Functional Characterization of Small Alarmone Synthetase and Small Alarmone Hydrolase Proteins from Treponema denticola

ABSTRACT The stringent response enables bacteria to survive nutrient starvation, antibiotic challenge, and other threats to cellular survival. Two alarmone (magic spot) second messengers, guanosine pentaphosphate (pppGpp) and guanosine tetraphosphate (ppGpp), which are synthesized by RelA/SpoT homologue (RSH) proteins, play central roles in the stringent response. The pathogenic oral spirochete bacterium Treponema denticola lacks a long-RSH homologue but encodes putative small alarmone synthetase (Tde-SAS, TDE1711) and small alarmone hydrolase (Tde-SAH, TDE1690) proteins. Here, we characterize the respective in vitro and in vivo activities of Tde-SAS and Tde-SAH, which respectively belong to the previously uncharacterized RSH families DsRel and ActSpo2. The tetrameric 410-amino acid (aa) Tde-SAS protein preferentially synthesizes ppGpp over pppGpp and a third alarmone, pGpp. Unlike RelQ homologues, alarmones do not allosterically stimulate the synthetic activities of Tde-SAS. The ~180 aa C-terminal tetratricopeptide repeat (TPR) domain of Tde-SAS acts as a brake on the alarmone synthesis activities of the ~220-aa N-terminal catalytic domain. Tde-SAS also synthesizes “alarmone-like” nucleotides such as adenosine tetraphosphate (ppApp), albeit at considerably lower rates. The 210-aa Tde-SAH protein efficiently hydrolyzes all guanosine and adenosine-based alarmones in a Mn(II) ion-dependent manner. Using a growth assays with a ΔrelAΔspoT strain of Escherichia coli that is deficient in pppGpp/ppGpp synthesis, we demonstrate that Tde-SAS can synthesize alarmones in vivo to restore growth in minimal media. Taken together, our results add to our holistic understanding of alarmone metabolism across diverse bacterial species. IMPORTANCE The spirochete bacterium Treponema denticola is a common component of the oral microbiota. However, it may play important pathological roles in multispecies oral infectious diseases such as periodontitis: a severe and destructive form of gum disease, which is a major cause of tooth loss in adults. The operation of the stringent response, a highly conserved survival mechanism, is known to help many bacterial species cause persistent or virulent infections. By characterizing the biochemical functions of the proteins putatively responsible for the stringent response in T. denticola, we may gain molecular insight into how this bacterium can survive within harsh oral environments and promote infection. Our results also expand our general understanding of proteins that synthesize nucleotide-based intracellular signaling molecules in bacteria.

Biophysical characterization of Tde-SAS. Recombinant Tde-SAS (molecular weight [MW], 51,318 Da), migrated with an apparent MW of ;220 kDa upon size-exclusion chromatography (SEC) analysis, indicating it formed a stable homotetramer in solution (Fig.  S2A). A recombinant protein that comprised only the catalytic domain (Tde-SAS 1-246 ; residues 1 to 246; MW, 32,557 Da) migrated with an apparent MW of ;160 kDa upon SEC analysis, suggesting it similarly adopted a homotetrameric arrangement (Fig. S2B). The purified Tde-SAS and Tde-SAS 1-246 proteins had A 260nm /A 280nm ratios of 0.71 6 0.12 and 0.53 6 0.11, respectively. This indicated that neither protein (obtained by heterologous expression in Escherichia coli) was complexed to RNA, as has previously been observed for long-RSH proteins (62).
The domain structure of Tde-SAS was probed using limited proteolysis with the nonspecific protease subtilisin (18,63). Two major protein fragments were formed that had apparent MWs of ;26 kDa and ;21 kDa (Fig. S2D). Peptide mass fingerprint (PMF) analysis revealed that these protein fragments respectively corresponded to the catalytic domain and the TPR domain (data not shown). This indicated that subtilisin had primarily cleaved the Tde-SAS protein within the interdomain "linker" region comprising residues ;230 to 250 (Fig. 2B).
Tde-SAS preferentially synthesizes ppGpp, and its catalytic activities are not notably modulated by alarmones. The results from initial sets of biochemical assays revealed that Tde-SAS catalyzed the synthesis of pppGpp, ppGpp, and pGpp from ATP 1 GTP, ATP 1 GDP, and ATP 1 GMP, respectively, with the concomitant production of AMP. Representative chromatograms of enzymatic product mixtures are shown in Fig. 3A. The optimal pH for ppGpp synthesis activities was approximately 8.8 to 9.2 (Fig. S3A). The specific molar activities of Tde-SAS were determined under standardized conditions to quantify the respective rates of pppGpp, ppGpp, and pGpp synthesis. This was defined in units of micromoles of (pp)pGpp product synthesized per minute per micromole of Tde-SAS protein (mmol Á min 21 Á mmol 21 ) based on monomeric protein concentrations, as previously described (40). The rate of ppGpp synthesis was 689 6 87 mmol Á min 21 Á mmol 21 , which was ;22-fold faster than the rate of pGpp synthesis (30.9 6 3.6 mmol Á min 21 Á mmol 21 ) and ;88-fold faster than the rate of pppGpp synthesis (7.8 6 1.6 mmol Á min 21 Á mmol 21 ) (Fig. 3B).
