Cyclic Di-AMP Homeostasis in Bacillus subtilis

Background: Bacillus subtilis encodes three diadenylate cyclases. Results: Cyclic di-AMP is essential for the viability of B. subtilis; however, excess c-di-AMP also harms the cells. The activity of the cyclases is subject to regulation. Conclusion: The control of c-di-AMP homeostasis is crucial for B. subtilis. Significance: c-di-AMP is the first essential signaling nucleotide in bacteria. The genome of the Gram-positive soil bacterium Bacillus subtilis encodes three potential diadenylate cyclases that may synthesize the signaling nucleotide cyclic di-AMP (c-di-AMP). These enzymes are expressed under different conditions in different cell compartments, and they localize to distinct positions in the cell. Here we demonstrate the diadenylate cyclase activity of the so far uncharacterized enzymes CdaA (previously known as YbbP) and CdaS (YojJ). Our work confirms that c-di-AMP is essential for the growth of B. subtilis and shows that an excess of the molecule is also harmful for the bacteria. Several lines of evidence suggest that the diadenylate cyclase CdaA is part of the conserved essential cda-glm module involved in cell wall metabolism. In contrast, the CdaS enzyme seems to provide c-di-AMP for spores. Accumulation of large amounts of c-di-AMP impairs the growth of B. subtilis and results in the formation of aberrant curly cells. This phenotype can be partially suppressed by elevated concentrations of magnesium. These observations suggest that c-di-AMP interferes with the peptidoglycan synthesis machinery. The activity of the diadenylate cyclases is controlled by distinct molecular mechanisms. CdaA is stimulated by a regulatory interaction with the CdaR (YbbR) protein. In contrast, the activity of CdaS seems to be intrinsically restricted, and a single amino acid substitution is sufficient to drastically increase the activity of the enzyme. Taken together, our results support the idea of an important role for c-di-AMP in B. subtilis and suggest that the levels of the nucleotide have to be tightly controlled.

To respond appropriately to changing environments, bacteria have evolved a variety of signaling strategies. Many of these strategies involve changes of gene expression programs. How-ever, quick responses may often be important for the bacterial cell. For this purpose, protein activities have to be modulated. This can occur by either covalent modification of the proteins (e.g. by phosphorylation or acetylation), degradation of the proteins, or non-covalent interaction with other proteins or low molecular weight effectors. The low molecular weight effectors may be either metabolites that are part of the normal metabolism or dedicated signaling molecules that are produced by the cell for the purpose of signal transduction (1).
Among the best studied signaling molecules produced by bacteria are the so-called autoinducers, either acylated homoserine lactones or peptides in Gram-negative and Gram-positive bacteria, respectively (2). In addition, specific signaling nucleotides have been discovered in all bacteria in which they have been searched (3). Although the autoinducers serve mainly for purposes of cell-cell communication (quorum sensing), the signaling nucleotides are used for intracellular signaling. In addition to cyclic AMP and (p)ppGpp that are involved in carbon catabolite repression and the stringent response, respectively, many bacteria also synthesize cyclic dinucleotides such as cyclic di-AMP (c-di-AMP) 2 and cyclic di-GMP (c-di-GMP). These nucleotides are often involved in the control of motility and biofilm formation i.e. in the switch between motile and sessile lifestyles (4). Their synthesis is catalyzed by dedicated diadenylate or diguanylate cyclases. These proteins always contain conserved catalytic domains that may be combined with additional domains for signal input and output. In addition, bacteria that produce cyclic dinucleotides also contain nucleotide-specific phosphodiesterases for the degradation of the molecules (5,6).
The synthesis, mode of action, and degradation of c-di-GMP have been studied in detail in many bacteria (for a review, see Ref. 4). In contrast, much less is known about the metabolism and physiological function of c-di-AMP. Diadenylate cyclase activity was first described for the DisA protein of Bacillus sub-* This work was supported by grants from the Deutsche Forschungsgemeinschaft (to J. S.). □ S This article contains supplemental Tables S1 and S2. 1 To whom correspondence should be addressed: Dept. of General Microbiology, Inst. of Microbiology and Genetics, Georg-August University Göttingen, Grisebachstr. 8, D-37077 Göttingen, Germany. Tel.: 49-551-393781; Fax: 49-551-393808; E-mail: jstuelk@gwdg.de. tilis (7). This octameric protein has two interdependent activities. (i) It binds DNA via its RuvA-like C-terminal DNA-binding domain and scans its integrity. (ii) It synthesizes c-di-AMP. If the protein arrives at branched DNA molecules present in Holliday junctions, then the catalytic activity is inhibited, and this reduction in c-di-AMP concentration results in the delay of sporulation (7,8). DisA contains a catalytic domain, called the diadenylate cyclase (DAC) domain (previously referred to as domain of unknown function, DUF147) (7). The discovery of this c-di-AMP-producing enzyme revealed that proteins with similar DAC domains are present in many bacteria, both Grampositive and Gram-negative, and archaea (5). This observation suggests that c-di-AMP might be a widespread signaling nucleotide. Moreover, DAC domains are coupled not only to RuvAlike DNA-binding domains but also to a wide variety of different domains of unknown function. These domains may control the signal in-and/or output of the proteins (5). The presence of DAC domains in so many different organisms and in varying domain arrangements indicates that c-di-AMP levels may respond to a number of distinct stimuli and that c-di-AMP may play an important role in the control of different cellular activities.
We are interested in the molecular biology of the Gram-positive model bacterium B. subtilis. Signaling in this bacterium involves a variety of transcription factors, alternative factors, RNA-mediated regulation via RNA-binding proteins or riboswitches, and protein phosphorylation. Moreover, signal transduction via small molecules plays an important role in B. subtilis. In this bacterium, small peptides control the initiation of sporulation and the induction of genetic competence (9). Moreover, ppGpp mediates the stringent response by inhibiting GTP synthesis and thus by differential control of transcription of mRNAs that use an A or a G as the first nucleotide (10 -12). The genome of B. subtilis encodes several potential diguanylate cyclases and potential c-di-GMP-specific phosphodiesterases (13,14). Very recently, c-di-GMP was found to control motility by binding the YpfA protein, which in turn interacts with and inhibits the motor protein MotA in B. subtilis (14). In addition to these well established signaling nucleotides, c-di-AMP synthesis by the DisA protein was recently discovered for the first time (7). In contrast to most other bacteria that contain only one diadenylate cyclase, B. subtilis encodes two additional proteins with DAC domains that may possibly be involved in c-di-AMP synthesis. Finally, the c-di-AMP-specific phosphodiesterase GdpP (previously referred to as YybT (15)) degrades cyclic di-AMP (16). Recently, c-di-AMP was implicated in the control of cell wall homeostasis in B. subtilis, and it was demonstrated that this signaling nucleotide is essential for the growth of the bacterium (15).
