The identification of the essential bacterial second messenger cyclic-di-AMP (c-di-AMP) synthesized by the DNA-integrity scanning protein A (DisA) has opened up a new and emerging field in bacterial signalling. To further analyse the diadenylate cyclase (DAC) reaction catalysed by the DAC domains of DisA, we crystallized Thermotoga maritima DisA in the presence of different ATP analogues and metal ions to identify the metal-binding site and trap the enzyme in pre- and post-reaction states. Through structural and biochemical assays we identified important residues essential for the reaction in the active site of the DAC domains. Our structures resolve the metal-binding site and thus explain the activation of ATP for the DAC reaction. Moreover, we were able to identify a potent inhibitor of the DAC domain. Based on the available structures and homology to annotated DAC domains we propose a common mechanism for c-di-AMP synthesis by DAC domains in c-di-AMP-producing species and a possible approach for its effective inhibition.

INTRODUCTION

Nucleotide second messengers are wide spread throughout all domains of life. Bacteria in particular are known to use cyclic AMP, cyclic GMP, cyclic-di-GMP and (p)ppGpp in order to regulate the most varied signalling pathways. In 2008, however, an additional secondary messenger was found: c-di-AMP (cyclic di-AMP). C-di-AMP was initially identified in the crystal structure of the Thermotoga maritima sporulation checkpoint protein DNA-integrity scanning protein A (DisA) [1] and was subsequently shown to be produced by a variety of enzymes containing a diadenylate cyclase (DAC) domain. DAC domain proteins are mainly found in Gram-positive bacteria belonging to the phyla Firmicutes and Actinobacteria, including pathogenic bacteria such as Listeria monocytogenes or Staphylococcus aureus. The DAC prototype protein DisA forms a homo-octameric complex, composed of two head-to-head tetrameric rings. The catalytic active sites are located at the interface between the two tetramers, whereby two opposing monomers form one reaction centre. These DAC domains are connected to the C-terminal DNA-binding HhH motifs by a long spine consisting of three antiparallel α-helices.

So far, little is known about the mode, function and regulation of bacterial c-di-AMP signalling. DisA was found to bind to branched DNA, such as replication intermediates or stalled replication forks and its enzymatic activity is thereupon inhibited, suggesting a role in DNA-damage signalling [1,2]. In agreement with these findings, it was shown that reduced c-di-AMP levels cause a delay in sporulation in Bacillus subtilis, whereas an elevated c-di-AMP concentration promotes sporulation [3]. Additionally, in B. subtilis CdaS (c-di-AMP synthase S) was identified as a DAC domain protein expressed only during sporulation. The third DAC protein in B. subtilis, CdaA c-di-AMP synthase A, is activated through interaction with CdaR (c-di-AMP synthase A regulator) and is presumably involved in control of cell wall biosynthesis [4]. Additional studies showed that c-di-AMP is involved in the regulation of cell wall characteristics and cell size in S. aureus and B. subtilis [5,6]. Although the DAC domain is widespread, the pathways regulated by c-di-AMP are so far not well understood. A genome-wide screen identified several proteins in S. aureus (KtrA, CpaA, KdpD and PstA) that act as c-di-AMP receptors, thereby linking c-di-AMP to potassium transport [7,8]. Very recently, the mode of c-di-AMP recognition by PstA/DarA, which is structurally related to the class of PII-like proteins, was elucidated [912], suggesting that additional cellular processes are regulated by c-di-AMP. Moreover, it was shown that c-di-AMP is a high-affinity ligand for the ydaO riboswitch, explaining why c-di-AMP affects such a wide range of processes and linking c-di-AMP to the regulation of gene expression involved in peptidoglycan synthesis, germination and osmotic shock response in bacteria [13,14]. Although the details of c-di-AMP signalling are poorly understood, it is nevertheless obvious that the cellular levels of c-di-AMP need to be tightly regulated. Total knockouts of DAC domain proteins in B. subtilis, L. monocytogenes and Streptococcus pneumoniae are lethal [6,15,16]. Similarly, an increase or decrease in c-di-AMP concentrations by overexpression or knockouts of DAC domain proteins or c-di-AMP-degrading phosphodiesterases respectively, severely impairs bacterial growth and affects the virulence of pathogenic bacteria [4,15,16]. More specifically, it was shown that a decrease in the cellular c-di-AMP concentration renders bacteria more sensitive to β-lactam antibiotics [6]. Based on this, DAC domain proteins might prove to be promising targets for antibiotic therapies.

In the present study, we have structurally analysed the reaction mechanism of T. maritima DisA for the synthesis of c-di-AMP. We crystallized DisA in presence of different substrate analogues and metal ions and are thus able to show the pre- and post-catalytic states of the reaction cycle. We were able to pinpoint the essential divalent cation-binding site and identified catalytically important residues. Using in vitro activity assays we analysed the impact of mutations in the active site. We were additionally able to identify a potent commercially available DAC inhibitor. A structure- and sequence-based comparison of the active sites of known DAC domains suggests that our findings can probably be directly transferred to other c-di-AMP-producing enzymes and help to understand their activity and regulation.

EXPERIMENTAL

Cloning, expression and protein purification

His-tagged T. maritima DisA was cloned and recombinantly expressed as reported previously [1]. Primers used for protein mutant generation are included in Supplementary Table S1. For protein purification, cells were lysed by sonication in buffer A (50 mM Tris/HCl, 300 mM NaCl and 10 mM imidazole, pH 8.0). After centrifugation, the supernatant was applied to an Ni-NTA (Ni2+-nitrilotriacetic acid) column (Qiagen) and extensively washed with buffer A and buffer W (50 mM Tris/HCl, 1 M NaCl and 20 mM imidazole, pH 8.0). Protein was eluted using buffer B (50 mM Tris/HCl, 300 mM NaCl and 250 mM imidazole, pH 8.0) and DisA-containing fractions were pooled and dialysed against SEC buffer (20 mM Tris/HCl and 200 mM NaCl, pH 8.0). The protein was applied to a Superdex 200 preparative-grade size-exclusion column (GE Healthcare) and the fractions were analysed by SDS/PAGE. Fractions containing only DisA were pooled and concentrated. The protein was flash-frozen in liquid nitrogen and stored at −80°C until usage.