Analogous sets of assays were performed for the Tde-SAS 1-246 protein under identical conditions. Chromatographic analysis of enzymatic product mixtures indicated that Tde-SAS 1-246 had catalytic activities that were equivalent to those of Tde-SAS but were more potent (Fig. 3A). The respective specific molar rates of pppGpp, ppGpp, and pGpp synthesis by Tde SAS  were subsequently determined to be ;3to 10-fold faster than those of Tde-SAS (Fig. 3B). Both Tde-SAS 1-246 and Tde-SAS synthesized ppGpp considerably faster than pGpp or pppGpp, suggesting that their substrate utilization patterns were equivalent (i.e., GDP ) GMP . GTP).  (TpecSAS, 415 aa, WP_147612650). The locations of the four synthesis motifs (Syn1 to Syn4) and four TPR motifs (TPR1 to TPR4) are indicated with brackets. The putative linker region is highlighted with a blue box. The C terminus of the Tde-SAS  protein is highlighted with an arrow. The SAS/small alarmone hydrolase (SAH) classification system used here was defined as in reference 31. The alignment figure was prepared using ESPript 3.0 (88). The phylograms were prepared using ITOL (94).
Alarmone Metabolism by Treponema denticola Proteins Microbiology Spectrum Micromolar concentrations of Zn 21 ions were previously shown to enhance the alarmone-synthesizing activities of Sa-RelP, while Fe 21 /Fe 31 ions played a putative structural role (42). Therefore, we assayed the ppGpp-synthesizing activities of Tde-SAS  in the presence of supplementary Fe 31 or Zn 21 ions (over the concentration range 0 to 32 mM). The results clearly indicated that neither of these metal ions had significant effects on the rate of ppGpp synthesis by Tde-SAS   (Fig. S4).
The addition of alarmones have previously been shown to modulate the synthetic activities of RelQ-and RelP-family SAS proteins (14,(40)(41)(42)64). Therefore, the Michaelis-Menten kinetic parameters for ppGpp synthesis were determined for Tde-SAS in the absence of alarmone, as well as in the presence of 100 mM pppGpp or ppGpp (Fig. 4B). In the absence of (p)ppGpp, the K m(GDP) value was 5.29 6 2.61 mM, with a corresponding k cat value (turnover number) of 13.77 6 4.13 s 21 , giving a catalytic efficiency (k cat /K m ) value of 2.60 6 0.51 mM 21 s 21 . The Hill coefficient (h) was 1.35 6 0.37, indicating that there may be low levels of positive cooperativity. (Note that h . 1 indicates product-mediated stimulation of enzymatic activities, also known as positive cooperativity or feed-forward control.) When 100 mM ppGpp or pppGpp was added, the corresponding K m(GDP) values increased slightly to 6.59 6 2.80 and 6.03 6 2.28 mM, respectively. There was little effect on the corresponding turnover numbers, catalytic efficiencies, or h coefficients. These data indicated that added (p)ppGpp had slight stimulatory effects on the catalytic activities of Tde-SAS at a concentration of 100 mM. Additional sets of assays containing 200 mM (allosteric) ppGpp or pppGpp led to only ;10% enhancements in rates of alarmone synthesis (Fig. S5).
As supplementary (p)ppGpp had minor effects on the ppGpp-synthesizing activities of Tde-SAS, the kinetic parameters for pGpp and pppGpp synthesis were determined in the absence of (p)ppGpp (Fig. 4). While the K m value for GTP (6.09 6 1.01 mM) was reasonably similar to that for GDP, the K m(GMP) value was ;2-fold higher (11.68 6 1.69 mM). The k cat values for ppppGpp and pGpp synthesis by Tde-SAS were 25-fold and 10-fold lower than that for ppGpp, respectively. Thus, the catalytic efficiencies for pGpp or pppGpp synthesis by Tde-SAS were ;20to 30-fold lower than that for ppGpp synthesis. The h coefficients for pGpp and pppGpp synthesis were almost identical to that for ppGpp synthesis, indicative of very low-level product-mediated stimulation of enzymatic activities.
