Structural and Biochemical Analysis of the Essential Diadenylate Cyclase CdaA from Listeria monocytogenes*

Background: Listeria monocytogenes CdaA is an essential diadenylate cyclase. Results: CdaA activity depends on manganese and cobalt ions. Conclusion: CdaA has an unusual requirement for metal cofactors. Significance: Characterization of essential enzymes is important for developing novel antibiotics. The recently identified second messenger cyclic di-AMP (c-di-AMP) is involved in several important cellular processes, such as cell wall metabolism, maintenance of DNA integrity, ion transport, transcription regulation, and allosteric regulation of enzyme function. Interestingly, c-di-AMP is essential for growth of the Gram-positive model bacterium Bacillus subtilis. Although the genome of B. subtilis encodes three c-di-AMP-producing diadenlyate cyclases that can functionally replace each other, the phylogenetically related human pathogens like Listeria monocytogenes and Staphylococcus aureus possess only one enzyme, the diadenlyate cyclase CdaA. Because CdaA is also essential for growth of these bacteria, the enzyme is a promising target for the development of novel antibiotics. Here we present the first crystal structure of the L. monocytogenes CdaA diadenylate cyclase domain that is conserved in many human pathogens. Moreover, biochemical characterization of the cyclase revealed an unusual metal cofactor requirement.

Bacteria use second messenger molecules like the cyclic and linear nucleotides cAMP and (p)ppGpp, respectively, as well as the cyclic dinucleotides c-di-GMP 3 and c-di-AMP to regulate and coordinate cellular processes in response to extra-and intracellular stimuli (1,2). The functions of cAMP, (p)ppGpp, and c-di-GMP have been studied extensively in several bacteria (3)(4)(5). However, fascinating novel discoveries (e.g. the control of Streptomycetes development by c-di-GMP (6)) reveal that the cellular functions of second messengers, even of well studied signaling molecules like c-di-GMP, are not yet fully explored. Moreover, some bacteria, such as the cyanobacterium Synechocystis PCC6803, are capable of producing the cyclic nucle-otide cGMP (7). However, although the cyclase that synthesizes cGMP has been biochemically and structurally characterized, the physiological role of the nucleotide remains to be elucidated (8).
The second messenger c-di-AMP was coincidently identified in 2008 (9) in the course of the structural analysis of the DNA integrity scanning protein A (DisA), an enzyme involved in proper genome replication and maintenance of genome integrity in the Gram-positive model organism Bacillus subtilis (10,11). A detailed biochemical characterization of DisA revealed that the nucleotide-binding domain of the protein is a diadenylate cyclase (DAC) that converts two molecules of ATP to c-di-AMP (9). It was also observed that the DAC activity of DisA is strongly inhibited by the presence of non-standard DNA secondary structures like Holliday junctions (9). A recent study emphasized a function of DisA in controlling genome replication during outgrowth of B. subtilis spores to prevent propagation of damaged DNA to the germinating cells (12). Meanwhile, homologs of DisA have been identified in several other bacteria like Mycobacterium tuberculosis (13)(14)(15). It is interesting to note that some bacteria possess more than one DAC-containing protein (14,16). In addition to the disA gene, the genome of B. subtilis harbors the cdaA and cdaS genes that code for the DACs CdaA and CdaS, respectively (17). Whereas the disA and cdaA genes are both constitutively expressed during vegetative growth, the cdaS gene encodes the sporulation-specific DAC CdaS that was shown to be required for efficient germination of B. subtilis spores (18).
Although c-di-AMP was discovered only recently, several targets have been identified that bind to the signaling nucleotide. The first target that was shown to bind c-di-AMP is the DNA-binding transcription factor DarR from Mycobacterium smegmatis (19). Cyclic di-AMP stimulates the DNA-binding activity of DarR, which in its nucleotide-bound form acts as a repressor and prevents transcription of three target genes. In three recent systematic approaches, several additional protein targets of c-di-AMP that are widely distributed among other bacteria were identified in Staphylococcus aureus, Listeria monocytogenes, and B. subtilis (20 -22). As yet, most of the mechanistic details of how c-di-AMP affects the activity of its targets are unknown. However, it was shown that c-di-AMP controls potassium uptake in S. aureus and Streptococcus pneumoniae (20,23) and allosterically regulates the activity of the L. monocytogenes pyruvate carboxylase (21). Moreover, c-di-AMP is capable of binding to a conserved riboswitch that controls the expression of the ydaO and ktrAB genes (24,25), of which the latter encodes a potassium transporter in B. subtilis (26). As stated above, c-di-AMP also directly controls potassium uptake (20,23). Thus, c-di-AMP is the first signaling molecule that regulates the expression of a gene and controls the activity of the encoded protein.