Crystallization

One microlitre of DisA (8 mg·ml−1) in SEC buffer with 2 mM of nucleotide and 20 mM MgCl2 was mixed with 1 μl of reservoir solution [30–32.5% (v/v) MPD (2-methyl-2,4-pentanediol), 200 mM ammonium acetate and 100 mM Tris/HCl, pH 8.0–8.3). Crystals were grown within 7 days at 15°C through hanging-drop vapour diffusion. Reservoir solution containing 35% (v/v) MPD was used as cryoprotectant prior to flash-cooling crystals in liquid nitrogen. For crystals containing manganese, 0.2 M MnCl2 was added to the cryo-solution and crystals were soaked for a few seconds.

Crystallographic data collection, processing and refinement

Diffraction data were collected at the beamlines SLS X06SA (Paul-Scherrer-Institute, Villigen, Switzerland) and PETRA-3 P14 (EMBL c/o DESY Hamburg, Germany) at 100 K. Diffraction data were processed using XDS and XSCALE [17]. Molecular replacement was performed using the apo DisA structure (PDB code 3c1z) as a search model in PHASER [18] within the CCP4 suite [19]. Structure refinement comprised automatic refinement steps using PHENIX [20] and manual building in COOT [21]. All structures show typical statistics for the resolution (Table 1). The coordinates and structure factors have been deposited in the PDB with the accession codes 4yvz, 4yxj and 4yxm. Figures of crystal structures were generated using PyMOL [22], the tunnels shown in Figure 3(A) were computed and displayed with CAVER Analyst [23].

Table 1
Crystal parameters, data collection and refinement statistics
 TmaDisA 3′-dATP/Mn2+ TmaDisA ApCpp TmaDisA D75N c-di-AMP 
Crystal data    
 Space group P421P421P421
 Molecules per ASU 
 Unit cell parameters    
  a,b,c (Å) 107.49, 107.49, 168.79 108.25, 108.25, 166.40 108.55, 108.55, 165.92 
  α,β,γ (°) 90, 90, 90 90, 90, 90 90, 90, 90 
Data collection statistics    
 Diffraction source PETRA3-P14 SLS X06SA SLS X06SA 
 Wavelength (Å) 1.23953 1.00149 1.00000 
Data processing statistics    
 Resolution range (Å) 168.8–2.50 (2.56–2.50) 50–2.55 (2.61–2.55) 50–2.25 (2.31–2.25) 
 Number of observed reflections 497086 (33161) 186053 (6649) 625199 (45367) 
 Number of unique reflections 65492 (4613) 32830 (2248) 47778 (3483) 
 Completeness (%) 99.6 (94.5) 99.2 (93.9) 100 (100) 
 Multiplicity 7.6 (7.2) 5.7 (3.0) 13.1 (13.0) 
 Mean II 15.9 (2.8) 15.5 (2.1) 18.3 (1.6) 
Rmeas (%) 11.6 (113.3) 14.7 (73.7) 11.9 (217.2) 
Refinement    
 Resolution (Å) 168.8-2.50 48.41-2.55 49.28-2.25 
 Number of used reflections 65492 32808 47724 
Rwork (%)/Rfree1 (%) 16.42/21.89 18.99/24.33 18.86/23.73 
 Number of non-hydrogen atoms (total) 5775 5840 5797 
 Protein 5600 5592 5602 
 Ligand 60 (3′-dATP)/4 (Mn2+62 44 (c-di-AMP)/8 (MPD) 
 Water 111 186 143 
 Average B factors (Å2   
  Wilson B factor 53.7 46.0 54.8 
  Overall 51.5 47.3 60.3 
  Protein 52.8 46.2 55.9 
 Ligands/water 3′-dATP 34.9 ApCpp 50.0 c-di-AMP 48.7 
 Mn2+ (active site) 44.1 Waters 46.7 MPD 85.9 
 Mn2+ (surface) 97.6  Waters 55.2 
 Waters 46.3   
 RMSDs    
 Bond lengths (Å)/angles (°) 0.008/1.103 0.009/1.125 0.007/1.046 
 Ramachandran plot analysis    
  Favoured (%) 97.8 97.0 98.0 
  Allowed (%) 2.2 3.0 2.0 
  Disallowed (%) 
 PDB identifier 4yvz 4yxj 4yxm 
 TmaDisA 3′-dATP/Mn2+ TmaDisA ApCpp TmaDisA D75N c-di-AMP 
Crystal data    
 Space group P421P421P421
 Molecules per ASU 
 Unit cell parameters    
  a,b,c (Å) 107.49, 107.49, 168.79 108.25, 108.25, 166.40 108.55, 108.55, 165.92 
  α,β,γ (°) 90, 90, 90 90, 90, 90 90, 90, 90 
Data collection statistics    
 Diffraction source PETRA3-P14 SLS X06SA SLS X06SA 
 Wavelength (Å) 1.23953 1.00149 1.00000 
Data processing statistics    
 Resolution range (Å) 168.8–2.50 (2.56–2.50) 50–2.55 (2.61–2.55) 50–2.25 (2.31–2.25) 
 Number of observed reflections 497086 (33161) 186053 (6649) 625199 (45367) 
 Number of unique reflections 65492 (4613) 32830 (2248) 47778 (3483) 
 Completeness (%) 99.6 (94.5) 99.2 (93.9) 100 (100) 
 Multiplicity 7.6 (7.2) 5.7 (3.0) 13.1 (13.0) 
 Mean II 15.9 (2.8) 15.5 (2.1) 18.3 (1.6) 
Rmeas (%) 11.6 (113.3) 14.7 (73.7) 11.9 (217.2) 
Refinement    
 Resolution (Å) 168.8-2.50 48.41-2.55 49.28-2.25 
 Number of used reflections 65492 32808 47724 
Rwork (%)/Rfree1 (%) 16.42/21.89 18.99/24.33 18.86/23.73 
 Number of non-hydrogen atoms (total) 5775 5840 5797 
 Protein 5600 5592 5602 
 Ligand 60 (3′-dATP)/4 (Mn2+62 44 (c-di-AMP)/8 (MPD) 
 Water 111 186 143 
 Average B factors (Å2   
  Wilson B factor 53.7 46.0 54.8 
  Overall 51.5 47.3 60.3 
  Protein 52.8 46.2 55.9 
 Ligands/water 3′-dATP 34.9 ApCpp 50.0 c-di-AMP 48.7 
 Mn2+ (active site) 44.1 Waters 46.7 MPD 85.9 
 Mn2+ (surface) 97.6  Waters 55.2 
 Waters 46.3   
 RMSDs    
 Bond lengths (Å)/angles (°) 0.008/1.103 0.009/1.125 0.007/1.046 
 Ramachandran plot analysis    
  Favoured (%) 97.8 97.0 98.0 
  Allowed (%) 2.2 3.0 2.0 
  Disallowed (%) 
 PDB identifier 4yvz 4yxj 4yxm 

