Characterization of c-di-AMP signaling in the periodontal pathobiont, Treponema denticola

SUMMARY

Pathobionts associated with periodontitis, such as Treponema denticola, must possess numerous sensory transduction systems to adapt to the highly dynamic subgingival environment. To date, the signaling pathways utilized by T. denticola to rapidly sense and respond to environmental stimuli are mainly unknown. Bis-(3’-5’) cyclic diadenosine monophosphate (c-di-AMP) is a nucleotide secondary messenger that regulates osmolyte transport, central metabolism, biofilm development, and pathogenicity in many bacteria but is uncharacterized in T. denticola. Here, we studied c-di-AMP signaling in T. denticola to understand how it contributes to T. denticola physiology. We demonstrated T. denticola produces c-di-AMP and identified enzymes that function in the synthesis (TDE1909) and hydrolysis (TDE0027) of c-di-AMP. To investigate how c-di-AMP may impact T. denticola cellular processes, a screening assay was performed to identify putative c-di-AMP receptor proteins. This approach identified TDE0087, annotated as a potassium uptake protein, as the first T. denticola c-di-AMP binding protein. As potassium homeostasis is critical for maintaining turgor pressure, we demonstrated that T. denticola c-di-AMP concentrations are impacted by osmolarity, suggesting that c-di-AMP negatively regulates potassium uptake in hypoosmotic solutions. Collectively, this study demonstrates T. denticola utilizes c-di-AMP signaling, identifies c-di-AMP metabolism proteins, identifies putative receptor proteins, and correlates c-di-AMP signaling to osmoregulation.

INTRODUCTION

Periodontitis is a chronic inflammatory disease of the periodontium that destroys tooth-supporting structures. It is one of the most prevalent human disorders, as >40% of adults 30 years or older in the US have some form of periodontitis (Eke et al., 2018). Nearly 11% of the global population has severe periodontitis, the primary cause of edentulism, causing poor mastication, low self-esteem, and reduced quality of life (Kassebaum et al., 2014). The chronic inflammation that results from periodontitis is a risk factor for multiple systemic diseases such as cardiovascular disease, diabetes, Alzheimer’s disease, and oral and gastrointestinal cancers (Olsen et al., 2018; Sanz et al., 2018; Werber et al., 2021; Winning & Linden, 2017). There is no cure for periodontitis, and therapies to constrain disease progression are expensive (>$350 billion globally), painful, and frequently ineffective, highlighting a need for novel therapeutic approaches (Botelho et al., 2022; Righolt et al., 2018).

Complex, dysbiotic communities of bacteria are the etiologic elements of periodontitis (Hajishengallis et al., 2023; Lamont et al., 2023). Treponema denticola is an anaerobic spirochete and an obligate colonizer of the human subgingival sulcus (Ellen & Galimanas, 2005). T. denticola, and other oral Treponema spp., are highly associated with the development and progression of periodontitis (Abusleme et al., 2013; Curtis et al., 2020). The bacterial population in the gingival sulcus alters the surrounding ecological environment. Importantly, the microbiome composition and biogeography of the biofilm are also shaped by the surrounding ecology (Mark Welch et al., 2020; Proctor et al., 2020). The ability of oral bacteria like T. denticola to sense and respond to their surrounding environment is critical to thrive in the highly dynamic niche of the subgingival crevice. The T. denticola genome encodes for multiple two-component systems and nucleotide secondary messengers that mediate adaptive response (Bian et al., 2013; Frederick et al., 2011; Patel et al., 2021).

Linear and cyclic nucleotides are critical secondary messengers that bridge the perception of external stimuli with changes to physiological processes in bacteria (Hengge et al., 2023; Yoon & Waters, 2021). Some of the most significant and best characterized nucleotide secondary messengers are guanosine (penta)tetra-phosphate ((p)ppGpp), cyclic (3’-5’)-adenosine phosphate (cAMP), bis-(3’-5’) cyclic diguanosine monophosphate (c-di-GMP) and bis-(3’-5’) cyclic diadenosine monophosphate (c-di-AMP) (Irving et al., 2021; Jenal et al., 2017; Yin et al., 2020). Nucleotide secondary messengers control core physiological processes in bacteria, including balancing central metabolism, growth rate, cell membrane and cell wall homeostasis, stress responses, sporulation, motility and chemotaxis, biofilm development, and regulation of virulence (Hengge et al., 2023; Yoon & Waters, 2021). Our understanding of T. denticola nucleotide secondary messengers is in its infancy. Recently, TDE1711 was identified as a small alarmone synthetase that synthesizes the stringent response signal ppGpp and, to a much lower extent, ppApp (Wang et al., 2023). The same study identified TDE1690 as a small alarmone hydrolase that degrades all guanosine- and adenosine-based alarmones. As a motile organism, T. denticola has also utilizes c-di-GMP signaling. The T. denticola genome encodes multiple proteins with putative functions in c-di-GMP metabolism and sensing and was shown to synthesize c-di-GMP during in vitro growth (Frederick et al., 2011; Kostick et al., 2011). DgcA (TDE0125) is the only characterized diguanylate cyclase synthesizing c-di-GMP (Patel et al., 2021). Deletion of TDE0214, a PilZ-domain protein that binds c-di-GMP, resulted in reduced biofilm development, defects in motility, and reduced virulence in a murine abscess model (Bian et al., 2013). The role of c-di-AMP signaling has never been studied in T. denticola, but its genome encodes for proteins with highly conserved motifs that suggest it may utilize c-di-AMP (Seshadri et al., 2004).