The growing resistance of pathogenic bacteria to the majority of antibiotics that are used in both community and health care settings is an enormous threat to infected patients (27,28). Moreover, some isolates of S. aureus are even resistant to multiple antibiotics (29). Unfortunately, over the past 25 years, only a handful of new antibiotics have been launched (30,31). Therefore, there is an urgent demand for the identification of essential targets in bacterial pathogens that are suitable for the development of novel antibiotic substances to fight multiresistant isolates (32). As described above, c-di-AMP plays a central role in bacterial physiology. In several bacteria, c-di-AMP homeostasis seems to be also crucial for cell wall metabolism, cell division, and cell size control (17,(33)(34)(35). Moreover, some bacteria like the prominent human pathogens S. aureus, S. pneumoniae, and L. monocytogenes possess only the DAC CdaA, which is essential for their growth (36 -38). The observation that CdaA is essential and that c-di-AMP does not exist in humans makes the DAC an excellent target for novel antibiotics.
So far, only the structures of the DACs DisA and CdaS from Thermotoga maritima and Bacillus cereus, respectively, are available (9) (PDB code 2FB5 for the B. cereus enzyme) (39). By contrast, no structure of a DAC is available that belongs to the most abundant CdaA family. However, structural information about these DACs is crucial because they are highly conserved and essential in many pathogens (14). In this study, we report the structural and biochemical characterization of a truncated variant of the L. monocytogenes DAC CdaA. The structural characterization of CdaA might be helpful for in silico and in vitro screenings to identify compounds that specifically inhibit the essential enzyme. Unlike other DACs, CdaA activity seems to strictly depend on manganese or cobalt ions.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-The Escherichia coli strains DH5␣ (40) and BL21(DE3) (Stratagene) were used for cloning procedures and for protein overproduction, respectively. E. coli was grown in LB medium, and transformants were selected on LB plates containing ampicillin (100 g/ml) and 17 g/liter Bacto agar (Difco). The L. monocytogenes wild-type strain EGD-e (laboratory collection) was cultivated in BHI medium (Sigma-Aldrich).
DNA Manipulation-Transformation of E. coli was performed using standard procedures (40). Plasmid DNA was extracted using the Nucleospin extraction kit (Macherey and Nagel, Düren, Germany). Commercially available restriction enzymes, T4 DNA ligase, and DNA polymerases were used as recommended by the manufacturers. DNA fragments were purified using the PCR purification kit (Qiagen, Hilden, Germany). DNA sequences were determined by the dideoxy chain termination method (SeqLab, Göttingen, Germany). Chromosomal DNA of L. monocytogenes was isolated as described previously (41).
Analysis and Quantification of Cyclic Dinucleotide Pools-Intracellular c-di-AMP pools were determined by the liquid chromatography-coupled tandem mass spectrometry (LC-MS/ MS) method (18,44). Quantification of c-di-AMP was performed as described previously (18).
Protein Purification-E. coli BL21(DE3) was used as the host for the production of the truncated protein that was purified using the Streptactin:Strep-tag purification system. The cultures were grown in 1 liter of LB medium at 37°C. Gene expression was induced by the addition of isopropyl 1-thio-␤-Dgalactopyranoside (final concentration 1 mM) to logarithmically growing cultures (A 600 of 0.5). Cells were collected 3 h after induction, and the cell pellets were resuspended in 8 ml of disruption buffer (10 mM Tris/HCl, 200 mM NaCl, pH 7.5). The cells were disrupted by using a French press (18,000 p.s.i., 138,000 kilopascals; Spectronic Instruments, G. Heinemann, Schwäbisch Gmünd, Germany), and the crude extracts were passed over a 1-ml Streptactin column (IBA, Göttingen, Germany). The column was washed five times with 2.5 ml of disruption buffer, and the proteins were eluted from the column with 2.5 ml of elution buffer containing 2.5 mM desthiobiotin (Sigma-Aldrich). The Bio-Rad dye-binding assay was used to determine the protein concentrations. Bovine serum albumin was used as the standard.
Assay for Monitoring Diadenylate Cyclase Activity in Vitro-The in vitro DAC activity assays were carried out in buffer containing 10 mM Tris/HCl, pH 7.5, and 0.1% bovine serum albumin. Depending on the assay conditions, different amounts of ATP/MgCl 2 (equimolar), divalent metal ions, and purified DAC were mixed to a final volume of 400 l in 1.5-ml reaction tubes. The reaction mixtures were incubated for 4 h at 37°C with agitation. Reactions were stopped by heating the samples for 5 min at 95°C. Next, the samples were incubated for 5 min on ice and centrifuged for 10 min at 20,800 ϫ g and 4°C. 350 l of the supernatant were added to 1.4 ml of extraction solution (acetonitrile/methanol mixture (1:1, v/v)). The extraction mixture was stored at Ϫ20°C overnight. Samples were then centrifuged at 20,800 ϫ g and 4°C, and then the supernatants were transferred to fresh vials and dried in a SpeedVac. The dried supernatants were solved with 100 or 200 l of H 2 O (depending on the expected amount of c-di-AMP). After repeated centrifugation and the addition of the internal standard 13 C, 15 N-c-di-AMP, part of the supernatants was analyzed by LC-MS/MS.
