AMP by enzyme-linked immunoassay (ELISA)

Preparation of the metabolome was performed as described previously [10]. Cell pellets were washed with PBS, resuspended in 300 µl of ice-cold extraction solvent containing acetonitrile/methanol/water (2:2:1, v/v/v), and incubated on ice for 15 min. Suspensions were then heated at 95 °C for 10 min and cooled again on ice. Following centrifugation at 20,800 × g at 4 °C for 10 min, the supernatants were collected, and the pellets were extracted twice with the solvent at room temperature without heating. Combined supernatants from three extraction steps were stored at -20 °C overnight to facilitate protein precipitation, then centrifuged to separate the protein precipitate. All residual pellets were collected for protein quantification using the Qubit Protein BR Assay Kit (Cat# A50668) and Qubit 4 Fluorometer (ThermoFisher Scientific), and supernatants were dried in a Speed-Vac (ThermoFisher Scientific) at 40 °C. Dried samples were dissolved in 50 µl sample solvent and vortexed for 30 s. Intracellular c-di-AMP levels were quantified using a competitive c-di-AMP ELISA Kit (Cayman Chemical, Ann Arbor, MI, USA). Each biological sample was assayed at a minimum of two dilutions, with each dilution measured in technical triplicate. For each well, 50 µL of sample was added, followed by 50 µL of tracer solution. After 15 min, 50 µL of the anti-c-di-AMP antibody was added. The plate was sealed and incubated for 2 h at room temperature on an orbital shaker. Wells were washed five times with the supplied Wash Buffer, and 175 µL of TMB (3,3′,5,5′-tetramethylbenzidine) substrate solution (Sigma-Aldrich, St. Louis, MO, USA) was added for color development. The reaction was stopped by adding 75 µL of HRP Stop Solution, and absorbance was measured at 450 nm on a microplate reader. Standard curves were generated by plotting %Bound/Maximum Bound (%B/B₀) against serial dilutions of c-di-AMP and fitted using a four-parameter logistic regression. Sample concentrations were interpolated from the standard curve based on their calculated %B/B₀ values.

LC-MS quantification of ATP and pApA

Intracellular ATP and pApA levels were quantified using LC-MS. Metabolites were extracted as previously described [10] with ice-cold acetonitrile/methanol/water (2:2:1, v/v/v). Dried extracts were reconstituted in LC-MS-grade water prior to analysis. Measurements were performed on a Dionex Ultimate 3000 UPLC system coupled to an ISQ EM single quadrupole mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). ATP and pApA separation were achieved by reverse-phase chromatography on an EC 50/3 Nucleodur Pyramid C18 column (Macherey-Nagel, Germany), maintained at 30 °C, with a 20 µL injection volume. Chromatographic conditions followed those described by Burhenne et al. [15] with minor modifications: eluent A, 10 mM ammonium acetate with 0.1% formic acid; eluent B, methanol. Gradient elution was applied as follows: 0.0–5.0 min, 0% B; 5.0–6.3 min, 0–10% B; 6.3–8.3 min, 10% B; 8.3–11.0 min, 10–30% B; 11.0–11.1 min, 30 − 0% B; and 11.1–15.0 min, 0% B, at a flow rate of 0.4 mL/min. Mass spectrometric detection was performed in positive heated electrospray ionization (HESI) mode with selected ion monitoring (SIM) for ATP (m/z 508.00) and pApA (m/z 677.1). HESI parameters were as follows: vaporizer temperature, 227 °C; ion transfer tube temperature, 300 °C; spray voltage, + 3000 V/−2000 V; sheath gas, 42.9 psig; auxiliary gas, 4.8 psig; and sweep gas, 0.5 psig. ATP and pApA standards (Sigma-Aldrich, Darmstadt, Germany) were used to generate a seven-point calibration curve (0.41–100 nM; pM). Data acquisition and quantification were performed with Chromeleon Console software, and ATP and pApA levels were normalized to total protein content.

Whole membrane fraction biochemical assays for PDE_Pg functional characterization

Preparation of whole membrane fractions for enzymatic assays was performed using a previously described protocol with minor modifications [16]. Briefly, 200 mL cultures of WT and Δpde_Pg strains were grown under the conditions described above to mid-exponential phase (OD₆₀₀ ≈ 0.7) and harvested for preparation of whole membrane fractions. Cell pellets were washed three times in cold PBS buffer and resuspended in ice-cold buffer containing 50 mM HEPES (pH 7.5), 150 mM NaCl, 10% glycerol, 1 mM MgCl₂, and EDTA-free protease inhibitor cocktail, followed by cell disruption using a cell disruptor. Cellular debris were removed by centrifugation at 10,000 × g for 15 min at 4 °C, and the clarified supernatant was subjected to ultracentrifugation at 150,000 × g for 90 min at 4 °C. Membrane pellets were washed once with the same buffer, re-pelleted by ultracentrifugation as described above, and gently resuspended in the assay buffer containing 0.05% n-Dodecyl-β-D-maltoside (DDM, ThermoFisher Scientific) for mild solubilization overnight at 4 °C. Membrane protein concentrations were determined using Qubit™ Protein Assay Kit (Thermo Scientific) prior to biochemical assays. For ATP-dependent c-di-AMP synthesis assays, equal amounts of membrane protein (20 µg) were incubated with 1 mM ATP in reaction buffer containing 1 mM MnCl₂ or MgCl₂ or FeSO₄ at 37 °C in a final reaction volume of 100 µL for 30 and 60 min [17, 18]. For exogenous c-di-AMP hydrolysis assays, identical membrane preparations were incubated with 50 µM c-di-AMP under the same reaction conditions. Reactions were terminated by addition of ice-cold acetonitrile to achieve approximately 70% final solvent concentration, followed by overnight incubation at -20 °C for protein precipitation. Samples were centrifuged to remove precipitated proteins, and clarified supernatants were analyzed by LC-MS for detection and quantification of ATP, c-di-AMP, and pApA. Metabolite analysis was performed using an Agilent 1290 Infinity II UHPLC system coupled to a 6546 LC/Q-TOF mass spectrometer (Agilent Technologies) equipped with an EC 50/3 Nucleodur Pyramid C18 column (Macherey-Nagel, Germany), using described chromatographic conditions as above with minor modifications to the mass spectrometric parameters for improved sensitivity.

Untargeted metabolomic profiling using LC-MS/MS

Bacterial cells cultured anaerobically overnight at 37 °C in TSBHK medium, as described above, were harvested by transferring aliquots into microcentrifuge tubes and centrifuging them at ~ 6700 rpm for 1 min. The resulting pellets were washed once with BHIY medium, combined, and resuspended in 1 mL of fresh BHIYHK medium to generate a concentrated cell stock. Afterwards, cells were inoculated into BHIYHK medium at an initial OD₆₀₀ of ~ 0.15 and grown anaerobically for 1 h to ensure that the culture reached a metabolically active and physiologically balanced state prior to metabolite extraction. Cultures were then taken out from the anaerobic chamber, cooled on ice for 15 min, and collected by centrifugation at ~ 6700 rpm for 20 min. Metabolome was extracted using a previously established protocol [10] with ice cold acetonitrile/methanol/water (2/2/1, v/v/v) as extraction solvent. The metabolites obtained were reconstituted in LC-MS-grade sample solvent prior to injection and analysis, and protein concentrations were determined using Qubit™ Protein Assay Kit (Thermo Scientific) for normalization. For untargeted metabolomics, LC-MS/MS analysis was performed on a Vanquish UPLC system coupled with an Orbitrap Exploris 120 mass spectrometer (ThermoFisher Scientific, Waltham, USA) with HESI that operated at both positive and negative ion modes. LC separation of intracellular metabolites was achieved on XBridge™ Premier BEH Amide column (2.1 × 150 mm, 2.5 μm particle size, Waters Corporation, Massachusetts, USA) using a 25 min gradient with a binary mobile phase system composed of solvent A (95:5 water: acetonitrile containing 20 mM ammonium hydroxide and 20 mM ammonium acetate, pH 9.4) and solvent B (acetonitrile) [19]. The flow rate was 150 µl/min, column temperature was maintained at 25 °C and 5 µl of each sample was injected separately for the analysis. Mass spectrometric acquisition was performed in full scan and data-dependent MS/MS (DDA, Top N, ddMS2) modes under the control of Xcalibur software (ThermoFisher Scientific). Full-scan (MS1) spectra were acquired at a resolution of 120,000 K at m/z 200, ensuring high mass accuracy (within a 5 ppm window) for precursor ions. Data-dependent MS2 spectra were acquired at a resolution of 60,000 at m/z 200 to provide sufficient detail for fragment assignment during product ion detection and identification. Raw data files generated on Orbitrap Exploris 120 system were converted into the mzML standard format using ProteoWizard (https://proteowizard.sourceforge.io/). Feature detection and chromatographic deconvolution were performed using open-source software MzMine 4.7.29 (https://github.com/mzmine), which provides modular workflows for peak detection, deisotoping, alignment, and feature grouping. Feature annotation was conducted using MSnLib (ms2-pos-latest, Zenodo:11163380) and further validated by comparison with GNPS spectral libraries to enhance confidence in metabolite identification. For statistical and functional interpretation, the processed data were analyzed using MetaboAnalyst 6.0 software (https://www.metaboanalyst.ca/). On this platform, univariate analyses, including fold-change analysis, t-tests, volcano plots, and heat maps, were used to identify significantly altered metabolites, while PCA and PLS-DA were employed to assess global metabolic differences across experimental conditions.

RNA extraction and RNA-seq analysis

Cells were grown in BHIYHK overnight as explained above. They were pelleted inside anaerobic chamber using a microfuge (6000 RPM) for 2 min. RNA extraction was performed using Direct-zol™ RNA MiniPrep Kit (Zymo Research) as we performed previously [20]. Cell pellets were immediately resuspended in 600 µL TRI reagent^® (Zymo Research) inside the anaerobic chamber. After homogenization, samples were removed from the chamber, incubated at room temperature for 5 min, and clarified by centrifugation (14,000 × g, 5–10 min). The supernatant was mixed with 100% ethanol and applied to RNA spin columns. On-column DNase I digestion (two sequential treatments at 30 °C for 1 h and 30 min) was performed to remove contaminating DNA, followed by multiple washes (using RNA wash and prewash buffers). RNA was eluted in 15 µL of nuclease-free water, quantified by NanoDrop, and stored at -80 °C until use. Other preparation and sequencing steps were performed at University of Louisville Sequencing Technology Center. Typical RNA yields ranged from 500 to 4000 ng/µL. All samples exhibited RNA Integrity Numbers (RIN) of 8.5-9.0. Library preparation was performed using the Illumina^® Stranded Total RNA Prep, Ligation with Ribo-Zero Plus (Illumina, 20040525), combined with the Ribo-Zero Plus Microbiome depletion kit (Illumina, 20072062) for rRNA removal. Unique dual indexes were added using IDT^® for Illumina^® RNA UD Indexes Set A (Illumina, 20040553). The procedure followed the Illumina Stranded Total RNA Prep, Ligation with Ribo-Zero Plus, Ribo-Zero Plus Microbiome Reference Guide (Document #1000000124514 v03). RNA purity was first assessed using the Nanodrop 2000 Spectrophotometer (Thermo Scientific). Quantitative determination was performed with the Qubit RNA HS Assay Kit (ThermoFisher, Q32855). RNA integrity was evaluated using the Agilent 2100 Bioanalyzer with the RNA Nano 6000 kit (Agilent, 5067 − 1511), using the Prokaryote Total RNA Nano assay module. For each sample, total RNA was normalized to 300 ng in nuclease-free water prior to library preparation. Libraries were generated following the standard Illumina protocol with one modification: an additional 1× AMPure bead cleanup was performed at the end of library construction to remove trace adapter dimers and improve library purity. Final libraries were quantified using the Qubit 1X dsDNA HS Assay Kit (ThermoFisher, Q33231). Fragment size distributions and average library insert size were assessed with the Agilent 5200 Fragment Analyzer using the Standard Sensitivity NGS Fragment Analysis Kit (Agilent, DNF-473-0500). Libraries were normalized to 2 nM in RSB+Tween buffer, re-quantified, and pooled at equimolar concentrations. The final pooled library was re-verified by Qubit and Fragment Analyzer prior to sequencing. The pooled 2 nM library was diluted to 750 pM, spiked with ~ 5% PhiX, and loaded onto a NextSeq 2000 instrument following the NextSeq 2000 System Guide (Illumina Document #1000000109376 v06). Two sequencing runs were performed. Run 1: P2-100 Cycle Cartridge (Illumina, 20046811), Single-end 101-bp read, with two 10-bp index reads; Run 2: P2-200 Cycle Cartridge (Illumina, 20046812), Paired-end 2 × 101-bp reads, with two 10-bp index reads. Base calling and FASTQ file generation were performed using BaseSpace DRAGEN BCL Convert v3.0.1.

Proteomics and posttranslational modification analysis

Bacterial pellets were lysed in 500 µL of urea buffer (8 M urea, 100 mM NaH₂PO₄, 10 mM Tris, pH 8.5) supplemented with a HALT protease and phosphatase inhibitor cocktail (Thermo Fisher Scientific, without EDTA) using Rino tubes containing stainless steel beads (0.9–2 mm; Next Advance, New York, USA). Samples were homogenized twice for 5 min at 4 °C, transferred to 1.5-mL tubes, sonicated (3 × 5 s at 30% amplitude with 5-s rests; Fisher Sonic Dismembrator) to disrupt nucleic acids, and centrifuged at 4 °C. Protein concentrations were determined by bicinchoninic acid (BCA) assay, and aliquots were stored at -80 °C. For digestion, 100 µg protein aliquots were reduced with 5 mM dithiothreitol for 30 min at room temperature, alkylated with 10 mM iodoacetamide for 30 min in the dark, and digested overnight with Lys-C (1:25, Wako, Osaka, Japan). Samples were diluted with 50 mM NH₄HCO₃ to < 2 M urea and further digested overnight with trypsin (1:25, Promega). Resulting peptides were desalted with Oasis HLB cartridges (Waters) and dried under vacuum. Peptides were resuspended in 0.1% trifluoroacetic acid (TFA), separated on a Waters Charged Surface Hybrid (CSH) column (150 μm × 15 cm; 1.7 μm), and analyzed on an EVOSEP LC system coupled to a Q-Exactive Plus Hybrid Quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific) using a preset 44-min gradient. Full MS scans were acquired over 400–1600 m/z (AGC target 3 × 10⁶; 100 ms maximum injection time) at 70,000 resolution (m/z 200), followed by 20 data-dependent HCD MS/MS scans (isolation width 1.6 m/z; collision energy 28%; AGC target 1 × 10⁵; 100 ms injection time) at 17,500 resolution. Dynamic exclusion was set at 30 s, and precursors with + 1 or ≥ + 7 charge states were excluded. Raw data were processed with MaxQuant (Andromeda search engine) against the UniProt P. gingivalis database (31,426 sequences). Carbamidomethyl-cysteine (+ 57.0215 Da) was set as a fixed modification, and methionine oxidation (+ 15.9949 Da), asparagine/glutamine deamidation (+ 0.9840 Da), and protein N-terminal acetylation (+ 42.0106 Da) were variable modifications (up to five per peptide). Only fully digested tryptic peptides with up to two missed cleavages were considered. Mass tolerances were ± 20 ppm before and ± 4.5 ppm after recalibration, with 0.05 Da MS/MS tolerance in the Orbitrap and 0.6 Da in the ion trap. False discovery rates for peptides, proteins, and sites were set at 1%. Label-free quantification was performed using razor and unique peptides, with “match between runs” enabled (0.7-min retention time match window, 20-min alignment space). No enrichment strategies were applied for phosphopeptides, acetylated peptides, or other PTMs. Phosphorylation and acetylation were searched as potential sites at peptide level, while oxidation and deamidation were included only as variable modifications to support peptide identification and were not interpreted as biologically regulated PTMs.

Phosphoproteomic analysis was performed using Proteome Discoverer 2.1 with fully tryptic specificity, ± 20 ppm precursor and ± 0.05 Da fragment tolerances, and variable modifications including methionine oxidation, asparagine/glutamine deamidation, phosphorylation (+ 79.9663 Da), and N-terminal acetylation, with carbamidomethyl-cysteine as a fixed modification. Up to three variable modifications per peptide were allowed. Filtering with Percolator ensured a peptide-level false discovery rate of 1%. Proteomics and posttranslational analysis were performed at Emory Integrated Proteomics Core, Emory University, Atlanta, GA, United States.

Alanine substitution of diadenylate cyclase (dac) Ser-226

Site-directed mutagenesis of dac (PGN_0523) at the codon encoding Ser-226 was performed to generate an alanine-substituted variant, dac (S226A). A linear DNA construct carrying the dac (S226A) allele was assembled and introduced into P. gingivalis strain 381 and its isogenic mutant 381 Δpde_Pg (ΔPGN_0521) via homologous recombination. DNA fragments containing 1-kb regions flanking dac were amplified from P. gingivalis 381 genomic DNA, whereas the promoterless tetracycline resistance gene (tetQ) was amplified from plasmid pT-COW [21]. The open reading frame of the dac(S226A) variant was synthesized by Integrated DNA Technologies, Inc. and used as a PCR template. Primer sequences are provided in Supplementary Table 11. PCR amplification was carried out using Q5^® High-Fidelity 2X Master Mix (New England BioLabs; Cat. # M0492S) with 25 cycles of 98 °C for 30 s, 57 °C for 30 s, and 72 °C for 10 min. The NEBuilder HiFi DNA Assembly Master Mix (New England BioLabs; Cat. # E2621) was then used to assemble the 1-kb upstream region, dac(S226A), tetQ cassette, and 1-kb downstream region into a linear construct according to the manufacturer’s protocol. Electrocompetent P. gingivalis cells were prepared from 15 mL of an actively growing culture (OD₆₀₀ = 0.5–0.7), washed twice with sterile electroporation buffer (10% glycerol, 1 mM MgCl₂, 4 °C), and resuspended in 0.5 mL of the same buffer. Cell aliquots (100 µL) were mixed with ~ 2 µg of assembled DNA, transferred to sterile 0.2-cm gap cuvettes, and electroporated at 2.5 kV for 5.5 ms using a Bio-Rad Gene Pulser. Following incubation on ice for 5 min, the cells were transferred to anaerobic chamber and recovered in 1 mL pre-reduced TSBHK broth for ~ 16 h. Then, these were plated on BHIYHK agar containing tetracycline (1 µg/mL). Transformants were confirmed by genomic DNA isolation (Wizard^® Genomic DNA Purification Kit; Promega), PCR verification across flanking regions, and Sanger sequencing (Eurofins Genomics).