C-di-AMP is synthesized from two molecules of ATP by diadenylate cyclases (DACs) and is degraded by phosphodiesterases (PDEs) to either a single molecule of pApA or two molecules of AMP (Commichau et al., 2019). Across bacteria and archaea, many DAC enzymes share the conserved HDG and RHR motifs, which are essential for synthesizing c-di-AMP (Galperin, 2023). The CdaA-type is the most widespread family of DAC and comprises three N-terminal transmembrane domains followed by a cytoplasmic C-terminal DAC domain (Rosenberg et al., 2015). C-di-AMP synthesis by CdaA is regulated by the CdaR regulatory protein co-transcribed with CdaA (Gibhardt et al., 2020; Rismondo et al., 2016). The CdaR protein consists of an N-terminal transmembrane domain and multiple tandem repeats of the YbbR (PF07949) domain. While the exact mechanism of regulation remains unclear, the transmembrane domains mediate the interactions between CdaA and CdaR (Gundlach et al., 2015). The T. denticola ATCC 35405 genome encodes for a single CdaA-family DAC (TDE1909) adjacent to a YbbR-domain containing protein (TDE1908) (Seshadri et al., 2004).

C-di-AMP levels must be tightly regulated as c-di-AMP is both essential for growth but is also toxic when allowed to accumulate (Commichau et al., 2019). Hydrolysis of c-di-AMP is again carried out by several conserved families of specific PDEs that harbor either a DHH/DHHA1 domain or an HD domain (Commichau et al., 2019; Yin et al., 2020). The PgpH family of PDEs comprises an N-terminal 7TMR-HDED domain followed by 7 transmembrane helices and a cytoplasmic C-terminal HD domain (Huynh et al., 2015). T. denticola encodes a single gene annotated as a PgpH-type PDE (TDE1241) (Seshadri et al., 2004). Both the GdpP and DhhP-family PDEs utilizes the DHH/DHHA1-type PDE domain that can degrade c-di-AMP into pApA and further degrade pApA into AMP. GdpP-family PDEs are membrane-bound via an N-terminal transmembrane domain and often harbor a Pern-Arnt-Sim (PAS) domain for environmental sensing with a degenerate GGDEF domain (Rallu et al., 2000). The DhhP family of PDEs was first described in B. burgdorferi consisting of cytoplasmic DHH/DHHA1 domains, lacking the transmembrane, PAS, and GGDEF-like domains of the GdpP family (Ye et al., 2014).

The work reported here seeks to elucidate the role of c-di-AMP signaling in T. denticola. The results of this study demonstrate that T. denticola produces intracellular c-di-AMP during in vitro growth and identify TDE1909 as a DAC synthesizing c-di-AMP. We then sought to identify c-di-AMP receptor proteins that may suggest the physiological processes or pathways regulated by c-di-AMP. These analyses identified that TDE0087, a protein involved in potassium uptake, binds c-di-AMP. This led to the observation that c-di-AMP levels are elevated in hypoosmotic conditions, suggesting that c-di-AMP negatively regulates potassium transport. Finally, we identified a DhhP-like phosphodiesterase that degrades c-di-AMP to pApA. Collectively, this work established c-di-AMP signaling as a signal transduction system in T. denticola that may contribute to its success as a human pathobiont.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

T. denticola strains ATCC 35405, ATCC 33520, SP50, and SP55 used in this study were grown in new oral spirochete (NOS) medium under anaerobic conditions (5% H₂, 5% CO₂, 90% N₂) at 37°C. ATCC 35405 and 33520 are common laboratory strains, while SP50 and SP55 are low-passage, clinical isolates. Strains were purchased from the ATCC or a kind gift from J. Christopher Fenno at the University of Michigan. The strains in this study are described in Supplemental Table 1.

Quantification of Intracellular c-di-AMP.

T. denticola 35405, 33520, SP50, and SP55 cultures were prepared by inoculating 30 ml of sterile, pre-reduced NOS and grown to mid-log phase (4 days) in anaerobic conditions at 37 °C. As a control, E. coli BL21 (DE3) cultures were made by inoculating 30 ml of sterile Luria-Bertani (LB) broth for 18 hr in aerobic conditions at 37 °C. All cultures were centrifuged at 5,000x g for 10 min to pellet cells. The supernatant was discarded, and the pellet was frozen and stored at −80 °C until processing. Frozen pelleted cell samples were then submitted to the VCU Massey Comprehensive Cancer Center Lipidomics and Metabolomics Shared Resource (LMSR) to quantify intracellular c-di-AMP. The frozen cell pellets were thawed and resuspended in 120 µl 80% cold methanol before sonication and mixed with 25 µM (¹³C,¹⁵N)-c-di-AMP (Biolog) and centrifuged. Methanol was collected after centrifugation and dried using a speed-vac concentrator, and the resulting pellets were reconstituted in 150 µl H₂O for LC-MS/MS analysis. Intracellular levels of c-di-AMP in samples were then assessed via mass spectroscopy with a 0.02 – 10 µM c-di-AMP curve in place for quantification. The total protein of pelleted cellular material was evaluated using a Qubit 4 Fluorometer (Invitrogen) for normalization. Data reported are the average and standard deviation of 5 biological replicates for each T. denticola strain and 2 replicates for E. coli.

Generation and purification of recombinant proteins.