In this work, we have analyzed the activity of the so far unknown diadenylate cyclases of B. subtilis, YbbP and YojJ. Our results indicate that both proteins have enzymatic activity. Moreover, we demonstrate that the activity of YbbP is modulated by a protein-protein interaction with the modulator protein YbbR. The activity of the sporulation-specific diadenylate cyclase YojJ is self-restricted. Full YojJ activity and accumulation of c-di-AMP due to the inactivation of the phosphodiesterase gene gdpP result in impaired growth. Thus, the cells need a certain level of c-di-AMP, and strongly reduced or increased amounts of the nucleotide seem to be deleterious for the cell. Based on our results, YbbP, YojJ, and YbbR were renamed cyclic di-AMP synthase A (CdaA), cyclic di-AMP synthase S, sporulation-specific (CdaS), and cyclic di-AMP synthase A regulator (CdaR), respectively. These designations will be used hereafter.
E. coli was grown in LB medium. B. subtilis was grown in LB medium or in C minimal medium (19) supplemented with carbon sources and auxotrophic requirements (at 50 mg/liter) as indicated. CSE medium is C medium supplemented with potassium succinate and potassium glutamate (6 and 8 g/liter, respectively). LB and SP (per liter: 8 g of nutrient broth, 1 mM MgSO 4 ⅐7H 2 O, 13.4 mM KCl, 0.5 mM CaCl 2 , 10 M MnCl, and 2.4 mg of ammonium ferric citrate) plates were prepared by addition of 17 g of Bacto agar/liter (Difco) to LB and SP, respectively (17,20).
DNA Manipulation-Transformation of E. coli and plasmid DNA extraction were performed using standard procedures (17). All commercially available plasmids, restriction enzymes, T4 DNA ligase, and DNA polymerases were used as recommended by the manufacturers. DNA fragments were purified from agarose gels using the QIAquick PCR purification kit (Qiagen, Hilden, Germany). DNA sequences were determined using the dideoxy chain termination method (17). Chromosomal DNA of B. subtilis was isolated as described (20).
Transformation and Phenotypic Analysis-Standard procedures were used to transform E. coli (17), and transformants were selected on LB plates containing ampicillin (100 g/ml). B. subtilis was transformed with plasmid DNA according to the two-step protocol (20). Transformants were selected on SP plates containing the appropriate antibiotics.
In B. subtilis, amylase activity was detected after growth on plates containing nutrient broth (7.5 g/liter), 17 g of Bacto agar/ liter (Difco), and 5 g of hydrolyzed starch/liter (Connaught). Starch degradation was detected by sublimating iodine onto the plates.
Quantitative studies of lacZ expression in B. subtilis were performed as follows. Cells were grown in C minimal medium containing succinate, glutamate, and sugars as indicated. Cells were harvested in the logarithmic growth phase at an A 600 of 0.6 -0.8. ␤-Galactosidase-specific activities were determined with cell extracts obtained by lysozyme treatment as described previously (20). One unit of ␤-galactosidase is defined as the amount of enzyme that produces 1 nmol of o-nitrophenol/min at 28°C.
Construction of Deletion and Complementation Strains-Deletion of the cdaA, cdaR, and cdaS genes was achieved by transformation with PCR products constructed using oligonucleotides (see supplemental Table S1) to amplify DNA fragments flanking the target genes and intervening antibiotic resistance cassettes (21) as described previously (22).
To obtain regulated ectopic expression of cdaS, we used plasmid pGP1959. This plasmid was obtained as follows. The cdaS gene was amplified using the primer pair FX46/FX47 and digested with XbaI and KpnI. The fragment was cloned into vector pGP888 (23) linearized with the same enzymes. Plasmid pGP1959 was then linearized with ScaI and used to transform the relevant B. subtilis mutants. In the resulting strains, the cdaS gene under the control of the xylose-regulated promoter is ectopically integrated into the lacA locus.

Construction of Integrative Vectors That Allow the Ectopic Expression of Proteins Carrying a Strep Tag in B. subtilis-To
facilitate the purification of proteins that are fused to a Strep tag directly from B. subtilis, we constructed two plasmids that allow the integration of the expression cassette into the lacA gene. For the fusion of the Strep tag to the N terminus of the target protein, the integration vector pGP882 (23) was linearized with BamHI and SmaI and ligated in a three-arm ligation to a kanamycin resistance cassette and to a PCR product corresponding to the constitutive promoter of the degQ Hy gene, the DNA coding for the Strep tag, and a multiple cloning site. The resistance gene was amplified from plasmid pDG780 (21) using the primers KG46 and KG47 and digested with SmaI and EcoRI.
The promoter-Strep tag fragment was obtained by PCR using plasmid pGP380 (24) and the primer pair M13-fwd/HE307 and subsequent digestion with EcoRI and BglII. The resulting plasmid was pGP1459. For the fusion of the Strep tag to the C terminus of the target protein, we constructed plasmid pGP1460 essentially in the same way. However, the fragment covering the promoter of the degQ Hy gene, the multiple cloning site, and DNA coding for the Strep tag was obtained with pGP382 (24) and the primer pair M13-fwd/HE308.
Construction of Strains Containing Tagged Proteins-To facilitate the analysis of protein-protein interactions involving CdaA and CdaR, we constructed strains expressing CdaA and CdaR carrying a triple FLAG tag and a Strep tag, respectively, at their C termini. To express CdaA fused to a triple FLAG tag at its native locus, we used plasmid pGP1966. This plasmid was obtained by amplification of about 300 bp of the 3Ј-end of the cdaA gene using the primer pair FX62/FX63 and insertion of the PCR product into the vector pGP1087 (23) linearized with KpnI and HindIII. For the ectopic expression of CdaR carrying a C-terminal Strep tag, we used plasmid pGP1969. This plasmid was constructed by cloning the cdaR gene (amplified using the primer pair FX56/FX57) into the integration vector pGP1460 linearized with XbaI and PstI. Prior to transformation of the relevant B. subtilis strains, plasmid pGP1969 was digested with NotI.
