P. gingivalis A7436 and ATCC 33277 (33277) wild-type strains, crp^Pg gene deletion mutant strains (A7436Δcrp^Pg and 33277Δcrp^Pg ), and complemented strain constructed in the A7436 genetic background (Δcrp^Pg +Crp^Pg-HA) ( Supplementary Table S1 , Supplementary Figure S1 ) were grown anaerobically at 37°C (Whitley A35 anaerobic workstation; Bingley, UK) on Schaedler blood agar (ABA) plates (Biomaxima, Lublin, Poland) as previously described (Śmiga et al., 2023). After 3-5 days of culturing on ABA plates, bacteria were transferred into a basal medium (BM) composed of 3% tripticase soy broth (Becton, Dickinson, Sparks, MD, USA) and 0.5% yeast extract (Biomaxima, Lublin, Poland), supplemented with 0.5 µg/ml menadione (Fluka, Munich, Germany), 0.05% cysteine (Sigma-Aldrich, St. Louis, MO, USA), and 7.7 µM hemin chloride (Fluka), resulting in iron and heme-replete conditions (Hm medium). To starve bacteria from iron and heme, BM medium was supplemented with 160 μM 2,2-dipyridyl (Sigma-Aldrich) to chelate iron, and hemin chloride was not added, resulting in iron and heme-depleted conditions (DIP medium). Respective strains were grown with appropriate antibiotics ( Supplementary Table S1 ). In most experiments, planktonic bacteria were maintained under anaerobic conditions in culture tubes. To assess the growth curves, bacterial growth was monitored by measuring optical density at 600 nm (OD₆₀₀) over time every 1 h using 96-well plates and a Stratus plate reader (Cerillo, Charlottesville, VA, USA).

Streptococcus gordonii ATCC 10558 and Prevotella intermedia 17 were cultured as described previously (Śmiga et al., 2015; Ślęzak et al., 2020). E. coli Rosetta 2 (DE3) was cultured under standard aerobic conditions.

A7436Δcrp^Pg and 33277Δcrp^Pg mutant strains were generated by replacing the majority of the crp^Pg gene in A7436 and 33277 wild-type strains with erythromycin resistance cassette from Bacteroides fragilis (ermF) ( Supplementary Figure S1 , Supplementary Table S1 ). Briefly, the flanking regions of the crp^Pg gene and the ermF gene were amplified by PCR and primers listed in Supplementary Table S2 . The obtained fragments were ligated using the NEBuilder HiFi DNA Assembly Cloning Kit (New England Biolabs, Ipswich, MA, USA), resulting in the linear DNA construct that was introduced into P. gingivalis A7436 and 33277 strains by electroporation (Simpson et al., 2000). Selection of mutants was carried out on ABA plates supplemented with 3 µg/ml erythromycin (Carl Roth, Karlsruhe, Germany) and homologous recombination between linear construct and chromosomal DNA was verified using PCR, RT-PCR ( Supplementary Figure S1 ), and DNA sequencing (Microsynth Seqlab GmbH, Gottingen, Germany).

To generate A7436Δcrp^Pg complemented strain (Δcrp^Pg +Crp^Pg-HA), producing Crp^Pg protein with HA-tag at the C terminus, the crp^Pg gene with its native promoter was amplified by PCR and primers listed in Supplementary Table S2 . Then, the DNA fragment was cloned into XhoI and BamHI restriction sites of a pTIO-tetQ plasmid, resulting in a pTIO-tetQ+crp^Pg plasmid. pTIO-tetQ plasmid was obtained from the original pTIO-1 plasmid (Tagawa et al., 2014) by replacing the ermF resistance cassette with the tetQ resistance cassette (Śmiga et al., 2019a). The cloned DNA sequence was verified by DNA sequencing (Microsynth Seqlab). The pTIO-tetQ+crp^Pg plasmid was introduced into P. gingivalis A7436Δcrp^Pg mutant strain by electroporation. Selection of a Δcrp^Pg +Crp^Pg-HA strain was carried out on ABA plates supplemented with 2 µg/ml tetracycline (Carl Roth). The crp^Pg gene expression was verified at transcript and protein levels using RT-PCR and Western blotting, respectively ( Supplementary Figure S1 ).

To overexpress and purify Crp^Pg protein with an N-terminal tag (6×His and maltose-binding protein; 6×His-MBP), the P. gingivalis A7436 crp^Pg gene (PGA7_RS07240) was amplified using PCR and primers listed in Supplementary Table S2 and cloned into XmnI and BamHI restriction sites of pMAL c5x_His plasmid (Śmiga et al., 2019a). DNA sequencing (Microsynth) was carried out to verify the cloned DNA sequence.