In Vitro Cross-linking of the Diadenylate Cyclase-Crosslinking experiments were performed with glutardialdehyde (Sigma-Aldrich) to study the multimerization of the DAC. For this purpose, 50 pmol of the purified protein were incubated with 0 -0.2% glutardialdehyde for 1 h at room temperature. The cross-linking reactions were quenched by the addition of glycine (Sigma-Aldrich) to a final concentration of 200 mM. Subsequently, the samples were heated for 10 min at 95°C in Laemmli buffer. Finally, the proteins were separated by 12% SDS-PAGE and visualized by silver staining as described previously (45).

Size Exclusion Chromatography and Multiangle Light Scattering (SEC-MALS) Analysis of the Diadenylate Cyclase-
The oligomerization status of the DACs ⌬80CdaA and ⌬100CdaA was determined by SEC-MALS using a setup of an S75 Superdex 10/300GL on an Äkta purifier (both from GE Healthcare) subsequent to a degasser (model 2003 from Biotech AB, Onsala, Sweden) for the buffer (50 mM Tris/Cl, pH 7.5, 150 mM sodiumchloride) in line with a miniDawn Treos multiangle light scattering system followed by an Optilab T-rEX RI detector (both from Wyatt Technology Europe). Data analysis was performed using the ASTRA version 6.1 software (Wyatt Technology).
Crystallization and Data Collection-The CdaA DAC domain was subjected to initial crystallization trials at a concentration of 9.4 mg/ml with a 5-fold molar excess of ATP. Best diffracting crystals grew in 6 -16% 1,6-hexanediol, 2 mM spermine, 20 mM MgCl 2 , 0.1 M Na-PIPES, pH 7.5. The oscillation images were collected at PETRA III (Beamline P13, EMBL, Hamburg, Germany) and processed with XDS (46). The scaling process revealed a tetragonal lattice and unit cell parameters of a ϭ b ϭ 130.7 Å, c ϭ 178.2 Å, ␣ ϭ ␤ ϭ ␥ ϭ 90.00°. Systematic absences indicated the space group to be either P4 3 2 1 2 or P4 1 2 1 2. The Matthews coefficient (V m ϭ 4.5 Å 3 /Da) suggested four molecules occupying the asymmetric unit corresponding to a solvent content of 72%. The data collection statistics are summarized below (see Table 1).
Structure Determination and Refinement-Initial phases were obtained by molecular replacement method with PHASER using the structure of the DAC CdaS from B. cereus (PDB code 2FB5) as the search model. Template candidate has been identified based on an HHPRED (47,48) search and trimmed to the last common C␥ atoms using phenix.sculptor prior to the MR search. The model was manually rebuilt using Coot (49) and refined with Phenix (50) using standard parameters. A random set of 5% of reflections generated in thin shells was excluded from refinement to monitor R free (51,52). The final model has been refined at a resolution of 2.8 Å to R and R free factors of 24.0 and 25.6%, respectively. The model consists of four protein molecules per asymmetric unit encompassing residues 0 -4 of the N-terminal Strep-tag and residues 1-157 (155 for molecule D) of the DAC domain, three water molecules, one ATP molecule, and one Mg ion per protein molecule and one 1,6-hexanediol. The refinement statistics are summarized below (see Table 1). Surface complementarity coefficients and solvent-accessible surface areas were calculated with the SC program using a 1.7-Å radius probe (53). Possible hydrogen bonds, salt bridges, and van der Waals contacts were detected with HBPLUS and CONTACSYM (54) using default parameters. Surface potentials were calculated with PyMOL using the APBS plugin (55). The quality of the model was assessed using MOLROBITY (56) as implemented in Phenix. Secondary structure predictions were performed using DSSP (57,58). Coordinates were superimposed with LSQKAB (59) from the CCP4 program suite (53) or as implemented in PyMOL. Structurebased sequence alignment was performed using Espresso (60) and optimized for presentation using ESPRIPT (61). Structure coordinates and structure factors of the CdaA-DAC domain from L. monocytogenes have been deposited as PDB entry 4RV7.