Detection of Protein-Protein Interactions-The isolation of protein complexes from B. subtilis cells was performed using Strep-protein interaction experiment technology (24). Briefly, growing cultures of B. subtilis were treated with formaldehyde (0.6%, w/v; 20 min) to facilitate cross-linking of interacting proteins (24). The Strep-tagged proteins and their potential inter-

c-di-AMP homeostasis in B. subtilis
action partners were then purified from crude extracts using a StrepTactin column (IBA, Göttingen, Germany) and desthiobiotin as the eluent. Interacting proteins were identified by Western blot analysis. Primary protein-protein interactions were identified by bacterial two-hybrid analysis (18). The bacterial two-hybrid system is based on the interaction-mediated reconstruction of Bordetella pertussis adenylate cyclase (CyaA) activity in E. coli. Functional complementation between two fragments (T18 and T25) of CyaA as a consequence of the interaction between bait and prey molecules results in the synthesis of cAMP, which is monitored by measuring the ␤-galactosidase activity of the cAMPcatabolite activator protein-dependent promoter of the E. coli lac operon. Plasmid p25-N allows the expression of proteins fused to the N terminus of the T25 fragment of CyaA, whereas pUT18C allows the expression of proteins fused to the C terminus of the T18 fragment (18). The plasmids constructed for the bacterial two-hybrid assay (see supplemental Table S2) were used for co-transformation of E. coli BTH101, and the proteinprotein interactions were then analyzed by plating the cells on LB plates containing 100 g/ml ampicillin, 50 g/ml kanamycin, 80 g/ml 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-Gal), and 0.5 mM isopropyl ␤-D-thiogalactopyranoside, respectively. The plates were incubated for a maximum of 72 h at 30°C.
Western Blotting-For Western blot analysis, proteins were separated by 12% SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF) membranes (Bio-Rad) by electroblotting. Rabbit anti-FLAG polyclonal antibodies (Sigma-Aldrich; 1:10,000) served as primary antibodies. The antibodies were visualized by using anti-rabbit immunoglobulin alkaline phosphatase secondary antibodies (Promega) and the CDP-Star detection system (Roche Diagnostics) as described previously (19).
Construction of Strains Carrying lacZ Fusions-Plasmid pAC6 (25) was used to construct transcriptional fusions of the cdaA or glmM control regions with the lacZ gene. The promoter regions were amplified using the primer pairs FX85/ FX87 and FX86/FX88, respectively. The PCR products for cdaA and glmM were digested with EcoRI/BamHI and MfeI/BamHI, respectively, and cloned into pAC6 linearized with EcoRI and BamHI. The resulting plasmids, pGP1980 and pGP1981, respectively (see Table 1 and supplemental Table S2 for details), were linearized with ScaI and used to transform B. subtilis 168.
Northern Blot Analysis-Preparation of total RNA and Northern blot analysis were carried out as described previously (26). Digoxigenin RNA probes were obtained by in vitro transcription with T7 RNA polymerase (Roche Diagnostics) using PCR-generated DNA fragments as templates. The primer pairs used to amplify DNA fragments specific for cdaA, cdaR, glmM, and glmS are listed in supplemental Table S1. The reverse primers contained a T7 RNA polymerase recognition sequence. In vitro RNA labeling, hybridization, and signal detection were carried out according to the instructions of the manufacturer (DIG RNA labeling kit and detection chemicals, Roche Diagnostics).
Reverse Transcription-PCR (RT-PCR)-B. subtilis 168 total RNA was converted to first strand cDNA using Maxima reverse transcriptase (Fermentas) following the manufacturer's protocol (ϩRT). To get rid of genomic DNA contaminations, the cDNA was digested with DNaseI (Fermentas) followed by purification with the RNeasy Plus minikit (Qiagen), which includes a genomic DNA eliminator column. As a negative control, an additional RNA sample (ϪRT) was treated equally, but the reverse transcription step was omitted. Samples (ϪRT and ϩRT) were subjected directly to regular PCRs using the primer pairs FX44/FX27, FX116/FX117, or FX118/119. The PCR products were detected by 2% agarose gel electrophoresis.
Expression of Recombinant Diadenylate Cyclases in E. coli-To assess the biochemical activity of the presumptive diadenylate cyclases, the enzymes were expressed in E. coli BL21(DE3). This organism does not produce c-di-AMP and is therefore well suited to analyze the production of the nucleotide (27). For this purpose, the cdaA and cdaS genes were amplified using the primer pairs FX68/FX69 and FX71/FX72, respectively, and cloned between the NdeI and BamHI sites of the expression vector pET28a (Novagen). The resulting plasmids were pGP1973 and pGP1974, respectively. The cdaS L44F mutant allele was amplified using chromosomal DNA of B. subtilis GP1334 and cloned in the same way as the wild type allele. The resulting plasmid was pGP1975. To test the role of CdaR in the control of the diadenylate cyclase activities of CdaA, CdaS, and DisA, the four genes were cloned into the compatible vectors pRSFDuet-1 (cdaR) and pET19b (cdaA, cdaS, and disA) (both vectors were from Novagen). Briefly, the cdaR gene was amplified using the primer pair FX91/FX70 and cloned between the BglII/NdeI sites of pRSFDuet-1. The cdaA, cdaS, and disA genes were amplified using the primer pairs FX68/FX69, FX71/ FX72, and FX111/FX112, respectively, and cloned between the NdeI/BamHI sites of pET19b. The resulting plasmids were pGP1984 (cdaR), pGP1970 (cdaA), pGP1972 (cdaS), and pGP2563 (disA).
Analysis of the Cyclic Dinucleotide Pools-The concentration of cyclic di-AMP was determined by a liquid chromatographycoupled tandem mass spectrometry method essentially as described previously (28). Briefly, E. coli cultures were grown in LB medium at 37°C. At an A 600 of 0.5-0.7, 1 mM isopropyl ␤-D-thiogalactopyranoside was added to induce the expression of the B. subtilis diadenylate cyclase genes and further incubated for 3 h. The culture (10 ml) was harvested by quick centrifugation at 4°C. An additional sample was taken for the determination of total protein amount for normalization purposes. The cell pellet was resuspended in 300 l of extraction buffer (acetonitrile/methanol/water mixture, 40:40:20, v/v/v) and frozen in liquid nitrogen. Samples were then treated by a heating step for 10 min at 95°C followed by centrifugation for 10 min at 20,800 ϫ g at 4°C. The extraction of the resulting pellet was repeated twice with 200 l of extraction mixture at 4°C omitting the heating step. Supernatants were pooled and stored at Ϫ20°C overnight. After centrifugation for 10 min at 20,800 ϫ g at 4°C, the supernatant was removed and dried in a SpeedVac.