Values in parentheses are for the highest resolution shell

*Rfree calculations 5% of the total number of reflections were used.

SAXS

Small-angle X-ray scattering (SAXS) data were collected at EMBL/DESY Hamburg beamline X33. Protein samples after size-exclusion chromatography and centrifugation were measured at different concentrations between 1 and 5 mg·ml−1. Before and after each sample the corresponding buffer was measured and used for buffer correction. No sample showed signs of aggregation or radiation damage, and scattering data were processed and analysed using programs of the ATSAS package [24] as described in [25]. Theoretical scattering curves of crystal structures were calculated using CRYSOL [26].

Size-exclusion chromatography-coupled static light scattering

Size-exclusion chromatography-coupled static light scattering [SEC-RALS] was performed using an ÄKTAmicro system (GE Healthcare Life Sciences) equipped with a RI-device and RALS detector (Viscotek/Malvern Instruments) with a Superdex S200 10/300 size-exclusion column (GE Healthcare). BSA (66 kDa) was used as standard protein for calibration. Analysis of data was performed using the Viscotek Software OmniSEC.

DAC activity assays

Photometric DAC assays were performed similarly to the description in [27] in an optimized reaction buffer [50 mM glycine/NaOH, pH 9.5, 50 mM NaCl, 10 mM MgCl2, 0.3 mM ATP and 0.1 m-unit of yeast pyrophosphatase (Thermo Scientific)] in a total volume of 50 μl. The ATP concentration was optimized in order to perform all assays under substrate-saturating conditions. The inhibitor 3′-dATP (Jena Biosciences) was used at the concentrations indicated (0–150 μM). The reactions were started by addition of 24 nM T. maritima DisA (monomer concentration) and incubated at 60°C for 30 min. Reactions were stopped by addition of 100 μl of BIOMOL Green reagent (Enzo) and absorbance at 620 nm was measured after 15 min in a platereader (Tecan M1000 Pro). The absorbance of a control without DisA was subtracted in order to calculate normalized activities. At least three independent experiments of all activity assays were performed to calculate S.D.s. The IC50 was determined using Prism (GraphPad Software).

RESULTS AND DISCUSSION

The DAC domain

The DAC domain is so far the only fold that has been show to specifically catalyse the reaction of two ATP molecules to c-di-AMP and both, its active site residues and fold, are highly conserved. Figure 1(A) shows a sequence comparison of the prototype T. maritima DisA (TmaDisA) N-terminus and the three encoded DACs of B. subtilis (DisA, CdaA and CdaS), in addition to DAC domains from various other species described in the literature. In good agreement with data from [1,28], highly conserved residues of two opposing subunits, such as the DGA (residues 75–77, numbering referring to Tma DisA) and RHR motifs (residues 108–110), form the reaction centre.

DAC domain conservation and overall characteristics of DisA

Figure 1
DAC domain conservation and overall characteristics of DisA

(A) Sequence alignment of DAC domains of selected DAC proteins from various organisms. Residues in TmaDisA that interact with substrate/product are marked by asterisks. (B) TmaDisA D75N is an octamer in solution as determined by SEC-RALS (MwSEC-RALS D75N=339 kDa) (C) SAXS of TmaDisA D75N (blue dots show the measured scattering data, black curve represents the theoretical scattering curve of an octameric DisA calculated with CRYSOL [26]).

Figure 1
DAC domain conservation and overall characteristics of DisA

(A) Sequence alignment of DAC domains of selected DAC proteins from various organisms. Residues in TmaDisA that interact with substrate/product are marked by asterisks. (B) TmaDisA D75N is an octamer in solution as determined by SEC-RALS (MwSEC-RALS D75N=339 kDa) (C) SAXS of TmaDisA D75N (blue dots show the measured scattering data, black curve represents the theoretical scattering curve of an octameric DisA calculated with CRYSOL [26]).

Previous in vivo experiments by Oppenheimer-Shaanan and colleagues [2] showed that GFP-labelled DisA D75N, demonstrated to be inactive in DAC assays, abolished foci formation in B. subtilis as observed previously for the wild-type. This suggests that DisA D75N might not form the octamers essential for its activity [3]. We crystallized the TmaDisA mutant D75N and both the overall structure of DisA D75N and the crystal packing are virtually identical with the wild-type DisA protein. To analyse the oligomeric state of DisA D75N in solution we performed SEC-RALS (Figure 1B) and SAXS experiments (Figure 1C). DisA D75N was eluted in a single peak with a molecular mass determined by SEC-RALS of MwrSEC-RALS D75N=339 kDa, showing that the mutant protein still forms an octamer under the conditions used (Mwtheoretical octamer=336 kDa). The molecular parameters determined by SAXS (MwSAXS=360 kDa, Rg=5.5±0.1 nm) and the shape of the scattering curve are only compatible with a homogenous octameric species (Rgtheoretical, unhydrated=5.34 nm). Thus, DisA D75N is still octameric in solution and the inactivity is instead due to changes in the active site.

