Biological and Biochemical Roles of Two Distinct Cyclic Dimeric Adenosine 3′,5′-Monophosphate- Associated Phosphodiesterases in Streptococcus mutans

Cyclic dimeric adenosine 3′,5′-monophosphate (c-di-AMP), a recently identified secondary messenger in bacteria, plays a role in several bacterial processes, including biofilm formation. It is enzymatically produced by diadenylate cyclase and cleaved by c-di-AMP phosphodiesterase. c-di-AMP is believed to be essential for the viability of bacterial cells that produce it. In the current study, the biochemical and biological roles of GdpP (SMU_2140c) and DhhP (SMU_1297), two distinct Streptococcus mutans phosphodiesterases involved in the pathway producing AMP from c-di-AMP, were investigated. Liquid chromatography-tandem mass spectrometry revealed that c-di-AMP was degraded to phosphoadenylyl adenosine (pApA) by truncated recombinant GdpP, and pApA was cleaved by recombinant DhhP to yield AMP. In-frame deletion mutants lacking the dhhP gene (ΔdhhP) and both the gdpP and dhhP genes (ΔgdpPΔdhhP) displayed significantly more biofilm formation than the wild-type and a mutant strain lacking the gdpP gene (ΔgdpP; p < 0.01). Furthermore, biofilm formation was restored to the level of the wild type strain upon complementation with dhhP. Optical and electron microscopy observations revealed that ΔdhhP and ΔgdpPΔdhhP mutants self-aggregated into large cell clumps, correlated with increased biofilm formation, but cell clumps were not observed in cultures of wild-type, ΔgdpP, or strains complemented with gdpP and dhhP. Thus, deletion of dhhP presumably leads to the formation of bacterial cell aggregates and a subsequent increase in biofilm production.


INTRODUCTION
Streptococcus mutans is a major etiological agent of dental caries in humans (Hamada and Slade, 1980), a biofilm-associated infectious disease and important health problem worldwide (Thenisch et al., 2006). This microorganism has the capacity to prevail in the complex microbial community of an oral biofilm under the low pH conditions responsible for tooth demineralization, and can adapt to the stressful conditions to which the cariogenic biofilms are exposed (Lemos and Burne, 2008).
c-di-AMP is enzymatically produced by diadenylate cyclases and cleaved by c-di-AMP phosphodiesterases. Diadenylate cyclases typically contain a diadenyl cyclase domain that is crucial for the synthesis of c-di-AMP from two molecules of ATP (Corrigan and Grundling, 2013), while phosphodiesterases degrading c-di-AMP are divided into two distinct classes. The first class is characterized by a catalytically active Asp-His-His/Asp-His-His-Ala (DHH/DHHA1) domain, whereas the second class contains a His-Asp (HD) domain in the active center (Huynh et al., 2015). DHH/DHHA1 domain-containing proteins responsible for c-di-AMP degradation are further divided into two subfamilies. The first subfamily comprises homologs of membrane-bound GdpP proteins that contain two N-terminal transmembrane helices, a Per-Arnt-Sim (PAS) domain, a degenerate GGDEF domain, and a catalytic DHH/DHHA1 domain (Rao et al., 2010). PAS domains usually function as sensory domains for detecting light, redox potential, and oxygen (Taylor and Zhulin, 1999). However, the domain appears to function for GdpP, since GdpP without the PAS domain was enzymatically equivalent to that with the domain (Bai et al., 2013). Proteins belonging to the other subfamily are homologs of the soluble DhhP protein from Borrelia burgdorferi that contains only a cytosolic core DHH/DHHA1 domain (Ye et al., 2014). It is interesting that Streptococcus pneumoniae harbors two DHH/DHHA1 domain-containing phosphodiesterases from both subfamilies, named Pde1 and Pde2 (Bai et al., 2013), which are orthologues of GdpP and Dhhp. The enzymatic activity of Pde1 is the same as that of membrane-bound GdpP proteins degrading c-di-AMP to pApA, whereas Pde2 cleaves both c-di-AMP and pApA to yield AMP. A recent study revealed that deletion of the gdpP gene (SMU_2140c) in S. mutans increased both the intracellular c-di-AMP concentration and biofilm production in the presence of sucrose (Peng et al., 2016). Since no ortholog of Pde2 in S. mutans, Dhhp (SMU_1297), has yet been fully investigated, the second putative phosphodiesterase should be characterized. In the current study, the S. mutans genes gdpP and dhhP were expressed in Escherichia coli, and the recombinant proteins were purified and enzymatically characterized. Markerless inframe deletion mutant strains, including a double mutant strain, were constructed, and phenotypic characterization of these mutant strains was performed to verify the functions of these proteins.