For the determination of c-di-AMP in cells of B. subtilis, the cultures (45 ml) were grown in SP medium supplemented with xylose (1%; to induce the ectopic cdaS L44F allele) at 37°C to an A 600 of 1.0. The cells were harvested by quick centrifugation at c-di-AMP homeostasis in B. subtilis 4°C. An additional sample was taken for the determination of total protein amount for normalization purposes. The cell pellet was resuspended in 200 l of H 2 O and lysed in a Micro Dismembrator S (Sartorius) for 3 min at 1,800 rpm. The resulting cell powder was resuspended in 800 l of extraction buffer and incubated on ice for 15 min. Samples were snap frozen in liquid nitrogen and subsequently treated by a heating step for 10 min at 95°C. After centrifugation for 10 min at 20,800 ϫ g, the supernatant was collected and stored at 4°C. The remaining sample mixture was resuspended in 200 l of extraction buffer, vortexed for 45 s, incubated on ice for 15 min, and centrifuged again. The supernatant was collected and pooled with the previous one. Once again, samples were resuspended in 200 l of extraction buffer, vortexed for 45 s, incubated on ice for 15 min, and centrifuged. All supernatants were pooled and stored at Ϫ20°C overnight. After centrifugation for 10 min at 20,800 ϫ g at 4°C, the supernatant was dried in a SpeedVac.
The dried supernatants were dissolved with 200 l of H 2 O. After repeated centrifugation and addition of the internal standard [ 13 C, 15 N]c-di-AMP, part of the supernatants was analyzed by LC-MS/MS.
Quantification of c-di-AMP by MS/MS-The chromatographic separation was performed on a Series 200 HPLC system (PerkinElmer Life Sciences) equipped with a binary pump system and a 200-l sample loop. A combination of column saver (2.0-m filter, Supelco Analytical), security guard cartridge (C 18 , 4 ϫ 2 mm) in an analytical guard holder (Phenomenex), and an analytical NUCLEODUR C 18 Pyramid RP column (50 ϫ 3 mm, 3-m particle size; Macherey-Nagel) temperature-controlled (Series 200 Peltier column oven, PerkinElmer Life Sciences) at 30°C was used. Eluent A consisted of 10 mM ammonium acetate and 0.1% (v/v) acetic acid in water, and eluent B was methanol. The injection volume was 50 l, and the flow rate was 0.4 ml/min throughout the chromatographic run. 100% A was used from 0 to 5 min followed by a linear gradient from 100% A to 70% A until 9 min. Re-equilibration of the column was achieved by constantly running 100% A from 9 to 13 min. The internal standard [ 13 C, 15 N]c-di-AMP and c-di-AMP were eluted with identical retention times of 9.1 min. The analyte detection was performed on an API 3000 triple quadrupole mass spectrometer equipped with an electrospray ionization source (AB SCIEX) using selected reaction monitoring (SRM) analysis in positive ionization mode. The following SRM transitions using a dwell time of 40 ms were detected: [ 13 C, 15 N]c-di-AMP, ϩ689/146 (quantifier) and ϩ689/345 (identifier); c-di-AMP, ϩ659/136 (quantifier), ϩ659/330 (identifier), and ϩ659/524 (identifier). The SRM transitions labeled as "quantifier" were used to quantify the compound of interest, whereas "identifier" SRM transitions were monitored as confirmatory signals. The quantifier SRM transitions were most intense and were therefore used for quantification. The mass spectrometer parameters were as follows: ion spray voltage, 5500 V; temperature, 350°C; nebulizer gas, 6 p.s.i.; curtain gas, 15 p.s.i. MS/MS was performed using nitrogen as the collision gas. The following collision energies were applied: 61 (ϩ689/ 146), 29 (ϩ689/345), 63 (ϩ659/136), 29 (ϩ659/330), and 33 eV (ϩ659/524).

RESULTS
Diadenylate Cyclases in B. subtilis-Diadenylate cyclases are characterized by the presence of a specific domain, the DAC domain (formerly known as DUF147) (5). The genome of B. subtilis encodes three proteins with a DAC domain that may be potential diadenylate cyclases, DisA, CdaA, and CdaS (5,13). The diadenylate cyclase activity of DisA has been demonstrated experimentally (7,8). disA is the promoter-distal gene of the hexacistronic ctsR-mcsA-mcsB-clpC-radA-disA operon. Moreover, a monocistronic disA mRNA is expressed by RNA polymerase containing the alternative factor M (29,30). Recent transcriptome studies show that cdaS is specifically expressed in sporulating cells, suggesting a role for c-di-AMP in spore formation (31). Finally, cdaA seems to be the first gene of a potential operon that also contains the downstream gene cdaR and the essential glmM gene (31).
Transcriptional Organization of the cda-glm Gene Cluster-For additional insights into the expression of the potential diadenylate cyclase CdaA, we performed Northern blot analyses. Only very faint bands were detectable using a probe against cdaA in RNA isolated from a culture grown in minimal medium containing glucose and glutamate (see Fig. 1A, CSE glc). Because the accumulation of the cdaA mRNA is negatively affected by the essential RNase Y (26), we compared the expression of the gene in a strain that allows xylose-controlled expression of RNase Y and thus depletion of the essential RNase in the absence of xylose. As shown in Fig. 1A, a major transcript of 4 kb was detectable upon RNase Y depletion. Because there is essentially no read-through from the upstream rsiW gene into cdaA (31), this transcript is likely to cover the cdaA, cdaR, and glmM genes. Indeed, the same band corresponding to a 4-kb transcript was detected using riboprobes specific for cdaR and glmM (see Fig. 1, B and C). Moreover, a faint 5.7-kb signal was detected with the glmM probe. This signal might correspond to a tetracistronic cdaAR-glmMS mRNA (see Fig. 1F). In addition, signals corresponding to mRNAs of 1.3 and 3.5 kb were observed with the glmM riboprobe. These signals probably correspond to monocistronic glmM and bicistronic glmMS transcripts, respectively. Using a riboprobe specific for glmS, a single 2.0-kb transcript corresponding to a monocistronic glmS mRNA was detected (Fig. 1D). This transcript was very abundant, thus preventing the concomitant detection of the bicistronic glmMS mRNA. The glmS mRNA was less abundant when the bacteria were grown in the presence of glucosamine, the precursor of glucosamine 6-phosphate. This latter metabolite acts as the trigger of the glmS ribozyme that initiates degradation of the transcript (32,33). To detect the transcript of the tetracistronic cdaAR-glmMS operon in a wild type strain, we used RNA from B. subtilis 168 for reverse transcription and subsequent PCR analysis. As shown in Fig. 1E, PCR products were obtained with primer pairs that allow the amplification of regions between the coding genes. This result supports the observations of the Northern blot analysis.