A partial TDE1909 gene from 35405, starting at amino acid 87, was commercially synthesized and cloned onto the pET-45b(+) plasmid (GenScript). The partial gene did not include the 3 transmembrane domains of TDE1909 to aid in protein expression. The removal of the transmembrane domains does not impact the DAC activity of previously characterized CdaA proteins (Rosenberg et al., 2015). The T. denticola 35405 TDE0027, TDE0087, TDE0294, and TDE1908 coding regions were amplified by PCR using primers listed in Supplemental Table 2 and cloned into pET-45b(+) (GE Healthcare). The resulting plasmids were transformed into E. coli BL21(DE3), and N-terminal 6x-histidine tagged proteins were purified using nickel-nitrilotriacetic acid agarose (Qiagen). Site-specific mutations were introduced into cdaA of pET-45b(+)-TDE1909 using a Q5 site-directed mutagenesis kit with the primers listed in Supplemental Table 2. The following modifications of CdaA in conserved motifs were created: G171A and S221A. All gene and plasmid sequences were confirmed on a fee-for-service basis. All recombinant proteins were soluble and purified under native conditions. After buffer exchange into phosphate-buffered saline (PBS), the purity of the recombinant proteins was assessed using SDS-PAGE and Coomassie staining, and concentration was determined by BCA assay.

Quantification of c-di-AMP by ELISA.

E. coli BL21 (DE3) cells containing either empty pET-45b(+), pET-45b(+)-TDE1908, or pET-45b(+)-TDE1909 were grown, followed by a standard IPTG induction of protein expression. Cells were harvested by centrifugation (5,000x g for 5 min) and lysed with B-PER Complete (ThermoFisher). The concentration of c-di-AMP in the lysates was determined using a c-di-AMP ELISA (Cayman Chemicals), and total protein was determined by a BCA assay (ThermoFisher). The experiment was repeated with 3 biological replicates and 3 technical replicates per condition.

Demonstration of diadenylate cyclase activity.

Recombinant protein TDE1909 and TDE1908 (5 µM) were incubated with 150 µM ATP in 300 µl total volume DAC buffer (10 mM Tris HCl pH 7.5, 50 mM NaCl, 0.5 mM EDTA, and 10 mM MnCl₂, 10 mM MgCl₂ or 10 mM CoCl₂) for 1 to 4 hrs at 37 °C. The reaction was stopped by boiling at 99°C for 3 min before centrifugation at 15,000x g for 2 min. The supernatant was filtered through a 0.22 µm PVDF membrane into a standard 300 µl HPLC vial. As controls, 150 µM ATP and 150 µM c-di-AMP without proteins were treated as described. Nucleotides were separated by reverse phase chromatography using a Supelco supelcosil LC-18-T columns on an Agilent 1260 infinity II system (Patel et al., 2021). Buffers consisting of 4 mM tetrabutyl ammonium hydrogen sulfate, 100 mM KH₂PO₄ (pH 5.9; Buffer A), and 100% methanol (Buffer B). A 0–50% buffer B gradient over 20 minutes separated the nucleotides.

Identification of c-di-AMP binding proteins.

T. denticola 35405 was grown to mid-log phase in NOS before centrifugation for 10 min at 5,000x g to pellet cells. The supernatant was discarded, and the pellet was frozen at −80°C before lysis. The pellet was resuspended in 10 ml of lysis buffer (PBS, 7.4 pH, 10 mM PMSF, 1x Protease/Phosphatase Inhibitor Cocktail (Cell Signaling) and 0.5% Triton X-100) and incubated on ice for 5 min before sonication at 5 sec intervals for 10 min. The sample was centrifuged at 5,000x g to pellet cellular debris. The supernatant was passed through a 0.22 µm PVDF filter. The cell-free supernatant was applied to either a c-di-AMP-embedded resin column or an ethanolamine control column. The columns were washed with 10 ml lysis buffer. Proteins were then eluted and collected from the columns using 9 M urea. Duplicate elutions from each column were provided to the Massey Comprehensive Cancer Center Proteomics Shared Resource for protein identification by LC-MS/MS.

Sample preparation and LC-MS/MS.

The samples were digested using commercially available Preomics iST sample clean-up protocol. The elutes were placed in a vacuum evaporator at 45°C until completely dried. Digests were then collected for mass LC-MS/MS analysis. LC-MS/MS analyses were performed using a Q-Exactive HF-X (Thermo Fisher Scientific) tandem mass spectrometer coupled to an Easy nLC 1200 (Thermo Fisher Scientific ) nanoflow UPLC system. The LC-MS/MS system was fitted with an Easy spray ion source and an Acclaim PepMap 75 µm x 2 cm nanoviper C18 3 µm x 100 Å pre-column in series with an Acclaim PepMap RSLC 75 µm x 50 cm C18 2 µm bead size (Thermo Fisher Scientific). The mobile phase consists of Buffer A (0.1% formic acid in water) and Buffer B (80% acetonitrile in water, 0.1% formic acid). The peptides were injected onto the above column assembly and eluted with an acetonitrile/0.1% formic acid gradient at a 300 nL/min flow rate over 1.6 hours. The nano-spray ion source was operated at 1.9 kV. The digests were analyzed using a data-dependent acquisition (DDA) method, acquiring a full scan mass spectrum (MS) followed by 15 tandem mass spectra (MS/MS) in the high energy C- trap (Dissociation HCD spectra). The data from the LC-MC/MS was then analyzed in Proteome Discoverer (v3.0) using the SEQUEST HT search algorithm and a custom T. denticola and contaminant protein database. Proteins were identified at a False-Discovery Rate (FDR) <0.01.

Binding of c-di-AMP to receptor proteins.