Transmission electron microscopy (TEM)

TEM imaging was performed as explained previously [10]. Cells were harvested and washed three times with ice-cold phosphate-buffered saline (PBS), after which a fixation solution (0.1 M phosphate buffer, pH 7.4, containing 2% paraformaldehyde and 2% glutaraldehyde in MilliQ water) was gently added to the cell pellet without resuspension. Samples were stored at 4 °C for at least 48 h to ensure complete fixation. The pellets were then washed three times with 0.1 M phosphate buffer (20 s each) and post-fixed with 1% osmium tetroxide for 2 h at room temperature. After osmium treatment, pellets were rinsed four times (10 min each) in phosphate buffer and stored overnight at 4 °C. For embedding, Durcopan resin was prepared by combining components A, B, C, and D. A 50:50 (v/v) mixture of resin and 200-proof EMS-grade ethanol was prepared in a specialized flask for the subsequent dehydration step. Pellets were transferred to glass Wheaton sample jars (VWR, Cat# 225536) and dehydrated through a graded ethanol series (50%, 70%, 95%, and 100% ethanol, 15 min each step), with 100% ethanol step repeated twice. Dehydrated samples were then infiltrated with resin: first with a 1:1 resin/ethanol mixture for 1 h on a rotator wheel, followed by a 3:1 resin/ethanol mixture for 1 h, and finally with 100% resin. Pellets were placed in a compression chamber overnight, then transferred to plastic TEM capsules filled with pure resin. Capsules were compressed for 1 h to remove trapped air bubbles, and polymerization was carried out at 60 °C for 48 h. Polymerized blocks were sectioned using a Leica EM UC7 ultramicrotome to a thickness of ~ 70 nm. Sections were collected onto copper EM grids and stained sequentially: first with 8% aqueous uranyl acetate (5 min), rinsed by ten dips in each of three beakers containing deionized water, dried for 1 h, and then with 3% stabilized lead citrate (2 min). Grids were again washed by ten dips in three changes of deionized water and air-dried for 1 h. Prepared samples were examined using a Hitachi HT7700 Transmission Electron Microscope.

Quantitative TEM image analysis and texture feature extraction

TEM micrographs of WT and Δpde_Pg mutant P. gingivalis cells were acquired at a calibrated magnification corresponding to a 200 nm scale bar. All images were converted into 8-bit grayscale. TEM images were pre-processed using contrast-limited adaptive histogram equalization (CLAHE) to improve boundary definition. Cells were segmented using adaptive Gaussian thresholding, denoising, and distance-transform watershed separation, a combination of approaches widely applied in biomedical imaging pipelines [22, 23]. To improve local contrast and enhance cellular boundaries, contrast-limited adaptive histogram equalization (CLAHE; clip limit = 2.0, grid size = 8 × 8) was applied. Images were subsequently smoothed with a Gaussian filter (σ = 1.0-1.5 pixels) to reduce high-frequency noise while preserving structural detail. Cells were segmented using an adaptive Gaussian thresholding algorithm (block size = 51 pixels, constant = 2), followed by binary denoising and hole filling. A distance-transform watershed procedure was applied to separate adjacent or overlapping cells. Objects touching image borders were excluded. To ensure inclusion of only bona fide single-cell profiles, we applied morphological filters: (i) area within empirically defined limits (≥ 300 pixels² and ≤ 5000 pixels² at 200 nm scale), and (ii) solidity ≥ 0.90 to exclude irregular fragments or aggregates. The curated dataset comprised 83 WT and 42 mutant cells across multiple fields of view. From each segmented cell, standard morphometric descriptors were extracted using region-based analysis: cross-sectional area, perimeter, major and minor axis lengths (from best-fit ellipses), aspect ratio, and circularity defined as . Mean gray intensity was also calculated per cell as a first-order measure of average electron density. To quantify internal cytoplasmic organization, we implemented a gray-level co-occurrence matrix (GLCM)–based texture analysis [24–26]. Each segmented cell was resized to 64 × 64 pixels to standardize co-occurrence statistics across cells. GLCMs were computed for four orientations (θ = 0^∘,45^∘,90^∘,135^∘) and pixel offset distance d = 5, with 256 Gy levels. For each orientation, the GLCM was normalized: where  P(i,j) represents the probability of observing gray-level  i adjacent to gray-level  j. The correlation statistic was calculated as: , where are the means and are the standard deviations of the row and column marginal distributions. The correlation values from the four orientations were averaged to yield a single per-cell correlation score. Higher correlation values reflect ordered, coherent textures, whereas lower values indicate disordered or entangled cytoplasmic organization.

To account for the observation that both WT and mutant populations contained mixtures of round-like and rod-like cells, morphotypes were classified objectively using circularity and aspect ratio thresholds (round-like: circularity ≥ 0.85 and aspect ratio ≤ 1.2; rod-like: circularity < 0.85 or aspect ratio > 1.2). This stratification prevented pooling of morphologically distinct subpopulations and allowed direct comparison of morphotypes. For each descriptor, distributions of WT and mutant values were compared using the non-parametric Mann–Whitney U test, which does not assume normality of the underlying distributions and is therefore appropriate for imaging-derived cellular measurements that are often skewed, heteroscedastic, or contain outliers. The Mann–Whitney U test compares the rank ordering of values between two groups and is robust to deviations from normality, making it particularly suitable for small-to-moderate sample sizes commonly encountered in morphometric and texture-based analyses of electron microscopy datasets [27, 28]. All analyses were conducted in Python (v3.10) within an Anaconda environment using NumPy (v1.23), SciPy (v1.9), scikit-image (v0.20), OpenCV (v4.6), pandas (v1.5), and Matplotlib (v3.6).

Genomic DNA-binding protein assay and identification of co-precipitated proteome

Genomic DNA-binding proteins were isolated from overnight cultures of P. gingivalis grown anaerobically in 10 mL TSBYK medium. Cells were harvested by centrifugation at ~ 6000 rpm for 1 min inside anaerobic chamber, washed once with BHIYHK medium, resuspended, and used to inoculate 50 mL of BHIYHK at an initial OD₆₀₀ of 0.25. Cultures were incubated anaerobically at 37 °C for 2 h. Parallel cultures were prepared for crosslinking assays. Cells were crosslinked with Trump’s Fixative (Em-grade, Mauressac, France) [29], yielding a final formaldehyde concentration of 1% (w/v) by adding 500 µL fixative to 4.5 mL of culture, followed by anaerobic incubation for 30 min. Crosslinking was quenched with 0.25 M glycine for 10 min under anaerobic conditions with gentle agitation. Cells were removed from the anaerobic chamber, cooled on ice for 15 min, washed twice with ice-cold PBS, and collected by centrifugation at 4000 × g for 15 min. DNA-protein complexes were purified using a Genomic DNA Purification Kit (Thermo Scientific, Vilnius, Lithuania) according to the manufacturer’s instructions. Briefly, bacterial pellets were lysed in TE buffer with lysis solution, extracted with chloroform, and precipitated with the supplied precipitation solution. Complexes were further purified by ethanol precipitation, washed with 70% ethanol, and resuspended in sterile deionized water. DNA yield and purity were assessed spectrophotometrically. Purified complexes were incubated at 65 °C overnight in a Mini Block Heater to reverse crosslinks, followed by digestion with DNase I (1 mg/mL in 0.1 M sodium acetate, pH 5.5, with 10 mM MgCl₂). Released DNA-binding proteins were resolved by SDS-PAGE using Bolt™ 4–12% Bis-Tris Plus Mini Protein Gels (Thermo Scientific, Rockford, USA). In-solution proteins were subjected to LC-MS/MS for identification.

Identification of proteins candidates was performed at the IDeA National Resource for Quantitative Proteomics at the University of Arkansas for Medical Sciences (UAMS). For filter-aided sample preparation (FASP), proteins were reduced, alkylated, and digested on-filter with sequencing-grade modified trypsin (Promega) following established protocols [30]. For the Q-Exactive Plus workflow, peptides were resuspended in 0.1% trifluoroacetic acid (TFA) and separated on a Waters CSH C18 column (150 μm × 15 cm, 1.7 μm) using an EvoSep LC system with the 44-min preset gradient. Eluted peptides were analyzed on a Q-Exactive Plus Hybrid Quadrupole-Orbitrap mass spectrometer (Thermo). MS1 scans (400–1600 m/z, 70,000 resolution at m/z 200, AGC target 3e6, max injection time 100 ms) were acquired in profile mode, followed by up to 20 HCD MS/MS scans (isolation width 1.6 m/z, normalized collision energy 28%, AGC target 1e5, resolution 17,500). Dynamic exclusion was set at 30 s; precursors with charge states + 1 and ≥ + 7 were excluded. For the Orbitrap Fusion Tribrid workflow, FASP-derived peptides were separated on a Waters XSelect CSH C18 column (150 × 0.075 mm, 2.5 μm) with a 60-min linear gradient from 98:2 to 65:35 buffer A: B (buffer A = 0.1% formic acid, 0.5% acetonitrile; buffer B = 0.1% formic acid, 99.9% acetonitrile) using either an UltiMate 3000 RSLCnano or a nanoAcquity UPLC system. Electrospray ionization (2.4 kV) was followed by MS analysis in the Orbitrap at 240,000 resolution (375–1500 m/z, profile mode). HCD activation (28–31% collision energy, charge-dependent) was used for fragmentation, with MS/MS spectra acquired in the ion trap in centroid mode. Top-speed data-dependent acquisition was applied. Spectra from Q-Exactive runs were processed in MaxQuant (Andromeda search engine) against the UniProt P. gingivalis database (31,426 sequences). Search parameters allowed up to two missed cleavages, precursor mass tolerance of ± 20 ppm (pre-calibration) and ± 4.5 ppm (post-calibration), and fragment tolerance of 0.05 Da (Orbitrap) or 0.6 Da (ion trap). FDR was set at 1% for peptide spectral matches, proteins, and modification sites. Label-free quantification was performed with “match between runs” enabled (20 min RT alignment, 0.7 min match window). Spectra from Orbitrap Fusion runs were searched using Mascot (Matrix Science) with precursor ion tolerance of 3 ppm and fragment ion tolerance of 0.5 Da. Results were validated with Scaffold (Proteome Software), accepting peptides at < 1% FDR (local FDR algorithm) and proteins at < 1% FDR if supported by ≥ 2 unique peptides. Protein probabilities were assigned using Protein Prophet [31].

SDS-PAGE and silver staining

SDS-PAGE separation of the isolated c-di-AMP- and DNA-binding proteins was performed using precast Bolt™ 4–12% Bis-Tris Plus Mini Protein Gels (Invitrogen) according to the manufacturer’s instruction. Briefly, protein samples (3 µg/well) were mixed with sample loading buffer (Invitrogen), heated at 95 °C for 5 min, and loaded onto the gels. For molecular weight reference, 5 µL of PageRuler™ Plus Prestained Protein Ladder (Thermo Scientific, Cat. No. 26619) was included in each run. Electrophoresis was carried out at room temperature under a constant voltage of 200 V for 22 min using 1x Bolt™ MES SDS running buffer (Invitrogen). Following electrophoresis, proteins were visualized by silver staining with the Pierce™ Silver Stain Kit (Thermo Scientific, Cat. No. 24600) using the manufacturer’s protocol optimized for mass spectrometry compatibility. Gels were fixed in Silver Stain Fixative for 2 h and washed twice for 5 min each with ethanol and ultrapure water. Afterwards, gels were incubated for 1 min in Sensitizer Solution (1:500 Silver Stain Sensitizer: water), followed by two 1 min ultrapure water washes. Gels were stained with Silver Stain Solution (1:100 Silver Stain Enhancer: Silver Stain) for 5 min, followed by two quick ultrapure water washes (20 s). Gels were then developed with freshly prepared Developer Solution (1:100 Silver Stain Enhancer: Silver Stain Developer) until the desired band intensity was achieved (~ 2–3 min). The reaction was terminated with Stop Solution (5% acetic acid), the gels were briefly washed with ultrapure water, and the stained gels were imaged and analyzed using an iBright1500 Imaging System (Thermo Fisher Scientific).

Bioinformatic analyses, molecular docking, and dynamic simulation

Phyre2.2 (https://www.sbg.bio.ic.ac.uk/~phyre2/htmL/page.cgi?id=index) [32, 33], I-TASSER (at https://zhanggroup.org/) [34–36], and AlphaFold (https://alphafold.ebi.ac.uk/) [37, 38] servers were applied for structural modeling predictions. Multiple sequence alignment program Clustal Omega (https://www.ebi.ac.uk/jdispatcher/msa/clustalo) was used for protein sequence alignments. To investigate the structural and dynamic behavior of the protein (DAC_Pg) under phosphorylated conditions, molecular dynamics simulations were carried out using GROMACS (https://www.gromacs.org/). The protein was first phosphorylated at Ser-226, solvated, and assembled, preparing a complete system for Gromacs production run using CHARMM-GUI Solution Builder. Next, to relieve steric clashes and optimize the geometry of the solvated system, energy minimization was performed using the steepest descent algorithm. Position restraints were applied on the protein heavy atoms with force constants of 400.0 kJ/mol·nm² for backbone atoms and 40.0 kJ/mol·nm² for side-chain atoms. The minimization was set with an energy tolerance (emtol) of 1000.0 kJ/mol·nm and a maximum of 5000 steps. Nonbonded interactions were calculated with the Verlet cutoff scheme, with a neighbor list updated every 10 steps and a cutoff radius of 1.2 nm. Van der Waals interactions were treated with a force-switch modifier between 1.0 and 1.2 nm, while long-range electrostatics were computed using the Particle Mesh Ewald (PME) method with a real-space cutoff of 1.2 nm. All covalent bonds involving hydrogen atoms were constrained using the LINCS algorithm, ensuring numerical stability during minimization. This procedure ensured that the system reached a relaxed state before subsequent equilibration steps. Further, equilibration of the system was carried out under position restraints on protein heavy atoms (400.0 kJ/mol·nm² for backbone and 40.0 kJ/mol·nm² for side chains). Molecular dynamics was performed using a 1 fs timestep for 1 ns (1,000,000 steps). Temperature was maintained at 303.15 K using the V-rescale thermostat with separate coupling of solute and solvent groups (τ = 1.0 ps). Nonbonded interactions were treated with the Verlet cutoff scheme, with a 1.2 nm cutoff for van der Waals and electrostatic interactions, PME for long-range electrostatics, and force-switching between 1.0 and 1.2 nm for van der Waals terms. Constraints were applied to all bonds involving hydrogens using LINCS. The system was initialized with Maxwell-Boltzmann velocity distribution at 300 K. Afterwards, A 200 ns production simulation was performed using the leap-frog integrator with a 2 fs timestep (100,000,000 steps). The system was maintained at 300 K with the V-rescale thermostat applied separately to solute and solvent groups (τ = 1.0 ps) and at 1 bar using the C-rescale barostat with isotropic pressure coupling (τ = 5.0 ps, compressibility = 4.5 × 10⁻⁵ bar⁻¹). Nonbonded interactions were treated with the Verlet cutoff scheme, using a 1.2 nm cutoff for van der Waals and electrostatics, force-switching between 1.0 and 1.2 nm for van der Waals, and PME for long-range electrostatics. All bonds involving hydrogens were constrained with LINCS, and the simulation continued from equilibrated coordinates. Following molecular dynamics simulations, the trajectory of the protein model was analyzed to evaluate its structural stability, energetic profile, and dynamic behavior. Global conformational stability and compactness were examined using root mean square deviation (RMSD) and radius of gyration (Rg), while residue-level flexibility was assessed through root mean square fluctuation (RMSF). Solvent-accessible surface area (SASA) was calculated to monitor surface exposure relevant to molecular interactions and shielding. Principal component analysis (PCA) was employed to capture major conformational transitions and collective motions, and dynamic cross-correlation, along with network analysis, were applied to explore cooperative residue interactions.

For the molecular docking and dynamics study of PDE_Pg’s atypical behavior, the three-dimensional structure of the c-di-AMP (ligand) was retrieved from the PubChem database and converted to .pdb format. Energy minimization of the ligand was performed using the MMFF94 force field and the conjugate gradient optimization algorithm. The minimized structures were then converted to .pdbqt format. The target protein (PDE_pg) was modeled and prepared by removing all heteroatoms and water molecules using Discovery Studio. Molecular docking was performed using AutoDock Vina to generate 10 conformations for each ligand. Two iron atoms and one water molecule were added at a specific catalytic site, acting as a template protein (PDB: 4S1B), located approximately 2 Å away. Following docking, the ligand poses with the lowest binding energy were selected and analyzed for ligand-protein interactions using Discovery Studio 2021. All-atom molecular dynamics (MD) simulations were performed in explicit solvent for 100 ns to assess the structural stability of the best-docked conformations for each target. The simulations were conducted with GROMACS 2024 using the AMBER ff99SB-ILDN force field. Each system was solvated in a rectangular box with a 10.0 Å buffer distance from the protein surface in all directions, employing a three-site water model. The resulting trajectories were analyzed using built-in GROMACS utilities, supplemented by additional software tools where appropriate for specialized analyses.

Statistical analysis and applied software

Statistical analysis of RNA-seq data was performed by the program ‘Rockhopper’ [39]. This program models read count variation using Poisson distributions as a baseline and then applies a negative binomial model to account for the overdispersion typical of RNA-seq data. Significance values are calculated for each gene or transcript and adjusted for multiple testing using false discovery rate (FDR) correction based on the Benjamini-Hochberg procedure [39]. Phosphoproteomic data were analyzed in Proteome Discoverer 2.1, with peptide-spectrum matches rescored by Percolator, which uses a target–decoy strategy and machine-learning-based probability scoring. Statistical filtering was applied to control the peptide-level false discovery rate at 1%, ensuring high-confidence identifications. A non-linear regression logistic growth model (GraphPad Prism 10) was used to statistically compare bacterial growth rates. Illustrations were created using BioRender (https://app.biorender.com).

Results

Rhythmic-phasic c-di-AMP signaling defines growth-phase transitions and establishes PDE_Pg as a key regulator

Previously, we observed that c-di-AMP turnover in P. gingivalis does not follow the typical synthesis-degradation pattern observed in other bacteria. Instead, c-di-AMP homeostasis is primarily controlled at the synthesis stage, mediated by the atypical phosphodiesterase PDE_Pg [6, 10]. Deleting pde_Pg leads to abnormal cellular accumulation of ATP, a paradoxical reduction in cellular c-di-AMP, and disproportionately low levels of the intermediate pApA, yet these studies did not capture the temporal dynamics of c-di-AMP signaling. Notably, LC/MS analysis of the extracellular milieu of P. gingivalis strains did not detect c-di-AMP, indicating that, under the conditions examined, this species cannot be classified as an active c-di-AMP secretor. Here, we quantified intracellular c-di-AMP, ATP, and pApA across defined growth phases in P. gingivalis WT and the Δpde_Pg mutant using ELISA or LC/MS-based assays (Supplemental Fig. 1A-D). These measurements were integrated with nonlinear regression and logistic growth modeling to discern phase-specific signaling behavior.