The Northern blot analysis revealed that transcription of the cda-glm operon is initiated upstream of the cdaA gene and that a second promoter or a processing site might exist upstream of glmM. To address this issue, we fused the upstream regions of c-di-AMP homeostasis in B. subtilis cdaA and glmM to a promoterless lacZ gene and determined the promoter activities by assaying the corresponding ␤-galactosidase activities. B. subtilis GP1319 containing the promoterless lacZ gene served as a control. Only background ␤-galactosidase activity was determined for this strain (see Table 2). The strains GP1339 (cdaA-lacZ) and GP1340 (glmM-lacZ) were grown in minimal medium in the absence or presence of glucose or glucosamine. With the cdaA promoter, about 40 units of ␤-galactosidase were detected irrespective of the available carbon source. Such an activity is often observed for weak constitutive promoters (34). In contrast, no promoter activity was detectable upstream of the glmM gene (see Table 2). This finding is in good agreement with a recent bioinformatics search for promoter regions in B. subtilis that suggested a SigA-dependent promoter upstream of cdaA, whereas no promoter was predicted for glmM (31). Taken together, our data suggest that the cda-glm gene cluster is constitutively expressed from a promoter upstream of cdaA and that multiple RNase-and ribozyme-dependent processing events contribute to the expression of the glmMS suboperon.

Diadenylate Cyclase Activity of CdaA and CdaS-Both
CdaA and CdaS possess the conserved DAC domain that is thought to be required for c-di-AMP formation. To test whether these proteins do indeed exhibit diadenylate cyclase activity, we took advantage of the inability of E. coli to produce c-di-AMP (27). The cdaA and cdaS genes were cloned into the expression vector pET28a, and the amounts of c-di-AMP in E. coli BL21(DE3) carrying the corresponding plasmids pGP1973 and pGP1974, respectively, were compared with those of the same strain car- . RNA was isolated from B. subtilis 168 grown in CSE minimal medium supplemented with glucose (0.5%) (glc) or glucosamine (0.5%) (glcNH 2 ). For the depletion of the essential RNase Y, B. subtilis GP193 was grown in CSE supplemented with glucose and with (ϩ) or without (Ϫ) xylose (2.0%). 5 g of total RNA was separated by electrophoresis in 1.0% agarose gels, and after blotting, nylon membranes were hybridized to gene-specific riboprobes as indicated. Note that the probes cross-hybridized with the 16 and 23 S rRNAs. The sizes of 16 S rRNA and 23 S rRNA are indicated by arrows. Moreover, the exposure time for glmS (D) was reduced compared with the other blots, and thus, the 3.5-kb transcript was not detected. E, RT-PCR analysis. Total RNA was used as a template in the reverse transcription reaction with a random nonamer primer. Regular PCRs were carried out subsequently for the amplification of different parts of the mRNA transcript using primer pairs indicated in F. As a negative control, RNA template without addition of reverse transcriptase was used (ϪRT). Experiments were carried out with two biological replicate samples at least three times each. F, summary of the genetic and transcriptional organization.

c-di-AMP homeostasis in B. subtilis
rying the empty vector pET28a. As shown in Fig. 2, no c-di-AMP was present in the strain with the empty vector. In contrast, the expression of CdaA and CdaS resulted in the detection of 150 and 670 ng of c-di-AMP/mg of protein, respectively. This result provides unequivocal evidence for the diadenylate cyclase activity of both proteins. c-di-AMP Formation during Exponential Growth Is Essential in B. subtilis-To identify roles for the diadenylate cyclases, we deleted the three genes individually from the chromosome of B. subtilis. Moreover, we combined the mutations to construct double and triple mutants. The three single mutant strains GP983 (⌬cdaS), GP987 (⌬disA), and GP997 (⌬cdaA) were viable and did not exhibit an obvious phenotype during logarithmic growth. Similarly, the disA cdaS and cdaA cdaS double mutants, GP991 and GP989, respectively, grew like the wild type. In contrast, we were unable to construct a cdaA disA double mutant strain or a triple mutant devoid of all three diadenylate cyclases. This observation is in good agreement with the recent report that c-di-AMP is essential for the viability of B. subtilis (15). Moreover, our results suggest that the lack of cdaS expression during exponential growth might be the reason for our inability to obtain the cdaA disA double mutant that would have cdaS as the only gene encoding a diadenylate cyclase. To test this hypothesis, we expressed the cdaS gene ectopically under the control of a xylose-regulated promoter. In the presence of xylose, we were able to delete the chromosomal copies of all three diadenylate cyclase-encoding genes. The triple mutant with the ectopic cdaS copy, GP1327, grew like the wild type strain as long as the cdaS gene was expressed due to the presence of the inducer xylose. However, the strain did not grow in the absence of xylose. Thus, the expression of the cdaS gene was the limiting factor that prevented the construction of the cdaA disA double mutant. Moreover, these results clearly support the essential role of c-di-AMP for the physiology of B. subtilis.
Physical Interaction between CdaR and the Diadenylate Cyclase CdaA-The cdaA gene is located upstream of the cdaR gene, and the two genes are part of the cda-glm operon identified in this work (see above). The function of the cdaR gene or its protein product has so far not been identified. The CdaR protein contains a repeat of four similar domains (called YbbRlike domains (35)). Interestingly, in some ␦-proteobacteria and streptococci, DAC domains are organized in one protein with YbbR-like domains (5). This suggests that the YbbR domains might somehow control the activity of the DAC domain. Specifically, the genetic arrangement and co-expression of the cdaA and cdaR genes implies that B. subtilis CdaR might control the diadenylate cyclase activity of CdaA.