The C-terminal TPR domain inhibits alarmone production. The respective Michaelis-Menten kinetic parameters for alarmone synthesis by Tde-SAS 1-246 were determined in the absence of supplementary (p)ppGpp (Fig. 4), as preliminary assays indicated that the addition of 100 mM pppGpp or ppGpp had very minor effects on the rate of ppGpp synthesis (data not shown). The k cat value for ppGpp synthesis by Tde-SAS 1-246 (45.62 6 5.65 s 21 ) was ;4.5-fold higher than for pGpp synthesis (10.80 6 1.10 s 21 ) and ;14.5-fold higher than for pppGpp synthesis (3.11 6 0.45 s 21 ). The K m(GTP) and K m(GMP) values were both ;3fold higher than the K m(GDP) value. Correspondingly, ppGpp was synthesized with the highest catalytic efficiency (44.10 6 6.20 mM 21 s 21 ), which was ;14-fold higher than that of pppGpp synthesis (3.14 6 0.25 mM 21 s 21 ) and 41-fold higher than that of pGpp synthesis (1.07 6 0.13 mM 21 s 21 ).
The K m(GDP) value for Tde-SAS 1-246 (1.03 6 0.30 mM) was ;5-fold lower than the corresponding value obtained for Tde-SAS (5.29 6 2.61 mM). The catalytic efficiency for ppGpp synthesis by Tde-SAS 1-246 was ;18-fold higher than for Tde-SAS (in the absence of added alarmone) (Fig. 4D). Analogously, the K m values for GMP (3.44 6 0.6 mM) and GTP (2.91 6 0.75 mM) for Tde-SAS 1-246 were, respectively, ;3and 2-fold lower than the values determined for Tde-SAS. The resultant catalytic efficiencies for pGpp and  Table summarizing the kinetic parameters obtained for (pp)pGpp production by Tde-SAS and Tde-SAS 1-246 from the respective sets of assays shown in panels A to C: maximum reaction velocity (V max ), Michaelis constants for GTP (K m(GTP) ), GDP (K m(GDP) ), or GMP (K m(GMP) ) in millimolar units; turnover number (k cat ) in units of AMP (= (pp)pGpp) molecules formed per second, enzymatic catalytic efficiency (k cat /K m ). All reactions were performed in triplicate, reporting the means 6 standard deviation. See Materials and Methods for experimental details.
Alarmone Metabolism by Treponema denticola Proteins Microbiology Spectrum pppGpp synthesis by Tde-SAS 1-246 were, respectively, ;26and 12-fold higher than those of Tde-SAS. Taken together, these results indicated that the presence of the C-terminal TPR domain greatly reduced the overall rates and catalytic efficiencies of (pp)pGpp synthesis but did not greatly affect the substrate preference; GDP was by far the most efficiently utilized substrate. Tde-SAS (Tde SAS 1-246 ) synthesizes alarmone-like nucleotides (pp)pApp and ppIpp. Tde-SAS and Tde-SAS 1-246 were incubated with ATP, ATP 1 ADP, or ATP 1 AMP, respectively, under standardized conditions to determine their respective abilities to synthesize pppApp, ppApp, and pApp (adenosine 39-diphosphate, 59-phosphate). Representative chromatograms of product mixtures are shown in Fig. 5A.
Tde-SAS 1-246 synthesized ppApp most effectively, with a specific molar activity (15.7 6 1.8 mmol Á min 21 Á mmol 21 ) that was ;120and 70-fold higher than that for pppApp and pApp synthesis, respectively (Fig. 5B). Tde-SAS synthesized ppApp ;8fold slower than Tde-SAS 1-246 , while the rates of pppApp and pApp synthesis were very low under these conditions. The respective rates of ppApp synthesis by Tde-SAS and Tde-SAS 1-246 were ;160and 360-fold lower than those of ppGpp synthesis (Fig. 3B). (Note that the additional small (pp)pApp peaks labeled in the chromatograms correspond to the products of competing side reactions, with ppApp primarily formed from the low levels of ADP present in ATP solutions.) Tde-SAS 1-246 also catalyzed the synthesis of inosine-39,59-bis(diphosphate) (inosine tetraphosphate, ppIpp) from IDP and ATP (Fig. S6), as has previously been found for other long-RSH and SAS proteins (18,40). Tde-SAS exhibited low-level ppIpp synthesis activities under comparable conditions. ITP could not function as a pyrophosphate donor in place of ATP (data not shown).