Logistic fits revealed pronounced divergence between WT and Δpde_Pg (Fig. 1, Supplemental Table 1). The WT displayed a growth rate constant (k) of 0.2273 (95% CI: 0.1837–0.2799), whereas Δpde_Pg grew significantly more slowly, with k reduced to 0.0940 (95% CI: 0.0776–0.1129). This means that the WT strain divided approximately 2.4-fold faster than the Δpde_Pg mutant. Importantly, the confidence intervals did not overlap, confirming that the reduction in growth rate is statistically robust. Moreover, Δpde_Pg showed a delayed inflection point (Xint = 10.64 h) compared with WT (Xint = 4.40 h), reflecting a pronounced extension of the lag period before exponential proliferation. Given these kinetic differences, we next examined how the intracellular c-di-AMP pool fluctuates during growth.

Fig. 1.

Rhythmic-phasic c-di-AMP signaling module orchestrates growth-phase transitions and establishes PDE_Pg as a key regulator. Growth curves (top panel), intracellular c-di-AMP (middle panel), and pApA and ATP pools (bottom panels) of P. gingivalis WT and Δpde_Pg quantified over 51 h period. To rigorously assess growth kinetics, non-linear regression was applied using a logistic growth model (GraphPad Prism). The growth rate constant (k) reflects the proliferation rate for each condition. WT cells displayed distinct phasic c-di-AMP cycling over four phases of sampling at quantification (I to IV) corresponding to distinct phases of growth: an initial sharp decline in c-di-AMP pool (~ 2.4-fold) during early exponential phase (phase II), a nearly 4-fold rebound during mid-exponential growth (phase III), and a decline again (~ 2.2-fold) in stationary phase (phase IV). Δpde_Pg cells lacked this cycling, showing a prolonged lag phase (~ 18 h) and only modest c-di-AMP fluctuations, underscoring the necessity of PDE-mediated regulation. Δpde_Pg retained disproportionately high ATP levels compared with WT, consistent with impaired energy utilization and dysregulated signaling. Collectively, these findings support a reproducible rhythmic-phasic organization of c-di-AMP dynamics during growth-phase transitions in P. gingivalis, and PDE_Pg is critical for this temporal signaling behavior

This study revealed that the dynamics of c-di-AMP signaling in the WT P. gingivalis strain are not linear but occur in distinct temporal episodes (Fig. 1). At the beginning of inoculation (phase I), intracellular c-di-AMP levels were 2.6 µM/mg protein. By hour 6 (early exponential phase II), levels decreased sharply by ~ 2.4-fold to 1.1 ± 0.05 µM/mg protein. This was followed by a striking rebound, with c-di-AMP levels increasing nearly 4-fold to 4.2 ± 0.01 µM/mg protein at hour 18 (mid-exponential phase III), consistent with enhanced synthesis during periods of rapid growth and metabolic demand (Fig. 1). Then, by hour 44 (stationary phase IV), c-di-AMP levels declined again by ~ 2-fold, returning close to the baseline observed at time zero. This temporally structured pattern indicates that c-di-AMP signaling in WT is not monotonic but instead involves coordinated cycles of decline and replenishment associated with phase-specific regulation.

In contrast, the Δpde_Pg mutant lacked this rhythmic-phasic scheme entirely (Fig. 1). The mutant exhibited a markedly extended lag phase lasting ~ 18 h (corresponding to phases I-III in WT) during which c-di-AMP levels fluctuated only within a narrow range (2.3–2.7 µM/mg protein). Unlike WT, the mutant did not undergo the early-phase reduction in c-di-AMP required to initiate efficient entry into exponential growth. These data suggest that PDE_Pg enables regulation of the transient early-phase reduction and mid-log replenishment, as the observed signaling defect directly corresponds to the mutant’s impaired growth kinetics.

ATP and pApA measurements in pooled samples aligned closely with the dynamic patterns of c-di-AMP cycling and were consistent with our previous findings. The Δpde_Pg mutant maintained disproportionately high ATP levels and persistently low pApA levels compared with WT cells, a pattern indicative of impaired energy utilization and disrupted nucleotide-signaling homeostasis (Fig. 1). In WT cells, ATP levels declined sharply from 139 nM/mg protein at time zero (inoculation) to 48 nM/mg protein at 6 h, and further to approximately 9 nM/mg protein by 18 h, reflecting progressive ATP consumption during the lag-to-log metabolic transition. By contrast, Δpde_Pg retained markedly elevated ATP concentrations throughout this period, exhibiting ~ 6-fold and ~ 19-fold higher ATP levels than WT at 6 and 18 h, respectively. Thus, whereas WT cells rapidly consumed ATP in coordination with c-di-AMP dynamics and growth initiation, the mutant failed to engage in this coordinated metabolic expenditure, instead preserving an abnormally high ATP reservoir.

Temporal quantification of pApA showed that its pool size in WT tracked tightly with c-di-AMP dynamics, albeit at picomolar abundance (Fig. 1). The initial pApA concentration measured 7.73 pM/mg protein, decreased ~ 2.3-fold to 3.36 pM/mg at 6 h, and then rebounded to 8.91 pM/mg by 18 h. This strict parallelism between c-di-AMP and pApA levels rules out active PDE-mediated degradation as the basis for the observed rhythmic-phasic c-di-AMP dynamics. Instead, it provides additional support for the conclusion that c-di-AMP fluctuations in P. gingivalis arise from synthesis-level modulation and population-dependent dilution rather than from turnover via enzymatic hydrolysis.

These data suggest that the initial reduction of c-di-AMP from time zero to 6 h results from a combination of temporarily reduced c-di-AMP synthesis and dilution of the existing intracellular pool as the population begins to transition toward active growth. Between 6 and 18 h, as cells enter rapid division, the c-di-AMP pool is replenished to a higher level to meet the increased signaling demand associated with accelerated metabolic and cell-cycle activity. This interpretation is reinforced by the behavior of the Δpde_Pg mutant, which fails to exhibit any rhythmic-phasic c-di-AMP dynamics observed in WT cells. Because the mutant does not initiate active division during its prolonged lag phase, it does not undergo the population-driven dilution that contributes to the early-phase decrease in WT cells, nor does it mount the subsequent synthetic rebound observed during WT exponential growth. Overall, these findings demonstrate that PDE_Pg functions as the key driver of rhythmic-phasic c-di-AMP signaling. This regulatory process is fundamental to the coordination of proper growth-phase transitions in P. gingivalis. In its absence, these rhythmic-phasic c-di-AMP dynamics collapse, the ATP pool remains high, and classical growth synchronization breaks down.

Biochemical evidence from whole membrane fractions supports a regulatory pseudo-phosphodiesterase role for PDE_Pg rather than canonical c-di-AMP hydrolase activity

To complement our physiological observations identifying PDE_Pg as a central regulator of rhythmic-phasic c-di-AMP signaling in P. gingivalis, we performed comparative biochemical analyses of whole membrane fractions from WT and Δpde_Pg mutant, followed by LC-MS quantification of ATP, c-di-AMP, and pApA (Fig. 2A-B, Supplemental Table 2). To determine whether membrane-associated c-di-AMP synthetic activity was preserved in the preparations and whether PDE_Pg influences DAC_Pg-mediated c-di-AMP synthesis from ATP, whole membrane fractions were incubated with ATP (1 mM) and analyzed after 30 and 60 min (Fig. 2A, Supplementary Table 2). Both WT and Δpde_Pg membrane fractions synthesized c-di-AMP, accompanied by a time-dependent decline in ATP abundance in both strains. This observation confirms preservation of functional membrane-associated cyclase activity in the membrane preparations. However, WT membrane fractions consistently produced greater amounts of c-di-AMP than Δpde_Pg fractions. In WT samples, c-di-AMP levels increased from 5.43 µM at 30 min to 10.17 µM at 60 min, whereas the mutant accumulated only 3.72 µM and 6.41 µM, respectively (Fig. 2A, Supplemental Table 2). Given that DAC_Pg transcript levels are comparable between WT and Δpde_Pg strains, these data support a positive role for PDE_Pg in regulating c-di-AMP synthesis at the enzymatic level rather than through transcriptional control of DAC_Pg. No detectable pApA accumulation was observed under these conditions, except for a minimal signal detected in the 60- min WT sample (7.57 pM). This level was disproportionately low relative to the amount of c-di-AMP produced and further reflects minor background turnover rather than physiologically meaningful PDE_Pg-mediated phosphodiesterase activity. These findings are consistent with our previous intracellular metabolic quantifications, in which changes in pApA abundance were markedly disproportionate to changes in c-di-AMP levels in both WT and mutant strains [6, 10].

Fig. 2.

Whole membrane fraction-based biochemical assays support a non-canonical pseudo-enzymatic regulatory role of PDE_Pgin P. gingivalis. Membrane fractions prepared from P. gingivalis WT and Δpde_Pg were subjected to in vitro biochemical assay followed by LC-MS analysis. A ATP-dependent c-di-AMP synthesis assay. Membrane fractions were incubated with ATP (1 mM) in the presence of MnCl₂ (1 mM) at 37 °C for 30 and 60 min, followed by LC-MS analysis. Representative chromatograms show c-di-AMP synthesis in both strains upon ATP supplementation, with consistently higher c-di-AMP accumulation observed in WT membrane fractions than in the Δpde_Pg mutant over time, indicating that PDE_Pg positively regulates c-di-AMP production. No detectable pApA accumulation was observed under these conditions, except for a disproportionately low abundance in the 60 min WT membrane fraction. B Exogenous c-di-AMP hydrolysis assay. Membrane fractions were incubated with exogenous c-di-AMP (50 µM) under identical conditions and analyzed after 30 and 60 min. Both WT and Δpde_Pg membrane fractions exhibited comparable recovery of c-di-AMP (~ 45–46 µM), indicating minimal early turnover and no detectable genotype-dependent pApA accumulation across both strains at different time points. These findings indicate the absence of measurable canonical phosphodiesterase activity in in vitro biochemical assay. ND, not detected

We next evaluated the potential for c-di-AMP hydrolysis by incubating whole membrane fractions with exogenously supplied c-di-AMP (50 µM) under the same reaction conditions over time (Fig. 2B, Supplementary Table 2). WT and Δpde_Pg membrane fractions exhibited nearly identical c-di-AMP recovery profiles (46.37 µM and 45.49 µM, respectively, Fig. 2B, Supplementary Table 2), indicating minimal turnover and no detectable genotype-dependent difference in substrate handling under the employed assay conditions. Importantly, no detectable pApA accumulation was observed in either strain throughout the assay.

Together, these observations suggest that canonical phosphodiesterase-mediated degradation is unlikely to be a major route of c-di-AMP turnover in P. gingivalis, consistent with a model in which c-di-AMP homeostasis is regulated predominantly through control of synthesis rather than degradation.

Untargeted metabolomics reveals that deletion of PDE_Pg compromises peptide uptake, redox state, and envelop homeostasis

Given the tight coupling between PDE_Pg-dependent c-di-AMP signaling and the metabolic ATP/redox state, we performed untargeted metabolomic profiling of WT and the Δpde_Pg mutant. Because the WT strain transitions into the exponential phase much earlier than Δpde_Pg, cultures were harvested in a narrowly defined early phase-transition window (~ 1-hour post-inoculation at OD₆₀₀ ≈ 0.15) to minimize growth-stage artifacts and capture the metabolic configuration associated with this critical physiological shift that precedes the initiation of cell division (Fig. 3A-B, Supplemental Fig. 1E-L, Supplemental Table 3). Under these controlled conditions, statistical analysis revealed a clear segregation of WT and Δpde_Pg metabolomes, indicating that the deletion of PDE_Pg is associated to reprogram nutrient uptake, central metabolism, envelope precursor pools, and redox-related metabolites. A prominent feature of the WT profile was the presence of multiple short, mostly hydrophobic or amphiphilic dipeptides and oligopeptides (Ile-Pro, Gly-Ile, Leu-Glu, Leu-Pro, Leu-Trp, Ser-Leu, Thr-Leu, PyroGlu-Val, and the hexapeptide Val-Leu-Pro-Val-Pro-Gln) that were consistently below the standard detection limit in the Δpde_Pg mutant (Fig. 3A-B, Supplemental Table 3). P. gingivalis is asaccharolytic and relies on proteolysis of environmental or host proteins and subsequent uptake of small peptides as its primary nutrient acquisition strategy [40–43]. Presumably, these peptides are converted to amino acids and subsequently fermented via a Stickland-like mechanism, in which balanced uptake of oxidative and reductive amino acids plays a central role in maintaining metabolic and redox equilibrium [44–46]. The WT-only peptide signature, therefore, is consistent with an intact proteolysis-peptide-uptake-catabolism axis under PDE_Pg regulation. Its loss in the Δpde_Pg mutant suggests that PDE_Pg and c-di-AMP signaling play critical roles during peptide uptake, consistent with the established role of c-di-AMP signaling in controlling uptake systems and coordinating cell-envelope stress responses in other bacteria [47–50]. This WT-specific pattern was reinforced by the observation that several key interconnected metabolic subpathways that regulate cellular energy/redox balance and osmoprotection, including the proline-GABA shunt, the arginine/polyamine axis, and the β-alanine/pantothenate/CoA synthesis pathway, were detectable only in WT cells. β-Alanine in anaerobes is primarily generated from aspartate decarboxylation and from peptide-derived aspartyl residues, while pantothenate is synthesized from β-alanine and pantoate [51–53]. The disappearance of both metabolites in Δpde_Pg therefore implies a break in nitrogen flow from peptide-driven aspartate, consistent with reduced peptide import. Limitation of pantothenate impairs CoA synthesis and, hence, fatty-acid and membrane phospholipid metabolism, consistent with the altered lyso-PC and PE profiles observed (Fig. 3A-B, Supplemental Table 3).

Fig. 3.

Integrated multi-omics analyses reveal coordinated physiological remodeling of PDE_Pg-dependent regulatory networks in P. gingivalis. A Schematic summary of metabolic distinctions between WT and Δpde_Pg. WT samples are enriched in peptides and metabolic sub-pathways critical for energy/redox balance and osmoprotection. These metabolites, designated as WT-only, were below the detection threshold in Δpde_Pg mutant cells. In contrast, Δpde_Pg cells exhibited elevated levels of free amino acids, modified amino acids, redox cofactors, and nucleotide-sugar precursor UDP-GlcNAc. In the WT, active PDE-mediated turnover of c-di-AMP maintains controlled signaling that is associated with efficient peptide transport (presumably via Opp/Dpp systems) and promotes peptide assimilation as the principal nutrient acquisition route. Imported peptides supply amino nitrogen and carbon skeletons for multiple interconnected pathways, including the proline-GABA shunt, the arginine-polyamine pathway, and β-alanine/pantothenate-dependent CoA biosynthesis. These fluxes sustain redox balance and membrane integrity under growth transition conditions. In contrast, Δpde_Pg cells exhibit repression of peptide uptake. This shift leads to the loss of peptide-derived metabolites, metabolic rigidity, and a compensatory buildup of free amino acids, redox modulators (β-NAD, betaine), and envelope-remodeling intermediates. Elevated N, N-dimethylarginine and methionine sulfoxide indicate redox imbalance and methylation stress. The pronounced reduction in PE (16:0/16:0) reflects disrupted phospholipid turnover and membrane composition. B Heatmap of significantly altered metabolites (p < 0.05, t-test) between WT and Δpde_Pg, including only components with signal intensities above the detection threshold. Each row represents one identified metabolite, hierarchically clustered according to relative abundance (red = higher; blue = lower). Together, these data demonstrate that PDE-dependent c-di-AMP signaling coordinates peptide uptake for catabolism, lipid/redox homeostasis, and that its disruption in Δpde_Pg rewires the metabolome toward a stress-adapted, energy-conserving state characterized by impaired peptide assimilation and enhanced cell-envelope remodeling. C RNA-seq (Rockhopper pipeline, analyzed in iDEP 2.0) identified differentially expressed genes (DESeq2, log₂FC ≥ 1, FDR < 0.01) after normalization and log₂(CPM + 4) transformation. The volcano plot highlights significantly upregulated (red) and downregulated (green) transcripts in the Δpde_Pg mutant relative to WT. Enriched modules include corrinoid/cobalamin metabolism, quinone/redox pathways, iron acquisition, envelope enzymes, and DNA-replication/repair factors, revealing a broad reorganization of energy/redox and growth-related functions. D Quantitative LC-MS/MS proteomics (Student’s t-test p < 0.05) show coordinated remodeling consistent with transcriptomic and metabolomic trends. Upregulated proteins (green), including flavodoxin (PGN_1209), NaMN-ATase (PGN_2007), pyridine-nucleotide disulfide oxidoreductase (PGN_0285), CaiA and ETF FixA, indicate reinforced NAD⁺/NADH and thiol-redox buffering. Conversely, downregulated proteins (red) include MurC (PGN_0628), β-barrel OMP (PGN_0156), ClpX (PGN_1550), PorN/M/W/Sov (T9SS), and DUF4835 (PGN_1783), signifying reduced envelope biogenesis, secretion, and ATP-dependent proteolysis. Together, these data demonstrate a compensatory shift toward redox stabilization and energy conservation accompanying high ATP, low c-di-AMP, and impaired peptide uptake. E Operonic organization. The dac-pde-cdaR(ybbR) cluster lies within a genomic neighborhood encoding folate, pantothenate/CoA, and secretion-associated genes (e.g., folP, coaE, yajC, ogt), linking c-di-AMP signaling to redox, biosynthetic metabolism, and DNA processing. This organization, together with multi-omics evidence, supports a model in which PDE_Pg-dependent c-di-AMP dynamics contribute to couple ATP/redox homeostasis with envelope maintenance, replication timing, and stress adaptation required for efficient lag-to-log transition and growth

In contrast, the Δpde_Pg mutant accumulated significantly higher levels of several free amino acids and related derivatives, including 4-hydroxyproline, L-methionine, methionine sulfoxide, valine, alloisoleucine, threonine, serine, tyrosine, glutamic acid, citrulline, urocanic acid, N,N-dimethylarginine, and asparagine butyl ester (Fig. 3A-B, Supplemental Table 3). These data suggests that the mutant reprograms metabolic state by maintaining or building intracellular pools of free amino acids when peptide-level import is compromised. Consistent with this finding, the pde mutant also showed significantly elevated levels of β-NAD and betaine. Both metabolites support redox balancing and methyl/one-carbon metabolism, respectively. Their accumulation suggests that the Δpde_Pg mutant is actively buffering an altered redox environment and preserving methylation capacity, both of which are processes through which c-di-AMP signaling is known to intersect, via regulation of cell-wall and osmotic homeostasis. Additionally, the presence of oxidized and methylated species, such as methionine sulfoxide and N, N-dimethylarginine, indicates that the mutant is operating under a more oxidative and methylation-demanding intracellular state. Another notable consequence of PDE_Pg deletion and c-di-AMP signaling dysregulation in Δpde_Pg was the increase in the nucleotide-sugar precursor uridine 5′-diphospho-N-acetylglucosamine (UDP-GlcNAc) together with several membrane-associated lyso-glycerophosphocholine species. UDP-GlcNAc is a high-value precursor for peptidoglycan and other envelope glycoconjugates. Furthermore, the parallel increase in lyso-glycerophosphocholine species and the significant decrease in the membrane phospholipid PE (16:0/16:0) (FC ≈ 9) in Δpde_Pg further indicate that the mutant cells require envelope remodeling to control stresses and support growth (Fig. 3A-B, Supplemental Table 3).