Recombinant Crp^Pg protein in fusion with N-terminal 6×His-MBP tag was overexpressed in E. coli Rosetta 2 (DE3) strain (Sigma-Aldrich). After heat-shock transformation, bacteria were grown in terrific broth (TB) with the addition of 35 µg/ml chloramphenicol (Carl Roth) and 100 µg/ml carbenicillin (A&A Biotechnology, Gdansk, Poland) (37°C, 220 rpm) until OD₆₀₀~0.8 was reached. Protein overexpression was induced by IPTG (Carl Roth) added to the final concentration of 0.5 mM, and the culture was incubated for 16 h at 16°C with shaking (220 rpm). Then, bacteria were centrifuged (4000×g, 20 min, 4°C) and the pellet was kept at -20°C until needed. Bacteria were suspended in 25 mM HEPES, pH 7.8, containing 300 mM NaCl and lysed with sonication (Sonopuls HD 4100; Bandelin, Berlin, Germany). Next, bacterial lysates were centrifuged (30000×g, 20 min, 4°C). 6×His-MBP-Crp^Pg recombinant protein was purified from the soluble fraction using TALON Superflow resin (Sigma). After removing the unbound fraction, the resin was washed with 25 mM HEPES, pH 7.8, containing 300 mM NaCl and 5 mM imidazole, followed by 25 mM HEPES, pH 7.8, containing 1 M NaCl and 5 mM imidazole. 6×His-MBP-Crp^Pg protein was eluted with 25 mM Tris-HCl, pH 7.0, containing 250 mM NaCl and 150 mM imidazole. Finally, the purified recombinant 6×His-MBP-Crp^Pg protein was subjected to proteolytic cleavage using Factor Xa in the presence of 2 mM CaCl₂ (New England Biolabs) for 48-72 h at 4°C. Released 6×His-MBP fusion protein was bound to amylose resin according to the manufacturer’s protocol (New England Biolabs). Purified Crp^Pg protein was concentrated using Amicon^® Ultra Centrifugal Filter (10 kDa MWCO; Millipore, Billerica, MA, USA) and stored in 25 mM Tris-HCl buffer, pH 7.0, containing 250 mM NaCl and 10% glycerol at -20°C until used. To measure Crp^Pg protein concentration, the empirical molar absorption coefficient was determined (ϵ₂₈₀ = 22514 M^-1 cm^-1) as reported by others (Eakanunkul et al., 2005).

The PgRsp protein was overexpressed and purified as previously reported (Śmiga and Olczak, 2019).

The ability of P. gingivalis to interact with immortalized human oral gingival keratinocytes (Gie-No3B11; ABM, Richmond, British Columbia, Canada) was analyzed as described previously (Śmiga et al., 2023). Keratinocytes were maintained in TM040 medium (ABM), supplemented with 2% heat-inactivated fetal bovine serum (FBS; Cytogen, Zgierz, Poland), 100 U/mL penicillin and 100 µg/mL streptomycin (Cytogen) at 37°C in an atmosphere of 5% CO₂. Cells were washed three times with PBS and suspended in a DMEM medium (Sigma Aldrich, Cat. No. D0822) without added serum and antibiotics. Bacteria cultured for 24 h in Hm medium were centrifuged (4000×g, 20 min, 4°C) and washed with PBS. Keratinocytes were incubated for 4 h at 37°C in an atmosphere of 5% CO₂ with P. gingivalis at the multiplicity of infection (MOI) of 100. The medium was collected, wells were washed three times with PBS, and a fresh DMEM medium was added to determine the bacteria present inside the cells and attached to them. To kill the external bacteria and to determine live bacteria inside the cells only, the DMEM medium was supplemented with 300 μg/mL gentamicin (Sigma-Aldrich) and 200 μg/mL metronidazole (Sigma-Aldrich). After 1 h, the wells were washed three times with PBS, and the cells were lysed with sterile distilled water. The cell lysates were used to prepare serial dilutions and subsequently plated on ABA plates. The plates were incubated for 7–10 days at 37°C under anaerobic conditions to determine colony-forming units (CFU/1 mL). The experiment was carried out at least three times using two independent biological samples, each sample examined in two technical repetitions.

Rgp and Kgp activities were measured as reported by others (Pomowski et al., 2017). Briefly, to 100 µl of 50 mM Tris-HCl buffer, pH 7.5, containing 150 mM NaCl (TBS), supplemented with 0.05% Tween 20, 5 mM CaCl₂ (TTBS), containing 10 mM L-cysteine hydrochloride, the latter neutralized with 10 mM NaOH, 10 µl of P. gingivalis whole cultures was added. Samples were incubated for 10 min at 37°C and the reaction was initiated by adding 100 µl of TTBS, supplemented with 1 mM Nα-Benzoyl-DL-arginine p-nitroanilide hydrochloride (BApNA; Sigma-Aldrich) for Rgp or N-(p-tosyl)-Gly-Pro-Lys 4-nitroanilide acetate salt (Sigma-Aldrich) for Kgp activity measurement. Samples were incubated for 2 h at 37°C and the reaction was monitored by measuring the absorbance at 405 nm over time using a GloMax Discover plate reader (Promega). Gingipain activity was normalized to the OD₆₀₀ of the bacterial culture and compared to the wild-type strains’ activity, the latter set as 100%.