Tde-SAH hydrolyzes (pp)pGpp and (p)ppApp in a Mn 2+ ion-dependent manner. Tde-SAH (TDE1690) belongs to the ActSpo2 family of RSH proteins (31). Tde-SAH and the other treponeme SAH homologues cluster separately from nontreponeme homologues in the ActSpo2 phylogenetic clade (Fig. 6A), the majority of which belong to taxa from the phylum Actinobacteria (31). A multiple sequence alignment of the amino acid sequences of Tde-SAH and seven previously described SAH homologues is shown in Fig. 6B. This includes the C. glutamicum RelH (CgSAH; RelH Cg ; Protein Data Bank [PDB] entry    (Fig. S2C), indicating that it was homodimeric, analogous to the RelH Cg and RelH Ll proteins (36,39). Tde-SAH specifically hydrolyzed pppGpp, ppGpp, and pGpp to form GTP, GDP, and GMP, respectively (Fig. 7A). Tde-SAH hydrolyzed (pp)pGpp in a Mn 21 ion-dependent manner. Other (divalent) metal ions were tested as potential cofactors, including Mg 21 , Co 21 , Zn 21 , Fe 31 , Ca 21 , and Ni 21 (Fig. 7D). Tde-SAH had lowlevel ppGpp-hydrolyzing activities in the presence of Mg 21 ions and slightly higher activities in the presence of Co 21 ions but negligible activities when Zn 21 , Fe 31 , Ca 21 , or Ni 21 ions were added (Fig. 7D). The hydrolytic activities of Tde-SAH were slightly higher in the presence of 1 mM Mn 21 compared to a combination of 1 mM Mn 21 and 10 mM Mg 21 , suggesting that Mn 21 and Mg 21 ions compete for the same protein-binding sites. The hydrolysis of ppGpp was optimal at pH 8.4 to 8.8 (Fig. S3B). Correspondingly, all subsequent biochemical assays were performed using pH 8.4 buffer containing 1 mM Mn 21 ions.
Further biochemical analysis using phosphate-release assays with/without the addition of a pyrophosphatase enzyme confirmed that Tde-SAH specifically hydrolyzed the diphosphate (PPi) unit from the 39-ribose position of (pp)pGpp (data not shown). Consequently, pyrophosphatase enzyme-coupled continuous spectrophotometric assays were performed to determine the Michaelis-Menten kinetic parameters for (pp)pGpp hydrolysis by Tde-SAH, analogous to previously described methods (40). The results indicated that Tde-SAH hydrolyzed pppGpp, ppGpp, and pGpp with similar catalytic efficiencies (Fig. 8). The k cat values were in the order ppGpp . pppGpp . pGpp, ranging from 0.79 6 0.07 to 0.54 6 0.04 s 21 . The K m values were fairly similar for all three alarmones, being lowest for ppGpp and ppGpp at ;100 mM. Correspondingly, Tde-SAH exhibited the highest catalytic efficiency for ppGpp hydrolysis (9.06 6 0.42 mM 21 s 21 ), with pGpp and pppGpp hydrolyzed ;40% and 60% less efficiently, respectively. There was some evidence of positive cooperativity for pGpp and ppGpp hydrolysis, with h coefficients of 1.45 6 0.24 and 1.89 6 0.25, respectively, but not for pppGpp.
Tde-SAH hydrolyzed pppApp/ppApp, to ATP/ADP 1 pyrophosphate, respectively, in a Mn 21 ion-dependent manner (Fig. 7B). pApp could not be enzymatically synthesized in sufficient quantities and was not tested. It may be noted that under the assay conditions used (pH 8.4, 1 mM Mn 21 ), (p)ppApp molecules are more prone to nonenzymatic (metal ionfacilitated) hydrolysis than (p)ppGpp. Consequently, peaks corresponding to ppApp and pApp are observed in the chromatograms of the control reactions. The specific molar rates of hydrolysis for (pp)pGpp and (p)ppApp were evaluated under identical conditions to quantify the substrate selectivity (Fig. 7C). The results indicated that Tde-SAH hydrolyzed (pp)pGpp ;2-fold faster than pppApp or ppApp under the conditions employed.
The Asp74 residue of Tde-SAH forms part of the HD motif (HD3; Fig. 6B), which has previously been shown to be essential (or critically important) for hydrolytic activity in other RSH homologues (24,38). The biochemical activities of the Tde-SAH D74A (His-Asp ! His-Ala) mutant were determined using assays analogous to those described above. the results revealed that the Tde-SAH D74A protein had undetectable hydrolytic activities against ppGpp or ppApp (Fig. 7E), confirming that Asp74 played an essential role in alarmone hydrolysis.  Tde-SAS synthesizes alarmones in vivo. The in vivo (p)ppGpp synthesis activities of Tde-SAS and Tde-SAS 1-246 were determined using a well established growth assay in E. coli DrelADspoT (CF1693) (23). The CF1693 mutant is deficient in (p)ppGpp synthesis and degradation, exhibits multiple amino acid auxotrophies, and therefore cannot grow in minimal medium (66). Complementation with a plasmid-based RSH protein capable of synthesizing (p)ppGpp to sufficient levels, but not to excessive (i.e., toxic) levels, restores growth in minimal medium (23,36). The wild-type E. coli MG1655 strain (CF1648) was included as a reference.