Consistent with these findings, the 2D PCA projection shows WT replicates are more metabolically dispersed during lag-to-log transition, reflecting greater cellular metabolic diversity consistent with active physiological remodeling during the lag-to-log transition. In contrast, Δpde_Pg replicates cluster tightly, consistent with a restricted, peptide-depleted metabolic state resulting from PDE_Pg deletion and dysregulated c-di-AMP signaling (Supplemental Fig. 1E-L). This pattern of metabolic shift suggests that, in P. gingivalis, the atypical PDE_Pg acts not only through its auxiliary role in c-di-AMP synthesis but also as a growth-stage nutrition/metabolic gatekeeper.

Differential transcriptomics of the ∆pde_pg mutant reveal coordinated shifts in redox metabolism, envelope homeostasis, and DNA replication

To gain mechanistic insight into how PDE_Pg-dependent c-di-AMP homeostasis shapes cellular physiology, we performed RNA sequencing in the Δpde_Pg mutant and compared its transcriptome with that of WT. Differential gene expression analysis (log₂FC > 1, q < 0.01), followed by pathway enrichment, identified extensive transcriptional remodeling in the absence of pde_Pg (Fig. 3C, Supplemental Fig. 3, Supplemental Table 4). Consistent with metabolomics and LC/MS data showing ATP accumulation, c-di-AMP depletion, and perturbed redox balance, the transcriptional response is consistent with a broad compensatory reprogramming of metabolic, redox, and replication pathways.

Energy and redox metabolism are extensively remodeled in the Δpde_Pg mutant

Deletion of pde_Pg (no detectable expression; log₂FC = -8.51, q = 0, Supplemental Table 4) triggered significant upregulation of genes associated with corrinoid metabolism, iron uptake, quinone biosynthesis, and CoA-linked energy flux, modules that together define a coordinated redox-rebalancing program.

Among the most upregulated genes were PGN_0456/PGN_0457, encoding the B₁₂-dependent methylmalonyl-CoA mutase (MCM), and PGN_0975, annotated as a MenG/UbiE-like demethylmenaquinone methyltransferase. MCM catalyzes the conversion of methylmalonyl-CoA to succinyl-CoA, providing an anaplerotic entry point into the TCA cycle, while MenG supports anaerobic quinone-dependent respiration (Fig. 3C, Supplemental Fig. 3, Supplemental Table 4). Their simultaneous upregulation suggests a compensatory rerouting of propionyl-CoA and branched-chain amino acid catabolism toward succinyl-CoA, a hallmark of redox-stressed states seeking to restore NAD⁺/NADH and redox balance.

Upregulation of PGN_1309 (FeoB-1; log₂ FC = + 1.10, q < 3.4 × 10⁻²⁷) further supports an increased demand for iron cofactors in redox and respiratory enzymes. In P. gingivalis, FeoB-mediated Fe²⁺ uptake underpins the activity of heme- and flavin-linked oxidoreductases essential for anaerobic energy metabolism. Additionally, our data shows the upregulation of PGN_1907 (TonB/BtuB-like by modeling) (Log₂ FC > 1.0; q < 3.49 × 10⁻⁶) and PGN_1221 (ATP: corrinoid adenosyltransferase), reinforcing the transcriptional evidence for enhanced cofactor turnover and cobalamin salvage (Supplemental Table 4).

Significant upregulation of PGN_1973 (phosphoglycerate mutase, PGM or GpmA) is also observed in the absence of PDE_Pg (Log₂ FC > 1.0; q < 6.77 × 10⁻⁵) (Fig. 3C, Supplemental Fig. 3, Supplemental Table 4). GpmA catalyzes the reversible interconversion of 3-phosphoglycerate (3-PGA) and 2-phosphoglycerate (2-PGA) in glycolysis and gluconeogenesis. This response may reflect an adaptive attempt to rebalance the intracellular redox state by enhancing substrate-level phosphorylation while avoiding excessive accumulation of reductants. In general, the metal-independent PGM functions as a metabolic “safeguard pathway” allowing bacteria to bypass Mn-dependence and maintain glycolysis when the host actively withholds manganese during infection [54]. While this has been shown in Staphylococcus aureus and Salmonella [54], it offers a parsimonious framework for P. gingivalis to sustain minimal energy flow and redox equilibrium under conditions of signaling dysfunction, nutrient limitation, and stress.

Collectively, our analyses of these transcriptional programs indicate that the Δpde_Pg mutant responds to energy/redox stress by activating alternative metabolic entry points, enhancing cofactor turnover, and rebalancing redox potential, consistent with the metabolomic signature of elevated ATP and defective peptide uptake in the mutant.

Envelope remodeling reflects adaptive stress and nutrient uptake defects

Genes associated with nucleotide-sugar metabolism and LPS/glycolipid modification were significantly upregulated in the ∆pde_Pg mutant. These include PGN_0518 (putative ribulose-phosphate 3-epimerase; racemase/epimerase activity; Log₂ FC = 1.16; q = 6.60 × 10⁻²⁰) and PGN_0547 (putative dTDP-4-dehydrorhamnose 3,5-epimerase; Log2 FC = 1.00; q = 9.00 × 10⁻⁷) (Supplemental Fig. 3, Supplemental Table 4). These enzymes regulate the availability of pentose-phosphate and dTDP-deoxyhexose intermediates, which serve as critical precursors for cell-surface glycans and LPS sugar moieties. Additional upregulated genes in this module in our data are PGN_0850 and PGN_1627 (homologously classified as ArnT-family-like) that fit the same theme of LPS or glycolipid modifications [55]. In the absence of PDE_Pg, P. gingivalis appears to reprogram these metabolic fluxes to support envelope and LPS remodeling under stress, a process that, in P. gingivalis and in other bacteria, is known to be tightly regulated by c-di-AMP [6, 10, 56].

Replication and proteostasis modules reflect a checkpoint response to energetic imbalance

Among the most significantly induced transcripts in the ∆pde_Pg mutant were two adjacent loci, PGN_1620 and PGN_1621 (Log₂ FC = 3.25, 3.25; q = 1.70 × 10⁻⁶, 3.64 × 10⁻¹¹), as well as the gene pair PGN_0369/PGN_0370 (Log₂ FC > 1.0, q < 10^⁻12) (Fig. 3C, Supplemental Fig. 3, Supplemental Table 4). PGN_1620 and PGN_1621 encode short peptides (32 and 78 amino acids, respectively) that remain uncharacterized in the AlphaFold Protein Structure Database [37]. Secondary structure modeling with Phyre2/AlphaThread [32] predicted PGN_1621 as a potential single-stranded DNA (ssDNA)-binding protein, though with low confidence (~ 18%). Similarly, PGN_0369 is predicted to function as a single-stranded DNA-binding protein, while PGN_0370 shows strong structural homology to hemolysin-like proteins harboring cystathionine β-synthase (CBS) domains (AlphaFold pLDDT > 90%). CBS domains are ~ 60-amino acid regulatory modules widely distributed across bacteria, archaea, and eukaryotes. They act as intracellular sensors of adenosine-containing metabolites (ATP, AMP, FAD, and S-adenosylmethionine), allowing proteins to adjust activity in response to energy or redox fluctuations [57–59]. Whether these CBS domains in P. gingivalis directly bind c-di-AMP remains unknown, but the possibility is intriguing given the role of c-di-AMP in nucleotide signaling. Proteins of this type are thought to integrate virulence functions (e.g., hemolysin activity or membrane perturbation) with metabolic regulation, fine-tuning pathogenic potential in accordance with cellular energy state [60].

In addition, PGN_0516, encoding a putative Hsp70 DnaK suppressor protein (Log₂ FC = + 2.28, q = 1.06 × 10⁻²⁷), was among the most significantly upregulated genes (Fig. 3C, Supplemental Table 4). Hsp70 DnaK is a central chaperone for protein folding and proteostasis, and its suppression can activate alternative stress-response pathways [61]. The strong induction of PGN_0516 suggests that deletion of PDE_Pg heightens proteotoxic stress in the ∆pde_Pg mutant, requiring compensatory adjustment of the chaperone network. This aligns with the altered redox and metabolic environment inferred from other upregulated pathways, as protein folding is particularly sensitive to redox imbalance and ATP availability.

Together with the significant downregulation of PGN_1070, annotated as a radical-SAM/Fe-S oxidoreductase-like enzyme, the transcriptional pattern points to a reorganization of DNA replication/repair processes, proteostasis, and redox enzyme prioritization in the absence of PDE_Pg. Such adjustments are consistent with the broader roles of c-di-AMP in coordinating DNA integrity, protein homeostasis, and stress management documented in Firmicutes [62–66], and suggest that P. gingivalis relies on PDE_Pg-dependent c-di-AMP signaling to balance replication, folding stress, and energy/redox homeostasis during growth transitions.

Conversely, PGN_2000, encoding a DnaB-like replicative helicase, was downregulated in the absence of PDE_Pg, indicating potential replication fork slowdown or impaired initiation (Fig. 3C, Supplemental Table 4). DnaB is a ring-shaped homohexameric ATPase that plays an indispensable role in bacterial DNA replication. By providing continuous ssDNA substrates, DnaB ensures efficient progression of replication machinery, linking ATP-dependent helicase activity to genome duplication and cell cycle progression [67, 68]. This transcriptomic shift connects PDE_Pg and c-di-AMP signaling to a checkpoint-like transcriptional state that controls replication and growth, a molecular correlate of the extended lag phase observed phenotypically in the mutant. In other bacteria, c-di-AMP has been shown to influence DNA integrity checkpoints, replication fork stability, and stress adaptation [1, 2, 63, 69, 70].

Transcriptional fine-tuning via small ORFs (sORFs) and regulatory factors

Significant downregulation of PGN_0970 (ECF σ subfamily) in the mutant suggests reprioritization of extracytoplasmic stress regulons. Moreover, significant downregulation of several sORFs (PGN_0854, PGN_1322, PGN_1720) track with nearby DNA repair/protein-folding loci in our dataset (e.g., adjacency to mutY for base-excision repair) (Supplemental Table 4). Recent ribosome profiling studies [71] have underscored that sORFs are widely translated in prokaryotes, including in P. gingivalis [20, 72], and can encode short peptides with important regulatory or structural roles. In our dataset, at least eight sORFs were differentially expressed between the ∆pde_Pg mutant and WT. Many of these sORFs flank loci involved in DNA repair or protein secretion, implying that they contribute to envelope and replication homeostasis during energy and redox imbalances.

Proteomic remodeling in Δpde_Pg reveals energy-redox compensation, envelope adaptation, and replication stress

To define how the loss of PDE_Pg-dependent c-di-AMP signaling reshapes the proteome, we performed quantitative LC-MS/MS-based proteomics comparing Δpde_Pg and WT cells. Significantly altered proteins (p < 0.05, Student’s t-test) were classified by functional ontology. The overall proteomic landscape mirrored metabolomic and transcriptomic trends, revealing coordinated remodeling across energy/redox metabolism, envelope structure, and replication-proteostasis control (Fig. 3D, Supplemental Table 5). These changes align with the biochemical evidence of elevated ATP, reduced c-di-AMP, defective peptide uptake, and an altered redox state, supporting a role for PDE_Pg as a key integrator of metabolic flux and growth-phase progression.

Energy and redox metabolism are rewired to buffer ATP and redox imbalance in the ∆pde_Pg mutant

Deletion of PDE_Pg induced a broad compensatory response in redox and bioenergetic enzymes. Levels of proteins supporting anaerobic electron transfer, NAD⁺/NADH balance, and thiol-disulfide exchange were markedly elevated, including flavodoxin (PGN_1209; Δ = +1.40, Student’s t-test, mutant vs. WT), nicotinate-nucleotide adenylyltransferase NaMN-ATase/NadD (PGN_2007; Δ = +1.15), pyridine nucleotide-disulfide oxidoreductase (PGN_0285; Δ = +0.82), and acyl-CoA dehydrogenase CaiA (PGN_0799; Δ = +0.80) together with electron transfer flavoprotein (ETF) FixA (PGN_0801; Δ = +0.54) (Fig. 3D, Supplemental Table 5). Moreover, concomitant increase in the abundance of PGN_1738 (putative long-chain-fatty-acid-CoA ligase; Δ = +1.00, q = 0) suggests lipid/acyl-flux tuning that often accompanies shifts in electron transfer systems (Fig. 3D, Supplemental Table 5).

Flavodoxin is a well-characterized soluble FMN-binding electron carrier that can substitute for ferredoxin under iron-limited or oxidative stress, thereby supporting low-potential redox reactions in anaerobic environments [73–75]. Elevated flavodoxin levels are widely recognized as a biomarker of iron deficiency in unicellular organisms [76]. In the ∆pde_Pg mutant, the observed increase in flavodoxin abundance, together with RNA-seq evidence of FeoB-1 upregulation, strongly suggests activation of an iron-homeostasis process to stabilize redox homeostasis under altered physiological conditions. NaMN-ATase catalyzes adenylation of NaMN to nicotinic acid adenine dinucleotide (NaAD) in NAD biosynthesis, indicating reinforcement of the NAD pool to support redox metabolism [77]. Similarly, the increase in pyridine nucleotide–disulfide oxidoreductase (PGN_0285) suggests enhanced thiol/disulfide redox cycling, a pathway linked to antioxidant defense, DNA synthesis, and protein folding. Such enzymes are known to sustain glutathione- and NAD(P)H-dependent redox balance, ensuring continuity of metabolic and replication processes during stress [78, 79].

The increased abundance of putative acyl-CoA dehydrogenase (PGN_0799) and FixA in the ∆pde_Pg mutant suggests rerouting of electrons through β-oxidation and ETF-linked dehydrogenase systems-metabolic circuits that generate reducing power while maintaining redox poise when primary energy coupling is dysregulated. This coupling is well documented in both aerobic and anaerobic bacteria, where ETF systems act as metabolic hubs balancing lipid catabolism with ATP generation [80–82].

Together, these findings indicate that the Δpde_Pg mutant engages a redox-compensation network to manage excessive ATP and restore electron balance. This adaptive expansion of flavodoxin-NAD-ETF systems enables basal energy turnover under conditions of inefficient ATP utilization, aligning with metabolomic evidence of high intracellular, dysregulated c-di-AMP signaling, and disrupted redox coupling.

Envelope and secretion systems undergo stress-responsive remodeling

Consistent with impaired peptide uptake and redox/osmotic stress observed metabolically, Δpde_Pg displayed major remodeling of envelope-related proteins. A probable glycoprotease (PGN_0802; Δ = +2.23) was strongly induced, while multiple structural and secretion components were downregulated, including MurC (PGN_0628; Δ = -1.83), β-barrel OMP (PGN_0156; Δ = -1.09), lipoprotein (PGN_1535; Δ = -1.16), and HagA adhesin (PGN_1733; Δ = -1.03), along with core Type IX Secretion System (T9SS) elements PorN, PorM, PorW, and Sov (Δ ≈ -0.5 to -0.7).

This pattern signifies a strategic energy-conservation mode, reducing de novo peptidoglycan and outer-membrane scaffolding while upregulating proteolytic turnover to remodel existing surface components. Upregulation of PGN_0802 (glycoprotease) reflects enhanced surface protein processing to compensate for impaired T9SS export and peptide transport, thereby maintaining envelope integrity under stress.

Additionally, the decline of PGN_1783 (DUF4835) and PGN_1642 (PapD-like chaperone) implicates weakened envelope-replication coupling, consistent with their genomic proximity to dnaA (PGN_1786), dnaQ (PGN_1785), and pantothenate/CoA biosynthetic genes and pathways, which are closely tied to energy/redox metabolism. These adjustments further support the metabolomic and transcriptomic evidence that Δpde_Pg cells experience envelope stress and compromised peptide assimilation, leading to compensatory remodeling that is metabolically and energetically conservative.

Nucleotide metabolism and replication initiation are constrained in the ∆pde_Pg mutant

Proteomic downregulation of guanylate kinase (Gmk; PGN_1460; Δ = -1.09) directly links PDE_Pg-dependent c-di-AMP dysregulation to reduced GTP/dGTP availability. Since GTP fuels ribosome activation, rRNA synthesis, FtsZ polymerization, and DNA replication [83, 84], reductions in Gmk abundance likely contribute to the delayed initiation of exponential growth. This is consistent with RNA-seq evidence of DnaB helicase suppression and upregulation of primosomal/ssDNA-binding modules (PGN_0369–0370, PGN_1620–1621), collectively suggesting that the mutant operates under a replication checkpoint-like state driven by nucleotide imbalance and energy excess. Additionally, a set of replication-repair and protein quality-control factors were differentially expressed, highlighting a shift in proteostasis strategy. PriA-like protein (PGN_1979; Δ = +1.07) was elevated, reflecting activation of replication-restart mechanisms that rescue stalled forks under energy/redox stress. In contrast, ClpX (PGN_1550; Δ = -1.31), the ATPase component of the ClpXP protease, was markedly reduced, indicating suppression of ATP-dependent proteolysis. Reduced ClpX together with transcriptional upregulation of DnaK-suppressor (PGN_0516) suggests that Δpde_Pg cells shift from canonical ATP-driven proteolysis to energy-sparing chaperone-mediated folding systems to buffer protein damage. This transition complements the mutant’s high ATP yet low c-di-AMP condition, in which excess ATP does not translate into efficient energy utilization.