Pre-reaction states

We solved the structure of DisA in complex with a non-reactive ATP analogue (3′-dATP, cordycepin triphosphate) and MnCl2 in order to trap the enzyme in the pre-reaction state and to identify the metal-binding site. Manganese was chosen as a substitute for magnesium to allow us to identify the metal ion and distinguish it from well coordinated water molecules by its anomalous signal. The ion is octahedrally coordinated with all six coordination positions occupied by oxygen ligands (Figure 2A). The three phosphate groups from 3′-dATP together with Asp75 from the adjacent subunit and two water molecules are all located between 2.0 and 2.5 Å (1 Å=0.1 nm) from the manganese, in good agreement with its ideal coordination distance. The 3′-dATP phosphate groups are bent around the metal ion, with the β- and γ-phosphate being additionally held in place via hydrogen bonds with Arg108 and His109 from the highly conserved RHR motif and Arg130. Thr107 interacts with the α-phosphate and additionally contacts N7 of the adenine (also contacted by Leu94) and Thr111 forms a hydrogen bond with N6. The ribose is contacted at O4′ by Arg108 and at 2′-OH by backbone interactions with Gly76. The γ-phosphate is mainly polarized by Ser127, Arg128 and Arg130 facilitating the nucleophilic attack on the α-phosphate (Supplementary Table S2; Supplementary Figure S1).

DisA active site

Figure 2
DisA active site

(A) Pre-reaction state of the cyclase reaction with 3′-dATP and a Mn2+-ion in the active site. The Mn2+ ion is shown with its anomalous difference density at 4σ, the two facing DAC domains are shown in light and dark blue. Some residues have been omitted for clarity. (B) In vitro diadenylate cyclase assays with normalized activities of selected active-site mutations. Error bars represent S.D. for n=3 independent experiments. Mutations of amino acids interacting with the substrate or the metal ion lead to a strong decrease in activity, whereas easier accessibility of the active site results in higher activity R(128–130)A. (C) Superposition of the DisA/3′-dATP (blue, 3′-dATP shown as lines) and the DisA–ApCpp structure (grey, ApCpp shown as sticks). The Mn2+ ion has been omitted for clarity. Note that the ApCpp structure does not contain a divalent metal ion and thus the triphosphates are in elongated conformation. (D) Close-up of the active site of the TmaDisA D75N mutant crystallized in the presence of ATP/Mg2+. Even though the D75N mutant is inactive in in vitro assays, c-di-AMP is present in the active site (c-di-AMP with annealed composite omit map at 1σ)

Figure 2
DisA active site

(A) Pre-reaction state of the cyclase reaction with 3′-dATP and a Mn2+-ion in the active site. The Mn2+ ion is shown with its anomalous difference density at 4σ, the two facing DAC domains are shown in light and dark blue. Some residues have been omitted for clarity. (B) In vitro diadenylate cyclase assays with normalized activities of selected active-site mutations. Error bars represent S.D. for n=3 independent experiments. Mutations of amino acids interacting with the substrate or the metal ion lead to a strong decrease in activity, whereas easier accessibility of the active site results in higher activity R(128–130)A. (C) Superposition of the DisA/3′-dATP (blue, 3′-dATP shown as lines) and the DisA–ApCpp structure (grey, ApCpp shown as sticks). The Mn2+ ion has been omitted for clarity. Note that the ApCpp structure does not contain a divalent metal ion and thus the triphosphates are in elongated conformation. (D) Close-up of the active site of the TmaDisA D75N mutant crystallized in the presence of ATP/Mg2+. Even though the D75N mutant is inactive in in vitro assays, c-di-AMP is present in the active site (c-di-AMP with annealed composite omit map at 1σ)

In summary, the pre-reaction state shows a highly coordinated arrangement of the two ATP (-analogues) in almost optimal distance for the nucleophilic attack of the 3′-OH on the α-phosphate of the neighbouring ATP. The α-phosphate is well coordinated and additionally stabilized by the positive dipole of helix α5. In good agreement with these observations, point mutations of selected active-site residues (Asp75, RHR motif 108–110, Arg130, Thr107, Thr111) led to significantly decreased DAC activity of DisA in vitro (Figure 2B; Supplementary Figure S2), proving the biological relevance of the structural features observed.

As 3′-dATP lacks the attacking hydroxy group, we crystallized DisA in the presence of the non-hydrolysable ATP analogue ApCpp {adenosine-5′-[(α,β)-methyleno]triphosphate}. The structure of the DisA–ApCpp complex is highly similar to the structure with 3′-dATP (overall RMSD 1.2 Å) and only the phosphate groups adopt a slightly different orientation (Figure 2C), presumably because the ApCpp complex structure lacks the metal ion coordinated by the phosphates. The distance between the 3′-hydroxy group and the α-phosphate of 4.6–4.9 Å is consequently relatively large. The α-phosphate, however, probably moves closer towards the 3′-OH when interacting with a metal ion (as observed in the 3′-dATP structure), facilitating the nucleophilic attack, as implied by the structure with 3′-dATP and MnCl2.

Interestingly, our attempts to crystallize DisA D75N, shown to be fully inactive in DAC assays, with the native substrate ATP/Mg2+ unexpectedly produced crystals with unambiguous density for c-di-AMP bound in the active site (Figure 2D). Obviously, the much longer time scale of crystallization (compared with in vitro assays) allows the reaction to occur to a significant extent. Probably the correct orientation of the ATP nucleotides still takes place due to the high number of stabilizing interactions within the DAC site, even though the main interaction partner for ion coordination (carboxy group of Asp75) is missing. Apparently, even imperfect binding of the nucleotides and their respective orientation is sufficient for the reaction to take place on the long time scale of crystallization.