Genes encoding the Tde-SAS, Tde-SAS 1-246 , Tde-SAH, and S. aureus RelP (Sa-RelP) proteins were respectively cloned into the medium-copy number and arabinose-inducible pBAD33 plasmid (67) and established in the E. coli DrelADspoT strain. The pBAD33-Sa-RelP and empty pBAD33 plasmids were included as positive and negative controls, respectively. Cultivation in MOPS minimal medium lacking arabinose led to modest increases in growth rates for the pBAD33-Tde-SAS, pBAD33-Tde-SAS 1-246 , and pBAD33-Sa-RelP complemented strains, compared to the pBAD33-Tde-SAH and pBAD33 strains (Fig. 9A). Analogous experiments performed in MOPS minimal medium containing arabinose (0.2%) led to notable increases in growth rates for the strains complemented with the pBAD33-Tde-SAS, pBAD33-Tde-SAS 1-246 , and pBAD33-Sa-RelP plasmids, indicative of (p)ppGpp production to subtoxic levels (Fig. 9B). In contrast, there were negligible changes in the growth rates of the strains containing the empty pBAD33 and pBAD33-Tde-SAH plasmids. The growth rate of the wild-type CF1648 (MG1655) strain complemented with an empty pBAD33 plasmid remained essentially unchanged in absence and presence of inducer, which were considerably higher than those of all the complemented DrelADspoT strains under all conditions employed. Taken together, these results strongly support the premise that both the Tde-SAS and Tde-SAS 1-246 proteins can synthesize (p)ppGpp in the E. coli DrelADspoT strain to levels that can maintain its effective growth in minimal medium.
Tde-SAH can hydrolyze (p)ppGpp produced by Tde-SAS in vivo. Further sets of growth complementation experiments were performed to investigate the in vivo alarmone-hydrolyzing activities of Tde-SAH, as well as the Tde-SAH D74A mutant. In these experiments, two plasmids were stably coestablished in E. coli DrelADspoT: (i) a highcopy number, isopropyl b-D-1-thiogalactopyranoside (IPTG)-inducible pGEX plasmid (Amp R ) containing the Tde-SAH wild-type or Tde-SAH D74A mutant genes or an empty pGEX plasmid and (ii) a pBAD33 (Cm R ) plasmid containing the Tde-SAS gene or an empty pBAD33 plasmid. Growth in MOPS minimal medium was measured over 10 h at 37°C, adding both the IPTG (0.2 mM) and arabinose (0.2%) inducers at the 2.5-h point. Growth curves are shown in Fig. 9C.
The E. coli DrelADspoT strains complemented with the pBAD33-Tde-SAS 1 pGEXempty and the pBAD33-Tde-SAS 1 pGEX-Tde-SAH D74A pairs of plasmids grew most effectively. This supported the premise that the Tde-SAH D74A mutant could not hydrolyze the (p)ppGpp synthesized by Tde-SAS. Complementation with the pBAD33-empty 1 pGEX-empty or the pBAD33-empty 1 pGEX-Tde-SAH plasmids resulted in similarly low growth rates. However, complementation with pBAD33-Tde-SAS 1 pGEX-Tde-SAH led to growth rates intermediate between these two levels. Taken together, these results suggest that Tde-SAS synthesized sufficient quantities of (p)ppGpp, which survived long enough within the cell to mediate various growth-promoting stringent response effects before being degraded by Tde-SAH (Fig. 9C).
More than 20 different families (phylogenetic lineages) of SAS proteins have been identified (31), but only a small number have yet been studied. Tde-SAS represents the first protein in the DsRel (SAS-TPR) family to be functionally characterized. Below, we compare and contrast the characteristics and biochemical activities of Tde-SAS with homologues from other SAS families.
Tde-SAS preferentially utilizes GDP for the synthesis of ppGpp. Broadly speaking, this is similar to the majority of RelP and RelQ homologues studied to date, under the majority of experimental conditions tested (14,29,32,(40)(41)(42)64). C. glutamicum RelS synthesizes pppGpp most efficiently but exhibits complex kinetic behaviors (36). The bifunctional RNase-SAS RelZ protein from M. smegmatis has been reported to synthesize pGpp most efficiently (45). The recently described RelQ homologue from Clostridiodes (Clostridium) difficille (RelQ Cd ) is a notable outlier, having the ability to utilize GDP and GTP for the production of pGpp, putatively via an intermediate phosphohydrolase step (35). We found no evidence that Tde-SAS catalyzes any competing phosphotransferase or phosphohydrolase processes throughout our experimental work.