Proposed integrated model of PDE_Pg function from multi-omics data

Multi-omics integration suggests that deletion of PDE_Pg disrupts c-di-AMP-mediated coordination of energy metabolism, ATP/redox balance, and cell-cycle control. The Δpde_Pg mutant exhibits excessive ATP accumulation, reduced c-di-AMP levels, and impaired peptide uptake, triggering compensatory rewiring across metabolic, envelope, and replication systems. Transcriptomic and proteomic data indicate rerouting of central carbon and corrinoid-dependent flux (via MCM and related transporters), along with reinforcement of NAD⁺/NADH and thiol-redox buffering through elevated flavodoxin, NaMN-ATase, and ETF-linked dehydrogenases. Concurrently, envelope and secretion systems shift toward energy conservation, with reduced MurC, OMPs, and T9SS components but increased proteolytic processing (PGN_0802). Replication and proteostasis modules exhibit a checkpoint-like pattern, with upregulation of primosome/ssDNA-binding and DnaK-suppressor proteins, concurrent with ClpX and Gmk downregulation, linking redox and nucleotide imbalance to delayed DNA replication and division. Collectively, these data define PDE_Pg as a key integrator that couples c-di-AMP signaling with ATP/redox homeostasis, nutrient assimilation, and replication timing in P. gingivalis. Its loss decouples energy generation from utilization, forcing a redox-compensatory, energy-conserving state that underlies the prolonged lag-to-log transition phenotype and growth defect.

Representative posttranslational modification (PTM) profiling highlights the role of PDE_Pg in controlling DAC_Pg phosphorylation and selective remodeling of replication and envelope-associated proteins

To extend the multi-omics framework linking PDE_Pg to c-di-AMP-dependent metabolic and replication control, we next compared the PTM landscape between WT and Δpde_Pg mutant proteomes. LC-MS/MS mapping revealed that loss of PDE_Pg causes selective remodeling of phosphorylation and acetylation patterns, consistent with the integrative transcriptomic and proteomic signatures of replication stress, translation adaptation, and envelope remodeling (Fig. 4A, Supplemental Table 6).

Fig. 4.

Posttranslational modification landscape in P. gingivalis identifies PDE_Pgand DAC_PgSer-226 as components of a candidate regulatory circuit associated with phasic c-di-AMP signaling and growth-phase transitions. A Posttranslational modification (PTM) analysis by LC-MS/MS (label-free quantification; Andromeda/MaxQuant search against the P. gingivalis Uniprot database) was used to map phosphorylation (P) and acetylation (Ac) events. The number of modification events per unique protein is shown for each strain, with significant decreases in the Δpde_Pg mutant indicated by black arrows. WT exhibited a balanced phosphorylation distribution across Ser/Thr/Tyr residues (11/7/10) sites, including unique DAC_Pg Ser-226), and a diverse acetylome (12 proteins). By contrast, Δpde_Pg showed a Thr-enriched, Tyr-deficient phosphorylation pattern (19 proteins), and a limited acetylome (4 proteins). B Sequence alignment (Clustal Omega) between DAC_Pg and DAC homologues from other oral microbiome species. Asterisks mark the SEE stretch involved in catalysis, and arrows highlight DAC_Pg-specific PTM site in the mutant as Ser-226 (phosphorylated in Δpde_Pg). Together, these findings indicate that disruption of PDE_Pg is associated with altered posttranslational regulatory landscapes and identify DAC_Pg Ser-226 as a candidate adaptive phosphoregulatory site linked to phasic c-di-AMP dynamics and growth-phase-associated physiological remodeling. Ser: serine, Thr: threonine, Tyr: tyrosine, Met: methionine, Asn: asparagine; Gln: glutamine, Lys: lysine, Cys: cysteine; Pg: P. gingivalis, Pi: Prevotella intermedia, Tf: Tannerella forsythia, Td: Treponema denticola, Vp: Veillonella parvula, So: Streptococcus oralis, Ss: S. sanguinis, Sg: S. gordonii, Fa: Filifactor alocis. We next assessed the effects of Ser-226 alanine substitution on growth rates (top panels) and intracellular c-di-AMP/ATP pools (bottom panels) of P. gingivalis WT and Δpde_Pg, with and without pantothenate supplementation. To rigorously assess growth kinetics, non-linear regression was applied using a logistic growth model (GraphPad Prism). The growth rate constant (k) reflects the proliferation rate for each condition. C Compared to WT strain, WT_{(DAC, S226A)} mutant exhibited impaired early-phase depletion and failed to replenish c-di-AMP during exponential growth, resulting in progressive decline across phases. D ∆pde_{Pg(DAC, S226A)} mutant showed partial rescue, with shortened lag phase and c-di-AMP cycling patterns resembling WT_{(DAC, S226A)}, though lacking early-phase depletion. Together, these results identify Ser-226 of DAC_Pg as a regulatory resistor hotspot whose interplay with PDE activity establishes the transient depletion-replenishment cycles required for optimal growth-phase transitions. E Pantothenate-treated WT cells showed comparable early growth with the untreated condition but diverged at the exponential phase, with attenuated c-di-AMP replenishment and reduced final biomass (~ 0.7 vs. 1.5 OD₆₀₀). WT_{(DAC, S226A)} displayed impaired c-di-AMP depletion between phases III and IV under untreated conditions, resulting in impaired exponential growth. Pantothenate treatment further disrupted c-di-AMP cycling, leading to an initial accumulation followed by sharp depletion and significant growth failure. F Δpde_Pg mutant showed a prolonged lag phase and dysregulated c-di-AMP cycling; pantothenate treatment improved early-phase growth and partially normalized c-di-AMP dynamics by tempering excessive accumulation. ∆pde_{Pg(DAC, S226A)} strain displayed similar dysregulated c-di-AMP profiles and growth in both treated and untreated conditions, indicating that pantothenate responsiveness is lost when Ser-226 is substituted. Together, these results establish PDE_Pg as a redox coupler linking energy/redox flux to c-di-AMP synthesis and Ser-226 of DAC_Pg as a resistor-like modulator that stabilizes synthesis timing, forming a two-node circuit that coordinates signaling, ATP turnover, and redox homeostasis during growth-phase transitions. Data represent mean ± SD (n = 3)

WT exhibited a balanced phosphorylation profile across serine (Ser), threonine (Thr), and tyrosine (Tyr) residues, with comparable numbers of modified sites (11 Ser, 7 Thr, and 10 Tyr; 28 total) (Fig. 4A, Supplemental Table 6). This distribution aligns with the diverse activities of bacterial kinases that govern envelope biogenesis, secretion systems, transporter activity, and central metabolism [85, 86]. In contrast, the Δpde_Pg mutant displayed a marked redistribution of phosphorylated proteins, with site-specific changes most evident as a reduction in Tyr phosphorylation (5 vs. 10 in WT) and selective Thr enrichment. Strikingly, diadenylate cyclase (DAC_Pg) was uniquely phosphorylated at Ser-226 (Fig. 4A-B, Supplemental Table 6). Ser-226 resides within the highly conserved “SEE” catalytic motif that defines the active center of DAC enzymes (Fig. 4B). Its phosphorylation in the Δpde_Pg background identifies this residue as a key regulatory hotspot, directly linking PDE_Pg to phospho-modulation of the cyclase and dynamic control of c-di-AMP synthesis. This modification provides a mechanistic node connecting altered signaling flux with replication timing and growth synchronization. Beyond DAC_Pg, additional phosphorylation targets mapped to pathways governing DNA replication and translation (e.g., XseA, HD-domain helicase PGN_1964, AspS, and ribosomal protein uL1) as well as envelope remodeling and glycan modification (PGN_0345, PGN_1493, ManB, and PGN_1725). Phosphorylation of surface and virulence-associated proteins, including TraK and FimE, further indicates remodeling of adhesion and conjugation systems. Redox-linked enzymes, such as succinyl-CoA:3-ketoacid-CoA transferase and magnesium chelatase, were also modified, consistent with altered metabolic coupling. Together, these findings reveal a broad phospho-signaling rewiring in Δpde_Pg mutant encompassing genome maintenance, envelope adaptation, and redox metabolism. The transition from a balanced Ser/Thr/Tyr profile in WT to a Thr-enriched, Tyr-deficient pattern in the mutant highlights the significance of PDE_Pg in providing a permissive condition for selective phosphorylations, which are critical for replication, metabolism, and cell envelope stability, an adaptive strategy to sustain growth [87–89].

Acetyl-lysine mapping revealed condition-specific acetylomes. WT carried a broader set of acetylated proteins (12 vs. 4 in Δpde_Pg) enriched in envelope/signaling and virulence functions (OmpH; histidine kinase; RgpA/B), redox/metabolism (FAD-binding FR-type), protein folding (PPIase), and transcriptional machinery (RNA polymerase subunit) (Fig. 4A, Supplemental Table 6). In contrast, Δpde_Pg displayed a smaller, more focused acetylome centered on nucleotide metabolism and cell-division/translation nodes, notably 5’-nucleotidase/2′,3′-cyclic-nucleotide 2′-phosphodiesterase (PGN_0282), a SPOR-domain septal PG-binder (PGN_1498), and rRNA methyltransferase G (PGN_1096), with a GLUG-domain protein (PGN_2080). These findings suggest that PDE_Pg influences targeted control of cell division, ribosome biogenesis, and nucleotide flux (Fig. 4A, Supplemental Table 6).

Collectively, the deletion of PDE_Pg is associated with broad PTM reprogramming, with selective phosphorylation and acetylation changes reallocating regulatory control toward genome maintenance, translation, and envelope stabilization. Phosphorylation of DAC_Pg at Ser-226 emerges as a central molecular signature of this adaptation, defining a compensatory “PTM-buffered” state that sustains survival but delays growth-phase progression. Together with omics datasets, these results position PDE_Pg as a key integrator of c-di-AMP signaling, energy/redox balance, and posttranslational control mechanisms that coordinate replication and envelop biogenesis during phasic growth transitions.

Sustaining rhythmic-phasic c-di-AMP signaling requires coordinated action of PDE_Pg and Ser-226 of DAC_Pg to couple signaling with the cellular bioenergetic/redox state

In this study, we demonstrated PDE_Pg drives rhythmic-phasic c-di-AMP signaling and classical growth phase transitions by P. gingivalis (Fig. 1). Since deletion of pde_Pg uniquely leads to phosphorylation of DAC_Pg at Ser-226 (Fig. 4A-B), we hypothesized that c-di-AMP synthesis is governed by a dynamic interplay between Ser-226 in DAC_Pg and PDE_Pg that generates transient, phase-specific decline-replenishment cycles essential for growth synchronization. To test this hypothesis, we introduced an alanine substitution at Ser-226 in DAC_Pg (S226A) in both the WT (WT_{DAC (S226A)}) and Δpde_Pg (Δpde_{Pg,DAC (S226A)}) backgrounds, followed by assessment of growth rates using non-linear regression and a logistic growth model and temporal quantification of cellular c-di-AMP and ATP pools as explained above. Replacement of Ser-226 with alanine in WT DAC_Pg significantly altered the kinetics of growth-phase transition (Fig. 4C, Supplemental Table 7). The WT_DAC(S226A) strain exhibited a reduced growth rate constant (k = 0.165) compared with WT (k = 0.227), as well as a delayed inflection point (Xint = 6.1 h vs. 4.4 h), indicating slower entry into the exponential phase. The non-overlapping confidence intervals for k (WT: 0.184–0.280; WT_DAC(S226A): 0.154–0.177) confirm that this reduction is statistically robust. Importantly, the decreased slope of the logistic curve means that WT cells divide more rapidly, while WT_DAC(S226A) progresses through exponential growth at a markedly slower pace. Consistent with this impaired transition, intracellular c-di-AMP levels in WT_DAC(S226A) declined only modestly (~ 1.3-fold) during early growth, in contrast to the pronounced ~ 2.4-fold reduction observed in WT. This attenuated decline further reinforces the notion that proper, phasic reduction of c-di-AMP is essential for triggering the metabolic and regulatory switches required for efficient progression through the growth phase. Strikingly, the substitution of Ser-226 also impaired the ability of DAC_Pg to replenish the c-di-AMP pool during subsequent growth (Fig. 4C). Instead of the robust resynthesis observed in WT, the WT_{DAC (S226A)} strain exhibited a progressive depletion of c-di-AMP, with reductions of ~ 4.3-fold and ~ 6.5-fold at 18 and 44 h, respectively. Concomitantly, ATP dynamics were markedly altered in WT_DAC(S226A) (Fig. 4C). While WT cells rapidly consumed their ATP reservoir in coordination with c-di-AMP dynamics (139 → 48 → 9 nM/mg protein at 0, 6, and 18 h, respectively), WT_DAC(S226A) retained disproportionately high ATP levels across the same intervals (179 → 140 → 32 nM/mg protein). At hours 6 and 18, ATP pools were ~ 3-fold and ~ 3.6-fold higher in WT_DAC(S226A) compared to WT, respectively.

Remarkably, replacement of Ser-226 with alanine in DAC of the Δpde_Pg background markedly shortened the prolonged lag phase otherwise observed in the Δpde_Pg mutant carrying the native DAC, thereby significantly improving the early growth-phase transition (Fig. 4D, Supplemental Table 7). Logistic regression analysis confirmed that, in contrast to the Δpde_Pg mutant, the Δpde_{Pg,DAC(S226A)} strain displayed a substantially higher growth rate (k = 0.161, 95% CI: 0.146–0.178) and a markedly earlier inflection point (Xint = 6.22 h), closely matching the kinetics of the WT_DAC(S226A) strain. In addition, the carrying capacity (YM = 1.53) of Δpde_{Pg,DAC(S226A)} exceeded that of Δpde_Pg (YM = 1.41), underscoring that the Ser-226 substitution partially rescues the growth defect of Δpde_Pg by restoring timely entry into exponential proliferation and enhancing overall growth efficiency. Consistently, the phasic c-di-AMP cycling pattern of Δpde_{Pg,DAC(S226A)} closely resembled that of WT_DAC(S226A), with Δpde_{Pg,DAC(S226A)} strain failing to execute early-phase decline but instead exhibiting a modest increase followed by progressive decline across later phases (Fig. 4D). The slight early increase likely contributed to the extended lag phase of Δpde_{Pg,DAC(S226A)} compared with WT_{DAC (S226A)}.

The Δpde_{Pg,DAC(S226A)} strain showed a distinctly different ATP profile: ATP was accumulated excessively by hour 6 (480 nM/mg protein, ~ 1.7-fold higher than Δpde_Pg), before sharply dropping to ~ 33 nM/mg protein at hour 18 (~ 5.3-fold lower than Δpde_Pg) (Figs. 1 and 4D). These dynamics indicate that the Ser-226 substitution not only alters c-di-AMP dynamics but also profoundly reshapes cellular energy handling, producing an initial delay in ATP accumulation followed by significant depletion during exponential growth.

Together, these results demonstrate that Ser-226 functions as a critical regulatory node within DAC_Pg, tuning c-di-AMP synthesis to the cell’s metabolic state. Substitution of Ser-226 with alanine partially compensates for the absence of PDE_Pg by dampening or desensitizing DAC activity, thereby restoring a c-di-AMP decline sufficient to shorten the prolonged lag phase and improve early growth-phase transitions in Δpde_Pg. However, because the S226A variant cannot generate the precise early-phase depletion pulse normally produced by PDE_Pg, the rescue is incomplete. Thus, Ser-226 acts as a modulatory site that adjusts late-phase replenishment and metabolic rebound, whereas PDE_Pg appears to provide the indispensable early-phase depletion cue. Together, these two regulatory elements form a coupled circuit that coordinates rhythmic-phasic c-di-AMP dynamics, ATP utilization, and redox balance to ensure properly synchronized growth-phase transitions in P. gingivalis.

To further probe the mechanistic coupling between signaling and redox/metabolism, we supplemented cultures with pantothenate, an essential precursor for CoA, a key metabolic currency that links acyl transfer, energy flux, and redox balance [90], and evaluated its impact on the dynamics of phasic c-di-AMP cycling. Indeed, the genomic clustering of c-di-AMP regulatory genes with those involved in CoA metabolism, together with our multi-omics data showing reduced pantothenate production in Δpde_Pg, strongly supports a direct coupling between disrupted c-di-AMP dynamics and impaired metabolic/redox homeostasis (Fig. 3A and D-E).

Growth rate analysis revealed that pantothenate-treated WT cells exhibited a comparable or slightly faster growth rate constant (k = 0.382, 95% CI: 0.291–0.505) relative to untreated WT controls (k = 0.227, 95% CI: 0.184–0.280), along with an earlier inflection point (Xint = 2.62 h vs. 4.40 h), indicating that pantothenate supplementation accelerated the lag-to-log transition (Fig. 4E, Supplemental Table 7). However, despite entering exponential growth earlier, pantothenate ultimately restricted sustained proliferation from mid-exponential phase, reducing the carrying capacity nearly two-fold (YM = 0.799, 95% CI: 0.770–0.828; OD₆₀₀: ~0.7) compared with untreated cells (YM = 1.441, 95% CI: 1.400-1.486; OD₆₀₀: ~1.5). During the first 6 h, both treated and untreated WT cultures reduced their c-di-AMP pool by ~ 2.4-fold. Between 6 and 18 h, both groups replenished their c-di-AMP pool, although pantothenate-treated cells did so slightly more slowly (3.3-fold vs. 3.8-fold) (Fig. 4E). Between 18 and 44 h, treated cells exhibited an altered phasic trajectory, reducing the c-di-AMP pool by only ~ 1.25-fold compared with ~ 2-fold in untreated cells.

In the context of Ser-226 substitution, pantothenate supplementation profoundly altered growth and c-di-AMP dynamics in the WT_DAC(S226A) strain (Fig. 4E). Treated cultures no longer conformed to logistic growth (R² = 0.327 vs. 0.993 untreated), and parameter estimates were unreliable, indicating collapse of normal proliferation. Unlike the modest early reduction seen in untreated WT_DAC(S226A), pantothenate-treated cells showed an initial 1.2-fold increase in c-di-AMP, followed by a sharp 5.3-fold decline at 18 h that persisted through 44 h. This temporally incoherent cycling mirrors the growth failure and indicates that pantothenate exacerbates the instability of the Ser-226 variant. These findings support the conclusion that Ser-226 functions as a regulatory resistor, buffering DAC activity against redox imbalance and maintaining coherent c-di-AMP flux.