Post-reaction state

The first structural report of DisA described the product state in which c-di-AMP is bound, even though no nucleotide was added to the crystallization condition [1]. This indicates that the reaction product is tightly bound in the active site with very slow off-rate kinetics, even though c-di-AMP is less well coordinated than ATP, since most interactions in the reaction centre occur with the three phosphate groups rather than the adenine or sugar moieties (Figure 2D; Supplementary Table S3). In comparison, the product state only shows coordination of c-di-AMP by the DGA motif (residues 75–77), Thr107 and Thr111. We analysed the accessible surface and cavities of DisA in order to identify the possible substrate and product release paths (Figure 3A). A likely reason for the slow release of c-di-AMP from the active site is the size of the tunnel that leads from the reaction centre to the surface. This has a bottleneck diameter of approximately 7.8 Å (Supplementary Figure S3), which is just large enough to allow c-di-AMP to pass through. To exit the DisA molecule, c-di-AMP has to first move to the centre of DisA and then to the side, passing the loop connecting β5 and β6 (Arg128–130) of the DAC domain and finally to the surface, covering a total distance of approximately 40 Å. In Figure 3(A), one of the possible paths c-di-AMP needs to follow is indicated. Since DisA is a symmetrical octamer, c-di-AMP from any active site can of course exit by any of the eight tunnels. Due to the fact that we only observe snapshots in our crystal structures, we cannot exclude a certain degree of ‘breathing’ of the octameric assembly that might lead to other exit pathways. To test for flexibility of DisA we calculated a simulation of protein structure fluctuation [29] of one monomer and found only small predicted movements in loops on the surface, showing that DisA is very rigid and thereby indicating that the tunnels observed in the static crystal structure are probably similar in solution (Supplementary Figure S4). However, upon mutating the three arginine residues 128–130 in the loop that c-di-AMP needs to pass to exit the molecule to less bulky amino acids R(128–130)A, the activity of DisA increases approximately 2-fold, even though Arg130 is a major factor in coordination of the γ-phosphate of ATP and the single mutation (R130A) is almost completely inactive (Figure 2B). This indicates that the rate-limiting step of c-di-AMP synthesis by DisA is not defined by the reaction itself, but rather by the accessibility of the active site.

Reaction states of DisA

Figure 3
Reaction states of DisA

(A) DAC domains in the octameric assembly with tunnels shown in blue (calculated with CAVER [23]), the RRR loop (residues 128–130) shown in black and c-di-AMP in red. Orange arrows show one possible path c-di-AMP has to follow in order to exit the active site. (B) Superposition of apo- (black), pre- (3′-dATP blue, ApCpp grey) and post-reaction state (green) of TmaDisA monomers (left panel) and their per-residue RMSD compared with the apo-structure (right panel). Whereas almost all residues have very low RMSDs with respect to the apo-structure, a short loop region on the surface (indicated by an asterisk) has slightly higher deviations, probably because of flexibility. (C) Superposition of L. monocytogenes CdaA (green) and TmaDisA (blue) DAC domains bound to ATP/Mg2+ and 3′-dATP/Mn2+ respectively (RMSD 1.25 Å)

Figure 3
Reaction states of DisA

(A) DAC domains in the octameric assembly with tunnels shown in blue (calculated with CAVER [23]), the RRR loop (residues 128–130) shown in black and c-di-AMP in red. Orange arrows show one possible path c-di-AMP has to follow in order to exit the active site. (B) Superposition of apo- (black), pre- (3′-dATP blue, ApCpp grey) and post-reaction state (green) of TmaDisA monomers (left panel) and their per-residue RMSD compared with the apo-structure (right panel). Whereas almost all residues have very low RMSDs with respect to the apo-structure, a short loop region on the surface (indicated by an asterisk) has slightly higher deviations, probably because of flexibility. (C) Superposition of L. monocytogenes CdaA (green) and TmaDisA (blue) DAC domains bound to ATP/Mg2+ and 3′-dATP/Mn2+ respectively (RMSD 1.25 Å)

Structural comparison of the DAC reaction states and DAC domains

A comparison of apo-, pre- and post-reaction states in a superposition shows no major structural changes of DisA (Figure 3B). Based on the structures, no small-scale movements of domains or loops can be observed, supporting the idea that binding and coordination of ATP/Mg2+ seems to be sufficient for the DAC reaction to take place. To test our hypothesis that DAC domains share this reaction site, we superimposed the DAC domains of DisA (3′-dATP complex) and the recently reported structure of L. monocytogenes CdaA in complex with ATP [28] (Figure 3C). Both DAC domains show virtually identical arrangement of motifs and also the nucleotides superimpose very well (RMSD 1.25 Å) as previously observed for a DAC domain from Bacillus cereus [1]. The CdaA construct used in [28] obviously does not form active dimers in DAC-to-DAC orientation and thus crystallized in presence of the native substrate ATP/Mg2+, thereby supporting our pre-reaction-state structure containing 3′-dATP (see above).

Inhibition of DisA

C-di-AMP synthesis is essential for bacteria, as shown through numerous failed attempts to knockout all DAC domain proteins in different species. Similarly, a decrease in the cellular c-di-AMP concentration renders bacteria more sensitive towards β-lactam antibiotics [6]. These findings suggest that DAC domain proteins may be promising targets for antibiotic therapy. Cordycepin (3′-deoxy adenosine) is a natural adenosine analogue produced by the fungus Cordyceps, which has been identified to offer a large variety of medically beneficial effects, such as anti-tumour, anti-inflammatory and anti-bacterial effects [30]. Inside the cell, cordycepin is phosphorylated to 5′-mono-, di- or tri-phosphate and subsequently interferes with different essential pathways [31]. Since cordycepin triphosphate (3′-dATP) is a non-reactive substrate analogue for DisA that traps the enzyme in the pre-reaction state (see above), we investigated its effect on the in vitro DAC activity. We found commercially available 3′-dATP to be an effective inhibitor of DisA with an inhibitory constant (IC50) of 3 μM when 26 nM DisA and 300 μM ATP were used (Figure 4A). Interestingly, it has already been shown that cordycepin isolated from Cordyceps fungi is able to inhibit growth of Clostridium species with similar efficiency as conventional antibiotics such as tetracycline and chloramphenicol [32]. This anti-bacterial effect is probably due to cordycepin affecting various essential cellular pathways as a nucleoside analogue, probably also including those that require DAC activity. 3′-dATP is the second DisA inhibitor to be identified following bromophenol thiohydantoin (TH) [33]. In the bromophenol TH activity assays, significantly different concentrations of DisA and ATP were used for determination of the inhibitory constant, thus the inhibitory effectiveness of 3′-dATP and bromophenol TH cannot be directly compared between our study and [33]. The two inhibitors are furthermore likely to have different modes of action. The crystal structure shows that, whereas 3′-dATP is a competitive inhibitor that binds in the same position as ATP (Figure 2A), bromophenol TH seems to be an allosteric inhibitor that binds close to a tryptophan residue. The exact binding site for bromophenol-TH remains unknown and it might therefore be specific for DisA. In contrast, the DAC domains from different proteins are highly conserved in their active-site residues and we thus postulate that 3′-dATP has the potential to inhibit not only DisA, but also other DAC domain proteins.