Following identifying putative binding proteins, recombinant proteins for TDE0087 and TDE0294 were produced using the techniques described above. CabP, a c-di-AMP binding protein from S. pneumoniae cloned into pET-28 was kindly provided by Dr. Bai at Albany Medical College as a positive control (Bai et al., 2014). CabP was expressed and purified using the standard approaches described above. Each protein was diluted in carbonate buffer (12.5 mM sodium bicarbonate and 87.5 mM sodium carbonate) at 10 µg/ml, applied to wells of microtiter plates, and incubated at 4°C overnight. The buffer was decanted, and the wells were washed 3 times with 200 µl PBST (PBS+ 0.05% Tween 20). Wells were then blocked with 1% bovine serum albumin (BSA) in PBST. The blocking buffer was decanted, and the wells were washed 3 times with 200 µl PBST. The wells were overlayed with either 100 µl 125 nM biotinylated-c-di-AMP (Biolog) diluted in binding buffer (50 mM Tris-HCl (pH 8.0), 750 mM KCl, 2.5 mM EDTA, 0.5% Triton-X 100, 1 mM DTT) or binding buffer alone and incubated for 2 hr before washing 3 times with PBST. 100 µl of HRP-conjugated streptavidin (1:10,000 dilution in PBST) was added to each well and incubated for 1 hr. The wells were washed 5 times with PBST, and binding was detected using TMB-One Step substrate (ThermoFisher), and the reaction stopped by adding 2 M H₂SO₄. Absorbance was measured at 405 nm using a SpectraMax iD3 (Molecular Devices). Each condition was repeated with 8 technical replicates per plate, and the entire experiment was repeated in triplicate. One-way ANOVA analyzed the average binding with Tukey’s post hoc test.

RESULTS

T. denticola produces intracellular c-di-AMP.

Analysis of the T. denticola genome revealed that T. denticola encodes a single CdaA-like diadenylate cyclase and one annotated c-di-AMP-specific phosphodiesterase, suggesting it may utilize c-di-AMP signaling. To initialize the characterization of c-di-AMP signaling in T. denticola, we sought to determine if T. denticola produces intracellular c-di-AMP using LC-MS/MS and (¹³C,¹⁵N)-c-di-AMP and an internal standard. We analyzed cell lysates from 5 independent cultures of two laboratory strains (ATCC 35405 and 33520) and two low-passage clinical isolates (SP50 and SP55) (Figure 1). As a negative control, the intracellular levels of c-di-AMP from the laboratory E. coli strain BL21(DE3) (n=2), which does not utilize c-di-AMP signaling, were also investigated (Rosenberg et al., 2015). LC-MS/MS analyses detected c-di-AMP in all strains of T. denticola ranging between 2–8 µM c-di-AMP per mg of total protein and failed to detect any c-di-AMP from E. coli, as expected. We observed the ATCC 33520 strain produced more c-di-AMP than the other strains of T. denticola. No significant differences in c-di-AMP levels was observed between the other strains. The natural variation in basal c-di-AMP levels between strains is not well studied.

Figure 1. T. denticola produces c-di-AMP during in vitro growth.

Each T. denticola strain was grown to mid-log phase in NOS prior to harvesting cells. The E. coli laboratory strain BL21(DE3) served as a negative control. Nucleotides were extracted from cell pellets in 80% methanol and c-di-AMP was quantified by LC-MS/MS using an internal standard. As expected, c-di-AMP was not detected (ND) in the E. coli control samples. Data was normalized to total protein. Data are the average with standard deviation for the T. denticola strains (n=5) and duplicate cultures from E. coli. Data was analyzed by ANOVA with Tukey’s post-hoc test.

Identification of TDE1909 as a CdaA-type DAC.

TDE1909 is annotated as a CdaA-like diadenylate cyclase and is the only protein predicted to produce c-di-AMP in T. denticola (Seshadri et al., 2004). The adjacent gene, TDE1908, encodes a protein with an N-terminal transmembrane domain and three YbbR domains, and is a homolog of CdaR, the regulator of CdaA activity in many bacteria. To investigate the function of TDE1909, we cloned and purified a recombinant protein of TDE1909 that removed the three N-terminal predicted transmembrane helices (the recombinant protein started at TDE1909 residue 87). We cloned the full-length TDE1909 gene but failed to induce protein in E. coli until the transmembrane domains were removed, consistent with the CdaA protein from L. monocytogenes (Rosenberg et al., 2015). The CdaA (TDE1909) and CdaR (TDE1908) proteins were expressed in E. coli, which does not utilize c-di-AMP signaling, and the production of c-di-AMP was quantified using an ELISA (Cayman Chemicals). C-di-AMP was only detected in cells expressing CdaA (TDE1909) but not in cells expressing CdaR (TDE1908) or with empty pET-45b(+) vector, suggesting TDE1909 is a DAC (hereafter referred to as CdaA) responsible for the production of c-di-AMP (Figure 2A). To investigate the activity of CdaA further, purified recombinant protein was incubated with 150 µM ATP for either 1 or 4 hr at 37°C and analyzed the reaction for the depletion of ATP and/or the synthesis of c-di-AMP by reverse phase high-pressure liquid chromatography (RP-HPLC). Here, we observed that CdaA converted ATP to c-di-AMP while CdaR did not (Figure 2B).

Figure 2. TDE1909 synthesizes c-di-AMP.

(A) The genes for TDE1909 and TDE1908 were cloned into pET-45b(+) and transformed into E. coli BL21(DE3) cells. Cells with empty pET-45b(+) served as a negative control. Protein expression was induced with IPTG prior to determining the intracellular levels of c-di-AMP by ELISA (Cayman Chemicals). Data are the average and standard deviation from replicate cultures (n=3) analyzed by ANOVA with Dunnett’s post-hoc test. (B) Recombinant proteins for TDE1909 and TDE1908 were incubated with 150 µM ATP for 1 hr and reactions products were analyzed by RP-HPLC. Purified ATP and c-di-AMP were fractionated as controls to determine elution profiles for each nucleotide. Data shown is representative of triplicate experiments.