RESULTS
Purification and Structure Determination of ⌬100CdaA-It was suggested previously that the DACs of the CdaA family are membrane-bound enzymes containing three N-terminally located ␣-helices that form a transmembrane (TM) domain ( Fig. 1) (14). The TM domain could hamper the purification of the enzyme. Indeed, previous attempts to express and purify the full-length B. subtilis and L. monocytogenes CdaA proteins failed. 4 Therefore, the truncated L. monocytogenes cdaA allele was cloned for heterologous production in E. coli and for in vitro characterization of the DAC. For this purpose, the ⌬300cdaA allele encoding the N-terminally truncated ⌬100CdaA protein was introduced into the E. coli expression vector pGP172 (see "Experimental Procedures"). The truncated DAC carries an N-terminal Strep-tag for affinity purification and lacks the predicted TM domain and the coiled-coil motif that separates the TM domain from the cyclase domain (Fig. 1). The truncated ⌬100CdaA protein was purified from E. coli, and the evaluation of the elution fractions by SDS-PAGE revealed that the protein can be purified in high amounts and with high purity (data not shown). Next, the protein was subjected to crystallization in the presence of ATP. The crystals obtained diffracted up to 2.8 Å resolution (Table 1, top). The structure was solved by means of molecular replacement using the model of the B. cereus DAC CdaS that was identified by performing an HHPred server search (PDB code 2FB5). The CdaA-DAC domain structure was refined at 2.8 Å and yielded a model of good quality with reasonable data statistics (see Table  1, bottom). ⌬100CdaA exhibits an overall globular fold with the long N-terminally located helix (␣1) flanking the core (Fig. 2, A  and B). A slightly twisted central ␤-sheet, made up of seven mixed-parallel and antiparallel ␤-strands (␤1-␤7), forms the core of the globular part. Both sides of the ␤-sheets are flanked by a total of five ␣-helices (␣1-␣5), resulting in the observed globular shape. In addition to the complete sequence of ⌬100CdaA, the structure contains 3-4 residues of the N-terminal Strep-tag, which are well defined in the electron density map and precede the N-terminally located ␣-helix.
Comparison of the DAC Domains of CdaA, DisA, and CdaS-Because c-di-AMP synthesis is thought to require the close proximity of two ATP molecules, the active sites of two DAC domains have to be face-to-face orientated. In order to make an educated guess about the structural arrangement of the dimer assembly, we compared the structure of ⌬100CdaA with the most prominent hits obtained by HHPred (Fig. 3, A-C). The first hit that was already used for molecular replacement is the sporulation-specific B. cereus DAC CdaS (PDB code 2FB5), which exists as a trimer in the asymmetric unit. Unfortunately, the analysis of the crystal packing and contacts to symmetryrelated neighboring molecules revealed that there is also no proper arrangement in a face-to-face manner allowing for c-di-AMP formation (PDB code 2FB5). A recently suggested model positions six molecules in a circular arrangement with dimers interacting in a face-to-face manner, connected via their hydrophobic tails (18). The second hit was the DAC DisA from T. maritima (PDB code 3C23, ATP analog-bound; 3C1Z, apoform, 3C1Y c-di-AMP-bound). Interestingly, DisA forms a homo-octamer, thereby arranging pairs of molecules in such a way that two active sites face each other. This arrangement is compatible with binding of two ATP molecules and subsequent c-di-AMP formation (PDB codes 3C21 and 3C23).
Interestingly, there are four ⌬100CdaA molecules present in the asymmetric unit but in an arrangement incompatible with the formation of c-di-AMP (PDB code 4RV7), which would require a face-to-face orientation of two active sites to be able to form the cyclic sugar phosphate backbone ring (9). The observed assembly within an asymmetric unit is probably due to the crystallization conditions and probably does not reflect the situation in aqueous solution (see below). However, superposition of either two ⌬100CdaA or CdaS DAC molecules with the DAC domains of two corresponding DisA molecules resulted in a dimer model of CdaA with a high degree of overlap (Fig. 3D).
The fully refined overall structure of the ⌬100CdaA DAC domain exhibited a prominent unoccupied electron density (Fig. 2C), which, based on its shape, could be interpreted as a molecule of ATP that was present in the crystallization condition. This ATP molecule is bound in a well defined cavity made up by helix ␣4 and strands ␤1 and ␤5 as well as the loops connecting ␣1 and ␤1, ␣3 and ␤3, ␤4 and ␣4, and ␤5 and ␤6 (Fig.  2B). The bottom of this pocket is formed by the hydrophobic patch of residues G(A/G)L 131 I of strand ␤1 conserved in all three known structures of CdaA, CdaS (PDB code 2FB5), and DisA (PDB code 3C21) (see Fig. 2D). The structure-based sequence alignment with the other two structures found by structure similarity also reveals the conservation of residues in two more distinct patches (Fig. 2D). One patch is made up by the loop connecting ␣3 and ␤3, which harbors the active site, a highly conserved D 171 GA motif followed by three hydrophobic residues (Fig. 2, C and D). The aspartic acid Asp-171 localizes in  a Values for the data in the highest resolution shell are shown in parentheses. b The CC1 ⁄ 2 is the correlation coefficient between two randomly selected half-data sets as described (71). c R free ϭ Α Test ʈF o ͉ Ϫ ͉F c ʈ/Α Test ͉F o ͉, where "Test" is a test set of about 5% of the total reflections randomly chosen and set aside prior to refinement for the complex.
