Structural and functional analysis of the D-alanyl carrier protein ligase DltA from Staphylococcus aureus Mu50

D-Alanylation of the teichoic acids of the Gram-positive bacterial cell wall plays crucial roles in bacterial physiology and virulence. Deprivation of D-alanine from the teichoic acids of Staphylococcus aureus impairs biofilm and colony formation, induces autolysis and ultimately renders methicillin-resistant S. aureus highly susceptible to antimicrobial agents and host defense peptides. Hence, the D-alanylation pathway has emerged as a promising antibacterial target against drug-resistant S. aureus. D-Alanylation of teichoic acids is mediated via the action of four proteins encoded by the dlt operon, DltABCD, all four of which are essential for the process. In order to develop novel antimicrobial agents against S. aureus, the D-alanyl carrier protein ligase DltA, which is the first protein in the D-alanylation pathway, was focused on. Here, the crystal structure of DltA from the methicillin-resistant S. aureus strain Mu50 is presented, which reveals the unique molecular details of the catalytic center and the role of the P-loop. Kinetic analysis shows that the enantioselectivity of S. aureus DltA is much higher than that of DltA from other species. In the presence of DltC, the enzymatic activity of DltA is increased by an order of magnitude, suggesting a new exploitable binding pocket. This discovery may pave the way for a new generation of treatments for drug-resistant S. aureus.

1. Introduction

Staphylococcus aureus is a Gram-positive bacterial human pathogen that is capable of infecting almost all human tissues and is responsible for life-threatening infections including pneumonia, endocarditis and toxic shock syndrome (Lowy, 1998; Tong et al., 2015; Balasubramanian et al., 2017). S. aureus is the leading cause of hospital-acquired infections, with substantial mortality and morbidity (Magill et al., 2014; Fry & Barie, 2011). More importantly, S. aureus is notorious for its ability to acquire resistance to a variety of antimicrobials (Chambers & DeLeo, 2009). The emergence of methicillin-resistant S. aureus (MRSA) and vancomycin-resistant S. aureus (VRSA) has made S. aureus-associated infections more life-threatening and more difficult to treat. This poses a serious public health threat (Lee et al., 2018; Smith et al., 1999). Therefore, there is an urgent demand for novel targets and therapeutic agents to address this challenge. In an effort to develop a new antibiotic to combat multidrug-resistant S. aureus, we focused on the D-alanyl carrier protein ligase DltA from S. aureus subspecies Mu50 to provide detailed structural information that will support a rationale for novel antibiotic drug design.

The bacterial cell envelope has a unique, highly complex three-dimensional structure that plays a crucial role in maintaining cell structure and ensuring cell viability by acting as an initial barrier against external environmental stress (Seltmann & Holst, 2013). Most pathogenic bacteria, including S. aureus, have devised various exquisite ways of modifying their cell envelope to avoid the host defenses and to successfully infect the host. Teichoic acids (TAs) are phosphate-rich negatively charged glycopolymers that are found on the surface of a wide range of Gram-positive bacterial cell walls (Brown et al., 2013; Percy & Gründling, 2014). TAs are connected to either peptidoglycan (wall teichoic acids; WTAs) or to the cytoplasmic membrane (lipoteichoic acids; LTAs) (Xia et al., 2010) and play fundamental roles in bacterial physiology, resistance to antibiotics, immune evasion and pathogenesis, making TAs a promising antibacterial target (D'Elia et al., 2006; Weidenmaier et al., 2004; Atilano et al., 2011; Kristian et al., 2003; Brown et al., 2012). S. aureus confers a net charge compensation for anionic TAs through the D-alanylation process via the action of four proteins encoded by the dlt operon: DltABCD (Koprivnjak et al., 2006). Deprivation of D-alanine from the TAs has been shown to induce autolysis, impair biofilm formation and inhibit normal cellular growth, and ultimately renders S. aureus highly susceptible to antimicrobial agents (Wecke et al., 1996; Koprivnjak et al., 2006; Walter et al., 2007; Gross et al., 2001; Collins et al., 2002; Peschel et al., 2000; Coupri et al., 2021). Thus, the D-alanylation pathway has been suggested as an attractive target for the development of novel antibiotics to treat infectious diseases caused by S. aureus (Weidenmaier et al., 2003). D-Alanylation of TAs is a multistep process that starts in the cytoplasm, and the D-alanyl carrier protein DltA catalyzes two sequential reactions in the initial step of the process: (i) adenylation of D-alanine at the expense of ATP and (ii) thioesterification of the 4′-phosphopantetheinyl (Ppant) prosthetic group of the D-Ala carrier protein DltC (Wecke et al., 1996). Through the adenylation and thioesterification reactions catalyzed by DltA, D-alanine is loaded onto the Ppant-modified conserved serine residue of DltC. D-Alanine is then transferred across the plasma membrane and incorporated into TAs through the action of two membrane-bound acyltransferases, DltB and DltD, although the detailed molecular mechanism remains elusive.