ATP measurements further clarified these effects. In WT, pantothenate increased ATP levels modestly at 6 h (47.88 → 63.71 nM/mg protein) and more substantially by 18 h (8.96 → 26.69 nM/mg protein), indicating enhanced CoA-dependent energy mobilization while preserving the characteristic WT decline during log-phase growth. Pantothenate similarly increased ATP in WT_DAC(S226A) (6 h: 140.49 → 160.17 nM/mg protein; 18 h: 32.01 → 51.25 nM/mg protein), magnifying the elevated ATP state associated with the S226A substitution. These patterns reinforce the notion that the Ser-226 variant elevates ATP utilization thresholds, consistent with a DAC enzyme desensitized to redox-dependent feedback.

In the Δpde_Pg background, pantothenate treatment markedly improved its prolonged lag-to-log phase and overall growth rate (Fig. 4F). Treated Δpde_Pg cells exhibited a higher growth rate (k = 0.141, 95% CI: 0.141–0.169) and a markedly earlier inflection point (Xint = 6.50 h) compared with the untreated condition, indicating a restoration of timely entry into exponential proliferation. The carrying capacity was also slightly higher in the treated mutant (YM = 1.46 vs. 1.41 in the untreated), further underscoring improved growth efficiency. Although Δpde_Pg retained a restricted, low-amplitude c-di-AMP trajectory in both conditions, pantothenate treatment shifted the timing of c-di-AMP decline to occur earlier within this narrow dynamic range. Whereas untreated Δpde_Pg cells accumulated c-di-AMP modestly (1.17-fold at 18 h), pantothenate-treated cells displayed an early increase followed by a progressive 1.7-fold decline by 44 h, aligning more closely with WT temporal patterns (Fig. 4F). Importantly, pantothenate produced a dramatic correction of ATP metabolism in Δpde_Pg. Untreated mutant cells accumulated ATP excessively (6 h: 284.27; 18 h: 170.88 nM/mg protein), while pantothenate supplementation sharply lowered ATP to 119.36 nM/mg protein at 6 h and 46.38 nM/mg protein at 18 h, a ~ 2.4-3.7-fold reduction (Fig. 4F). This shift toward a WT-like ATP trajectory coincides with improved growth kinetics and supports the conclusion that pantothenate partly compensates for the loss of PDE_Pg by restoring CoA-dependent energy flux and improving the coherence of c-di-AMP cycling. These data further indicate that an early decline in both ATP and c-di-AMP serves as a key metabolic signal enabling efficient growth-phase transitions, highlighting PDE_Pg as the primary redox-signal coupler linking energy/redox status to ATP homeostasis, c-di-AMP synthesis, and the cell cycle.

In contrast, pantothenate had little effect on the Δpde_{Pg,DAC(S226A)} strain. Growth rate, inflection point, and carrying capacity remained nearly identical in treated and untreated cultures (Fig. 4F). Untreated Δpde_{Pg,DAC(S226A)} cells grew with a rate constant of k = 0.161 (95% CI: 0.146–0.178) and an inflection point of Xint = 6.22 h, while treated cultures exhibited a nearly identical growth rate (k = 0.144, 95% CI: 0.117–0.178) and inflection point (Xint = 6.93 h). Both conditions showed markedly better growth than Δpde_Pg carrying native DAC, confirming the compensatory role of Ser-226 substitution in restoring growth-phase transitions (Fig. 4F). ATP dynamics, however, differed sharply. Pantothenate-treated Δpde_{Pg,DAC(S226A)} cells exhibited extremely high ATP at 6 h (480.3 → 646.32 nM/mg protein), followed by a dramatic ~ 13-fold drop at 18 h (646.3 → 51.7 nM/mg protein), compared with only a 2.6-fold reduction in Δpde_Pg (Fig. 4F). Although this restored a WT-like ATP nadir at 18 h, pantothenate did not improve the growth of this strain, indicating that Ser-226 and pantothenate converge on the same regulatory axis and are therefore not additive.

Together, these results show that pantothenate can partially normalize c-di-AMP dynamics and improve growth when PDE_Pg is absent, but the amplitude and nature of its effects depend critically on the Ser-226 regulatory node. The combined ATP and c-di-AMP datasets define PDE_Pg as the primary redox-signal coupler and Ser-226 of DAC_Pg as a redox-resistant tuning site, forming a two-node feedback circuit that maintains rhythmic-phasic c-di-AMP dynamics, optimal ATP flux, and synchronized growth-phase transitions.

Transmission electron microscopy (TEM) revealed marked ultrastructural differences between WT and ∆pde_pg cells

Given the paradoxical role of PDE_Pg, where its deletion unexpectedly elevates intracellular ATP levels, reduces rather than increases c-di-AMP concentrations, disrupts peptide uptake and energy/redox homeostasis, and causes a pronounced lag-to-log growth delay, we examined cellular ultrastructure by TEM to assess whether these physiological perturbations were accompanied by measurable morphological changes. TEM analysis revealed that mutant populations frequently display larger, irregularly contoured morphotypes, often with heterogeneous internal organization and variable rod- versus round-like appearances (Fig. 5, Supplemental Fig. 4A-D). To move beyond qualitative impressions, we employed a quantitative image analysis workflow with scale calibration to extract per-cell descriptors in physical units (nm, nm²) across multiple datasets acquired at a magnification of 200 nm. TEM images were pre-processed using contrast-limited adaptive histogram equalization (CLAHE) to improve boundary definition. Cells were segmented using a combination of approaches widely applied in biomedical imaging pipelines, including adaptive Gaussian thresholding, denoising, and distance-transform watershed separation [22, 23]. A gray-level co-occurrence matrix (GLCM) based texture analysis was employed to evaluate the mutant’s cytoplasmic heterogeneity, entangled internal structures, and disrupted subcellular organization in comparison to the WT. GLCM correlation focuses on the correlation parameter as a measure of local intensity co-variation, and provides a sensitive descriptor of how smooth neighboring pixel intensities vary across a structure: higher values reflect ordered, coherent textures, whereas lower values indicate irregular, disorganized patterns [24–26]. A non-parametric Mann-Whitney U test was used to compare groups, given the non-Gaussian distribution of per-cell texture features. Because visual inspection suggested that WT and mutant populations contained a mixture of round-like and rod-like morphotypes, we classified cells objectively using circularity and aspect-ratio thresholds. This stratification avoided pooling morphologically distinct subpopulations and allowed more biologically meaningful comparisons.

Fig. 5.

Quantitative ultrastructural analysis reveals broader phenotypic alterations associated with disruption of PDE_Pg. Representative transmission electron microscopy (TEM) of osmium tetroxide- and uranyl acetate-stained P. gingivalis cells revealed marked structural alterations in Δpde_Pg compared to WT. Boxplots summarize quantitative image analysis: boxes represent interquartile ranges, horizontal lines indicate medians, crosses mark means, and dots denote individual cells. Mutant cells displayed significantly larger median cross-sectional areas (19,454 nm² vs. 5,849 nm²; p = 3.9 × 10⁻¹⁰) and increased perimeters (570 nm vs. 320 nm; p = 6.8 × 10⁻¹⁰), consistent with cell enlargement and irregular outlines. GLCM (gray-level co-occurrence matrix)-based texture analysis revealed a significant reduction in correlation values in mutants (0.466 vs. 0.820; p = 0.0028), indicating disrupted cytoplasmic organization and heterogenous internal architecture. These features align with impaired growth-phase transitions, ATP accumulation, and redox imbalance observed in Δpde_Pg cells

Within the rod-like class, which predominated in mutants, cells displayed significantly larger dimensions. The median cross-sectional area of mutant rods was 6.8 × 10⁴ nm² compared to 5.9 × 10⁴ nm² in WT rods (p = 0.041) (Fig. 5, Supplemental Fig. 4A-D, Supplemental Table 8). Mutant rods also had longer major axes (median 338 nm vs. 310 nm, p = 0.048) and higher perimeters (median 1,310 nm vs. 1,200 nm), consistent with elongated and irregular outlines. Crucially, the analysis revealed a statistically significant reduction in GLCM correlation in mutant cells (median = 0.22, n = 10) compared to WT cells (median = 0.41, n = 11), with a Mann-Whitney U statistic of 98.0 (p = 0.0028) (Fig. 5, Supplemental Fig. 4A-D, Supplemental Table 8). This reduction indicates that mutant cells exhibit markedly lower pixel-level intensity co-variation, consistent with increased cytoplasmic heterogeneity, entangled internal structures, and disrupted subcellular organization.

Together, these results quantitatively confirm that Δpde_Pg cells display enlarged, morphologically heterogeneous, and structurally disordered architecture relative to WT, consistent with broader physiological perturbation observed across our complementary multi-omics datasets. While TEM does not establish a specific mechanistic basis for these changes, the observations provide orthogonal phenotypic evidence that disruption of PDE_Pg is associated with broader cellular physiological remodeling, supporting the model that PDE_Pg-dependent rhythmic-phasic signaling is essential for maintaining cytoplasmic organization and synchronized growth-phase transitions in P. gingivalis.

DNA-binding protein profiling identifies candidate adaptive nucleoprotein signatures associated with PDE_Pg disruption

To explore whether disruption of PDE_Pg-dependent c-di-AMP homeostasis alters DNA-associated regulatory networks that may underlie the prolonged lag-phase phenotype of the mutant, we characterized the DNA-binding proteome of P. gingivalis using a genomic DNA co-precipitation assay performed with and without chemical crosslinking. This approach enables recovery of native DNA-protein complexes and their subsequent identification by SDS-PAGE and LC-MS/MS. However, this method can capture both direct DNA-binding proteins and proteins indirectly associated with DNA through multiprotein complexes, the resulting datasets were interpreted as exploratory profiles of DNA-associated proteomic states rather than definitive identification of sequence-specific DNA-binding factors. Distinct protein patterns were observed between crosslinked and native preparations, and notably between WT and Δpde_Pg strains (Supplemental Fig. 5; Supplemental Table 9). The native WT sample displayed a unique high-molecular-weight profile that was absent in both the Δpde_Pg mutant and crosslinked conditions, indicating that dynamic DNA-protein interactions are disrupted when PDE_Pg is lost. Mass spectrometry of the natively co-precipitated WT complex identified four gDNA-associated proteins: HutH (PGN_1638), GapA (PGN_0173), a TonB-dependent receptor (PGN_0463), and PGN_0369, annotated as a primosomal protein PriB/ssDNA-binding protein. GapA and PriB corresponded to prominent bands at ~ 36 kDa and ~ 18 kDa, respectively, which were substantially reduced in the mutant background. Transcriptomic analyses additionally confirmed altered priB expression in the ΔpdePg strain, suggesting broader physiological adaptation involving DNA-associated processes. PriB, a core primosomal factor, cooperates with PriA to reload stalled replication forks [91, 92], while GapA (a GAPDH homologue) is known to “moonlight” in DNA repair and redox coupling between metabolism and replication [93]. Although these findings do not provide direct evidence for altered replication initiation or specific genome maintenance mechanisms, the differential recovery of PriB- and GapA-associated DNA complexes suggests that disruption of PDE_Pg-dependent c-di-AMP homeostasis is accompanied by remodeling of DNA-associated stress-adaptation pathways, potentially linking nucleotide signaling, redox state, and replication-associated processes during growth-phase transitions.

Molecular simulation suggests conformational consequences associated with conditional phosphorylation of DAC_Pg at Ser-226

Our PTM profiling identified conditional phosphorylation of DAC_Pg at Ser-226, located adjacent to a highly conserved ‘SEE’ motif that mediates catalytic activity (Fig. 4B). This is a significant finding, as earlier studies provided little experimental evidence for posttranslational regulation of c-di-AMP synthases [94, 95], with the exception of a recent report describing phosphorylation of L. monocytogenes DacA at Thr-369 under bile stress [96]. Given that Ser-226 in DAC_Pg emerged as a critical regulatory spot through site-directed mutagenesis and associated physiological phenotypes, including altered growth behavior and loss of pantothenate-mediated rescue under PDE_Pg-disrupted conditions, we examined whether its conditional substitution with a phosphorylated serine residue disrupts local hydrogen-bonding networks and alter protein conformational dynamics. To this end, we first compared the structural homology of P. gingivalis DAC with that of other species and then performed molecular alignment and all-atom molecular dynamics (MD) simulations to explore possible conformational consequences associated with this conditional modification.

Comparative alignment of DAC_Pg with its homologues from other species highlights both conservation and divergence in motifs central to c-di-AMP synthesis (Fig. 4B). While the overall structural folds superimpose closely, immediately adjacent to the catalytic region lies the hallmark ‘SEE’ motif (loop) (Fig. 6A). Structural modeling using Phyre2-AlphaThread, followed by superposition of DAC_Pg and DAC_Lm (PDB: 6HVL and 8C4O), revealed a key divergence in the RGR stretch of DAC_Pg following the SEE motif. The RGR stretch replaces the conserved TGG motif in DAC_Lm. This substitution appears to favor the formation of a longer β-sheet and a correspondingly shorter SEE loop in DAC_Pg relative to DAC_Lm (Fig. 6A). We hypothesized that this site-specific variation could influence DAC conformation upon phosphorylation. To test this, DAC_Pg was remodeled into a phosphorylated form (p-DAC_Pg) at Ser-226 and subsequently re-aligned with DAC_Lm. Remarkably, phosphorylation at this site enabled the RGR stretch of p-DAC_Pg to align seamlessly with the TGG motif of DAC_Lm, with both adopting a well-matched SEE loop integrated into a β-sheet of equivalent length (Fig. 6A).

Fig. 6.

Phosphorylation at Ser-226 drives structural remodeling of DAC_Pg in molecular dynamics simulations. A Structural superimposition of DAC_Pg (green) with two PDB-resolved models of DAC_Lm (ATP-bound, PDB: 6HVL; c-di-AMP–bound, PDB: 8C4O) (yellow and brown, respectively) reveals a high degree of overall homology, including conserved folds, catalytic residues, and loops required for c-di-AMP synthesis. However, a notable divergence is observed adjacent to the conserved SEE motif: DAC_Pg contains an ‘RGR’ stretch in place of the canonical ‘TGG’ motif found in DAC_Lm (left model). Phyre2-AlphaThread indicates that this substitution results in a longer β-sheet and a shorter SEE loop in DAC_Pg compared to DAC_Lm, suggesting a species-specific adaptation in structural dynamics. To probe the impact of posttranslational modification, Ser-226 of the SEE motif in DAC_Pg experimentally identified as phosphorylated in the Δpde_Pg mutant was modeled in silico as phosphoserine (P-S226) using Phyre2-AlphaThread. Remarkably, phosphorylation remodeled the local conformation such that the RGR stretch adopted a fold closely resembling the TGG loop of DAC_Lm, producing a SEE loop-β-sheet arrangement of equivalent length (right model). This finding supports the hypothesis that Ser-226 phosphorylation acts as a conditional regulatory switch, remodeling local folding propensities and potentially modulating DAC function in c-di-AMP signaling. B RMSD shows comparable global stability, indicating that phosphorylation does not grossly destabilize the backbone structure. C RMSF highlights increased residue-level flexibility in p-DAC_Pg, particularly in the N-terminal region. Yellow-marked residues correspond to those participating in catalytic activity, while the red P marks Ser-226, the experimentally validated phosphorylation site. D SASA indicates that DAC_Pg compacts (150–160 nm²), while p-DAC_Pg remains solvent-exposed (165–175 nm²). E Rg confirms stable compaction of DAC_Pg (~ 2.05 nm) versus persistent variability in p-DAC_Pg. F PCA projections show limited conformational sampling by DAC_Pg versus broad, dynamic exploration by p-DAC_Pg. Together, these results show that Ser-226 phosphorylation may influence DAC_Pg conformational flexibility and local structural organization, suggesting a PTM-based mechanism for safeguarding c-di-AMP synthesis under PDE deficiency. A detailed interpretation of the MD data is provided in Supplemental Text 1

We then performed molecular dynamics simulations to compare the structural stability and conformational dynamics of DAC_Pg versus modeled p-DAC_Pg over 200 ns. The analyses included root mean square deviation (RMSD) and root mean square fluctuation (RMSF), along with measurements of solvent accessible surface area (SASA), radius of gyration (Rg), and principal component analysis (PCA) projections. A detailed interpretation of the MD data is provided in Supplemental Text 1 and Supplemental Table 10. Briefly, both proteins exhibited comparable overall stability, with mean RMSDs of 1.07 nm (DAC_Pg) and 0.97 nm (p-DAC_Pg), reaching maxima of 1.31 nm and 1.28 nm, respectively (Fig. 6B). The differences are within natural fluctuation ranges, indicating similar global stability. RMSF analysis revealed clear residue-specific effects of phosphorylation. p-DAC_Pg displayed elevated flexibility throughout several regions, particularly around Ser-226, where RMSF increased sharply, reflecting disruption of local hydrogen bonds and new electrostatic interactions. Enhanced fluctuations in distal regions (~ 20–50, 70–100 aa) suggest long-range allosteric propagation of this modification (Fig. 6C). Consistent with these changes, p-DAC_Pg remained more solvent-exposed (SASA ≈ 165–175 nm²) and dynamically expanded compared with DAC_Pg, which compacted to ≈ 150 nm² after 50 ns (Fig. 6D). Similarly, Rg analyses showed that DAC_Pg rapidly stabilized at ≈ 2.05 nm, whereas p-DAC_Pg fluctuated between 2.10 and 2.30 nm, reflecting reduced compactness and greater conformational flexibility (Fig. 6E). Principal component analysis further confirmed that DAC_Pg occupies a relatively confined conformational space, while p-DAC_Pg has a broader distribution along PC1-PC2, denoting greater structural plasticity (Fig. 6F). Collectively, these data demonstrate that phosphorylation at Ser-226 enhances flexibility, solvent exposure, and conformational heterogeneity of DAC_Pg, features that may influence catalytic behavior and regulatory interactions under specific conditions. Mechanistically, this dynamic remodeling raises the possibility that posttranslational modification, either as permanent or transient, may fine-tune local folding of DAC_Pg as part of an adaptive compensatory response during signaling or cellular perturbation. Considered alongside the experimental identification of conditional Ser-226 phosphorylation and associated physiological phenotypes, these findings suggest that DAC_Pg posttranslational modification may contribute to adaptive regulatory responses within the c-di-AMP signaling network.