close proximity to the ribose and the ␣-phosphate of the neighboring subunit in the CdaA dimer model (Fig. 3D), well suited for catalysis (Fig. 4A). This region positions the adenosine moiety in a conformation highly similar to the adenosine moieties of c-di-AMP in DisA (PDB code 3C21; Fig. 4B). In the dimer model of CdaA, the two ATPs would be oriented similarly to the ATP analogs in the DisA dimer (PDB code 3C23; Fig. 4A). The loop subsequent to ␤5 contains a short patch of 2-3 residues subsequent to the highly conserved Ser-222 (see Fig. 2C). This patch interacts with the ␤and ␥-phosphate and is involved in coordination of the magnesium ion. The other conserved patch, a GXR 203 HRXA motif, resides in helix ␣4 and harbors an arginine, Arg-203, which is involved in coordination of the ribose and the phosphates (see Fig. 2D). Within the long loop connecting the preceding strand ␤4 with helix ␣4, a leucine is located (Leu-188) whose carbonyl group interacts with the exo-cyclic N6 of the adenosine moiety (see Fig. 2, C and D). Taken together, the identical arrangement of the residues within the structures used in the structure-based sequence alignment seems to be required for binding of ATP and catalysis, which explains the high conservation of the residues and their function.
In Vivo Activities of Truncated Diadenylate Cyclases-As described above, four molecules are present in the asymmetric unit in an arrangement incompatible with c-di-AMP formation. As shown in Fig. 1, the ⌬100CdaA protein lacks the TM domain and the predicted coiled-coil (CC) motif that separates the TM and DAC domains from each other. The lack of either of the domains or both together might be the reason for the molecular arrangement of the ⌬100CdaA molecules. To evaluate whether the TM domain and the CC motif are important for enzyme catalysis, we assessed the in vivo activities of the truncated DACs ⌬80CdaA and ⌬100CdaA. Whereas the latter var- Overall structure of the truncated L. monocytogenes DAC ⌬100CdaA and conservation of structural motifs. A, the secondary structure of ⌬100CdaA consists of seven ␤-sheets and five ␣-helices. The protein is shown in schematic mode with the individual secondary structure motifs labeled and colored in rainbow coloring from the N to C terminus. B, the ⌬100CdaA protein is of globular shape composed of a twisted central ␤-sheet flanked by the five ␣-helices on both sides. The elements of the secondary structure are labeled and colored as in A, and the surface of the ⌬100CdaA monomer is depicted in gray. For better orientation, the ATP molecule is shown in ball and stick mode (carbon in yellow, phosphates in orange, oxygens in red, and nitrogens in blue). C, F o Ϫ F c omit map (green) of the ATP binding site, which is positioned in a shallow cavity on ⌬100CdaA (surface representation, gray). Important active site residues involved in coordination of ATP are labeled (carbon in pink, phosphates in orange, oxygens in red, and nitrogens in blue). The numbers of residues indicated in the truncated DAC variants correspond to the full-length CdaA protein (UniProt code Q8Y5E4). A Mg 2ϩ ion is depicted in green. D, sequence alignment of the DACs from L. monocytogenes (CdaA_Lmo), B. cereus (CdaS_Bce; PDB code 2FB5), B. subtilis (DisA_Bsu), and T. maritima (DisA_Tma; PDB code 3C23). The residues in the conserved DGA and next to the RHR motifs that were mutated in the ⌬80CdaA mutants, Asp-171, Gly-172, and Thr-202, are labeled with red arrows.