Consistent with the observation that TDE1909 is a CdaA-type DAC, we performed a sequence alignment of the DAC domain of TDE1909 to the CdaA homologs from S. pneumoniae and L. monocytogenes and the HDG, RHR, and SEET motifs that are critical for DAC activity are perfectly conserved in TDE1909 (Figure 3A) (Bai et al., 2013; Rosenberg et al., 2015). The predicted structure of CdaA from AlphaFold2( 82.3 average pLDDT) was compared to the solved structure of CdaA from L. monocytogenes (LMO2120; PDB: 8C4O) by overlaying the catalytic domains of both proteins in PyMol (Figure 3B) (Neumann et al., 2023). The predicted structure of CdaA from T. denticola closely overlapped the known structure of the homolog from L. monocytogenes, allowing for a prediction of metal coordination and ATP binding by CdaA.

Figure 3. Sequence and structural comparison of TDE1909 to CdaA proteins.

(A) The protein sequence for TDE1909 was aligned to the CdaA sequence from L. monocyotgenes (LmCdaA) and S. pneumoniae using Clustal X. Highly conserved regions involved in metal coordination in the binding site are highlighted in red boxes. (B) The predicted structure of TDE1909 (cyan) from AlphaFold2 was aligned to the structure of LmCdaA (gray) co-crystallized with ATP (PDB: 8C4O) using PyMol. The Co⁺² ions are magenta, the ATP bound to LmCdaA is blue and the conserved regions on both TDE1909 and LmCdaA are red.

Based on the sequence and structural homology with proteins from S. pneumoniae and L. monocyotogenes, structure-function analyses of CdaA were initiated by examining metal ion requirements and site-directed substitutions within predicted metal binding motifs that likely impact DAC activity of CdaA. Because all DACs require metal ions as cofactors for activity, CdaA metal-dependent DAC activity was examined and the highest ctivity was observed when Mn⁺² was present, but Co⁺² also supported DAC activity while Mg⁺² did not allow for the synthesis of c-di-AMP (Figure 4A). Structure-function analyses of CdaA from T. denticola was initiated by substituting the Gly171 adjacent to the HDG catalytic domain with alanine. Based on the LM2120 structures, we anticipated that H169 and Asp170 in CdaA would coordinate the metal ion. The G171A substitution was predicted to abolish DAC activity. Based on crystallographic studies of LMO2120, Glu223 likely coordinates Mn⁺², so the S221 residue was substituted with alanine. Here, we anticipated that targeting the residue adjacent to ATP binding but not directly shown in any study to coordinate metal and ATP binding would attenuate DAC activity. The G171A and S221A CdaA substitutions were incubated with ATP and RP-HPLC was performed to assess activity. Interestingly, both G171A and S221A substitutions are essential for the synthesis of c-di-AMP by CdaA (Figure 4B). These data strongly suggest that the T. denticola CdaA binds metal ions and ATP utilizing the same molecular mechanism described for the CdaA protein in L. monocytogenes (LMO2120).

Figure 4. TDE1909 requires Mn⁺² or Co⁺² for DAC activity.

(A) Recombinant TDE1909 was incubated with 150 µM ATP for 4 hr in buffer containing 10 mM of MnCl₂, CoCl₂, or MgCl₂ and reactions products were analyzed by RP-HPLC. Purified ATP and c-di-AMP were fractionated as controls to determine elution profiles for each nucleotide. (B) Residues G171 and S221 were substituted with alanine by Q5 site-directed mutagenesis (NEB) and recombinant proteins with amino acids substitutions were purified and incubated with 150 µM ATP for 4 hr. The native TDE1909 protein was also run as a control along with purified ATP and c-di-AMP as controls.

Data shown is representative of duplicate experiments.

Identification and characterization of c-di-AMP receptor proteins in T. denticola.

Having established that T. denticola produces c-di-AMP and identifying CdaA as the DAC responsible for c-di-AMP synthesis, we next wanted to identify c-di-AMP receptors in T. denticola that may be effectors of c-di-AMP-mediated signaling. Here, T. denticola ATCC 35405 was grown, and cell-free lysates were generated, which were passed over a resin column with c-di-AMP immobilized or an ethanolamine control column. Duplicate elutions were collected from both columns and identified the proteins in solution by LC-MS/MS compared to a custom database of T. denticola ATCC 35405 and common contaminant proteins. The number of peptide-spectrum matches (PSMs) were utilized to semi-quantitatively determine whether a peptide was more abundant in elutions from the c-di-AMP column or the control column. The 10 T. denticola proteins with the largest difference in PSMs between the c-di-AMP and the control columns are shown in Figure 5 and the complete dataset is available in Supplemental Table 3. This approach was utilized as a screen to identify the most promising c-di-AMP receptors, and all putative receptors require additional binding studies to verify c-di-AMP binding. However, several important insights support the validity of this approach. Three of the most likely c-di-AMP receptors are involved in potassium uptake (TDE0087, TDE0294, and TDE2438), the most ubiquitous function of c-di-AMP signaling across bacteria (Quintana et al., 2019; Zarrella et al., 2020). Additionally, TDE0322 is a P-II family regulatory protein, a class of protein already characterized as c-di-AMP receptors (Choi et al., 2015).

Figure 5. Ten putative c-di-AMP receptor proteins with the largest difference in PSMs detected between the experimental and control columns.

T. denticola lysate was either passed through an experimental column with c-di-AMP immobilized or a control column with ethanolamine. Columns were extensively washed prior to elution of all proteins with 5 mL of 9M urea. Proteins in duplicate elutions from each column were identified by LC-MS/MS. The total peptide-spectrum matches (PSMs) from the c-di-AMP (blue) and ethanolamine (yellow) columns are shown for the 10 T. denticola proteins with the largest difference in PSMs detected.