If CdaR regulates the activity of CdaA, one would expect that the two proteins interact physically in the cell. To test this hypothesis, we attempted to purify CdaR with its attached potential interaction partners. For this purpose, we constructed strain GP1331 that ectopically encodes a variant of CdaR carrying a C-terminal Strep tag for affinity purification. Moreover, this strain codes for a CdaA variant labeled with a FLAG tag. Parallel cultures of strain GP1331 were grown in C minimal medium with glucose as the carbon source, and the cross-linker formaldehyde was added to one of the cultures to fix the interaction between CdaR and its possible partners. CdaR-Strep was then purified from protein extracts of these cultures, and the elution fractions were analyzed by SDS-PAGE and subsequent Western blot analysis with antibodies recognizing the FLAG tag to detect the CdaA protein. As shown in Fig. 3A, CdaA-FLAG was present in the crude extract of GP1331. Similarly,

c-di-AMP homeostasis in B. subtilis
CdaA-FLAG was detected in the elution fractions of Streptagged CdaR even when CdaR had been purified from the culture that had not been treated with the cross-linker. When CdaR was purified from the cross-linked cultures, a second larger protein was recognized by the antibodies. This band corresponds to CdaA dimers that can be resolved by prolonged heating of the samples. To exclude nonspecific detection of FLAG-tagged CdaA, two different controls were performed. First, we used strain GP1332 that expresses CdaR-Strep and a FLAG-tagged version of the DEAD box RNA helicase CshA. Again, CshA-FLAG was present in the crude extract. However, no CshA was detectable in the elution fractions of CdaR-Strep (see Fig. 3B). Second, strain GP1333 encoding Strep-tagged CshA and FLAG-tagged CdaA served as a control to exclude nonspecific binding of CdaA-FLAG to the StrepTactin affinity matrix. When CshA-Strep was purified, no CdaA-FLAG was present in the elution fraction (see Fig. 3C). This result demonstrates that the observed co-purification of CdaR and CdaA is specific and that it is indeed the result of a physical interaction between the two proteins. The co-purification of CdaA with CdaR even in the absence of the cross-linker suggests that the interaction is rather strong.
The co-purification of two proteins indicates that they are part of common protein complexes but does not exclude the possibility that the interaction is indirect. To test whether CdaA and CdaR are capable of interacting directly, we used the bacterial adenylate cyclase two-hybrid system (18). As shown in Fig. 3D, a primary interaction between the two proteins was observed, thus confirming our in vivo results.
CdaR Stimulates the Diadenylate Cyclase Activity of CdaA-The observed physical interaction between CdaA and CdaR supports the idea that CdaR might control the activity of the diadenylate cyclase CdaA. To get a first impression of the role of CdaR, we used a genetic approach. For this purpose, we constructed three isogenic strains with deletions of disA and cdaS and the ectopic inducible cdaS allele. In these strains, we deleted either cdaA, cdaR, or both genes. As described above, strain GP1327 with the deletion of all three diadenylate cyclaseencoding genes was unable to grow in the absence of the inducer xylose that allowed CdaS-catalyzed synthesis of c-di-AMP. In contrast, strain GP1328 that has a functional cdaA gene but a deletion of cdaR was able to grow in the absence of xylose. This result allows two important conclusions to be drawn. First, the construct used here does not interfere with the

. Detection of in vivo interactions between CdaA and CdaR by Western blot (A-C) and bacterial two-hybrid (D) analyses.
Cells were grown in CSE minimal medium supplemented with glucose in the absence and presence of the cross-linker formaldehyde (FA). The protein complexes were isolated from B. subtilis GP1331 (A) and GP1332 (B) with an ectopically encoded CdaR protein carrying a C-terminal Strep tag or GP1333 (C) with CshA encoded in the native locus carrying a C-terminal Strep tag. A C-terminal FLAG tag was attached to the putative interaction partner CdaA. 28 l of the elution fractions from each purification without (ϪFA) or with (ϩFA) cross-linking by formaldehyde and 15 g of untreated crude extracts (CE) were analyzed by 12% SDS-PAGE. After electrophoresis and blotting onto a PVDF membrane, interaction partners were detected by an ␣-FLAG antibody. As a control, CshA and CdaA (B and C, respectively) were detected. For the bacterial two-hybrid analysis (D), the cdaR and cdaA genes were cloned into p25-N. In addition, cdaR was cloned into the plasmid pUT18C. Plasmid p25-N allows the expression of the selected enzymes fused to the N terminus of the T25 domain of the adenylate cyclase. Plasmids pUT18C allows the expression of the selected proteins fused to the C terminus of the T18 domain of the B. pertussis adenylate cyclase. The E. coli transformants were incubated for 72 h at 30°C. The degradation of X-Gal (blue color) indicates the presence of a functional adenylate cyclase because of the interaction of the two proteins of interest.

c-di-AMP homeostasis in B. subtilis
expression of the essential downstream glmM gene, thus excluding the possibility that cdaA was required due to possible polar effects. Second, CdaA had retained its diadenylate cyclase activity even in the absence of CdaR. For strain GP1329 with a deletion of both cdaA and cdaR genes, xylose was again required to allow growth of the bacteria. Taken together, these observations suggest that CdaR might act as a negative effector of CdaA. Alternatively, CdaA might a priori have diadenylate cyclase activity that can be enhanced upon interaction with CdaR.
To test the role of CdaR for the activity of CdaA more directly, we co-expressed the two proteins in E. coli and determined the concentration of c-di-AMP of this strain and of strains expressing either CdaA or CdaR alone (see Table 3). In this experiment, we detected 237 ng of c-di-AMP/mg of protein in the strain carrying the cdaA gene. This is in good agreement with earlier observations (see above and Fig. 2). No c-di-AMP was detectable in the strain expressing the cdaR gene. In the strain expressing both cdaA and cdaR, a more than 20-fold increase in the c-di-AMP concentration (5,256 ng of c-di-AMP/mg of protein) was observed. Because all three diadenylate cyclases of B. subtilis share the conserved DAC domain, we could not exclude the possibility that CdaR might also stimulate the activities of CdaS and DisA. To address this question, we determined the formation of c-di-AMP by CdaS and DisA in the presence or absence of CdaR. As shown in Table 3, CdaR did not affect the activity of either CdaS or DisA. Taken together, these results demonstrate that CdaR specifically stimulates the enzymatic activity of CdaA.