Biofilm formation was analyzed using 96-well plates (Corning, NY, USA) or plates coated with S. gordonii or P. intermedia. A fresh Hm medium was inoculated with S. gordonii or P. intermedia to the OD₆₀₀ = 0.3. Then, 100 μl of bacterial cultures were added per well and incubated at 37°C under anaerobic conditions for 24 h to create a biofilm composed of S. gordonii or P. intermedia. Unbound bacteria were washed 3 times with 200 µl of TBS. P. gingivalis was grown for 24 h in Hm medium, centrifuged (4000×g, 20 min, 20°C), and washed with TBS. Then, bacteria were resuspended in TBS to OD₆₀₀ = 2, and 100 μl of bacterial suspension was added per well pre-coated with S. gordonii or P. intermedia, or directly to uncoated wells. Samples were incubated for 1 h at 37°C and unbound bacteria were washed 3 times with TBS. Biofilm formation was visualized using 3 methods:

PDE activity was determined as previously reported (Lin et al., 2016) with minor modifications. Briefly, P. gingivalis was grown to the early stationary phase (~24 h) in Hm medium, centrifuged (4000×g, 20 min, 20°C), and washed with TBS. Bacteria were suspended in TBS, supplemented with a protease inhibitor cocktail (Bimake, Houston, TX, USA), and stored at -80°C. Bacterial samples were thawed and lysed by sonication (Bandelin). The protein content was measured in the lysates using Roti Nanoquant (Carl Roth, Karlsruhe, Germany). For the PDE activity measurement, the samples containing 3 µg of proteins in 50 mM Tris-HCl buffer, pH 8.5, supplemented with 1 mM MnCl₂ were placed in a 96-well plate. The reaction was started by adding bis(p-nitrophenyl) phosphate sodium salt (Sigma-Aldrich) at the final concentration of 5 mM. The samples were incubated at 37°C and the absorbance at 405 nm (A₄₀₅) was measured over time using a GloMax Discover plate reader (Promega). PDE activity was determined as the change in absorbance at 405 nm over 60 min (ΔA_405nm/60 min).

Nucleotide contents were determined in P. gingivalis grown for 6 and 24 h in Hm medium. Samples containing approximately 2×10⁹ bacterial cells (1 ml of bacterial culture at OD₆₀₀ equal to 2) were centrifuged (6000×g, 10 min, 4°C), washed once with PBS, and stored at -80°C.

To determine ATP and ADP contents, the luminescence-based ADP/ATP Ratio Assay Kit (Sigma-Aldrich, catalog no: MAK135) was used. The bacterial pellets were suspended in 200 µl of PBS supplemented with 0.2% SDS, incubated for 10 min at 60°C, and then subjected to the procedure provided by the manufacturer.

To determine NAD+ and NADH contents, the colorimetric-based NAD/NADH Assay Kit (Sigma-Aldrich, catalog no: MAK468) was used according to the manufacturer’s protocol. Lysis of bacterial pellets was performed using buffers supplied by the manufacturer.

RNA isolation, reverse transcriptase reaction (RT), and quantitative PCR (qPCR) were performed as previously described (Śmiga et al., 2023). Relative quantification of the transcript was calculated using the double delta method (ΔΔCq) and P. gingivalis 16S rRNA as a reference gene. All primers used in this study are listed in Supplementary Table S2 . At least three independent experiments were run in triplicate for the target and reference genes.

P. gingivalis was grown to the mid-exponential growth phase (OD₆₀₀ = 0.5-0.6). Then, bacteria were pelleted (6000×g, 10 min, 4°C), suspended in 20 µl of 75% ethanol, immediately frozen in liquid nitrogen, and stored at -80°C. RNA isolation, sequencing, and gene expression analysis were performed by Novogene (Cambridge, United Kingdom) using standard company protocols. Gene expression experiment was performed using three independent biological replicates. Fold changes >1.5 or <-1.5 with adjusted P value (P_adj) <0.05 were considered significant. The expression of selected genes was validated using RT-qPCR with total RNA extracted from the same bacterial cultures used for RNA-seq analysis.

Bacterial lysates were prepared with a protease inhibitor cocktail (Bimake) and standardized to OD₆₀₀ or to the amount of protein determined using Roti Nanoquant (Carl Roth). Samples were separated using SDS-PAGE and either stained with Coomassie Brilliant Blue G-250 (CBB G-250) or transferred onto nitrocellulose membranes (Millipore), and probed with mouse anti-HA (0.5 µg/ml; Thermo Fisher Scientific, Waltham, MA, USA), rabbit anti-HmuY (1:10000) (Śmiga et al., 2015), or rabbit anti-RgpB (0.5 µg/ml) (Cusabio, Houston, TX, USA) antibodies. The complexes were further probed with goat horseradish peroxidase (HRP)-conjugated anti-mouse (1:10000; Promega, Madison, WI, USA) or anti-rabbit (1:10000; Sigma-Aldrich) IgG antibodies. Alternatively, the membranes were incubated with lectins (Vector Laboratories, Newark, CA, USA) as previously described (Śmiga et al., 2019b, 2024a). Formed complexes were visualized using chemiluminescence staining (PerkinElmer, Waltham, MA, USA or Thermo Fisher Scientific) and the ChemiDoc imaging system (Bio-Rad, Hercules, CA, USA).