Based on prior studies demonstrating c-di-AMP binds to TrkA-like proteins involved in potassium uptake, TDE0087 and TDE0294 were prioritized for validation studies (Bai et al., 2014; Commichau et al., 2018; Gundlach et al., 2019; Pham et al., 2018; Quintana et al., 2019; Zarrella et al., 2020). Recombinant TDE0087 and TDE0294 proteins were tested for the ability to bind c-di-AMP using an ELISA-based approach. CabP, a TrkA-family protein from S. pneumoniae served as our positive control (Bai et al., 2014; Underwood et al., 2014). We observed that TDE0087 bound c-di-AMP similarly to the CabP control, while no significant binding was detected to either TDE0294, CdaA, or BSA (Figure 6A).

Figure 6. TDE0087 is a c-di-AMP receptor and osmolarity impacts c-di-AMP levels in T. denticola.

(A) Recombinant TDE0087 and TDE0294 were tested for binding to biotinylated c-di-AMP. CabP from S. pneumoniae and BSA served as positive and negative controls, respectively. TDE0087 bound c-di-AMP (****p<0.0001) while no significant (ns) binding was observed for TDE0294. Data are representative of at least 3 independent experiments and analyzed by one-way ANOVA with Tukey’s post-hoc test. (B) T. denticola 35405 was grown and placed in a 4:1 mixture of NOS and PBS at various osmotic strengths for 1 hr before harvesting cells for c-di-AMP quantification by LC-MS/MS. Data are the average and standard deviation from c-di-AMP quantified from 5 independent cultures normalized for total protein. Data were analyzed by one-way ANOVA with Tukey’s post-hoc test (****p<0.001, **p<0.01).

Levels of c-di-AMP respond to changes in osmolarity.

Based on the published literature, the results of the LC-MS/MS screen, and validation c-di-AMP binding studies, we hypothesized that intracellular c-di-AMP levels would respond to changes in osmolarity. C-di-AMP levels do not change between T. denticola grown in NOS broth or NOS diluted with PBS (4:1) (data not shown). This suggested reducing the nutrient availability in 80% NOS broth did not impact c-di-AMP signaling. To investigate c-di-AMP signaling in response to osmolarity, mid-log phase T. denticola was exposed to 80% NOS diluted with PBS of various ionic strengths (0.5x, 0.75x, 1x, 1.25x, and 1.5x) for 1 hr before collecting lysates for c-di-AMP quantification. C-di-AMP concentrations were observed to increase with reduced osmotic strength (Figure 6B). These data further support the role of c-di-AMP in regulating osmolarity in T. denticola.

Characterization of TDE_0027 as a phosphodiesterase that hydrolyzes c-di-AMP.

Analysis of the T. denticola genome identified three genes (TDE0027, TDE1479, and TDE2122) that encode proteins with DHH/DHHA1 domains, consistent with DhhP-family phosphodiesterases (Konno et al., 2018; Kundra et al., 2021; Ye et al., 2014). Interestingly, TDE0027 was identified as a likely c-di-AMP binding protein in the LC-MS/MS screen (Figure 5), warranting further characterization. To investigate whether TDE0027 is a c-di-AMP phosphodiesterase, recombinant TDE0027 was incubated with bis-para-nitrophenyl phosphate (bis-pNPP), a generic colorimetric substrate for PDE activity containing a diester bond, or para-nitrophenyl phosphate (pNPP), a similar substrate for phosphatase activity with a monoester bond. TDE0027 hydrolyzed bis-pNPP but not pNPP, and that hydrolysis was Mn⁺²-dependent while Mg⁺² did not support the PDE activity of TDE0027 (Figure 7A). The hydrolysis of bis-pNPP was measured with increasing concentrations of TDE0027 and monitored the hydrolysis over 30 min (Figure 7B–C). To characterize the specific reaction catalyzed by TDE0027, recombinant TDE0027 was incubated with either c-di-AMP or pApA and the products generated by RP-HPLC were determined by comparing to the retention times (Rt) to several purified standards (9.3 min for AMP, 14.2 min for pApA, and 15.9 min for c-di-AMP). TDE0027 fully hydrolyzed c-di-AMP to pApA after 4 hours as no c-di-AMP peak remained at ~16 min and the only detectable peak was at 14.2 min (Figure 7D). When the reaction condition was repeated with the inclusion of 10 mM EDTA to chelate the Mn⁺² from the reaction buffer, most of the substrate remained and only a small peak was observed for pApA. We did not observe the generation of AMP when TDE0027 was incubated with c-di-AMP as the substrate. To further validate that TDE0027 only hydrolyzes c-di-AMP into pApA, TDE0027 was incubated with pApA as the substrate, and no significant loss of the substrate was observed following a 4 hr incubation along with no detectable peak at 9.3 min, corresponding to AMP. These results demonstrate that TDE0027 is a Mn⁺²-dependent phosphodiesterase that degrades c-di-AMP to pApA. Hereafter, TDE0027 is referred to as DhhA.

Figure 7. TDE0027 is a phosphodiesterase that hydrolyzes c-di-AMP into pApA.

(A) Recombinant TDE0027 was incubated with either bis-pNPP or pNPP in buffer containing MnCl₂ or MgCl₂ for 30 min. Hydrolysis of both substrates was monitored at 405 nm. Data are the average from 3 replicates with standard deviation analyzed by one-way ANOVA and Dunnett’s post-hoc test. (B) Hydrolysis of bis-pNPP after 30 min was measured at 405 nm with increasing amounts of TDE0027 enzyme (n=3). (C) The hydrolysis of bis-pNPP was monitored at 405 nm every 5 min with various amounts the TDE0027 enzyme (n=3). (D) The products generated after incubating TDE0027 with c-di-AMP and pApA for 4 hr were identified by RP-HPLC. As a control, TDE0027 was incubated with c-di-AMP in a buffer containing 10 mM EDTA. Purified AMP, c-di-AMP and pApA were also fractionated to determine retention times in the column. Data are representative of duplicate experiments.