Isolation and Characterization of a Hyperactive Variant of CdaS-As mentioned above, B. subtilis GP1327 was unable to grow in the absence of the inducer xylose i.e. if no active diadenylate cyclase was present. However, after prolonged incubation, this strain gave rise to small colonies that grew even in the absence of xylose. We hypothesized that this suppression might result from (i) an inactivation of the xylR gene that would lead to constitutive cdaS expression, (ii) a mutation in the promoter region that would lead to constitutive promoter activity, or (iii) a mutation within the cdaS gene that allows the production of sufficient c-di-AMP even at very low cdaS expression levels. Of these possibilities, the inactivation of xylR seemed rather unlikely because the strain GP1327 carries two copies of xylR, one at the native locus and a second copy at the ectopic lacA site. Thus, we sequenced the p xylA -cdaS copies present in the lacA locus of two suppressor mutants to distinguish between the second and third possibilities. In one mutant, GP1334, we found a single substitution, 130C3 T, in the coding sequence of the cdaS gene. This mutation results in a replacement of Leu-44 of CdaS by Phe (L44F). In the second suppressor mutant, GP1348, two mutations that both affect the binding site for XylR were detected. An extra A is inserted at the end of the spacer between the two parts of the palindrome that is recognized by XylR, and the A at position 4 of the second unit of the palindrome is replaced by a T. Together, these mutations affect the length of the spacer and disturb the palindrome. Both spacer length and palindrome quality are important for efficient repression by XylR (36,37). Thus, the two mutants indeed had the two expected types of mutations. Because the mutation affecting the expression of CdaS was not of primary interest in the context of this study, we focused our attention on the mutant cdaS allele. We first confirmed that the suppressor phenotype did indeed result from the mutation of cdaS and not from secondary mutations. For this purpose, chromosomal DNA of the suppressor mutant GP1334 was used to transform strain GP991 that is deleted for disA and cdaS. Selection for resistance against kanamycin, tetracycline, and chloramphenicol allowed the isolation of clones that had the three chromosomal genes encoding the diadenylate cyclases deleted and the mutant cdaS copy at the lacA locus. These clones were tested for growth in the absence of xylose to verify that the mutation in cdaS was causative for the suppressor phenotype. Indeed, all clones were able to grow in the absence of xylose.
The findings reported above suggest that the mutant CdaS L44F protein is capable of producing sufficient c-di-AMP even at very low expression. To test this hypothesis, the mutant allele was cloned into the expression vector pET28a, and the formation of c-di-AMP was assayed in the heterologous host E. coli. Although the expression of wild type CdaS resulted in the detection of 670 ng of c-di-AMP/mg of protein, 56,400 ng of c-di-AMP/mg of protein was detected in the E. coli strain expressing CdaS L44F . Thus, the mutation resulted in a nearly 100-fold increase in c-di-AMP production. This result is in excellent support of the idea that the mutant enzyme exhibits increased activity.
Accumulation of c-di-AMP Inhibits Growth and Leads to Aberrant Cell Morphology-As shown above, c-di-AMP is required for the growth of B. subtilis. The availability of the cdaS L44F allele allowed us to investigate the consequences of increased c-di-AMP accumulation. Because B. subtilis does also possess the c-di-AMP-specific phosphodiesterase GdpP, it was possible that the increased amount of c-di-AMP could be degraded by the latter enzyme. Therefore, we included gdpP mutants in our study. We first determined the intracellular concentrations of c-di-AMP in the wild type strain GP1173 and the isogenic cdaS L44F gdpP mutant GP1344. In the wild type, we detected 5.6 Ϯ 2.82 ng of c-di-AMP/mg of protein. In contrast, a concentration of 17.9 Ϯ 2.49 ng of c-di-AMP/mg of protein was determined for GP1344. Thus, the intracellular concentration of c-di-AMP was indeed increased upon expression of the hyperactive cdaS allele and deletion of the phosphodiesteraseencoding gene gdpP.

c-di-AMP homeostasis in B. subtilis
To study the consequences of c-di-AMP accumulation for bacterial growth and morphology, we used a set of four strains that were wild type for the diadenylate cyclases and that additionally contained either the cdaS wild type or the cdaS L44F mutant allele under the control of the xylose promoter. Moreover, each cdaS allele was also combined with a deletion of the gdpP gene to prevent degradation of c-di-AMP. When the cells were cultivated in the absence of xylose (i.e. when the ectopic cdaS alleles were not expressed), the growth was similar for the four tested strains (see Fig. 4). However, strain GP1344 encoding the hyperactive CdaS diadenylate cyclase variant in the absence of the phosphodiesterase GdpP exhibited a somewhat slower growth (generation time of 53.6 versus 48.5 min for the isogenic strain with wild type CdaS (GP1343)). The differences became more drastic when the ectopic cdaS alleles were induced with xylose. Although the growth of the strain expressing wild type cdaS in the presence of the phosphodiesterase (GP1341) was not affected, the isogenic strain GP1343 from which the gdpP gene was deleted grew somewhat more slowly (see Fig. 4) (generation time of 49.7 min for GP1341 versus 60.0 min for GP1343). Similarly, expression of the mutant allele cdaS L44F in the presence of a functional phosphodiesterase (GP1342) had only a slight effect (76.4 min for GP1342 versus 49.7 min for GP1341). However, when the hyperactive CdaS variant was expressed in the absence of the functional phosphodiesterase GdpP (GP1344), growth was severely impaired (see Fig. 4; generation time of 89.1 min). These data suggest that the accumulation of excess c-di-AMP is detrimental to the normal growth of B. subtilis.
For a first glimpse of how c-di-AMP might interfere with the growth of B. subtilis, we observed the morphology of the strains GP1173 (no accumulation of additional c-di-AMP) and GP1344 (isogenic construct with strong accumulation of c-di-AMP due to the expression of the hyperactive diadenylate cyclase CdaS L44F and the absence of the phosphodiesterase GdpP). During the logarithmic growth phase, we observed the formation of curled non-separated cell filaments in the strain accumulating c-di-AMP (GP1344 in the presence of xylose) (see Fig. 5). Interestingly, the curled morphology disappeared when the cells entered sporulation (Fig. 5). In this phase, the cells were short with a significant portion already containing spores, a phenotype indistinguishable from that of the control strain GP1173. This wild type-like morphology of the sporulating cells may result from either the acquisition of a suppressor mutation that prevents the accumulation of c-di-AMP or from the insensitivity of sporulating cells to increased amounts of c-di-AMP. To distinguish between these two possibilities, we tested the phenotype of cells that had undergone sporulation after a new growth cycle. Again, exponentially growing cells exhibited the curled filament morphology, whereas the sporulating cells were short and contained spores (data not shown). Thus, the consequences of c-di-AMP accumulation are different during the different phases of growth: although a process related to cell division seems to be impaired during logarithmic growth, this is obviously not the case during sporulation.