The ~100 kDa and ~120 kDa gel sections were excised from electrophoretic lanes corresponding to samples from 33277 and 33277Δcrp^Pg . Sections were then prepared for bottom-up ESI-LC-MS. Analyses were carried out on an M-class Acquity nanoUPLC connected to a Synapt XS HDMS (Waters, Milford, MA, USA), and the data was analyzed with the PLGS v3.0.3 software. Sample preparation, LC-MS conditions, and data processing protocols were described in detail in our previous work (Śmiga et al., 2024b). For the glycopeptide search via oxonium ion identification, the N- and O-linked Variable Glycosylation Modification was added to the PLGS search workflow.

The formation of protein dimers was analyzed using cross-linking of Crp^Pg with formaldehyde (Sigma-Aldrich). 10 μM Crp^Pg was incubated in PBS for 30 min at room temperature. Alternatively, 1 mM cAMP was added to the sample. Then, samples were incubated for 1 h at 37°C with 0.2% formaldehyde. To visualize dimers, all samples were analyzed using SDS-PAGE and CBB G-250 staining.

To examine whether the Crp^Pg protein binds cyclic nucleotides or di-nucleotides, the affinity chromatography with selected nucleotides immobilized on agarose resin was used: 2-AHA-cAMP agarose, 8-AHA-cAMP-agarose, 5’AHC-2’,3’-cAMP-agarose, 2’-AHC-cAMP-agarose, 2’-AHC-cGMP-agarose, 2-AH-cGMP-agarose, 2’-AHC-cCMP-agarose, 4-AH-cCMP-agarose, 2’-AHC-c-di-GMP-agarose, 2’-AHC-c-di-AMP-agarose (Biolog Life Science Institute, Supplementary Figure S2 ). 5 μM Crp^Pg protein in 25 mM Tris-HCl buffer, pH 7.0, containing 250 mM NaCl and 5 mM MgCl₂. The protein solution was added to the appropriate resin, incubated for 30 min, and samples were centrifuged (500×g, 2 min). To wash out unbound protein, the resin was incubated for 10 min with an initial buffer and centrifuged. The protein-binding process was monitored by SDS-PAGE and staining with CBB G-250.

In addition, the binding of the purified Crp^Pg to fluorescent cAMP analog (2-Aza-ε-cAMP; Biolog Life Science Institute) was analyzed. Briefly, 50 µM 2-Aza-ε-cAMP was added to the 20 µM protein solution and the sample was incubated for 1 h at room temperature. Alternatively, 250 μM cAMP was additionally added to the sample for competitive binding with 2-Aza-ε-cAMP. To verify whether the protein can bind the 2-Aza-ε-cAMP, unbound nucleotide was removed using Amicon Ultra-4 Centrifugal Ultracel-10KDa filter units (Millipore) and washing once with PBS. After the final concentration, the sample volume was adjusted to the initial volume, and 200 µl of the obtained sample was used for fluorescence measurement using a GloMax Discover plate reader (Promega) with a wavelength of excitation and emission set at 365 nm and 500–550 nm, respectively.

The LightShift Chemiluminescent EMSA Kit (Thermo Fisher Scientific) was used to examine Crp^Pg binding to the selected DNA probes according to the manufacturer’s protocol. Biotin-labeled DNA fragments were amplified by PCR using primers listed in Supplementary Table S2 . Purified biotin-labeled DNA probes (1 ng for the entire promoters sequences or 0.33 ng for promoters’ fragments per sample) were added to the binding buffer, containing 2.5% glycerol, 5 mM MgCl₂, 50 ng/μL poly (dI-dC), 50 mM KCl, and 0.05% NP-40. Alternatively, various concentrations of the purified Crp^Pg protein, non-biotinylated DNA (100 ng for the entire promoter or 33 ng for shorter fragments per sample), and 1 mM nucleotides (cAMP, cGMP, c-di-AMP, c-di-GMP) were added. Samples were incubated for 20 min at room temperature and then subjected to electrophoresis on pre-run (for 30 mins) 6% polyacrylamide gels for 80 min at 200 V in 0.25× TBE buffer (1× TBE buffer, pH 8.6, contains 25 mM Tris, 25 mM boric acid, 0.5 mM EDTA). Subsequently, samples were transferred onto a nylon membrane (Bionovo, Legnica, Poland) by wet transfer in 0.5× TBE buffer for 30 min at 380 mA and cross-linked for 10 min using UV radiation at 254 nm. Labeled DNA was visualized using the Chemiluminescent Nucleic Acid Detection Module Kit (Thermo Fisher Scientific) and the ChemiDoc MP imaging system (Bio-Rad).

Phylogenetic analyses and search for Crp^Pg (accession number WP_005873831) homologous proteins were performed using BLASTP (https://www.ncbi.nlm.nih.gov/). Protein sequence comparison was performed using The Clustal Omega (Madeira et al., 2022), Sequence Manipulation Suite: Ident and Sim (Stothard, 2000), and Jalview (Waterhouse et al., 2009). The phylogenetic tree was created using Simple Phylogeny (Madeira et al., 2022) and visualized using iTOL (Letunic and Bork, 2021). Theoretical Crp^Pg function was predicted using I-TASSER (https://zhanggroup.org/), COACH (Yang et al., 2013a, b), COFACTOR (Zhang et al., 2017; Roy et al., 2012), and EDock (Zhang et al., 2020) tools based on the known Crp^Pg three-dimensional structure (PDB ID: 2GAU) and amino acid sequence (accession number WP_005873831). Visualization of protein structures was performed with UCSF Chimera (https://www.cgl.ucsf.edu/chimera) (Pettersen et al., 2004).