The K m(GDP) value for Tde-SAS (5.29 6 2.61 mM) is similar to that of C. glutamicum RelS (6.1 mM) (36) but is considerably higher than the values previously reported for M. smegmatis RelZ and Firmicutes RelP and RelQ homologues, which are in the range 0.1 to 1.7 mM (14,40,41,45,64). However, removal of the TPR domain (as in the Tde-SAS 1-246 protein) dramatically increased (pp)pGpp production levels. It increased the rate of (pp)pGpp synthesis ;3-fold and reduced K m(GDP) more than 6-fold (to ;1 mM), making it comparable to the above-mentioned SAS proteins. Putatively, this would greatly increase ppGpp production at GDP concentrations typically found within bacterial cells (13,46,(72)(73)(74). However, removing the TPR domain did not alter substrate selectivity, with GDP strongly preferred over GTP or GMP. Thus, the TPR domain appears to act like a brake, repressing the overall alarmone-synthesizing activities of the Tde-SAS catalytic domain by reducing the effective substrate binding affinity (increasing K m ) and reducing the overall reaction rate (decreasing k cat ).
Both Tde-SAS and Tde-SAS 1-246 were capable of synthesizing sufficient quantities of (p)ppGpp to restore the growth of the E. coli DrelADspoT (CF1693) strain in minimal medium, without inducing toxic effects (due to the overproduction of (p)ppGpp) (Fig. 9). While these sets of experiments demonstrated in vivo alarmone synthesis activities in a model bacterial system, they do not directly show that Tde-SAS (or Tde-SAS 1-246 ) synthesizes alarmones within the native T. denticola host. It should also be noted that this E. coli reporter system is semiquantitative in nature and cannot be used to accurately evaluate levels of alarmone production or hydrolysis. These are notable limitations of our study.
Proteins containing TPR motif domains are widely distributed throughout prokaryotes, in which they typically function as protein "interaction modules" that promote the formation of multiprotein complexes, function as chaperones, or regulate the extracellular export of proteins or exopolysaccharides (60,(75)(76)(77)(78)(79). While TPR motif domains comprise multiple (typically 3 to 16) tandem repeats of a 34-aa structural motif that adopts a distinctive helix-turn-helix conformation, there is considerable sequence heterogeneity in their sequence composition (60,75). There are high levels of sequence conservation within the four respective TPR motifs in SAS-TPR homologues encoded by diverse Treponema species, indicative of conservation of structure and/or function (Fig. 2). We speculate that the TPR domain of Tde-SAS and other treponeme SAS-TPR (DsRel) homologues play analogous regulatory roles in the stringent response (see below).
Tde-SAS synthesized three different adenine nucleotide-based alarmone products: pppApp, ppApp, and pApp, albeit at greatly differing rates (Fig. 5). The pattern for AMP/ ADP/ATP utilization (as diphosphate acceptor) was equivalent to that of GMP/GDP/GTP, i.e., ADP was strongly preferred for ppApp synthesis, just as GDP was preferred for ppGpp production. Notably, the rates of (pp)pApp production were tens to hundreds of times lower than the corresponding rates of (pp)pGpp production. Tde-SAS 1-246 synthesized ppApp ;8-fold faster than Tde-SAS, indicating that the TPR domain of Tde-SAS repressed (pp)pGpp and (pp)pApp production by the catalytic domain via the same mechanism.
The physiological functions and molecular sources of (p)ppApp within bacterial cells remains poorly understood (1,31,38,46,47,49,80). Several phylogenetic lineages of SAS proteins (PhRel, CapRel, PhRel2, and FaRel) are prolific (p)ppApp synthesizers, leading to potent cytotoxic effects (31,47). In addition to functioning as toxins, (p)ppApp directly modulates enzymatic activities (e.g., PurF) and alters transcriptional activities in a manner notably different from that of (p)ppGpp (47,(81)(82)(83). Thus far, Rel Mex is the only long-RSH protein shown to possess notable (p)ppApp synthesis activities (49,50). The majority of SAS and Rel (long-RSH) proteins characterized to date do not appear to function as sources of (p)ppApp (1,5,38,46,49,50). The putative in vivo ppApp-synthesizing abilities of Tde-SAS remain to be verified. However, Tde-SAS synthesizes ppApp rather slowly in vitro, and results from our growth rate assays indicate that Tde-SAS does not function as a "Tox-SAS" in E. coli. Furthermore, the presence of (pp)pApp in T. denticola cells remains unknown. It is conceivable that Tde-SAS synthesizes low levels of (p)ppApp as an alarmone-like transcriptional modulator or for other regulatory purposes in T. denticola cells (3,(81)(82)(83). In these regards, our study has notable limitations, and these important issues require future investigation.
Tde-SAH contains the HD1 to HD6 motifs identified in diverse SAH, SpoT, and bifunctional long-RSH homologues (Fig. 6B) (27). The strictly conserved HD diad (His73 and Asp74) are predicted to play a key role in binding the catalytic Mn 21 ion within the Tde-SAH active site (20,24,38,39). Mutation of Asp74 to alanine abrogated the alarmonehydrolyzing activities of Tde-SAH (Fig. 7E), putatively by disrupting Mn 21 cofactor binding. This is consistent with results from the in vivo growth rate assays, in which the E. coli DrelADspoT (CF1693) strain complemented with Tde-SAS, and Tde-SAH D74A grew considerably faster than the strain complemented with Tde-SAS and Tde-SAH. This is putatively due to the inability of the Tde-SAH D74A mutant to hydrolyze the (p)ppGpp synthesized by Tde-SAS in the cell. We have elucidated the X-ray crystal structure of the Tde-SAH protein, and hence a more detailed mechanistic description of this protein will be provided elsewhere (M. Wang and R.M. Watt, unpublished data).