iant that was used for crystallization lacks the TM domain and the CC motif, the newly constructed ⌬80CdaA variant only lacks the TM domain (Fig. 1). Because E. coli does not produce c-di-AMP, this organism is well suited to analyze the production of the nucleotide in vivo (14,17,35). Therefore, both enzymes were expressed in E. coli (see "Experimental Procedures"). The amounts of c-di-AMP in in the bacteria harboring the plasmids pBP119 and pBP33 encoding the truncated DACs ⌬80CdaA and ⌬100CdaA, respectively, were compared with those produced by the same strain carrying the empty plasmid pGP172. As expected, no c-di-AMP was detectable in cells of the control strain that carries the empty vector (Fig. 5A). In contrast, 31,019 ng and 4,772 ng of c-di-AMP/mg of protein were detected in cells carrying the plasmids pBP33 (⌬100CdaA) and pBP119 (⌬80CdaA), respectively (Fig. 5A). ⌬100CdaA and ⌬80CdaA were expressed in similar amounts, as shown by Coomassie-stained SDS-polyacrylamide gels of 100/A 600 l from the cultures (Fig. 5B). Although the shorter DAC variant was 6.5-fold more active than the longer variant, the results clearly indicate that both truncated DACs form c-di-AMP in vivo. This suggests that the DACs must be transiently face-to-face orientated in vivo to allow the formation of c-di-AMP.
Oligomerization of the Truncated Diadenylate Cyclases-Although the precise mechanism of the cyclase reaction remains to be elucidated, structural and biochemical analyses of the DAC DisA from T. maritima revealed that two cyclase domains must come into close proximity with each other to allow the conversion of two molecules of ATP to c-di-AMP (9). To determine the oligomerization of the truncated L. monocytogenes DACs, we performed an in vitro cross-linking experiment. For this purpose, the purified proteins ⌬80CdaA and ⌬100CdaA were incubated with different amounts of the crosslinker glutardialdehyde, and the oligomerization was analyzed by SDS-PAGE. As shown in Fig. 6A, the untreated proteins exclusively exist as monomers. By contrast, ⌬80CdaA as well as ⌬100CdaA formed dimers when the proteins were treated with 0.1 or 0.2% glutardialdehyde. Thus, both enzymes are capable of forming dimers in the presence of a cross-linker that helps to stabilize oligomeric states of proteins under denaturating conditions (Fig. 6A).
We also performed SEC-MALS with ⌬80CdaA and ⌬100CdaA (Fig. 6B). The determined masses were 38.3 and 46.9 kDa for ⌬100CdaA and ⌬80CdaA, respectively. This almost perfectly matches the doubled mass of 42.5 and 47.2 kDa calculated for ⌬100CdaA and ⌬80CdaA, respectively, supporting the idea of a more or less stable dimer formation of both variants. To conclude, despite the lack of the TM domain in the DAC ⌬80CdaA or the TM domain and the CC motif in the ⌬100CdaA protein  (see Fig. 1), both enzymes are capable of forming dimers, which is the prerequisite for c-di-AMP formation. The fact that the buffer that was used for size exclusion chromatography was not supplemented with divalent ions suggests that the metal ion is required for catalysis and not for dimerization of the enzyme. Moreover, the observation shows that the assembly observed within an asymmetric unit is due to the crystallization conditions and does not exist in aqueous solution.
In Vitro Activity of the L. monocytogenes Diadenylate Cyclases-Next, we performed in vitro enzyme assays with the DAC ⌬80CdaA protein that was purified from E. coli (see "Experimental Procedures"). The truncated DAC showed no activity in a buffer containing 10 mM MgCl 2 (see Fig. 7A). The shorter ⌬100CdaA variant also did not produce c-di-AMP under these conditions (data not shown). It has been shown that the regulatory protein CdaR stimulates the CdaA homolog from B. subtilis (17). However, the fact that the DACs ⌬80CdaA and ⌬100CdaA produced significant amounts of c-di-AMP in E. coli (see Fig. 5A), although the organism does not contain a CdaR-like protein, suggests that the L. monocytogenes DAC does not require a protein as a cofactor for being active in vitro.
To identify the cofactor that enables the DAC ⌬80CdaA to produce c-di-AMP in vitro, we performed a metal ion screening with the divalent chloride salts of magnesium, cobalt, copper, nickel, zinc, manganese, and calcium. As shown in Fig. 7A, the ⌬80CdaA enzyme only produced c-di-AMP when either Mn 2ϩ or Co 2ϩ was present in the reaction mixture. The shorter DAC variant ⌬100CdaA required the same cofactors (data not shown). Unlike other DACs, such as DisA from M. tuberculosis and T. maritima, that synthesize c-di-AMP in the presence of Mg ions (9,13), the activity of the L. monocytogenes DAC unequivocally depends on Mn 2ϩ or Co 2ϩ .