Aberrant cell morphologies are often observed with mutants impaired in cell wall metabolism. Recently, a depletion of c-di-AMP was shown to cause a defect in peptidoglycan biosynthesis that could be suppressed by the addition of magnesium (15). To test whether an excess of c-di-AMP also causes a defect in peptidoglycan metabolism, we tested the effect of an addition of magnesium on the morphology of the bacteria that accumulate c-di-AMP. As shown in Fig. 6, the addition of magnesium did not affect the (normal) morphology of strain GP1173 that can degrade c-di-AMP. In contrast, the curly appearance of strain GP1344 that accumulates an excess of c-di-AMP was abolished at a high magnesium concentration. However, many cells did not divide properly as seen by the formation of long cell chains, suggesting that magnesium cannot completely counteract the effect of a massive c-di-AMP accumulation. Thus, our results suggest that increased levels of c-di-AMP lead to a defect in the peptidoglycan-synthesizing machinery.

DISCUSSION
In this work, we have demonstrated that the intracellular concentration of the signaling nucleotide c-di-AMP has to be adjusted to a certain level. Amounts of the nucleotide that are too low or too high are disadvantageous for the cell.
The genetic evidence presented here and in a previous study (15) demonstrates that the three diadenylate cyclases of B. subtilis can replace each other in the generation of a c-di-AMP pool sufficient for growth. However, the three enzymes seem to have distinct functions. DisA scans the DNA and stops c-di-AMP production if it encounters problems with DNA integrity such as Holliday junctions. The reduced DisA-mediated c-di-AMP synthesis results in a delay of sporulation (8,38). This specific function is rather unlikely to be involved in the essential role of c-di-AMP. The diadenylate cyclase CdaS is specifically expressed during sporulation most likely by RNA polymerase containing the late forespore-specific factor G (31). In agreement with this observation, CdaS was unable to provide the cell with sufficient c-di-AMP in the absence of DisA and CdaA unless it was expressed from a regulated promoter that is also active during exponential growth. Thus, the function of CdaS seems to be limited to the spore.
Several lines of evidence suggest that CdaA is implicated in the control of cell wall biosynthesis. First, CdaA is encoded in an operon with proteins involved in cell wall biosynthesis. Interestingly, the genetic clustering of CdaA-related diadenylate cyclase genes with the essential glm genes involved in the generation of glucosamine 1-phosphate, a key precursor for cell wall biosynthesis, is conserved in most ␦-proteobacteria and firmicutes (with the notable exception of cell wall-less mollicutes). Second, a link of at least one of the diadenylate cyclases to cell wall metabolism is also supported by the recent study of Luo and Helmann (15). These authors report that accumulation of c-di-AMP due to the inactivation of the gdpP gene encoding the specific phosphodiesterase results in increased resistance to cell wall antibiotics such as ␤-lactams and conclude that c-di-AMP plays an essential role in peptidoglycan homeostasis. Third, this conclusion is supported by our observation that severe overproduction of c-di-AMP interferes with cell morphology in a Mg 2ϩ -dependent manner (see Fig. 6). This is reminiscent of the phenotypes of other mutations that affect cell wall synthesis (39 -42).
The obvious functional specialization of the three diadenylate cyclases suggests that the c-di-AMP synthesized by the individual proteins is active in a time-and compartment-specific manner. This is rather obvious for CdaS, which is expressed only in the forespore, suggesting that c-di-AMP produced by CdaS has a spore-specific function. In the vegetative cell, DisA is associated to the DNA, whereas CdaA contains three transmembrane domains and is associated to the membrane (43). Thus, the enzymes seem to form distinct c-di-AMP pools in a temporally and spatially ordered manner. In this way, the c-di-AMP may also be close to its potential target proteins.

c-di-AMP homeostasis in B. subtilis
A similar hypothesis has been proposed for c-di-GMP, which can be produced by more than a dozen different proteins in a single bacterial cell (44).
The rather specialized functions of DisA and CdaS suggest that CdaA is the major player in c-di-AMP production in B. subtilis and is key for the essential function of the nucleotide. In this context, the conserved genetic linkage with the essential glmM and glmS genes is noteworthy. The idea of a crucial function for CdaA is further supported by the observation that the single diadenylate cyclases of the Gram-positive pathogens Listeria monocytogenes, Staphylococcus aureus, and Streptococcus pneumoniae are highly similar to CdaA (including the domain organization) and that the encoding genes are essential (45)(46)(47). Moreover, the genes homologous to cdaA and cdaR are also linked to glmM and glmS in those organisms. The intimate genetic, transcriptional, and functional linkage of the cdaAR and glmMS genes suggests that they form a conserved module, the cda-glm module.
As mentioned above, the c-di-AMP concentration has to be tightly controlled to ensure that it does not fall below or exceed a certain physiological concentration. In the case of CdaA, the enzyme has a basal activity, and this activity can be increased by a specific regulatory interaction with the CdaR protein. CdaR proteins are widespread in bacteria; however, this is the first report of their function. These proteins consist of a repeated conserved domain (the YbbR domain); B. subtilis CdaR contains four such domains. The structural analysis of YbbR domains from Desulfitobacterium hafniense (35) revealed a striking similarity to the ribosomal protein L25. Because L25 proteins bind the 5 S ribosomal RNA (48,49), it is tempting to speculate that CdaR might also bind RNA and that this binding might in turn control the interaction with and activation of CdaA.
The activity of CdaS seems to be limited to keep the c-di-AMP levels at a physiologically acceptable level. Interestingly, a single amino acid substitution is sufficient to increase the activity of the protein by a factor of 100. The structure of CdaS can be inferred by modeling based on the known structure of the homologous protein from Bacillus cereus (see Fig. 7). The protein consists of three identical subunits, and each monomer contains two long ␣-helices at the N terminus that are followed by the rather globular DAC domain. Interestingly, the mutation resulting in the hyperactive CdaS protein is located in the loop that connects the two N-terminal helices. This might result in a repositioning of the two helices with respect to each other and the catalytic DAC domain. Based on this finding, it is tempting to speculate that the N-terminal helices are involved in the control of the activity of the DAC domain and thus in the sporespecific synthesis of c-di-AMP. Similarly, inhibitory protein domains have been found in the B. subtilis transcription factor RocR and the alternative factor 54 (51,52).
Future studies will focus on the distinct molecular mechanisms controlling the c-di-AMP levels in B. subtilis. Moreover, we will search for the receptors of c-di-AMP and thus for the precise cellular functions of this fascinating signaling molecule.