DisA inhibition and regulation

Figure 4
DisA inhibition and regulation

(A) Inhibition of c-di-AMP synthesis by the non-reactive nucleotide 3′-dATP under substrate-saturating conditions (300 μM ATP, 26 nM DisA) shows an IC50 of 3 μM (R2=0.997). Error bars represent S.D. for n=4 experiments. (B) TmaDisA F57R mutant shows salt-dependent dissociation. The mutant (blue, no salt buffer; green, 10 mM NaCl buffer) elutes at higher volumes from a Superose 6 PC 3.2/30 column compared with wild-type TmaDisA (black curve) and thus is destabilized by salt, suggesting a dissociation from, e.g., tetramers to monomers.

Figure 4
DisA inhibition and regulation

(A) Inhibition of c-di-AMP synthesis by the non-reactive nucleotide 3′-dATP under substrate-saturating conditions (300 μM ATP, 26 nM DisA) shows an IC50 of 3 μM (R2=0.997). Error bars represent S.D. for n=4 experiments. (B) TmaDisA F57R mutant shows salt-dependent dissociation. The mutant (blue, no salt buffer; green, 10 mM NaCl buffer) elutes at higher volumes from a Superose 6 PC 3.2/30 column compared with wild-type TmaDisA (black curve) and thus is destabilized by salt, suggesting a dissociation from, e.g., tetramers to monomers.

Hypothesis on the regulation of DisA

DisA recognizes recombination intermediate DNA structures via its HhH motifs and upon binding these DNAs displays strongly reduced or abolished DAC activity. The current model for the down-regulation of DisA is based on the fact that the HhH domains at the DisA C-terminus need to rearrange in order to be able to bind to DNA. This rearrangement and rotation of the DNA-binding motif is probably translated by the helical spine domain leading to changes in the orientations of the DACs and thus to their inactivation. So far we were not able to crystallize a DisA–DNA complex or identify these DisA–Holliday junction complexes by EM, making it hard to predict which molecular rearrangements might lead to signalling or loss of DAC activity. In order to obtain a smaller version of DisA, we created a F57R mutant of TmaDisA that was supposed to form tetramers, as the F57R mutation disturbs the DAC dimer interfaces. However, depending on the salt concentration, this mutant also forms lower oligomeric species, as observed in SEC experiments (Figure 4B). This finding is in good agreement with the interactions between the helical spine domains being mainly ionic and suggests that once the DAC domains are disrupted by HhH-induced structural changes upon DNA binding, DisA might become unstable and dissociate. Dissociation and/or degradation of DisA as a result of the DNA complex formation would be a reasonable explanation of the measurable decrease in c-di-AMP levels in the cells upon DisA sensing DNA damage [3] as it would affect more than one DisA. It would probably be impossible for the cell to sense the presence of recombination intermediates by just a single DisA being down regulated, while 25–50 other DisA-octamers [34,35] and also CdaA and CdaS remain active in the cell. Of course, we might lack effector proteins that specifically recognize the DisA–DNA complex and amplify the signal, e.g. by RadA interaction [36,37] or up-regulation of a c-di-AMP phosphodiesterase. However, the exact mechanism of regulation remains to be shown, as there is currently no structural information available about the DisA–DNA or proposed DisA–RadA complexes.

In summary, our structural and biochemical analysis provides a model for the reaction mechanism of DisA that can probably be transferred to any other bacterial DAC domain protein because of structural and sequence similarity. Moreover, we were able to show that a commercially available nucleotide analogue that has also been shown to affect various other essential processes in the cell is also a potent inhibitor of c-di-AMP synthesis. The fact that c-di-AMP levels have been reported to influence, e.g. MRSA, (methicillin-resistant Staphylococcus aureus) antibiotic resistance, renders the c-di-AMP pathway an interesting new target for anti-microbial therapy.

AUTHOR CONTRIBUTION

Martina Müller and Tobias Deimling performed and evaluated experiments. Karl-Peter Hopfner supported experimental design and evaluation. Gregor Witte evaluated data, directed the work and prepared the manuscript with the help of Martina Müller.

We thank the staffs of SLS X06SA and EMBL Hamburg X33/P14 for on-site help and discussions, Robert Byrne for comments on the manuscript and the Hopfner group for discussions.

FUNDING

This work was funded by grants from the Deutsche Forschungsgemeinschaft [grant numbers WI3717/2-1 (to G.W.) and GRK1721 (to G.W. and K.-P.H.)]; M.M. and T.D. are supported by GRK1721.