Preparation of Recombinant GdpP and DhhP Proteins
Truncated gdpP and intact dhhP genes were amplified by PCR using primers listed in Supplementary Table S1 and S. mutans genomic DNA as a template, and cloned into the pCold ProS2 vector (Takara Bio, Otsu, Japan). The N-terminal region (108 amino acids), which were not contained in the recombinant GdpP, was determined as previously described (Bai et al., 2013). The resulting plasmids ( Table 1) were verified by DNA sequencing. E. coli BL21 cells harboring either plasmid were grown at 37 • C to absorbance at 595 nm (A 595 ) of ∼0.5. The bacterial culture was then cooled to 16 • C and isopropyl-β-Dthiogalactoside was added to a final concentration of 0.3 mM. After induction at 16 • C for 24 h, cells were sonicated and cell debris was removed by centrifugation at 50,000 × g and 4 • C for 1 h. The supernatant was loaded onto a Co-based TALON column (Takara Bio, Otsu, Japan) pre-equilibrated with TBS (Tris-HCl buffer, pH 7.5, supplemented with 300 mM NaCl). After the resin was extensively washed with TBS containing 10 mM imidazole, proteins were eluted using 250 mM imidazole in TBS. Each eluate was dialyzed against 10 mM Tris-HCl (pH 7.4) and stored at −20 • C following addition of an equal volume of 80% glycerol. The protein concentration was measured using a Pierce BCA Protein Assay Kit (Thermo Scientific, Tokyo, Japan), and the purity of samples was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

Liquid Chromatography Coupled With Tandem Mass Spectrometry (LC-MS/MS)
Adenosine monophosphate (AMP), pApA, c-di-AMP, and c-di-GMP were detected in reaction mixtures using a Nexera2 LC system (Shimadzu, Kyoto, Japan) connected to a triple quadrupole mass spectrometer (LC-MS 8040; Shimadzu) equipped with an electrospray ionization (ESI) source and operated in positive ion mode for pApA and negative ion mode for AMP, c-di-AMP, and c-di-GMP. Samples were separated using a reversed-phase Mastro SP column (2.1 × 100 mm; Shimadzu). The mobile phases were as follows: (A) 10 mM ammonium acetate/acetonitrile (90:10 v/v) and (B) 50 mM ammonium acetate:acetonitrile (80:20 v/v). A linear gradient from 0 to 60% solvent B over 9 min was applied at a flow rate of 0.5 ml min −1 . The A 254 value was monitored to quantify AMP, pApA, and c-di-AMP in reaction mixtures. ESI multiple reaction monitoring (MRM) mode was used to identify the compounds. The temperature of the ESI capillary was maintained at 300 • C. The drying gas was introduced at a temperature of 300 • C and a flow rate of 4 l min −1 . The nebulizer gas pressure was 10 psi. Each sample (1 µl) was injected into the LC capillary. All solvents used in these experiments were LC-MS grade from Wako Pure Chemicals (Osaka, Japan) or Sigma-Aldrich Japan.