c-di-AMP Production by ⌬80CdaA in the Presence of Co 2ϩ and Mn 2ϩ -To identify the best cofactor concentration for the truncated DAC ⌬80CdaA, we performed in vitro activity assays with different concentrations of the divalent chloride salts of  ⌬80CdaA and ⌬100CdaA, and the ⌬80CdaA mutants D171N, G172A, and T202N. A, detection of intracellular c-di-AMP in E. coli cells harboring the empty plasmid pGP172 (control) and the plasmids pBP33 and pBP119 that encode the truncated DACs ⌬80CdaA and ⌬100CdaA, respectively. The mutant variants ⌬80CdaA(D171N), ⌬80CdaA(G172A), and ⌬80CdaA(T202N) are encoded on plasmids pBP124, pBP125, and pBP126, respectively. The cells were grown in LB medium, and cell extracts for the determination of c-di-AMP by HPLC-MS/MS were prepared as  The elution volume and the molecular mass calculated from the scattered signal indicate dimer formation. The light scattered (LS) by the ⌬80CdaA and ⌬100CdaA proteins, which allows measurement of the molecular mass is shown in dark blue and red, respectively. The UV light detector signals for the ⌬80CdaA and ⌬100CdaA proteins are shown in light blue and orange, respectively. The short blue and red lines within the peaks indicate the estimated molecular masses for the ⌬80CdaA and ⌬100CdaA proteins, respectively. cobalt and manganese. As shown in Fig. 7B, when Mn 2ϩ ions were present in the reaction mixture, the ⌬80CdaA protein showed the highest DAC activity in the range between 0.75 and 20 mM. Below 0.75 mM and above 5 mM MnCl 2 , the activity of the enzyme decreased, and no c-di-AMP was produced below 0.5 mM and above 100 mM MnCl 2 . The in vitro activity assay that was performed with various amounts of CoCl 2 revealed that the ⌬80CdaA protein was active with low amounts of Co 2ϩ , and in the presence of this ion, the enzyme produced even more c-di-AMP than with MnCl 2 (Fig. 7B). With 20 mM CoCl 2 , the activity of the enzyme was strongly reduced, and no c-di-AMP was produced below 0.5 mM and above 20 mM CoCl 2 .
In conclusion, the truncated DAC ⌬80CdaA from L. monocytogenes preferentially uses Co 2ϩ to produce c-di-AMP from ATP.
Importance of Conserved Residues for the Cyclase Reaction-It has been suggested previously that the conserved basic and acidic side chains in the RHR and DGA motifs, respectively, are involved in the cyclase reaction (see Fig. 2D) (9). In order to test the importance of these residues for ATP binding and c-di-AMP formation, the residues Asp-171, Gly-172, and Thr-202 directly preceding the RHR motif were selected and replaced by Asn, Ala, and Asn, respectively, in the truncated DAC ⌬80CdaA. In the dimeric model, the active site asparagine, Asp-171, is involved in coordination of one ribose and phosphates of the neighboring ATP mediated by the metal ion. The additional methyl group of alanine replacing glycine at position 172 is expected to sterically interfere with binding of the adenosine and the sugar moiety of the ATP molecule (Figs. 2C and 4A). Moreover, the T202N mutation should disrupt the stabilization of the sugar ␣-phosphate through interaction between the asparagine and the ␣-phosphate. The three respective ⌬240cdaA mutant alleles were expressed in E. coli and tested for in vivo activity (see "Experimental Procedures"). All mutants expressed like the non-mutated DAC ⌬80CdaA (data not shown). In contrast to the non-mutated variant, the mutated enzymes were unable to form c-di-AMP in vivo (Fig.  5A). These results unequivocally indicate that the conserved residues are important for the cyclase reaction.

DISCUSSION
In the present study, we have structurally and biochemically analyzed truncated variants of the essential DAC of the human pathogen L. monocytogenes. Interestingly, the biochemical characterization of the DAC revealed that this enzyme requires either a divalent manganese or cobalt ion for in vitro activity. By contrast, other DACs like the DisA orthologs from T. maritima, B. subtilis, Bacillus thuringiensis, and M. tuberculosis were all shown to be active in vitro with magnesium ions (9,62,63). DisA from M. tuberculosis also produced c-di-AMP when manganese was provided as a cofactor (62,63). However, in none of the characterized DACs can magnesium be fully replaced by cobalt. Because L. monocytogenes is an invasive human pathogen that is able to penetrate and to survive in different cell types like macrophages (64), it has been suggested that competition between the pathogen and the host cell for divalent metal ions is important for survival (65). Therefore, mechanisms must have evolved in pathogenic bacteria, allowing them to survive in very specialized ecological niches like macrophages that are believed to be nutrient-restricted environments. For instance, one could imagine that an alteration of the cofactor requirement of the essential DAC CdaA by adaptive evolution might improve survival of L. monocytogenes inside the host cells. However, it remains to be elucidated whether the physiological role of cobalt is indeed important for pathogenesis of L. monocytogenes.