All experiments were conducted at least three times using a minimum of three biological replicates, with each sample tested in technical replicates. Unpaired Student’s t-test, one-way analysis of variance (ANOVA) with post hoc Tukey’s test, or two-way ANOVA with post hoc Sidac test (to analyze biofilm formation) was applied using GraphPad software (GraphPad Prism 8.0 Inc., San Diego, CA, USA). All results are shown as mean ± standard deviation (mean ± SD), and for analysis of growth curves as mean ± standard error (mean ± SE).

Preliminary theoretical analyses (Śmiga and Olczak, 2019) suggested that the protein (accession number WP_005873831) encoded in P. gingivalis A7436 and 33277 strains by PGA7_RS07240 and PGN_1569 genes, respectively (PG0396 gene in the reference P. gingivalis W83 strain), showed homology to proteins from the CRP/FNR superfamily. However, our previous phylogenetic analysis indicated that this protein is not grouped within families comprising proteins with known functions (Śmiga and Olczak, 2019). An updated phylogenetic analysis carried out in this study confirmed and extended the previous findings ( Figure 1A ). BLASTP search demonstrated the close similarity to the uncharacterized proteins encoded within the closely related species from the Bacteroidota phylum, such as Porphyromonas macacae, Parabacteroides merdae, Tannerella forsythia, Bacteroides uniformis ( Figures 1A, B , Supplementary Figure S3 ), with amino acid sequence identity and similarity higher than 43% and 60%, respectively ( Figure 1B ). Comparison of the amino acid sequence of the P. gingivalis WP_005873831 protein with selected sequences of characterized Crp proteins showed low identity/similarity: 18.0/34.8% for Escherichia coli Crp^Ec, 18.6/34.8% for Pseudomonas aeruginosa Vfr^Pa, 21.0/41.6% for Thermus thermophilus Crp^Tt, 24.9/46.8% for Mycobacterium tuberculosis Crp^Mt, 20.3/37.0% for Bacillus subtilis Crp^Bs, 17.9/31.7% for Xanthomonas campestris CLP^Xc, 16.6/34.9% for Deinococcus radiodurans DdrI^Dr ( Figure 1B ). In contrast, the three-dimensional structure of WP_005873831 protein (PDB ID: 2GAU; deposited by Rotella FJ, Zhang RG, Mulligan R, Moy S, and Joachimiak A; Midwest Center for Structural Genomics, MCSG; 2006, revised in 2011), possessing a typical N-terminal ligand-binding domain and a C-terminal DNA-binding domain, is highly similar to other known proteins from the CRP/FNR superfamily (Passner and Steitz, 1997; Agari et al., 2012; Wang et al., 2024). As in homologous proteins, both domains are connected by an α-helix involved in dimerization ( Figure 1C ). Moreover, comparison with proteins of known three-dimensional structures showed the highest structural similarity to Crp proteins ( Figure 1C ). Therefore, we classified P. gingivalis protein to the CRP/FNR superfamily and Crp family and termed it as Crp^Pg.

To determine the role of Crp^Pg in P. gingivalis, we examined two P. gingivalis strains, representatives of more (A7436) and less virulent (33277) strains, which differ in phenotypes and virulence potential (Laine and van Winkelhoff, 1998; Griffen et al., 1999; Dorn et al., 2000; Amano et al., 2000; Yoshino et al., 2007; Rodrigues et al., 2012; Ohya et al., 2016; Śmiga et al., 2019b; Seers et al., 2020; Śmiga et al., 2024a). The A7436 strain is encapsulated, produces fewer fimbriae, which is reflected in the formation of smaller biofilm structures and is connected more often with patients with periodontitis. In contrast, the ATCC 33277 strain lacks a capsule, forms larger biofilm structures due to its high fimbriation, and is connected more often with healthy periodontium. In addition, although both strains are fimbrated, they produce different types of the long fimbrial protein FimA (Nagano et al., 2013). Therefore, mutant strains lacking the functional crp^Pg gene were constructed in both A7436 (A7436Δcrp^Pg ) and 33277 (33277Δcrp^Pg ) strains ( Supplementary Figure S1 ).

To examine the influence of Crp^Pg protein on the global gene expression, RNA-seq analysis was carried out in the Δcrp^Pg mutant strain constructed in the A7436 genetic background. The expression of selected genes was validated using RT-qPCR. The inactivation of the crp^Pg gene resulted in 76 up-regulated and 60 down-regulated genes ( Figure 2A ). Genes whose expression decreased the most belong mainly to the group of hypothetical proteins, protein synthesis, and signal transduction/regulatory functions (including 5 helix-turn-helix transcriptional regulators) ( Figure 2B ; Supplementary Table S3 ). Among the most highly down-regulated genes in the Δcrp^Pg mutant strain are the genes encoding thioredoxin (TrxA; PG0034) and SDR family oxidoreductase (PG2069) ( Supplementary Table S3 ). Also, the expression of the gene encoding a SoxR homolog (SoxR reducing system RseC family protein/positive regulator of sigma(E); PG0302) and two genes encoding 4Fe-4S-binding proteins (PG1421 and PG1813) was decreased ( Supplementary Table S3 ). Another highly down-regulated gene encodes up-regulated in stationary phase protein A (UstA; PG0246) ( Table 1 ; Supplementary Table S3 ).