Construction of Deletion and Complemented Streptococcus mutans Strains
Markerless mutagenesis was performed to construct three in-frame S. mutans XC deletion mutants, namely, gdpP, dhhP, and gdpP dhhP (Table 1 and Supplementary  Figures S1, S2, S3). A two-step transformation protocol using an IFDC2 cassette was used as described previously (Xie et al., 2011). This cassette contains an erythromycin resistance gene and a gene encoding mutated phenylalanine tRNA synthetase that incorporates the toxic p-Cl-Phe into proteins during translation. Introduction of the IFDC2 cassette altered the phenotype of the wild-type (WT) strain from p-Cl-Phe resistance and erythromycin sensitivity to p-Cl-Phe sensitivity and erythromycin resistance. First, three DNA fragments (an upstream targeting sequence, the IFDC2 cassette, and downstream targeting sequence) were individually generated by PCR, ligated using an In-fusion HD Cloning kit (Takara Bio, Otsu, Japan), and inserted into the cloning vector pMCL200 (Nakano et al., 1995). S. mutans XC cells were transformed with the respective linearized plasmids as previously described (Yoshida et al., 2005). The sensitivity of transformants (intermediate strains gdpP-Int and dhhP-Int) to p-Cl-Phe was then verified (Supplementary Figures S1, S2, S3). In the second step, two DNA fragments (an upstream targeting sequence and a downstream targeting sequence) were linked using an Infusion HD Cloning kit and then subcloned into pMCL200. The linearized plasmids were used to transform gdpP-Int and dhhP-Int strains. The transformants were selected and patched onto fresh BHI plates containing 200 µg p-Cl-Phe ml −1 , and onto BHI plates containing 20 µg erythromycin ml −1 , to confirm loss of the IFDC2 cassette. Deletion of the gdpP and dhhP genes in the mutant strains, designated gdpP and dhhP, respectively, was confirmed by PCR and DNA sequencing. Similarly, a double mutant strain, gdpP dhhP was constructed in the gdpP background using the two-step transformation protocol, as described above (Supplementary Figure S3). All strains and plasmids used are listed in Table 1.
To facilitate genetic complementation studies of mutants of S. mutans, shuttle plasmids were constructed using pJY (Yang et al., 2009) with a constitutive promoter (CP25) (Hansen et al., 2001). After PCR-amplified gdpP and dhhP genes with a Shine-Dalgarno sequence (Shine and Dalgarno, 1974) were inserted into the BamHI and NcoI sites of pJY using an Infusion HD Cloning kit (Takara Bio, Otsu, Japan), the integrity of the cloned sequence was verified by DNA sequencing. The resulting plasmids, pSM2140 and pSM1297 (Table 1), were used to transform S. mutans gdpP and dhhP, respectively, to construct gdpP-com and dhhP-com (Table 1).

Biofilm Assay
Biofilm formation assays were performed on polystyrene plates as previously described (Yano et al., 2012) with several minor modifications. Bacterial cultures used for biofilm formation were prepared by inoculating overnight cultures of S. mutans strains into fresh BHI broth at a ratio of 1:20, and anaerobically incubated at 37 • C for 24 h. Culture aliquots (5 µl) were inoculated into 95 µl of fresh BHI broth in a 96-well flatbottomed polystyrene plate. To evaluate the effect of sucrose on biofilm formation, 1.0% sucrose was added to the media. After incubation at 37 • C for 24 h, bacterial cultures were gently discarded, and plates were carefully washed three times with phosphate-buffered saline (PBS) to remove non-adhered cells and then stained with 0.2% (w/v) Victoria Blue for 15 min. Bound dye was dissolved using 100 µl of 99% methanol, and biofilm volume was evaluated by measuring the absorbance at 595 nm (A 595 ). When A 595 values were >2.0, they were FIGURE 2 | Phosphodiesterase activity of rGdpP 109-657 -ProS2 and rDhhP-ProS2 by liquid chromatography (LC) analysis. Reaction mixtures (50 µl) contained 50 mM potassium phosphate buffer (pH 5.0 for GdpP, pH 7.0 for rDhhP-ProS2), 1 mM MnCl 2 , 100 µM substrate (c-di-AMP or pApA), and 1.2 µg ml -1 rGdpP 109-657 -ProS2 or 0.36 µg ml -1 rDhhP-ProS2. Reactions were incubated for 15 min and terminated by addition of TE-saturated phenol and chloroform, and an aliquot (1 µl) of the aqueous phase was injected onto the LC column. Each sample was separated by LC and monitored at 254 nm. Purified AMP, c-di-AMP, and pApA were used as standards. measured again after appropriate dilution. A blank value (medium alone) was subtracted from each value. Data are represented as mean ± standard deviation from 12 replicates in three independent experiments.

Morphological Analyses
To observe planktonic S. mutans cells, cultures were grown for 24 h, inoculated into fresh BHI broth at a ratio of 1:20, and incubated until the A 595 value was ∼0.3 (mid-log phase). Aliquots of streptococcal cultures (5 µl) were fixed and stained using a conventional Gram staining method for optical microscopy. Biofilm formation after 12 h was also assessed using scanning electron microscopy (SEM) as previously described (Konno et al., 2015).