Molecular simulations and structural modeling provide mechanistic insight into the atypical pseudoenzymatic behavior of PDE_Pg

Given the biochemical evidence from membrane fractions indicating that PDE_Pg lacks measurable canonical c-di-AMP PDE activity, we investigated the structural basis underlying this atypical PDE_pg behavior through structural modeling, sequence analysis, molecular docking, and molecular dynamics simulations guided by available crystallographic data [18]. Sequence alignment (Clustal Omega) and structural modeling (Phyre2) revealed that PDE_Pg possesses a multidomain organization: an N-terminal region with no apparent sequence homology, a central segment predicted to contain seven transmembrane (7TM) helices indicative of sensing/energy-coupled transport activity, and a conserved C-terminal domain showing strong structural homology to canonical PgpH/HD-type phosphodiesterases. The C-terminal region retains the conserved residues of the HD catalytic motif from Listeria monocytogenes (PDB: 4S1B), supporting its classification as an HD-domain PDE homolog (Supplemental Fig. 6A). I-TASSER analysis further suggested that the N-terminal domain shares structural similarity with components of the eukaryotic anaphase-promoting complex (APC), implying a possible role in cell-cycle regulation (Supplemental Fig. 6B-C). Superimposition of the AlphaFold2-predicted PDE_Pg model onto the L. monocytogenes PgpH/HD PDE structure in complex with c-di-AMP (PDB: 4S1B) demonstrated a high degree of structural congruence, indicating conservation of both the overall fold and the active-site architecture (Fig. 7A). Model validation using Ramachandran plot analysis confirmed high stereochemical quality, with 97% of residues residing in favored regions and only one in a disallowed region (Supplemental Fig. 7A, Supplemental Text 2).

Fig. 7.

Molecular docking and dynamics simulations support atypical ligand interaction behavior consistent with a noncanonical pseudoenzymatic regulatory function of PDE_Pg. A Structural superimposition of PDE_Pg (cyan) with PDB-resolved model of PDE_Lm (c-di-AMP and metal-bound, PDB: 4S1B) (green) reveals a high degree of overall homology, including conserved folds and catalytic residues. B Overall ribbon and close-up surface views of the docked PDE_Pg-c-di-AMP complex showing the HD-domain (green) with one c-di-AMP molecule (Blue). C Overall ribbon and close-up surface views of the docked PDE_Pg-c-di-AMP complex with one water molecule (blue sphere) and two Fe²⁺ ions (orange spheres) and bound c-di-AMP. D Two-dimensional interaction map of the initial docking pose showing extensive interaction network: His558, His559, and Asp604 coordinate Fe; Lys502, Thr510, and Tyr587 stabilize the ribose-phosphate backbone; Ile531, Ala605, and Ala608 provide π-alkyl interactions within the adenine pocket. The phosphate oxygens of c-di-AMP are coordinated by both Fe²⁺ ions through bidentate metal-acceptor interactions, resembling the coordination arrangement observed in canonical reference PgpH/HD PDE geometry, i.e., PDB: 4S1B. E Comparative representation of the experimentally resolved PDE_Lm-c-di-AMP-bound active site arrangement (PDB: 4S1B) used as the structural reference framework for comparative analysis. F-G Following 2D interaction analysis, a 100-ns MD simulation was performed to evaluate the stability of the PDE_Pg-c-di-AMP complex and assess active active-site conformational behavior over time. The conformational free-energy landscape converged to a low-energy basin, with the 67.40-ns snapshot representing an equilibrated, compact binding pose. During equilibration, c-di-AMP underwent a major local rearrangement, shifting from the hydrogen-bond-supported docking pose into a more stable, hydrophobically driven configuration. The ligand remained anchored by the two Fe centers via bridging water (HOH997). Still, it collapsed as the ligand shifted deeper into a hydrophobic pocket formed by Ile531, Ile628, Ile632, Ser520, Tyr566, and Phe567. Aromatic contacts from His519 and Ala605 further stabilized the adenine rings, while productive polar interactions required for catalysis were lost. This rearrangement did not consistently maintain the interaction geometry typically observed in catalytically characterized HD-domain PDEs. Thus, despite strong stable metal coordination and hydrophobic packing, the ligand becomes locked in a non-catalytic rearrangement, consistent with PDE_Pg functioning as a potential regulatory pseudoenzymatic c-di-AMP binder rather than an active phosphodiesterase

Members of the HD-domain phosphohydrolase family typically require two divalent metal ions to stabilize the catalytic pocket and activate a water molecule that bridges the metals for nucleophilic attack on the phosphodiester bond of c-di-AMP [18]. Because no experimentally resolved structure is currently available for PDE_Pg, we used the closest structurally characterized substrate-bound HD-domain c-di-AMP PDE template, the Fe-containing L. monocytogenes PgpH/HD structure (PDB: 4S1B) described by Huynh et al., as a comparative structural framework. Although Huynh et al. demonstrated that the Fe-bound form primarily represents a substrate-binding structural state rather than the physiologically active Mn²⁺-dependent catalytic configuration, this structure remains particularly informative because it captures the experimentally resolved enzyme-substrate architecture of a canonical HD-domain c-di-AMP PDE. Accordingly, it enabled comparative evaluation of conserved ligand-binding features, active-site organization, metal coordination behavior, and broader structural determinants potentially relevant to canonical catalytic versus atypical regulatory function. This structure (Fig. 7A) was therefore used as a template for our docking simulations and comparative analyses to determine how the predicted PDE_Pg HD-domain could coordinate canonical metal-associated ligand interactions within an HD-domain PDE framework.

Initial molecular docking of c-di-AMP into PDE_Pg was performed without explicit metal ions or water molecules (Fig. 7B), generating ten conformations per ligand. This unbiased approach enabled free exploration of the binding pocket, whereas standard docking algorithms often fail to capture realistic metal coordination geometries or water-mediated interactions in metalloproteins. The lowest-energy pose was then refined by incorporating two Fe²⁺ ions and a single water molecule at the experimentally resolved coordination geometry, positioned ~ 2 Å apart, mirroring the canonical coordination observed in reference L. monocytogenes PgpH/HD PDE (PDB: 4S1B) (Fig. 7C). This approach enabled comparative assessment of how the predicted PDE_Pg HD-domain accommodates canonical metal-substrate interaction architectures. 2D interaction mapping of the resulting PDE_Pg-c-di-AMP complex revealed a well-organized network of metal coordination, hydrogen bonding, and hydrophobic interactions stabilizing the ligand within the HD catalytic pocket (Fig. 7D). The phosphate oxygens of c-di-AMP were coordinated by both Fe²⁺ ions through bidentate metal-acceptor interactions, resembling the coordination arrangement observed in canonical reference PgpH/HD PDE geometry (Fig. 7E). The Fe centers were stabilized by His558, His559, and Asp604, which are key residues of the HD motif serving as metal-binding ligands, while Ala605, Ala608, Ser609, and Ile531 contributed π-amide, π-alkyl, and van der Waals contacts that define the adenine-binding pocket. Additional polar contacts with Lys502, Thr510, and Tyr587 anchored the ribose-phosphate backbone of c-di-AMP, further stabilizing the complex. Relative to the canonical PgpH/HD PDE reference architecture, however, the modeled water molecule failed to form the bridging coordination or hydrogen-bonding network typically observed in catalytically active HD-domain PDEs. Unlike the canonical 4S1B structure (Fig. 7E), where two Fe²⁺ ions are symmetrically bridged by water and precisely coordinated by His/Asp residues, PDE_Pg displayed incomplete coordination spheres (Fig. 7D). The Fe centers interacted only with His and Asp residues and the ligand phosphate oxygens, without establishing the canonical bridging water geometry typically associated with productive nucleophilic attack on the phosphodiester bond. Moreover, the catalytic residues His558 and Asp604 were spatially displaced, resulting in an expanded, distorted geometry of the catalytic pocket. Consequently, ligand engagement appeared shifted toward peripheral stabilizing interactions rather than an arrangement clearly aligned with efficient canonical catalytic turnover. Together, the modeled PDE_Pg configuration suggested altered active-site organization and differences in ligand accommodation that appeared less obviously aligned with efficient canonical catalytic turnover, consistent with a potential regulatory role within the c-di-AMP signaling network.

Following 2D interaction analysis, the PDE_Pg-c-di-AMP complex was subjected to a 100 ns molecular dynamics (MD) simulation to examine its dynamic stability and possible ligand-driven rearrangements within the predicted active site. This allowed relaxation of the ligand, protein, metal ions, and solvent, enabling sampling of physiologically relevant conformations to examine dynamic ligand-driven rearrangements and metal-associated structural behavior within the predicted active site. Analysis of conformational free-energy landscape (Fig. 7F and Supplemental Fig. 7B) showed convergence to a low-energy basin, with the 67.40 ns snapshot depicting a compact, energetically favorable binding pose stabilized around the iron centers and coordinating residues. This energetic minimum conformation was used to illustrate key interactions and guide subsequent functional analyses during the equilibrated phase of the simulation. Notably, a major local structural transition was observed as the complex reorganized the c-di-AMP binding mode within the PDE_pg active site into a more stable and hydrophobically driven pose. The corresponding 3D conformation and 2D interaction maps (Fig. 7F-G) highlight stabilized contacts and hydrophobic rearrangements that dominate ligand recognition and binding stability. Unlike docking, during MD under physiological conditions, c-di-AMP transitioned from an initial hydrogen bond-supported pose to a hydrophobic, shape-driven configuration, accompanied by progressive reduction in polar ligand-protein interactions and increased shape-complementary packing within the binding pocket. The 2D interaction map shows that the bicyclic ligand remains anchored by two Fe centers (Fe998 and Fe999), with a bridging water (HOH997) positioning the phosphate oxygen and stabilizing the coordination sphere (Fig. 7G). However, this hydrogen bond network collapses as the ligand shifts deeper into a hydrophobic, shape-complementary pocket formed by Ile531, Ile628, Ile632, Ser520, Tyr566, and Phe567, with additional π-mediated interactions from His519 and Ala605 stabilizing the adenine rings and further anchoring the ligand through aromatic contacts rather than catalytic hydrogen bonds. The resulting configuration shields the ligand from solvent and eliminates productive polar contacts, preventing proper positioning of the putative nucleophilic water (HOH997), which remains coordinated to Fe but is mispositioned relative to the scissile phosphate and appears unfavorably positioned for productive inline attack. Moreover, the absence of a canonical catalytic base (Asp/Glu) to deprotonate this water, combined with active-site steric crowding from His519, Tyr566, Phe567, and Ala605, prevents the formation of the geometry required for efficient SN2-like phosphodiester cleavage. Thus, although the ligand binds with high affinity through metal coordination, hydrophobic packing, and aromatic stacking, it adopts a configuration less clearly aligned with productive canonical phosphodiester hydrolysis. Supplemental Fig. 7B shows the 3D structural view highlighting residues that form hydrophobic contacts with the ligand, providing a clear depiction of the non-polar interaction landscape that stabilizes ligand binding within the predicted active site. Collectively, these rearrangements transformed the initial histidine-anchored, π-stacked docking pose into a π-π and van der Waals interactions-oriented, more stable, and hydrophobically driven pose, consistent with persistent ligand accommodation and a potential regulatory role rather than efficient canonical PDE turnover.

Quantitative MD parameters confirmed the structural stability and local reorganization of this complex state (Supplemental Fig. 7C-H, Supplemental Table 11). The backbone RMSD remained consistently low throughout the 100 ns simulation (0.235 ± 0.026 nm), with only a modest increase from the early (0–20 ns, 0.218 nm) to late (80–100 ns, 0.259 nm) stages, reflecting stable global folding with slow relaxation rather than structural drift. RMSF analysis indicated that most structured regions were rigid (remained within 0.05–0.15 nm), while flexibility was confined to three localized loop regions (residues 450–452, 512–521, and 573–575), suggesting local adaptation around the binding site. Global packing remained intact, as indicated by a slightly decreasing radius of gyration (1.74 to 1.87 nm with a mean of 1.80 ± 0.017 nm) and a ~ 7 nm² reduction in SASA (0–20 ns, averaged 127.0 nm², and 80–100 ns, averaged 120.0 nm²), highlighting gradual compaction and reduced solvent exposure of the binding interface. The most notable dynamic change was observed in ligand-protein hydrogen bonding, which decreased markedly from ~ 3.35 during the initial phase to ~ 0.39 in the final phase. This decline suggests that the ligand transitions from an initially hydrogen-bond-rich, polar pose to a more hydrophobic, shape-complementary binding mode as the pocket reorganizes. PCA revealed a single dominant conformational ensemble with continuous but non-bistable motions, indicating that the protein-ligand complex samples a stable energetic basin. The correlation between PC1 and PC2 was approximately 0.0, confirming that the dominant motions are independent, and ~ 68% of all frames clustered within 1 SD of each PC centroid, indicating different functional or local structural rearrangements. Thus, these MD metrics confirmed that the PDE_Pg-c-di-AMP complex remains globally stable, moderately compacts over time, and undergoes localized adjustments. Overall, these simulations suggest that while PDE_Pg retains structural competence for stable c-di-AMP recognition and ligand accommodation, its dynamic behavior appears more consistent with persistent ligand association than with the canonical catalytic organization observed in active HD-domain PDEs. When interpreted alongside our biochemical membrane fraction analyses demonstrating no measurable PDE_Pg-dependent canonical c-di-AMP hydrolysis, these findings are consistent in support with PDE_Pg functioning as a pseudoenzymatic regulatory component within the c-di-AMP signaling network rather than as a conventional degradative PDE.

To investigate the sequence-based determinants underlying the non-productive binding pose observed during MD equilibration, we performed a comparative alignment of PDE_Pg and the catalytically active PDE_Lm using Clustal Omega (Fig. 8). The most prominent alteration occurs in the loop immediately adjacent to the metal-coordinating Asp604. In PDE_Lm, this region is defined by the motif DSVEAAVR, a sequence enriched in hydrophobic and helix-stabilizing residues that help position the Asp residue to coordinate the metal-bound nucleophilic water. In PDE_Pg, this segment is replaced by DAIEASSR, introducing additional polar serine residues and a bulkier isoleucine side chain. These changes increase local polarity and conformational flexibility while reducing the hydrophobic packing that typically stabilizes the rigid two-metal catalytic framework. As seen in the MD trajectory, this more flexible loop permits collapse of the original hydrogen-bonding network and promotes the inward movement of the adenine rings into a hydrophobic pocket formed by Ile531, Ile628, Ile632, Ser520, Tyr566, and Phe567, shifting the ligand into a non-productive pose that favors π-stacking and van der Waals contacts over catalytic alignment. Additional substitutions reinforce this shift. The replacement D520 (PDE_Lm) → Ser520 (PDE_Pg) removes a potential hydrogen-bond donor/acceptor from the inner wall of the catalytic pocket, further reducing the polar scaffolding needed to orient the scissile phosphate. Similarly, Arg632 (PDE_Lm) → Ile632 (PDE_Pg) eliminates a positively charged residue that, in active PDEs, stabilizes phosphate groups through electrostatic interactions. Its substitution with a hydrophobic isoleucine reconfigures the pocket into a more non-polar environment, precisely matching the hydrophobic anchoring and metal-anchored but catalytically misaligned binding mode observed in the equilibrated MD structure. Collectively, these sequence-level deviations explain why PDE_Pg’s active site undergoes a solvent-driven rearrangement into a hydrophobic, shape-complementary, and catalytically inert pocket, consistent with a pseudoenzymatic function.

Fig. 8.

Pairwise alignment of PDE_Pg and PDE_Lm highlighting divergence within the catalytic HD domain. Clustal Omega alignment of P. gingivalis PDE_Pg and L. monocytogenes PDE_Lm (PDB: 4S1B) reveals both conservation and divergence across the HD catalytic motif. Sequence logos illustrate residue conservation, with the relative symbol size indicating the strength of conservation. Red asterisks denote canonical metal-coordinating residues. Red-labeled residues correspond to catalytically active, metal-binding positions in PDE_Lm, whereas green-labeled residues represent those identified in the PDE_Pg 2D interaction network after molecular docking, and purple-labeled residues indicate molecular contacts detected following molecular dynamics (MD) simulation. The loop adjacent to the metal-binding Asp residue, DSVEAAVR in catalytically active PDE_Lm, is replaced by DAIEASSR in PDE_Pg (red arrows/brackets), introducing two serine residues and a bulkier isoleucine that increase polarity and conformational flexibility while reducing hydrophobic packing around the metal center. Additional sequence shifts include D520→S520 and R632→I632, replacing charged residues with hydrophobic isoleucine at positions that flank the active site. Together, these substitutions disrupt the precise electrostatic environment and steric geometry required to position the metal-bound water for inline attack

Discussion

Our findings introduce a fundamentally new paradigm in bacterial second-messenger biology, demonstrating that c-di-AMP in P. gingivalis operates not as a steady-state metabolite but through temporally structured depletion-replenishment episodes, which we term “rhythmic-phasic c-di-AMP signaling.” Unlike the monotonic synthesis-degradation models described for classical c-di-AMP networks in L. monocytogenes, B. subtilis, and S. aureus, where DACs synthesize the nucleotide and HD-domain PDEs continuously degrade it to pApA [1–4, 12, 18, 97], our data suggests that P. gingivalis relies on rhythmic, phase-locked waves of c-di-AMP synthesis to synchronize metabolic reprogramming with growth phase transition. These cyclic episodes of early-phase decline followed by mid-phase resynthesis are essential for preserving the classical bacterial growth rate, ATP consumption patterns, and metabolic/redox equilibrium in this redox-sensitive proteolytic organism. Strikingly, this dynamic behavior is critically dependent on the atypical function of PDE_Pg, raising the question of whether this HD-domain protein functions as a conventional c-di-AMP PDE or instead serves a non-catalytic regulatory role in maintaining c-di-AMP homeostasis. Our previous studies demonstrated that deletion of pde_Pg paradoxically results in reduced intracellular c-di-AMP and disproportionately low pApA abundance under physiological conditions, yet those analyses did not capture the temporal dynamics of c-di-AMP signaling [6, 10]. In the present study, time-resolved growth-phase analyses revealed that WT cells undergo coordinated cycles of c-di-AMP depletion and replenishment, whereas Δpde_Pg cells exhibit dampened c-di-AMP dynamics and a delayed lag-to-log transition. Importantly, WT cells displayed coordinated ATP consumption accompanying rhythmic c-di-AMP fluctuations and growth initiation, whereas Δpde_Pg cells maintained disproportionately elevated ATP pools. Notably, pApA remained detectable only at extremely low levels throughout growth-phase analyses, occurring approximately a hundred thousand-fold lower than c-di-AMP abundance. Moreover, the limited pApA signal generally tracked with broader c-di-AMP fluctuations rather than exhibiting the independent accumulation pattern expected if phosphodiesterase-mediated turnover were a dominant regulatory mechanism. This disproportionately low abundance is therefore more consistent with minor background or residual hydrolytic activity than with physiologically meaningful catalytic turnover. Consistent with this interpretation, membrane-fraction biochemical assays further support the lack of measurable genotype-dependent pApA production or c-di-AMP hydrolysis. Collectively, these findings support a model in which PDE_Pg functions as a regulatory pseudo-phosphodiesterase that modulates c-di-AMP homeostasis through non-catalytic mechanisms rather than classical second-messenger degradation, consistent with the defining characteristics of pseudoenzymes, which are structural enzyme homologs that have lost or markedly diminished catalytic activity while retaining ligand-binding and regulatory functions [98, 99]. Our model aligns with and extends prior studies demonstrating that temporal changes in intracellular signaling molecules can arise predominantly through synthesis dynamics and growth-associated dilution rather than enzymatic degradation. For example, time-resolved analysis of (p)ppGpp metabolism in E. coli demonstrated that the decline of ppGpp levels during recovery from growth arrest is primarily explained by growth-driven dilution, with flux calculations showing that the return to basal ppGpp concentrations can be accounted for primarily by cell division rather than enzymatic hydrolysis [100, 101]. Our findings extend this principle to c-di-AMP signaling. The atypical PDE_Pg is required to orchestrate both the early decline and subsequent replenishment phases of rhythmic-phasic c-di-AMP signaling. Importantly, intracellular metabolite abundances were normalized to total cellular protein, thereby accounting for changes in biomass during planktonic growth. Consequently, the observed early reduction in c-di-AMP reflects a biological state in which c-di-AMP production becomes transiently arrested relative to the rate of biomass expansion and associated metabolic demand. In this context, the decline phase likely results from a temporary reduction in c-di-AMP synthesis coupled with effective dilution of the intracellular signal relative to newly generated cellular biomass, whereas the subsequent rebound reflects renewed synthetic activity that restores c-di-AMP abundance during exponential growth. The inability of the Δpde_Pg mutant to execute either phase indicates that PDE_Pg functions as a key coordinator of these dynamic transitions. More broadly, this mechanistic model provides a conceptual framework for explaining observations in other bacterial systems where fluctuations in second-messenger abundance occur despite little or no corresponding change in phosphodiesterase activity [100–102]. While secondary adaptive mutations arising during prolonged cultivation cannot be formally excluded, the reproducibility of the growth phenotype, metabolite dynamics, and multi-omics signatures, together with their restoration by genetic complementation, strongly support a direct role for PDE_Pg in regulating rhythmic-phasic c-di-AMP signaling and growth-phase progression.