The structural analysis of the truncated L. monocytogenes DAC CdaA, lacking the first 100 amino acids, revealed a structural arrangement in the crystals that is incompatible with a functional dimer of CdaA (PDB code 4RV7). Recently, a similar observation was made when the sporulation-specific DAC CdaS from B. cereus was structurally analyzed (PDB code 2FB5). In the case of CdaS, the arrangement of the molecules in the crystals is also incompatible with a functional DAC. However, the data obtained by the enzyme assays, in vitro crosslinking experiments, and SEC-MALS analysis show that CdaA forms dimers and suggest that DAC is functional as a dimer in solution (Figs. 6 and 7). Therefore, the observed structural arrangements of the B. cereus DAC CdaS and the truncated L. monocytogenes DAC CdaA is most likely a consequence of FIGURE 7. In vitro assay to identify the cofactor that is required for the DAC activity of ⌬80CdaA. A, the ⌬80CdaA enzyme (1 M) was incubated with the divalent chloride salts of magnesium, cobalt, copper, nickel, zinc, manganese, and calcium (10 mM each). The reaction mixtures were incubated for 4 h at 37°C, and the produced c-di-AMP levels were determined as described under "Experimental Procedures." The data points shown are mean values that were calculated from duplicates; the deviation from the mean does not exceed 30%. B, the ⌬80CdaA enzyme requires different amounts of the crystallization conditions used. However, the structures of the CdaA DAC domain as well as of the deposited structure of CdaS are highly similar to that of the DAC domain of DisA from T. maritima (9). The superposition of the DAC domains demonstrates the proper arrangement of the important residues within the three identified highly conserved patches. These patches that are required for ATP coordination and synthesis of c-di-AMP are present in all three structures compared (Figs. 2 (C and D) and 4 (A and B)). Taken together, our observations strongly argue for a similar arrangement of the dimers in the truncated DAC CdaA in a face-to-face fashion and an identical mechanism in c-di-AMP formation.
As illustrated in Fig. 1, the full-length DAC CdaA contains an N-terminal TM domain. It has been proposed that this domain acts as a sensor for the perception of stimuli from the cell membrane or cell wall and that the altered c-di-AMP production in turn affects essential cellular processes that remain to be uncovered (14). For CdaA from B. subtilis, it has been shown that the DAC activity is stimulated through a direct proteinprotein interaction with CdaR (17), a protein that is also present in L. monocytogenes. Thus, alternatively to the model proposed above, CdaR might serve as the sensor that perceives an environmental stimulus and transduces the information to CdaA, which changes its activity accordingly. However, the precise physiological role of the regulatory interaction between CdaA and CdaR is rather unclear.
In contrast to other bacteria like B. subtilis that possesses three DACs (17,18), L. monocytogenes synthesizes only the membrane-bound DAC CdaA. It is not yet clear why CdaA in L. monocytogenes and other bacteria is localized at the membrane. Recently, it has been shown that growth of an L. monocytogenes conditional cdaA mutant that was depleted for CdaA could be rescued by expressing the DAC domain of DisA from B. subtilis (35). Similarly, a B. subtilis ⌬disA ⌬cdaA mutant strain was able to grow when cdaS, encoding the sporulationspecific DAC CdaS, was overexpressed (17). These observations suggest that c-di-AMP production by a cytosolic DAC is per se sufficient to ensure survival of the bacteria. However, the situation might be completely different if the bacteria grow in their natural habitats. Under these conditions, in particular for bacteria having multiple DACs, the enzymes might perceive different signals and generate local c-di-AMP pools that trigger different downstream targets. Recently, it has indeed been suggested that signaling specificity of second messengers can be achieved by generating local pools through the localization of enzymes that are involved in synthesis and in degradation of the signaling molecule (16, 66).
As described above, the emergence of multiresistant human pathogens is a serious threat for humankind because some isolates are untreatable with the antibiotics that are currently used. Among the DACs, CdaA is the most abundant enzyme that is conserved and essential in several human pathogenic bacteria like L. monocytogenes and multiresistant S. aureus isolates (14, 36 -38). Moreover, c-di-AMP proves to be a second messenger that has multiple cellular targets like riboswitches, signal transduction proteins, and metabolic enzymes (21,22,24). In addition to this, perturbation of c-di-AMP homeostasis was shown to severely affect cell wall metabolism in Gram-positive bacte-ria (17,33) and to influence the susceptibility of bacteria to cell wall-targeting antibiotics (34,67). Therefore, the DAC CdaA might be a promising target for the development of novel antimicrobial substances to fight multiresistant pathogenic bacteria. Since the discovery of c-di-AMP, several methods for monitoring its production in small scale have been developed (68,69). These methods seem to be suitable for high throughput screenings to identify substances that inhibit bacterial DACs (70).