The genes whose expression increased the most belong mainly to the group of hypothetical proteins, protein synthesis, DNA/RNA processing, cell envelope, cellular processes, energy metabolism, biosynthesis of cofactors/prosthetic groups, and transport and binding proteins ( Figure 2B ; Supplementary Table S3 ). Examples of up-regulated genes involved in P. gingivalis virulence include those encoding fimbrial Mfa proteins (mainly Mfa1-Mfa3) (PG0178-PG0182) and Arg-specific gingipain A (RgpA; PG2024) ( Table 1 ; Supplementary Table S3 ). Among the most highly up-regulated genes in the Δcrp^Pg mutant strain is the hagC gene encoding one of the hemagglutinins; however, this result was not confirmed by RT-qPCR analysis ( Table 1 ). Using RNA-seq and RT-qPCR analyses we did not observe changes in the expression of other genes encoding hemagglutinins, namely HagA and HagB ( Table 1 ). Therefore, these findings strongly suggest that Crp^Pg does not regulate the hemagglutinin activity of P. gingivalis.

To verify whether the Crp^Pg influences the expression of Mfa proteins, the direct Crp^Pg interaction with the mfa promoter was examined ( Supplementary Figure S4A ; Supplementary Table S4 ). First, a ~250 bp DNA fragment containing the entire promoter was analyzed. After confirmation by EMSA that the Crp^Pg binds to this DNA (data not shown), the promoter region was divided into 5 smaller ~85 bp fragments ( Supplementary Figure S4A ). This allowed us to specify the promoter region recognized by the Crp^Pg protein (fragment 5 shown in purple in Supplementary Figure S4B ). As Crp proteins often act as autoregulators (Korner et al., 2003; Śmiga and Olczak, 2019), a similar analysis was performed for the crp^Pg promoter (data not shown) and its fragments ( Supplementary Figure S4 ). This allowed us to specify the promoter region recognized by the Crp^Pg protein (fragment 4 shown in red in Supplementary Figure S4B ). Analysis of both selected promoter fragments showed a shift in the migration of DNA ( Figure 2C ; Supplementary Figure S4B ), indicating DNA-protein complex formation. The Crp^Pg also tends to interact with DNA with lower affinity, visible as protein-DNA aggregates formed in the gel wells. Such behavior was also observed in the case of DNA fragments to which the Crp^Pg did not bind (e.g., promoters of sod and pgfur; our unpublished data).

By comparing mfa and crp^Pg promoter fragments to those recognized by selected CRP/FNR superfamily proteins it was possible to indicate the potential Crp^Pg-binding box which showed high similarity to the one bound by E. coli Crp^Ec ( Figure 2D ). Highly homologous sequences were also found in other promoters, including those for trxA, PG2008, and PG0682-PG0685 genes ( Supplementary Table S4 ) whose expression was changed in the crp^Pg mutant strain ( Supplementary Table S3 ). In contrast, similar sequences were identified within the sequences encoding PgFur and HagC, but not in their promoter regions, which may indicate the lack of regulation of the expression of these genes by Crp^Pg ( Table 1 ; Supplementary Table S4 ).

Further, we aimed to find whether Crp^Pg functions utilizing the binding of cAMP or other ligands. In silico analysis showed that the Crp^Pg protein has a potential ligand-binding pocket between the dimerizing α-helix and the N-terminal domain ( Figure 3A ) and the predictions indicated a possibility of cAMP binding (max C-score: 0.6). To verify this, we analyzed the binding of several ligands to the Crp^Pg protein. In the initial stage, we examined the binding of cyclic nucleotides or di-nucleotides immobilized on agarose resin. To exclude a physical blockage of the binding site in the protein, we used various ligands labeled in different positions of nucleobases or ribose ( Supplementary Figure S2 ). Additionally, we used a cAMP analog with fluorescent properties (2-Aza-ϵ-cAMP) ( Figure 3B ). We observed neither cAMP ( Figure 3B ) nor other ligands (data not shown) binding to the Crp^Pg protein. As a control, we used PgRsp, which did not bind 2-Aza-ϵ-cAMP. Moreover, by using EMSA we did not demonstrate the influence of cAMP, cGMP, c-di-AMP, or c-di-GMP on the Crp^Pg binding to the promoter sequences ( Supplementary Figure S4C ). Although the Crp^Pg protein may exist in the dimeric form, cAMP did not influence its dimerization ( Figure 3C ).