Since investigations to identify c-di-AMP receptors/effectors and downstream mechanisms in P. gingivalis and members of Bacteroidota are underway, our integrated omics datasets support pleiotropic physiological effects of PDE_pg-dependent c-di-AMP homeostasis. Hence, loss of PDE_Pg was associated with coordinated remodeling across multiple physiological processes. Transcriptomic and proteomic profiling demonstrate that, in the absence of PDE_Pg, P. gingivalis.

exhibits reprogramming of five coordinated axes of physiology, including (i) central carbon rerouting, (ii) redox and iron homeostasis, (iii) envelope and LPS remodeling, (iv) proteostasis and replication-repair adjustments, and (v) nucleotide imbalance. These changes were mirrored by extensive posttranslational remodeling, including altered phosphorylation of DNA repair and translation-associated proteins, acetylation of ribosome assembly factors, and a restricted landscape of oxidative modifications. Collectively, these findings indicate that disruption of PDE_Pg-dependent c-di-AMP homeostasis propagates through multiple interconnected regulatory layers that govern growth-phase progression. Among the affected pathways, PGN_2000, PGN_0369, and PGN_1885 are particularly notable because they provide insight into the links between c-di-AMP signaling, replication-associated processes, metabolic adaptation, and stress responses. PGN_2000 was significantly downregulated in the Δpde_Pg mutant and is predicted to encode a DnaB-like replicative helicase. Given the essential role of DnaB-family helicases in generating single-stranded DNA templates at replication forks, reduced expression of PGN_2000 may impair replication fork progression and contribute to delayed entry into active proliferation. This observation is consistent with growing evidence from B. subtilis, S. pneumoniae, and L. monocytogenes that disruption of c-di-AMP homeostasis influences DNA replication, genome stability, and cell-cycle-associated processes [1, 2, 63, 70, 103]. Complementing these findings, DNA-associated proteome profiling identified PGN_0369, annotated as a PriB/single-stranded DNA-binding protein, within DNA-associated complexes recovered from WT cells but altered in the Δpde_Pg background. PriB is a core component of the bacterial primosome and facilitates replication restart through recruitment of PriA and associated factors during recovery of stalled replication forks [91, 92]. Although our DNA co-precipitation approach does not establish direct mechanistic involvement in replication restart, the altered recovery of PriB-containing complexes in the mutant suggests that disruption of c-di-AMP homeostasis is accompanied by remodeling of DNA-associated stress-adaptation pathways linked to genome maintenance.

Interestingly, these observations parallel previous studies in B. subtilis, in which dynamic fluctuations in c-di-AMP pools were functionally coupled to DNA integrity surveillance during sporulation [70]. In that system, DNA damage inhibits DisA-mediated c-di-AMP synthesis, resulting in a transient reduction of intracellular c-di-AMP that delays developmental progression until favorable conditions are restored. Although P. gingivalis does not sporulate, both systems support a broader conceptual framework in which c-di-AMP signaling is functionally integrated with DNA integrity mechanisms that govern major cellular transitions. In B. subtilis, disruption of this coupling delays sporulation, whereas in P. gingivalis, disruption of PDE_Pg-dependent rhythmic-phasic c-di-AMP signaling is associated with prolonged lag-phase transition and extensive remodeling of replication-, metabolism-, and stress-associated pathways. Taken together, the multi-omics datasets converge on a model in which PDE_Pg functions as a key integrator that couples rhythmic-phasic c-di-AMP signaling to ATP/redox homeostasis, peptide acquisition and metabolism, envelope remodeling, and DNA-associated processes required for efficient growth-phase progression.

Moreover, our findings revealed crosstalk between PDE_Pg and DAC_Pg via Ser-226 within the SEE loop of DAC_Pg, thereby controlling these mechanisms. We identify Ser-226 as a regulatory “hotspot” for c-di-AMP synthesis, since it is also required to sustain phasic c-di-AMP replenishment during exponential growth. Notably, it becomes a potential phosphorylation target in the absence of PDE_Pg. This provides direct evidence that conditional posttranslational modification of DAC_Pg intersects with PDE_Pg-mediated control to fine-tune the timing and amplitude of c-di-AMP signals. This represents a significant advance in the field of signaling, as previous studies had rarely provided experimental evidence for the posttranslational regulation of DAC enzymes [94]. Only recent phosphoproteomic work in L. monocytogenes revealed phosphorylation of several cell envelope proteins under bile stress, including DacA at Thr-369 [95, 96]; the functional consequences of these modifications remained unresolved. The pairing of PDE_Pg and DAC_Pg suggests an evolved regulatory strategy in which P. gingivalis can exploit post-translational modification of the synthase as a compensatory mechanism to restore signaling balance when the regulatory function of PDE_Pg is disrupted. Such feedback control likely enhances the robustness of c-di-AMP signaling by maintaining homeostasis across fluctuating growth phases and environmental conditions. More broadly, this coupling through phosphorylation may represent an evolutionarily conserved logic in cyclic nucleotide networks, paralleling phosphorylation-dependent regulation of diguanylate cyclases in c-di-GMP systems [104, 105]. Our data support a model in which Ser-226 phosphorylation acts as an “adaptive switch” that helps preserve catalytic integrity when the protein interaction network is disrupted. Consistent with this, molecular dynamics simulations show that Ser-226 phosphorylation increases local flexibility, reduces compact packing, elevates solvent exposure, and broadens conformational sampling, collectively reshaping the dynamic landscape of DAC_Pg. Such phosphorylation-driven remodeling of short sequence motifs is a recurring regulatory strategy in other systems, where modification of intrinsically disordered segments or short linear motifs stabilizes latent, motif-like conformations and tunes catalytic or protein-protein interaction properties [106–108].

The metabolomics and pantothenate supplementation experiment further reinforces the model that the regulatory activities of PDE_Pg, rhythmic-phasic c-di-AMP signaling, ATP homeostasis, peptide acquisition, and redox-coupling metabolism are critically synchronized mechanisms in this species. Indeed, the presumable Stickland-type metabolism under anaerobic conditions, in which peptide-based nutritional acquisition provides a balanced supply of oxidative and reductive amino acids [44–46], plays a significant role in P. gingivalis, and perhaps other proteolytic pathogens. Previously, we showed that the metabolic plasticity of P. gingivalis enables it to utilize other carbon and energy resources, but in a manner highly dependent on peptide utilization, thereby balancing a shifted redox status [72]. Here, we showed that, in this context, PDE_Pg and Ser-226 of DAC_Pg form a two-node metabolic-signaling circuit that synchronizes c-di-AMP dynamics with the cellular bioenergetic/redox state. In Δpde_Pg cells along with deficient peptide acquisition, the pantothenate/CoA pathway, the proline-GABA shunt, and intermediates of arginine/lysine metabolism were markedly decreased to a level lower than the detectable threshold. In contrast, supplementation with pantothenate partly rebalanced these mechanisms by tempering the early accumulation of c-di-AMP and sustaining a steady, moderate rise during exponential growth. Hence, pantothenate-supplemented Δpde_Pg cells displayed an improved lag-to-log transition and enhanced overall growth rate, indicating pantothenate partly compensates for the absence of PDE_Pg-mediated early-phase c-di-AMP decline and shifts in redox-coupling metabolites as observed in metabolomics. Critically, this rescue effect was entirely absent in the Δpde_{Pg,DAC(S226A)} strain, demonstrating that the Ser-226 residue functions as a regulatory resistor node whose intact chemistry is required for pantothenate/redox responsiveness. Likewise, pantothenate treatment catastrophically destabilized the growth of WT_DAC(S226A) cells, driving incoherent c-di-AMP dynamics and the collapse of logistic growth behavior. These findings thereby position DAC_Pg via Ser-226 as an essential tunable element that senses and adjusts to redox/metabolic flux. This is consistent with broader principles in bacterial redox and cofactor biology, where CoA and pantothenate pathways interface with nucleotide signaling networks to coordinate growth with intracellular energy state [90, 109]. Together, within this framework, PDE_Pg emerges as the critical redox coupler, coupling phase-appropriate control of c-di-AMP and ATP pools with peptide uptake as the main energy- and redox-controlling mechanism. The early decline pulses in c-di-AMP and ATP pools are necessary signals for linking efficient Stickland-type metabolism to the initiation of cell division. As shown in Fig. 4, loss of both nodes, i.e., PDE_Pg or Ser-226 in DAC_Pg, in Δpde_{Pg,DAC(S226A)} desensitized this system to redox changes, preventing it from re-establishing c-di-AMP and ATP pools, leading to pantothenate insensitivity and also to improved growth rate. This explains why structural modeling supports a model wherein PDE_Pg harbors a central region resembling energy-coupling factor (ECF)-type transporters, as well as elements of the eukaryotic anaphase-promoting complex (APC). This raises the possibility that PDE_Pg may function as a key redox-sensing/coupling regulator of broader cell-cycle regulatory circuits, integrating metabolic state with replication timing and division. This hypothesis aligns with our multi-omics evidence showing that loss of PDE_Pg triggers extensive rewiring of replication, primosomal, and DNA repair modules, and with TEM images revealing that Δpde_Pg cells exhibit enlarged, irregular morphotypes and disordered cytoplasmic texture, consistent with impaired cell division and disrupted turgor homeostasis. As an essential second messenger, c-di-AMP coordinates cell wall biosynthesis, potassium/osmolyte transport, and DNA replication [2, 5, 48, 110], and its dysregulation is known to cause extended lag phases, cell enlargement, division defects, and osmotic sensitivity [97, 111, 112].

Our structural and molecular simulation analyses support a mechanistic framework explaining why PDE_Pg, despite harboring a conserved HD-domain fold, behaves as an atypical, maybe catalytically defective PDE that nonetheless binds c-di-AMP and participates extensively in regulatory signaling functions. Using closest structurally characterized Fe-bound L. monocytogenes PgpH/HD PDE (PDB: 4S1B) as comparative reference, this model captures the substrate-bound architecture of an HD-domain c-di-AMP PDE, enabling evaluation of conserved ligand-binding features, comparative active-site organization, metal coordination geometry, and broader structural determinants relevant to catalysis versus regulatory behavior. Within this framework, the integrated docking, MD simulation, and comparative sequence analyses reveal that it lacks the precise geometric and electrostatic features required to support metal-water-substrate chemistry [18]. Collectively, these data demonstrate that PDE_Pg is architecturally optimized for ligand recognition rather than catalysis, supporting it functions as a pseudoenzyme within the c-di-AMP network. This interpretation aligns with our biochemical analyses showing no measurable PDE_Pg-dependent c-di-AMP hydrolysis in membrane fractions, supporting PDE_Pg functions as a pseudo-phosphodiesterase within the c-di-AMP network. In this capacity, PDE_Pg likely modulates the local concentration, spatial distribution, or temporal flux of c-di-AMP, enabling the second messenger to interface with metabolic state, membrane-associated systems, or cell-cycle circuitry. This concept aligns with an emerging body of work demonstrating that many degenerate GGDEF, EAL, and HD-domain proteins act as noncatalytic hubs that fine-tune cyclic dinucleotide signaling, protein-protein interactions, and transcriptional programs rather than performing enzymatic turnover [113–118]. The presence of a predicted 7TM region and an APC-like function further supports the idea that PDE_Pg integrates environmental or membrane-derived physiological cues with intracellular signaling outputs. Our data strongly supports the conclusion that PDE_Pg is a functionally repurposed HD-domain protein whose catalytic apparatus has evolutionarily diverged toward a signaling-specific, non-hydrolytic role. This places PDE_Pg within the expanding class of pseudoenzymes that regulate, rather than catalyze, enzymatic reactions. In P. gingivalis, this pseudoenzymatic function appears central to maintaining rhythmic-phasic c-di-AMP dynamics, coordinating ATP/redox homeostasis, and aligning nutrient sensing with cell-cycle progression. Beyond P. gingivalis, this pseudoenzyme driven rhythmic-phasic signaling module may represent a conserved bacterial strategy for coordinating growth, metabolism, and pathogenesis.

Conclusions

In summary, our work uncovers a previously unrecognized principle of bacterial cell signaling in which P. gingivalis, known as a typical proteolytic keystone pathogen, utilizes rhythmic-phasic c-di-AMP dynamics, rather than classical synthesis-degradation turnover, to coordinate growth-phase transitions, nutrient acquisition, redox and metabolic homeostasis, and cell-cycle progression. Central to our findings is PDE_Pg, which we identify as a pseudoenzymatic regulator that, together with conserved serine residue (Ser-226) within DAC_Pg SEE loop, regulates these rhythmic-phasic dynamics. Rather than functioning as a degradative PDE, PDE_Pg appears to be associated during peptide uptake, c-di-AMP rhythms, ATP homeostasis, and redox balance, ensuring coordinated metabolic redox and cell-cycle progression during growth-phase transitions and pathogenesis. This mode of regulation aligns with an emerging principle in biology as pseudoenzymes act as metabolic and signaling integrators rather than reaction catalysts. Our findings highlight pseudoenzymes as key regulators of bacterial signaling and pathogenesis, while also identifying the rhythmic-phasic c-di-AMP signaling module as a new mechanism that synchronizes nutrient utilization with cell-cycle progression in proteolytic pathogens employing Stickland-like metabolism within the host.

Study limitations and future directions

Although the present study integrates complementary genetic, biochemical, multi-omics, structural, and physiological approaches to establish previously unrecognized model of rhythmic-phasic c-di-AMP regulation during bacterial growth and physiological adaptation in P. gingivalis, several limitations warrant acknowledgment. First, while native membrane-associated biochemical analyses, growth phase-resolved nucleotide profiling, molecular dynamics simulations, and the absence of detectable pApA collectively support a model in which PDE_Pg functions as a catalytically inactive HD-domain pseudoenzymatic regulatory protein, definitive classification as a pseudo-phosphodiesterase will require future biochemical characterization of purified recombinant PDE_Pg under membrane-mimetic conditions. Second, although phosphorylation of the conserved Ser-226 residue emerged as a prominent regulatory feature associated with the mutant and genetic analyses demonstrated its functional importance, the mediating pathways responsible for posttranslational modifications remain unknown. Future studies should determine how phosphorylation at this site is dynamically regulated under different physiological conditions. Third, our multi-omics analyses identified coordinated remodeling of metabolic, replication-associated, and envelope-related pathways that are consistent with altered c-di-AMP homeostasis, however, targeted functional studies including direct measurements of peptide transport, replication dynamics, and protein-protein interactions will further refine the mechanistic links between rhythmic-phasic c-di-AMP signaling and these downstream physiological processes. Finally, extending these findings to additional members of the phylum Bacteroidota and other proteolytic bacteria will determine whether pseudoenzymatic regulation of c-di-AMP signaling represents a broader regulatory strategy in other species. These future investigations will further clarify the molecular basis and physiological significance of rhythmic-phasic c-di-AMP signaling and its role in coordinating bacterial metabolism, stress adaptation, and growth-phase transitions.

Authors’ contributions

F.M. conceived and supervised the study, secured funding, designed the experimental framework, interpreted the data, and led the writing of the whole manuscript. S.A. performed untargeted metabolomics, LC-MS-based nucleotide quantification, genomic DNA co-precipitation assays, and molecular docking simulations, and was responsible for data analysis, interpretation, and editing of the manuscript. S.G. conducted growth-phase analysis across multiple time points, performed and curated transcriptomics data sets, and provided intellectual input to data interpretation, drafting, and overall manuscript development in coordination with F.M. R.S.S. and M.H. carried out bacterial culture experiments, prepared samples for metabolomics and proteomics analysis, and performed ELISA-based quantification. M.S. contributed to statistical analysis. P.B. conducted proteomics and assisted with data analysis and interpretation. D.A.A. and R.L. provided technical support, critically reviewed the data, and revised the original draft. All authors have read, reviewed, and approved the final manuscript.

Funding

This work was supported by NIH/NIDCR/NIGMS grants: R01DE033702, R03DE031854, R03DE031854-02S1, and GM125504 awarded to Fata Moradali.

Data availability

All data that supports the findings of this study are available within the article, and its Supplementary Information files or from the corresponding author upon request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.