Based on our results, we tentatively speculate that Tde-SAS and Tde-SAH modulate alarmone (and [p]ppApp) levels in T. denticola cells via the following mechanism. In its default conformation, Tde-SAS maintains a basal level of (pp)pGpp (and [p]ppApp) synthesis. While Tde-SAS synthesizes ppGpp most efficiently, the precise ratios of pppGpp/ppGpp/ pGpp produced would also be governed by the respective intracellular levels of GTP/GDP/ GMP. The alarmones (and [p]ppApp) produced would be rapidly hydrolyzed by Tde-SAH due to its efficient catalytic activities. We further speculate that the TPR domain of Tde-SAS forms binding associations with components of the transcription or translational machinery that are indicative of nutrient starvation or certain extracellular stresses. These proteinbinding events alter the conformation of the Tde-SAS catalytic domain, derepressing alarmone production (akin to Tde-SAS 1-246 ). Upon cessation of binding to the TPR domain, the catalytic domain returns to its default conformation, and alarmone production drops to basal levels. Further investigations are required to validate this proposed mechanism.
In brief conclusion, we have established the activities of two previously undescribed lineages of SAS and SAH protein to significantly enhance our molecular understanding of (pp)pGpp and (p)ppApp metabolism within bacterial systems. Taken together, our in vitro and in vivo data suggest that the activities of Tde-SAS and Tde-SAH may be sufficient for the metabolism of (pp)pGpp in the periodontal pathogen T. denticola.
Protein expression and purification. Proteins were expressed from the respective recombinant plasmids that had been established in E. coli BL21(DE3) (Invitrogen), as previously described (40,86). Briefly, 5-mL overnight cultures inoculated from single colonies were expanded into 500 mL of Terrific Broth (TB) containing kanamycin (50 mg/mL) and grown aerobically with shaking at 37°C. Protein expression was induced at OD 600 0.6 to 0.8 by the addition of 0.2 mM IPTG (GE Healthcare), and incubation was maintained at 25°C for 12 h. The cells were chilled on ice and collected by centrifugation (6,000 Â g, 4°C,10 min), the supernatant was discarded, and cell pellets were washed with cold phosphate-buffered saline (PBS; pH 7.4, 25 mL) and then stored at 270°C for future use or used directly.
Bioinformatic methods. The DNA and protein sequences were download from the NCBI and routinely manipulated using BioEdit version 7.2.0 (87). Amino acid multiple sequence alignments (.fas files) were constructed using the ClustalW program within BioEdit, visualizing the results in EsPript 3.0 (88). TPRpred in the MPI Bioinformatics Toolkit (89) was used to identify the TPR motif locations using default parameters. TPR structure modeling was performed using the SWISS-MODEL webserver using Tde-SAS residues 256 to 410 as the input sequence with default settings (90).
Representative SAS/SAH homologues were identified using the NCBI Basic Local Alignment Search Tool (BLAST) (91), searching the NCBI GenBank nonredundant (nr) amino acid sequences and PDB sequences (www.rcsb.org) (92), and were classified into families based on previously published studies (16,31). Phylogenetic relationships between protein amino acid sequences were inferred using GARLI v.2.0 using the default maximum likelihood (ML) approach (93). Constructed ML phylograms were visualized and edited using ITOL (94) to highlight branches from distinct SAS/SAH protein families. Details of the SAS and SAH protein sequences used for phylogenetic tree construction are summarized in Supplemental File 2.
Nucleotide product analysis and quantification. Enzymatic product mixtures were analyzed by anion-exchange chromatography (1 mL MonoQ 5/50 GL; GE Healthcare) using an AKTA purifier system using the following program: 100% buffer A (25 mM Tris-HCl, pH 8.0, 25 mM NaCl) for 3 CV, a linear gradient of 100% buffer A increasing to 44% buffer B (25 mM Tris-HCl, pH 8.0, 1 M NaCl) over 13 CV and then 100% buffer B for 3CV, at a flow rate of 2 mL/min. The UV absorption of the eluent was monitored at 254 nm. Nucleotides and alarmone products were identified based on their unique elution volumes, in comparison with reference standards, as previously described (40). Reference chromatograms of AMP, ADP, ATP, GMP, GDP, GTP, pGpp, ppGpp, pppGpp, ppApp, and pppApp standards run under identical conditions are shown in Fig. S1. The respective levels of AMP biproduct formed (which are equimolar to those of the corresponding enzymatic (pp)pGpp or (p)ppApp products) were quantified by measuring the respective peak areas on the chromatograms, which were then compared with a standard curve prepared from a set of AMP solutions of known concentrations (0 to 1,000 mM). The unit of activity was defined as the number of micromoles of AMP (or alarmone) synthesized per minute per micromole of protein (mmol Á min 21 Á mmol 21 protein). All reactions were performed in triplicate. The y axes of the chromatograms shown in the figures (plotted in units of milliabsorbance units, mAU) were routinely removed for the sake of clarity.