DISCUSSION

This is the first demonstration that Treponema produces the nucleotide secondary messenger c-di-AMP (Figure 8). Signaling via c-di-AMP was initially thought to be exclusive to Gram-positive bacteria, and while it is now recognized to contribute to some Gram-negative organisms, the best characterized c-di-AMP signaling systems remain predominantly in Gram-positive species (Yin et al., 2020). Our findings contribute to the growing evidence that c-di-AMP signaling is common to pathogenic spirochetes, which are Gram-negative in cell structure. Several studies have shown the Borrelia spirochetes associated with Lyme disease and relapsing fever produce c-di-AMP, and the genome of Leptospira interrogans possesses genes annotated to encode for c-di-AMP metabolizing enzymes (Jackson-Litteken et al., 2021; Savage et al., 2015; Ye et al., 2014). T. denticola and several other oral Treponema spp. are unique among the pathogenic spirochetes as strict anaerobes. C-di-AMP signaling has been demonstrated for C. difficile and P. gingivalis, but its impact on physiology and pathogenesis remains poorly characterized in anaerobic bacteria (Moradali et al., 2022; Oberkampf et al., 2022). Further detailed study of c-di-AMP signaling in T. denticola will be critical to understanding the signaling systems T. denticola uses to sense and respond to its dynamic environment but will also contribute significant new insights to c-di-AMP-mediated regulation in spirochetes and strict anaerobes.

Figure 8. Model of c-di-AMP signaling in T. denticola.

CdaA (TDE1909) synthesizes c-di-AMP from ATP and CdaA activity is likely regulated by CdaR (TDE1908). Intracellular levels of c-di-AMP are maintained by the PDE DhhA (TDE0027) and possibly PgpH (TDE1241). C-di-AMP binds to a putative potassium uptake protein, TrkA (TDE0087), and may inhibit the import of potassium into the cell under hypoosmotic stress.

In Firmicutes, the gene for CdaA is in a well-conserved operon with genes encoding for CdaR and GlmM, a phosphoglucomutase involved in peptidoglycan biosynthesis (Galperin, 2023). In addition to all three genes being co-transcribed, the proteins form a complex with both CdaR and GlmM regulating the DAC activity of CdaA (Gibhardt et al., 2020; Pathania et al., 2021; Tosi et al., 2019). The gene arrangement in T. denticola, and spirochetes in general, is distinct from the well-conserved operon in Firmicutes (Supplemental Figure 1). In T. denticola, the cdaA and cdaR genes are in a four-gene operon that includes TDE1910 (a 1-deoxy-D-xylulose-5-phosphate (DXP) synthase) and TDE1908 (a holo-(acyl-carrier-protein) synthase). Additionally, T. denticola lacks a homolog for glmM. This operon structure is conserved among Treponema spp. but distinct in Borrelia spp. which retain the holo-(acyl-carrier-protein) synthase and several other genes downstream of cdaA-cdaR but do not encode for a DXP synthase homolog. While unstudied in T. denticola, DXP is an essential bacterial metabolite that feeds into central metabolism and the biosynthesis of terpenoids, precursors for peptidoglycan biosynthesis (Bartee & Freel Meyers, 2018). It is worth considering that c-di-AMP is known to regulate cell membrane homeostasis, and the gene encoding CdaA in T. denticola is flanked by genes that may impact cell membrane homeostasis (Yin et al., 2020). It is unclear if the proteins encoded with CdaA are regulated by c-di-AMP or can influence c-di-AMP levels like GlmM. Future studies will assess if these protein functions are impacted by c-di-AMP signaling and if c-di-AMP contributes to cell membrane homeostasis in T. denticola.

CdaA is an attractive drug target as c-di-AMP signaling is essential to bacterial survival and c-di-AMP signaling is absent in mammals (Rosenberg et al., 2015). C-di-AMP signaling frequently regulates cell membrane and cell wall homeostasis (Corrigan et al., 2011; Witte et al., 2013). Disruption of c-di-AMP-mediated regulation of cell structure enhances the susceptibility of cell-wall targeting antibiotics (Luo & Helmann, 2012; Whiteley et al., 2017). Drugs designed to abolish CdaA activity could be an avenue for future stand-alone therapeutic or antibiotic adjuvants. Structure-function analysis of CdaA revealed Mn⁺² and Co⁺² coordination occurs through the conserved HDG and SEET motifs. The development of pharmacological inhibitors of DAC activity is based on LMO2120 crystallographic and binding studies (Neumann et al., 2023). Our data suggests future development of these inhibitors would likely abolish CdaA activity in T. denticola and be a potential avenue for future therapeutic development for periodontal disease.

Several proteins with no homology with known c-di-AMP binding proteins were identified by the LC-MS/MS screen. TDE0174 is annotated as a nicotinate phosphoribosyltransferase that catalyzes the first reaction in the synthesis of NAD and is the protein with the highest abundance in the c-di-AMP column. TDE1725 and TDE2293 are conserved hypothetical proteins that are uncharacterized in T. denticola but have homology with a protein annotated as a chromosomal segregation ATPase in T. pallidum. TDE0460 is uncharacterized but has protein domains that suggest it functions as a transporter that is induced during phosphate starvation, consistent with the role of c-di-AMP as a regulator of nutrient availability and osmolytes. TDE0870 is annotated as a putative 2’,3’-cNMP phosphodiesterase. As this is the first study to identify c-di-AMP receptors in a spirochete or a strict anaerobe, it is likely several of these proteins are novel c-di-AMP receptors, and our future studies will continue to explore the role of c-di-AMP signaling in T. denticola.