Abbreviations

     
  • ApCpp

    adenosine-5′-[(α,β)-methyleno]triphosphate

  •  
  • c-di-AMP

    cyclic-di-AMP

  •  
  • DAC

    diadenylate cyclase

  •  
  • DisA

    DNA-integrity-scanning protein A

  •  
  • SEC-RALS

    size-exclusion chromatography-coupled right-angle laser light scattering

  •  
  • TH

    thiohydantoin

  •  
  • TmaDisA

    DisA of Thermotoga maritima

References

References
1
Witte
G.
Hartung
S.
Büttner
K.
Hopfner
K.P.
Structural biochemistry of a bacterial checkpoint protein reveals diadenylate cyclase activity regulated by DNA recombination intermediates
Mol. Cell
2008
, vol. 
30
 (pg. 
167
-
178
)
[PubMed]
2
Bejerano-Sagie
M.
Oppenheimer-Shaanan
Y.
Berlatzky
I.
Rouvinski
A.
Meyerovich
M.
Ben-Yehuda
S.
A checkpoint protein that scans the chromosome for damage at the start of sporulation in Bacillus subtilis
Cell
2006
, vol. 
125
 (pg. 
679
-
690
)
[PubMed]
3
Oppenheimer-Shaanan
Y.
Wexselblatt
E.
Katzhendler
J.
Yavin
E.
Ben-Yehuda
S.
c-di-AMP reports DNA integrity during sporulation in Bacillus subtilis
EMBO Rep.
2011
, vol. 
12
 (pg. 
594
-
601
)
[PubMed]
4
Mehne
F.M.
Gunka
K.
Eilers
H.
Herzberg
C.
Kaever
V.
Stülke
J.
Cyclic di-AMP homeostasis in Bacillus subtilis: both lack and high level accumulation of the nucleotide are detrimental for cell growth
J. Biol. Chem.
2013
, vol. 
288
 (pg. 
2004
-
2017
)
[PubMed]
5
Corrigan
R.M.
Abbott
J.C.
Burhenne
H.
Kaever
V.
Gründling
A.
c-di-AMP is a new second messenger in Staphylococcus aureus with a role in controlling cell size and envelope stress
PLoS Pathog.
2011
, vol. 
7
 pg. 
e1002217
 
[PubMed]
6
Luo
Y.
Helmann
J.D.
Analysis of the role of Bacillus subtilis sigma(M) in beta-lactam resistance reveals an essential role for c-di-AMP in peptidoglycan homeostasis
Mol. Microbiol.
2012
, vol. 
83
 (pg. 
623
-
639
)
[PubMed]
7
Bai
Y.
Yang
J.
Zarrella
T.M.
Zhang
Y.
Metzger
D.W.
Bai
G.
Cyclic di-AMP impairs potassium uptake mediated by a cyclic di-AMP binding protein in Streptococcus pneumoniae
J. Bacteriol.
2014
, vol. 
196
 (pg. 
614
-
623
)
[PubMed]
8
Corrigan
R.M.
Campeotto
I.
Jeganathan
T.
Roelofs
K.G.
Lee
V.T.
Gründling
A.
Systematic identification of conserved bacterial c-di-AMP receptor proteins
Proc. Natl. Acad. Sci. U.S.A.
2013
, vol. 
110
 (pg. 
9084
-
9089
)
[PubMed]
9
Campeotto
I.
Zhang
Y.
Mladenov
M.G.
Freemont
P.S.
Gründling
A.
Complex structure and biochemical characterization of the Staphylococcus aureus cyclic di-AMP binding protein PstA, the founding member of a new signal transduction protein family
J. Biol. Chem.
2014
, vol. 
290
 (pg. 
2888
-
2901
)
[PubMed]
10
Choi
P.H.
Sureka
K.
Woodward
J.J.
Tong
L.
Molecular basis for the recognition of cyclic-di-AMP by PstA, a PII-like signal transduction protein
Microbiologyopen
2015
 
doi:10.1002/mbo3.243
[PubMed]
11
Gundlach
J.
Dickmanns
A.
Schroder-Tittmann
K.
Neumann
P.
Kaesler
J.
Kampf
J.
Herzberg
C.
Hammer
E.
Schwede
F.
Kaever
V.
, et al. 
Identification, characterization and structure analysis of the c-di-AMP binding PII-like signal transduction protein DarA
J. Biol. Chem.
2014
, vol. 
290
 (pg. 
3069
-
3080
)
[PubMed]
12
Müller
M.
Hopfner
K.P.
Witte
G.
c-di-AMP recognition by Staphylococcus aureus PstA
FEBS Lett.
2015
, vol. 
589
 (pg. 
45
-
51
)
[PubMed]
13
Gao
A.
Serganov
A.
Structural insights into recognition of c-di-AMP by the ydaO riboswitch
Nat. Chem. Biol.
2014
, vol. 
10
 (pg. 
787
-
792
)
[PubMed]
14
Ren
A.
Patel
D.J.
c-di-AMP binds the ydaO riboswitch in two pseudo-symmetry-related pockets
Nat. Chem. Biol.
2014
, vol. 
10
 (pg. 
780
-
786
)
[PubMed]
15
Bai
Y.
Yang
J.
Eisele
L.E.
Underwood
A.J.
Koestler
B.J.
Waters
C.M.
Metzger
D.W.
Bai
G.
Two DHH subfamily 1 proteins in Streptococcus pneumoniae possess cyclic di-AMP phosphodiesterase activity and affect bacterial growth and virulence
J. Bacteriol.
2013
, vol. 
195
 (pg. 
5123
-
5132
)
[PubMed]
16
Witte
C.E.
Whiteley
A.T.
Burke
T.P.
Sauer
J.D.
Portnoy
D.A.
Woodward
J.J.
Cyclic di-AMP is critical for Listeria monocytogenes growth, cell wall homeostasis, and establishment of infection
mBio
2013
, vol. 
4
 (pg. 
e00282
-
00213
)
[PubMed]
17
Kabsch
W.
Xds
Acta Crystallogr. D Biol. Crystallogr.
2010
, vol. 
66
 (pg. 
125
-
132
)
[PubMed]
18
McCoy
A.J.
Grosse-Kunstleve
R.W.
Adams
P.D.
Winn
M.D.
Storoni
L.C.
Read
R.J.
Phaser crystallographic software
J. Appl. Crystallogr.
2007
, vol. 
40
 (pg. 
658
-
674
)
[PubMed]
19
Winn
M.D.
Ballard
C.C.
Cowtan
K.D.
Dodson
E.J.
Emsley
P.
Evans
P.R.
Keegan
R.M.
Krissinel
E.B.
Leslie
A.G.
McCoy
A.
McNicholas
S.J.
, et al. 
Overview of the CCP4 suite and current developments
Acta Crystallogr. D Biol. Crystallogr.
2011
, vol. 
67
 (pg. 
235
-
242
)
[PubMed]
20
Afonine
P.V.
Grosse-Kunstleve
R.W.
Echols
N.
Headd
J.J.
Moriarty
N.W.
Mustyakimov
M.
Terwilliger
T.C.
Urzhumtsev
A.
Zwart
P.H.
Adams
P.D.
Towards automated crystallographic structure refinement with phenix.refine
Acta Crystallogr. D Biol. Crystallogr.
2012
, vol. 
68
 (pg. 
352
-
367
)
[PubMed]
21
Emsley
P.
Cowtan
K.
Coot: model-building tools for molecular graphics
Acta Crystallogr. D Biol. Crystallogr.
2004
, vol. 
60
 (pg. 
2126
-
2132
)
[PubMed]
22
Schrödinger
L.L.C.
2010
 