Statistical Analysis
Data were analyzed by one-way analysis of variance (ANOVA) followed by Tukey's post hoc tests at a significance level of p < 0.01.

Identification and Purification of Putative c-di-AMP Phosphodiesterases GdpP and DhhP
A homology search revealed that the amino acid sequences of GdpP (SMU_2140c) and DhhP (SMU_1297) from S. mutans are 56 and 68% identical with Pde1 and Pde2 from S. pneumoniae, respectively, and several functional domains are shared between the respective homologs ( Figure 1A). To evaluate the enzymatic functions of GdpP and DhhP in S. mutans, we attempted to purify the recombinant proteins following heterologous expression in E. coli but failed to obtain intact GdpP. Thus, genes encoding a truncated GdpP variant lacking transmembrane helices and a part of PAS domain (spanning amino acids 1-108), and an intact DhhP were cloned, and recombinant proteins were purified from E. coli lysates (Figure 1). The N-terminal hydrophobic transmembrane domains are dispensable for c-di-AMP phosphodiesterase activity (Corrigan et al., 2011;Rao et al., 2011;Bai et al., 2013). In addition, the PAS domain also did not affect the enzymatic function of GdpP (Bai et al., 2013). Each protein was generated as a fusion construct with a 23 kDa ProS2 tag that rendered the recombinant protein soluble and stable. The recombinant GdpP and DhhP were designated rGdpP 109−657 -ProS2 and rDhhP-ProS2, respectively. The molecular masses of the denatured fusion proteins were consistent with their predicted molecular masses (i.e., 84 kDa and 57 kDa for rGdpP 109−657 -ProS2 and rDhhP-ProS2, respectively; Figure 1B).
Enzymatic Characterization of rGdpP 109−657 -ProS2 and rDhhP-ProS2 From S. mutans as c-di-AMP Phosphodiesterases LC analysis of A 254 values revealed that incubation of rGdpP 109−657 -ProS2 with c-di-AMP resulted in the production of pApA (Figure 2), and this was further confirmed by MRM mode LC-MS/MS, in which transitions from m/z 677.4 (1+, precursor ion) to m/z 136.1 (product ion) and m/z 97.0 (product ion) were observed. However, pApA was not degraded by rGdpP 109−657 -ProS2, but incubation of rDhhP-ProS2 with pApA resulted in the production of AMP, and this was also confirmed by LC-MS/MS, in which the transition of m/z 346.05 (1−, precursor ion) to m/z 79.1 (product ion) was observed. Unlike Pde2 from S. pneumoniae, c-di-AMP was not cleaved by rDhhP-ProS2   (Figure 2), even when a large amount of rDhhP-ProS2 (100fold more than usual) was incubated with c-di-AMP for a long incubation time (12-fold longer than usual). This suggests that direct production of AMP from c-di-AMP by DhhP is negligible in S. mutans. Neither rGdpP 109−657 -ProS2 nor rDhhP-ProS2 cleaved c-di-GMP (data not shown), indicating that c-di-GMP is not a substrate for either of GdpP and DhhP. Reaction conditions for rGdpP 109−657 -ProS2 and rDhhP-ProS2 from S. mutans were optimized by quantifying pApA and AMP produced from c-di-AMP and pApA, respectively, in different pH, temperatures, and divalent ions (Figure 3).
Maximum activity of rGdpP 109−657 -ProS2 was observed at pH 5.0 and 37 • C, and Co 2+ was the most effective of the divalent metal ions tested. Mn 2+ also stimulated the enzyme with an optimal concentration of 1 mM, equivalent to ∼60% of the activity with Co 2+ . Interestingly, the enzymatic activity of rGdpP 109−657 -ProS2 in the presence of Zn 2+ or Ca 2+ was lower than that of the control, suggesting that these ions inhibit catalytic activity. rDhhP-ProS2 exhibited highest activity under slightly different conditions, with an optimal pH of 7.0 and an optimal temperature of 37 • C. Of the divalent metal ions tested, Mn 2+ was the most effective for stimulating enzymatic activity, and (B) Sucrose-independent and -dependent biofilm formation in 96-well polystyrene plates after 24 h. Data are expressed as the mean ± standard deviation from 12 replicates from three independent experiments. Data were analyzed by one-way analysis of variance (ANOVA) followed by Tukey's post hoc tests. * p < 0.01.
Kinetic parameters of rGdpP 109−657 -ProS2 and rDhhP-ProS2 enzymes were determined using the initial rate method, with end products (separated by LC) quantified using A 254 values ( Table 2).