The Crp^Pg protein structure is highly similar to the structure of DdrI^Dr ( Figures 1C , 3D ). Moreover, it shows more similarity and structure coverage to the ligand-binding domain of the Crp^Ec-cAMP complex (protein active form) than to the apo-Crp^Ec (not active form) ( Figure 3E ). Predicted cAMP-binding amino acids in the ligand-binding pocket in the Crp^Pg differ from those of classical cAMP-binding Crp proteins ( Figures 3A, F ). Our detailed theoretical analyses showed that the Crp^Pg protein and other closely related Crp homologs have the conserved arginine in the position relevant to R55 or R51 in DdrI^Dr or SdrP^Tt, respectively ( Figures 3D, F ; Supplementary Figure S3 ). However, in the dimeric state of the Crp^Pg protein, the side chain of glutamic acid (E142) located in the position relevant to Y113 in DdrI^Dr may block the entrance to the cAMP-binding site located in the second protein monomer ( Figure 3G ).

Subsequent experiments aimed at further understanding the potential mechanism of the Crp^Pg function employed the characterization of the effect of growth conditions on the expression of the crp^Pg gene. First, the crp^Pg transcript level was determined in both A7436 and 33277 wild-type strains. The highest crp^Pg expression was observed in the early growth phase (4 h of the culture), but a significant decrease (up to 6 and 26 times) in crp^Pg transcript levels monitored in the late exponential (10 h of the culture) and early stationary (24 h of the culture) growth phases were detected ( Figure 4A ). Using the Δcrp^Pg +Crp^Pg-HA complemented strain generated in the A7436 genetic background, in which the Crp^Pg protein in fusion with the HA-tag was produced under its native promoter, we confirmed that also Crp^Pg protein production is growth phase-dependent. We observed a significant decrease in Crp^Pg protein amount (>2 times) in bacteria grown for 24 h compared to analysis carried out in bacteria grown for 6 h ( Figure 4B ), which is in agreement with the transcript levels.

Following analysis of the potential regulation of the expression of P. gingivalis virulence factors by the Crp^Pg, we examined P. gingivalis growth alone or with other bacteria or host cells. First, we examined how crp^Pg deletion affects the phenotype of P. gingivalis, with a focus on the effect on its virulence potential. The inactivation of the crp^Pg gene did not influence the 33277Δcrp^Pg mutant strain proliferation ( Figure 5A ). The A7436Δcrp^Pg mutant strain grew slightly better than the A7436 wild-type strain, regardless of the growth conditions, but the differences were not statistically significant ( Figure 5A ). Its wild-type phenotype was restored when the Crp^Pg-HA protein was produced in the A7436Δcrp^Pg mutant strain ( Supplementary Figure S5A ). The analysis of the production and activity of the two groups of the most important virulence factors, namely hemophore-like HmuY protein and gingipains (Kgp and Rgp) showed no differences between the wild-type and mutant strains ( Figures 5B, C ).

Since Crp^Pg regulates mfa operon, we analyzed the influence of the crp^Pg gene on the ability of the bacterium to form biofilm. P. gingivalis mono-cultures were formed directly on an abiotic surface or on an abiotic surface first colonized by S. gordonii or P. intermedia. Using this simplified model, we aimed to examine the influence of the lack of the crp^Pg gene on P. gingivalis adherence and biofilm formation on an abiotic surface. To mimic the environment of the oral cavity, we employed plates pre-coated with S. gordonii or P. intermedia. Both species colonize the oral cavity earlier than P. gingivalis, form biofilms on an abiotic surface better than P. gingivalis alone, and play a crucial role in recruiting P. gingivalis into the biofilm. In addition, these bacteria were selected due to their proven interactions. P. gingivalis fimbral protein Mfa1 interacts with the S. gordonii SspA/B proteins, while P. gingivalis fimbral protein FimA binds to S. gordonii GAPDH (Kuboniwa and Lamont, 2010). In the interaction between P. gingivalis and P. intermedia, a crucial role is played by P. gingivalis gingipains (Kamaguchi et al., 2001). Biofilm formation was first determined using crystal violet staining ( Figure 5D , left panel). Although the inactivation of the crp^Pg gene resulted in a lower tendency to form biofilm structures on an abiotic surface in the case of both strains, especially the A7436 strain, the differences were not statistically significant. Biofilm formation was further examined in correlation to selected P. gingivalis surface proteins. Since no change in HmuY or Kgp production or activity, respectively, was demonstrated between wild-type and mutant strains, and both proteins function in part in cell-associated forms, the P. gingivalis biofilm formation was also determined using specific anti-HmuY antibodies or enzymatic reaction with a specific substrate for Kgp. When the biofilm was examined on the abiotic surface colonized first by S. gordonii or P. intermedia and related to the Kgp activity, only the Δcrp^Pg mutant strain constructed in the A7436 strain exhibited a lower ability to form biofilm structures ( Figure 5D , middle panel). When the formation of biofilm was correlated with the production of HmuY protein, inactivation of the crp^Pg gene in both A7436 and 33277 strains resulted in a lower ability to form biofilm structures when P. gingivalis was incubated on plates colonized first by S. gordonii ( Figure 5D , right panel). This effect was also observed in the case of the Δcrp^Pg mutant strain constructed in the 33277 strain, grown on an abiotic surface.