Qualitative assays for determining (pp)pGpp/(p)ppApp/ppIpp synthesis activities. Reaction mixtures (20 mL) containing 50 mM Tris-HCl, pH 8.8 (optimal pH), 150 mM NaCl, 10 mM MgCl 2 , 1 mM DTT, 5 mM ATP, 5 mM GMP/GDP/GTP/ADP/ATP/IDP (as indicated), and 250 nM protein were incubated at 37°C for 2 h, before being quenched by the addition of 2 mM EDTA and then snap-frozen in liquid nitrogen for future analysis. Each product mixture (20 mL) was diluted to 200 mL with Milli-Q water and analyzed by anion-exchange chromatography on a Mono Q column 5/50 GL (1 mL) as described above. All reactions were performed in triplicate.
Kinetic analysis of alarmone synthesis by Tde-SAS and Tde-SAS 1-246 . Kinetic parameters (V max , k cat , and K m ) were calculated using the Michaelis-Menten model, incorporating results from sets of assays performed in triplicate. The assays were performed at the pH value optimal for the Tde-SAS protein (Fig.  S3). Reaction mixtures (20 mL) contained 50 mM Tris-HCl, pH 8.8, 150 mM NaCl, 10 mM MgCl 2 , 1 mM DTT, 5 mM ATP, 0 to 10 mM GMP/GDP/GTP, 250 nM protein (Tde-SAS/Tde-SAS 1-246 ), with or without 100 mM pppGpp/ppGpp (as a putative allosteric modulator). Alarmone synthesis rates were determined by quantifying AMP levels as described above.
Evaluation of in vivo (p)ppGpp production using E. coli DrelADspoT reporter system. Cellular (p) ppGpp levels were evaluated using growth assays with the E. coli DrelADspoT (CF1693) strain (from Michael Cashel) (23), which was cultivated in MOPS minimal medium (1Â MOPS mixture, 1.32 mM K 2 HPO 4 , 1% glucose as carbon source) as previously described (31,36) with minor modifications. Plasmids and strains used are listed in Table S1. pBAD33 plasmids containing the Tde-SAS, Tde-SAS 1-246 , S. aureus RelP (Sa-RelP, NWMN_2405), or Tde-SAH gene or no insert (pBAD33-empty, negative control) were transformed into E. coli CF1693 and were propagated at 37°C in LB medium containing chloramphenicol (Cm, 25 mg/mL) to ensure plasmid maintenance. Wild-type E. coli MG1655 (CF1648) (23) transformed with pBAD33 was included as an additional control. Aliquots (10 mL) of overnight cultures (single colonies inoculated into 5 mL of LB 1 Cm, incubated at 37°C for 16 h) were diluted to a final OD 600 of 0.08 into MOPS minimal medium (990 mL) containing chloramphenicol (25 mg/mL) and 0.2% arabinose to induce protein expression. Aliquots (200 mL) were pipetted into 96-well plates, which were incubated at 37°C for 10 h with OD 600 readings taken every 10 min (immediately after a few seconds of automated agitation), using a SpectraMax M2e multilabel microplate reader (Molecular Devices).
Analogous sets of growth assays were performed in E. coli CF1693 transformed with pairs of pBAD33 and pGEX-4T1 plasmids respectively containing Tde-SAS, Tde-SAH, Tde-SAH D74A (hydrolytically inactive), or no genetic insert (negative control). Transformed E. coli CF1693 strains were routinely propagated at 37°C in LB medium containing ampicillin (100 mg/mL) and chloramphenicol (25 mg/mL) for stable plasmid maintenance. These strains were used in growth assays performed analogously to those described above, except using MOPS minimal medium containing ampicillin (100 mg/mL), chloramphenicol (25 mg/mL), arabinose (0.2%), and IPTG (0.2 mM) to induce protein expression. Throughout these experiments, we were fastidious with the cultivation and restreaking of this "(p)ppGpp zero" strain to reduce the likelihood of spontaneous suppressor (e.g., RNA polymerase) mutations occurring (66). Additional experimental details are included in the Supplemental Methods.
Data availability. All research materials and primary data sets are available upon request (Rory M. Watt, rmwatt@hku.hk).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.9 MB. SUPPLEMENTAL FILE 2, XLSX file, 0.1 MB.