Deletion of dhhP in S. mutans Results in Suppression of Growth
To examine the biological roles of GdpP and DhhP, three markerless in-frame mutant strains, in which the potentially confounding polar effect of the promoter of the antibiotic resistance gene on the downstream genes was eliminated, were constructed (Supplementary Figures S1, S2, S3). In addition, strains complemented with gdpP and dhhP were also constructed ( Table 1). Since c-di-AMP homeostasis plays an important role in cell growth in some bacteria (Mehne et al., 2013;Ye et al., 2014), the growth of S. mutans and its derivatives was examined by measuring A 595 values ( Figure 4A). The growth of the gdpP strain was similar to that of the WT strain, while the growth of dhhP and gdpP dhhP strains was slower than that of the WT. Furthermore, the slow growth of the dhhP strain was restored by complementation with dhhP ( Figure 4A).

Deletion of dhhP in S. mutans Results in Increased Biofilm Production
When S. mutans WT and derivative strains were incubated for 24 h in the absence of sucrose, biofilm formation was significantly more pronounced in strains dhhP and gdpP dhhP than in the other strains ( Figure 4B). By contrast, the amount of biofilm formed by the gdpP mutant in the absence of sucrose was not significantly different from that of the WT strain when incubated for 24 h ( Figure 4B). Importantly, the amount of biofilm produced by the dhhP strain was restored to WT levels upon complementation with dhhP ( Figure 4B). These findings revealed that the function of DhpP plays an important role in stimulating biofilm formation. Biofilm formation was not significantly different among strains in the presence of 1.0% sucrose ( Figure 4B).
Optical and electron microscopy observations showed that cells of both dhhP and gdpP dhhP strains clumped together to form large bacterial aggregates of ∼0.1 mm in diameter under both biofilm and planktonic conditions, whereas cell clumps were not observed in cultures of WT, gdpP, gdpP-com, or dhhPcom cells (Figure 5), revealing that lack of DhhP resulted in the formation of S. mutans aggregates.