The ability to interact with host cells (adhere to and invade host cells) was analyzed using a P. gingivalis-gingival keratinocytes co-culture model (Śmiga et al., 2023). These abilities increased in the case of the Δcrp^Pg mutant strain constructed in the A7436 genetic background ( Figure 5E ). The observed changes were eliminated when the Δcrp^Pg +Crp^Pg-HA complemented strain was examined ( Supplementary Figure S5B ). In the case of the Δcrp^Pg mutant strain constructed in the 33277 genetic background, although we detected a slightly increased interaction of this strain with the host cells, the difference was not statistically significant. However, we observed a higher ability of the 33277Δcrp^Pg strain to adhere to keratinocytes, similarly as in the case of the A7436Δcrp^Pg strain ( Figure 5E ). We also examined the expression of selected genes in bacteria grown for 4 h in the culture medium collected from 24-hour gingival keratinocyte cultures (DMEM^KER). Slightly increased crp^Pg , pgfur, and mfa1 gene expressions were observed when the A7436 strain was grown in DMEM^KER, whereas no changes were found in the case of the 33277 strain ( Figure 5F ). These results indicated that the external environment of host cells, which is enriched with proteins secreted by keratinocytes ( Figure 5G ) can induce crp^Pg -dependent gene expression.

In the next step of characterizing the Crp^Pg role in P. gingivalis, proteins present in whole bacterial cell lysates were analyzed by SDS-PAGE. The main differences were visible in the disappearance of the ~120-kDa band with a simultaneous intensification of the ~100-kDa band in the 33277Δcrp^Pg mutant strain compared to the 33277 wild-type strain ( Figure 6A , Supplementary Figure S6 ). In the A7436 and A7436Δcrp^Pg strains, only ~100-kDa band was visible ( Figure 6A , Supplementary Figure S6 ). Using MS we confirmed that a protein ascribed to ~100-kDa and ~120-kDa bands is a putative zinc carboxypeptidase (91.5 kDa) encoded by the PGA7_RS01035 (PGN_0335/PG0232) gene ( Supplementary Figure S6 ). This led us to conclude that the difference in the protein molecular mass may result from posttranslational modification, such as glycosylation. However, the search for glycosylated peptides after MS analysis was inconclusive. Reactivity with JACALIN, SNA, and MAL II lectins confirmed previously shown (Śmiga et al., 2024a) differences in glycosylation patterns between A7436 and 33277 strains ( Figure 6B ). Moreover, it showed that the deletion of the crp^Pg gene caused a change in the glycosylation pattern, mainly in the 33277 strain. Among P. gingivalis proteins modified by glycosylation is RgpB. To detect changes in RgpB glycosylation in the Δcrp^Pg mutant strains, we used anti-RgpB antibodies and determined the mass of produced RgpB. Similarly to the putative zinc carboxypeptidase, also in the 33277Δcrp^Pg mutant strain, RgpB had a lower mass than in the 33277 wild-type strain. Moreover, the observed RgpB mass corresponded to the mass of the protein produced in the A7436 wild-type and A7436Δcrp^Pg mutant strains ( Figure 6C ). Changes in the presented phenotype can partially be explained by the down-regulation of PG1135 (wbaQ, sugar transferase/phosphoglycosyltransferase), PG0119 (WecB/TagA/CpsF family glycosyltransferase), and up-regulation of PG0043 (family 20 glycosylhydrolase/beta-hexosaminidase) genes ( Supplementary Table S3 ), all involved in glycosylation and sugar metabolism.

Results from phenotypic and transcriptomic analyses suggested that Crp^Pg may regulate processes dependent on energy production. Indeed, in our transcriptomic analysis, we observed changes in the expression of genes involved in energy metabolism. The best example is a system composed of genes encoding V-type ATP synthase subunits (PG1803-PG1807 operon, encoding AtpD, AtpI, and AtpB) whose expression increased in the A7436Δcrp^Pg mutant strain ( Supplementary Table S3 ). Another example is a system composed of FtsX (FtsX-like permease family protein and two ABC transporter permeases) and FtsE (ABC transporter ATP-binding protein) proteins encoded by PG0682-PG0685 whose expression increased in the mutant strain ( Supplementary Table S3 ). Classical Crp proteins play a key role in regulating energy metabolism, a function often reflected in the levels of various signal molecules, including cAMP and c-di-AMP, as well as in the ratio of other critical nucleotides, such as ATP/ADP and NAD⁺/NADH. Therefore, we determined ATP, ADP, NAD⁺, and NADH contents ( Supplementary Figures S7A, B ). While NAD⁺ and NADH amounts did not change in both Δcrp^Pg mutant strains, the amount of ATP and ADP pull decreased in the A7436Δcrp^Pg mutant strain after 6 h of the culture (~20% decrease), and in the 33277Δcrp^Pg mutant strain after culture for 24 h (~29%). Moreover, the decrease was up to 20% or 70% for ATP or ADP amounts, respectively, in the 33277Δcrp^Pg mutant strain, whereas no changes were found in the case of the A7436Δcrp^Pg mutant strain.

Additionally, we determined PDE activity to demonstrate whether the lack of the functional crp^Pg gene affects the metabolism of cyclic nucleotides; however, no differences between the wild-type and corresponding mutant strains were observed ( Supplementary Figure S7C ). This is in agreement with the RNA-seq analysis where no change in gene expression whose products are involved in cyclic nucleotide metabolism was found.