TDE0294 was strongly enriched in the screen but ELISA-based validation determine TDE0294 does not directly bind c-di-AMP, suggesting this protein likely interacts with another c-di-AMP binding protein. Whether this is TDE0087 or a yet-to-be-identified receptor will be investigated in future studies. While binding of c-di-AMP to proteins involved in the regulation of osmolarity and potassium transport is well established, this report identified the first c-di-AMP receptor in a spirochete and suggests c-di-AMP likely regulates responses to changes in osmolarity in T. denticola (Commichau et al., 2018). This is particularly relevant to T. denticola as the sulcular crevice is known to be a dynamic environment with routine changes to pH, redox potential, microbial composition, and nutrient availability, especially as the periodontal environment transitions from health to disease (Diaz & Valm, 2020; Mark Welch et al., 2020; Mark Welch et al., 2016). While c-di-AMP regulation of osmolyte transport is nearly universal in bacteria, the mechanisms are variable. In Lactococcus lactis, c-di-AMP binding to the KupA and KupB inhibits K⁺ transport while c-di-AMP binding promotes K⁺ import by binding the CpaA protein in S. aureus (Mukkayyan et al., 2022; Quintana et al., 2019). This study suggests that c-di-AMP binding to TDE0087 likely inhibits K⁺ uptake, and reduced c-di-AMP will result in K⁺ uptake when T. denticola is in a hypertonic environment.

DhhA (TDE0027) was identified as a DhhP-family PDE, but T. denticola possesses genes that encode for two other DHH/DHHA1 domain-containing proteins (TDE1479 and TDE2122) that were not experimentally tested in this study. Interestingly, TDE1479 has homology with the nano-RNase, NrnA, from L. monocytogenes that preferentially degrades the linear dinucleotides pApA and pGpG to AMP and GMP, respectively (Gall et al., 2022). The T. denticola ATCC 35405 contains an annotated PgpH-like c-di-AMP phosphodiesterase (TDE1241). However, efforts to clone the N-terminal 7TMR-HDED domain and the C-terminal HD-domain of TDE1241 failed to produce active PDEs. The cloning strategy to separate each domain of the TDE1241 protein may prevent proper protein folding or may have eliminated some regulatory elements that influence TDE1241 activity. We will continue alternative cloning strategies to determine if TDE1241 is a c-di-AMP-specific PDE, or future studies may generate a TDE1241 knockout in T. denticola to assess its role in c-di-AMP signaling. The CdaA (TDE1909) and PgpH genes (TDE1241) are found in every strain of T. denticola and every species of Treponema in the human oral microbiome database (HOMD) (Supplemental Figure 2) (Escapa et al., 2018). The CdaA, DhhA, and PgpH homologs in ATCC 35405 and ATCC 33520 are 100%, 97%, and 99% identical, respectively, yet we observed variations in c-di-AMP levels between these strains. The mechanisms or regulation that explain the differences in observed c-di-AMP levels are unknown, and whether these differences are biologically relevant. Interestingly, DhhA (TDE0027) is absent in 3 of the T. denticola strains, and homologs are only present in T. putidum and T. medium (Escapa et al., 2018). This may suggest CdaA and PgpH were ancestral acquisitions that have co-evolved with treponemes, while DhhP was more recently acquired through horizontal gene transfer. The role each PDE may play in regulating c-di-AMP levels and T. denticola physiology will be determined in future studies. In addition to hydrolysis of c-di-AMP, L. monocytogenes and Lactococcus lactis utilize MDR-family transporters to secrete c-di-AMP to prevent toxic accumulation of c-di-AMP (Kaplan Zeevi et al., 2013; Pham et al., 2018). The gene encoding DhhA is adjacent to genes encoding a multidrug resistance (MDR)-family ABC-transporter (TDE0028-TDE0029). While this study did not explore the secretion of c-di-AMP by T. denticola, it is worth future investigation as both a means to reduce intracellular c-di-AMP levels, but it may also contribute to periodontal inflammation. STING, RECON, and DDX41 are host receptors for c-di-AMP that, upon stimulation, induce NF-kB and interferon responses (Cheng et al., 2022). The degree to which c-di-AMP promotes inflammation is evident by its development as a vaccine adjuvant (Ebensen et al., 2019; Skrnjug et al., 2014). Interestingly, c-di-AMP was found to be a strong inducer of pro-inflammatory cytokines in gingival epithelial cells, however the role of c-di-AMP in host-pathogen interactions that contribute to periodontitis are unknown (Elmanfi et al., 2019).

CONCLUSIONS

In this study, we demonstrated that T. denticola produces the nucleotide secondary messenger, c-di-AMP. We characterized the TDE1909 gene encodes for a CdaA-family DAC that coordinates Mn⁺² and Co⁺² ions to highly conserved motifs within the DAC domain of TdCdaA. We used c-di-AMP affinity column enrichment and LC-MS/MS to identify possible c-di-AMP receptor proteins and validated that TDE0087, a K⁺ uptake protein binds c-di-AMP. We observed that c-di-AMP levels were elevated when T. denticola was exposed to hypotonic media. We characterized TDE0027 as a DHH/DHHA1-domain containing PDE that hydrolyzes c-di-AMP into pApA. Collectively, this work describes a functional c-di-AMP signaling system in T. denticola contributing new insights into both c-di-AMP signaling in bacteria and defining a new sensory transduction system in T. denticola. Continuation of this work will validate additional c-di-AMP receptors characterize their role in regulation cellular processes because of c-di-AMP binding. While many c-di-AMP receptors have been identified in bacteria, the molecular details of how c-di-AMP binding impacts protein function remains largely unknown. As alluded to, c-di-AMP signaling is poorly described in both spirochetes and strict anaerobes, making the study of T. denticola a great opportunity to fill gaps in knowledge across multiple disciplines.

DATA AVAILABILITY STATEMENT:

The data that support the findings of this study are available from the corresponding author upon reasonable request.