DISCUSSION
The secondary messenger c-di-AMP was recently identified in bacteria and appears to be required for cell viability (Song et al., 2005;Woodward et al., 2010;Corrigan et al., 2011;Luo and Helmann, 2012;Mehne et al., 2013). c-di-AMP is synthesized by most of Gram-positive bacteria (Woodward et al., 2010;Corrigan et al., 2011;Kamegaya et al., 2011;Bai et al., 2013;Du et al., 2014), whereas the molecule is also present in some Gram-negative pathogens (Barker et al., 2013) and Spirochaetae (Ye et al., 2014). By contrast, the enzymes associated with c-di-GMP and its functions for the secondary messenger have been investigated in a diverse range of bacterial species (Jenal et al., 2017). GdpP and DhhP in S. mutans are orthologs of Pde1 and Pde2 in S. pneumoniae, respectively, despite sharing relatively low sequence identity. Homology analyses revealed that streptococcal species, including S. sobrinus, S. gordonii, S. salivarius, S. pyogenes, S. agalactiae, S. oralis, S. suis, and S. anginosus, contain one ortholog of GdpP and one ortholog of DhhP. Thus, the two distinct c-di-AMP phosphodiesterases characterized in this study appear to be present in most streptococci.
The rGdpP 109−657 -ProS2 appears to share the same substrate specificity as the Pde1 enzyme from S. pneumoniae (Bai et al., 2013). By contrast, the substrate specificity of rDhhP-ProS2 was different from that of Pde2 from S. pneumoniae, which produced AMP not from c-di-AMP but from pApA (Figure 2), whereas Pde2 from S. pneumoniae produced AMP both from c-di-AMP and pApA. Notably, the enzymatic activity of S. pneumoniae Pde2 to produce AMP from c-di-AMP was much lower (approximately 1/2000) than that of the same enzyme to produce AMP from pApA (Bai et al., 2013). Based on these previous findings and our current results (Figure 2), we conclude that DhhP has no detectable capacity to degrade c-di-AMP to produce AMP, and mainly cleaves pApA to AMP in S. mutans. It is probable that in S. mutans, c-di-AMP is degraded to the pApA intermediate by GdpP, and this is subsequently degraded to AMP by DhhP.
The optimal pH for the enzymatic activities of rGdpP 109−657 -ProS2 and rDhhP-ProS2 (pH 5.0 and 7.0, respectively) was lower than that of the orthologs listed above (mostly pH 8.0 or 8.5). This lower optimal pH for rGdpP 109−657 -ProS2 and rDhhP-ProS2 might be related to the ability of S. mutans to survive in more acidic environments (Hamilton and Buckley, 1991). DHH family proteins generally require a divalent cation for function (Yamagata et al., 2002). Indeed, Mn 2+ is indispensable for the enzymatic activity of many phosphodiesterases that act on c-di-AMP, including Pde2 from S. pneumoniae (Bai et al., 2013), CnpB from Mycobacterium tuberculosis , DhhP from B. burgdorferi (Ye et al., 2014), and GdpP from B. subtilis (Rao et al., 2011). These findings, together with our current results (Figure 3C), suggest that Mn 2+ might be the physiological metal ion of GdpP and DhhP proteins. Interestingly, the enzymatic activity of these enzymes was inhibited by Zn 2+ and Ca 2+ , consistent with previous reports of inhibition of phosphodiesterase activity by Zn 2+ and Ca 2+ in several bacteria (Rao et al., 2011;Ye et al., 2014;Tang et al., 2015).
Biofilm formation by gdpP was not significantly different from that of the WT strain (Figure 4), unlike other streptococcal species; deletion of gdpP resulted in increased biofilm production by S. suis (Du et al., 2014) and S. pneumoniae (Bai et al., 2013). Our data from three independently constructed mutant strains ( dhhP, gdpP dhhP, and dhhP-com) clearly demonstrated that deletion of the dhhP gene led to increased biofilm formation, probably due to aggregation of S. mutans (Figures 4B,  5). Bacterial aggregates of dhhP and gdpP dhhP strains could act as scaffolds on which thick biofilms are formed. However, the exact mechanism by which deletion of the dhhP gene is associated with the formation of cell clumps remains to be elucidated. Given the products of the rDhhP-ProS2 enzymatic reaction (Figure 2), exhaustion of AMP in bacterial cells and/or accumulation of pApA might affect biofilm formation. Otherwise, it would be possible that AMP has an inhibitory effect on biofilm formation. Deletion of gdpP, which leads to exhaustion of AMP in S. mutans cells, did not result in aggregation of the bacteria (Figure 5), suggesting that the concentration of AMP in cells might not be associated with formation of biofilms. Further studies are necessary to precisely elucidate the relationship between increased biofilm formation and deletion of dhhP since the intercellular level of c-di-AMP affects the expression of ∼700 genes in B. subtilis (Gundlach et al., 2016). Peng et al. (2016) reported that biofilm formation by a gdpP-deficient mutant of S. mutans UA159 was greater than the WT strain in the presence of sucrose. Contrary to this previous report, we showed that the amount of biofilm formed by S. mutans XC and its derivatives in the presence of 1.0% sucrose, which was higher than that in the absence of sucrose, was not significantly different among the strains tested ( Figure 4B). S. mutans glucosyltransferases, including GTF-I and GTF-SI, synthesize water-insoluble glucans from sucrose (Kuramitsu, 1993), thereby forming extraordinarily strong and thick biofilms (Bowen and Koo, 2011). Therefore, it might be difficult to precisely evaluate how gdpP and dhhP deletion affects biofilm formation because large amounts of insoluble glucans produced in the presence of sucrose could mask the relatively weak effect of gdpP and dhhP deletion on biofilm formation.
Growth as a biofilm is usually triggered to assist survival in inhospitable conditions, and deletion of the dhhP gene may constitute a stressful situation for bacteria. Interestingly, the DhhP protein identified as a phosphodiesterase in the current study was previously reported to be essential for the superoxide stress response (Zhang and Biswas, 2009). Together with the existing literature, our present findings indicate that DhhP protein and related molecules such as c-di-AMP, pApA, and AMP might play an important role in the response of S. mutans to stress conditions.