The PyMOL Molecular Graphics System, Version 1.3r1
23
Chovancova
E.
Pavelka
A.
Benes
P.
Strnad
O.
Brezovsky
J.
Kozlikova
B.
Gora
A.
Sustr
V.
Klvana
M.
Medek
P.
, et al. 
CAVER 3.0: a tool for the analysis of transport pathways in dynamic protein structures
PLoS Comput. Biol.
2012
, vol. 
8
 pg. 
e1002708
 
[PubMed]
24
Petoukhov
M.V.
Franke
D.
Shkumatov
A.V.
Tria
G.
Kikhney
A.G.
Gajda
M.
Gorba
C.
Mertens
H.D.
Konarev
P.V.
Svergun
D.I.
New developments in the program package for small-angle scattering data analysis
J. Appl. Crystallogr.
2012
, vol. 
45
 (pg. 
342
-
350
)
[PubMed]
25
Putnam
C.D.
Hammel
M.
Hura
G.L.
Tainer
J.A.
X-ray solution scattering (SAXS) combined with crystallography and computation: defining accurate macromolecular structures, conformations and assemblies in solution
Q. Rev. Biophys.
2007
, vol. 
40
 (pg. 
191
-
285
)
[PubMed]
26
Svergun
D.
Barberato
C.
Koch
M.H.J.
CRYSOL - a program to evaluate X-ray solution scattering of biological macromolecules from atomic coordinates
J. Appl. Crystallogr.
1995
, vol. 
28
 (pg. 
768
-
773
)
27
Chan
C.
Paul
R.
Samoray
D.
Amiot
N.C.
Giese
B.
Jenal
U.
Schirmer
T.
Structural basis of activity and allosteric control of diguanylate cyclase
Proc. Natl. Acad. Sci. U.S.A.
2004
, vol. 
101
 (pg. 
17084
-
17089
)
[PubMed]
28
Rosenberg
J.
Dickmanns
A.
Neumann
P.
Gunka
K.
Arens
J.
Kaever
V.
Stülke
J.
Ficner
R.
Commichau
F.M.
Structural and biochemical analysis of the essential diadenylate cyclase CdaA from Listeria monocytogenes
J. Biol. Chem.
2015
, vol. 
290
 (pg. 
6596
-
6606
)
[PubMed]
29
Jamroz
M.
Kolinski
A.
Kmiecik
S.
CABS-flex: Server for fast simulation of protein structure fluctuations
Nucleic Acids Res.
2013
, vol. 
41
 (pg. 
W427
-
W431
)
[PubMed]
30
Yue
K.
Ye
M.
Zhou
Z.
Sun
W.
Lin
X.
The genus Cordyceps: a chemical and pharmacological review
J. Pharm. Pharmacol.
2013
, vol. 
65
 (pg. 
474
-
493
)
[PubMed]
31
Tuli
H.S.
Sharma
A.K.
Sandhu
S.S.
Kashyap
D.
Cordycepin: a bioactive metabolite with therapeutic potential
Life Sci.
2013
, vol. 
93
 (pg. 
863
-
869
)
[PubMed]
32
Ahn
Y.J.
Park
S.J.
Lee
S.G.
Shin
S.C.
Choi
D.H.
Cordycepin: selective growth inhibitor derived from liquid culture of Cordyceps militaris against Clostridium spp
J. Agric. Food Chem.
2000
, vol. 
48
 (pg. 
2744
-
2748
)
[PubMed]
33
Zheng
Y.
Zhou
J.
Sayre
D.A.
Sintim
H.O.
Identification of bromophenol thiohydantoin as an inhibitor of DisA, a c-di-AMP synthase, from a 1000 compound library, using the coralyne assay
Chem. Commun.
2014
, vol. 
50
 (pg. 
11234
-
11237
)
[PubMed]
34
Maass
S.
Sievers
S.
Zuhlke
D.
Kuzinski
J.
Sappa
P.K.
Muntel
J.
Hessling
B.
Bernhardt
J.
Sietmann
R.
Volker
U.
, et al. 
Efficient, global-scale quantification of absolute protein amounts by integration of targeted mass spectrometry and two-dimensional gel-based proteomics
Anal. Chem.
2011
, vol. 
83
 (pg. 
2677
-
2684
)
[PubMed]
35
Muntel
J.
Fromion
V.
Goelzer
A.
Maabeta
S.
Mader
U.
Büttner
K.
Hecker
M.
Becher
D.
Comprehensive absolute quantification of the cytosolic proteome of Bacillus subtilis by data independent, parallel fragmentation in liquid chromatography/mass spectrometry (LC/MS(E))
Mol. Cell. Proteomics
2014
, vol. 
13
 (pg. 
1008
-
1019
)
[PubMed]
36
Gándara
C.
Alonso
J.C.
DisA and c-di-AMP act at the intersection between DNA-damage response and stress homeostasis in exponentially growing Bacillus subtilis cells
DNA Repair
2015
, vol. 
27
 (pg. 
1
-
8
)
[PubMed]
37
Zhang
L.
He
Z.G.
Radiation-sensitive gene A (RadA) targets DisA, DNA integrity scanning protein A, to negatively affect cyclic Di-AMP synthesis activity in Mycobacterium smegmatis
J. Biol. Chem.
2013
, vol. 
288
 (pg. 
22426
-
22436
)
[PubMed]