Here, we present a comprehensive structural and enzymatic characterization of the first protein in the D-alanylation pathway, DltA, from S. aureus. Based on comparisons of crystal structures, we detail the structural differences in the active site, as well as in the overall conformation of the enzyme, possibly explaining the discrepancies in catalytic parameters between S. aureus DltA and DltAs from other organisms. We believe that the characterization of DltA from the major human pathogen S. aureus will be valuable for future drug development and that its unique features as presented here will be exploitable when trying to rationally design specific antibiotics targeting DltA.

2. Materials and methods

2.1. Cloning, expression and purification

The cDNAs coding for S. aureus DltA (SaDltA; UniProt P68876) and S. aureus DltC (SaDltC; UniProt P0A018) were amplified by PCR and expressed as N-terminally His₆-tagged fusion proteins using the expression vector pCOLD1 (Takara Bio, Shiga, Japan). The cDNA coding for Escherichia coli AcpS (UniProt P24224) was amplified by PCR and cloned into vector pET-28a (NEB, Ipswich, Massachusetts, USA), resulting in the production of AcpS without an affinity tag (for co-expression with wild-type SaDltC), or into vector pMALc2X (for in vitro phosphopantetheinylation) (NEB). The S36A point mutation was introduced into SaDltC using a QuikChange site-directed mutagenesis kit (Agilent Technologies, Wilmington, Delaware, USA).

SaDltA and SaDltC S36A were expressed in E. coli BL21(DE3) cells (Invitrogen, Carlsbad, California, USA). The E. coli cells were grown at 37°C in LB medium until the OD₆₀₀ reached 0.5 and expression was then induced by the addition of 0.5 mM isopropyl β-D-1-thiogalactopyranoside. The culture was grown under inductive conditions at 15°C for 12 h. The cells were harvested by centrifugation at 8000g, resuspended in 50 mM Tris–HCl pH 8.0, 500 mM NaCl (buffer A) and lysed by ultrasonication. The resulting cell lysate was centrifuged at 20 000g for 1 h at 4°C. The supernatant was then bound to an Ni–NTA affinity column and eluted with buffer A supplemented with 300 mM imidazole. The His₆ tag was removed by overnight incubation with TEV protease at 4°C and the protein was further purified using a HiLoad 16/60 Superdex 200 prep-grade column (GE Healthcare, Chicago, Illinois, USA) equilibrated with 20 mM Tris pH 7.5, 100 mM NaCl, 1 mM DTT. The purity of the protein was analyzed by 15% SDS–PAGE and it was concentrated to 30 mg ml⁻¹ using Amicon Ultra 10 kDa molecular-weight cutoff centrifugal filters (Millipore, Burlington, Massachusetts, USA).

Wild-type (WT) SaDltC was co-expressed with AcpS in E. coli BL21(DE3) cells to produce uniformly Ppant-modified SaDltC. Recombinant plasmids containing WT SaDltC and E. coli AcpS were co-transformed into E. coli BL21(DE3) and were expressed and purified similarly to as described above. The WT SaDltC protein was purified using an Ni–NTA affinity column in 50 mM Tris pH 8.0, 500 mM NaCl and was eluted with 250 mM imidazole. The His₆ tag was removed by overnight incubation with TEV protease at 4°C and the protein was additionally purified using a HiLoad 16/600 Superdex 75 prep-grade column (GE Healthcare) in 20 mM Tris pH 7.5, 100 mM NaCl, 1 mM DTT. The recombinant AcpS for in vitro phosphopantetheinylation was produced and purified as described previously (May et al., 2005). Sequence alignments were generated with MAFFT (Katoh & Standley, 2013) and were illustrated using ESPript (Gouet et al., 2003). The conservation scores of the amino acids in sequence alignments were determined with Scorecons (Valdar, 2002).

2.2. Crystallization and structure determination

Purified SaDltA at a concentration of 30 mg ml⁻¹ in 20 mM Tris pH 7.5, 100 mM NaCl, 1 mM DTT was used for initial screening of crystallization conditions. Before screening, the protein was premixed with 1 mM adenosine 5′-triphosphate (ATP) and 1 mM MgCl₂ (Merck, Darmstadt, Germany). 1 µl protein solution was mixed with an equal volume of reservoir solution consisting of 15% PEG 4K, 0.16 M magnesium acetate, 0.1 M sodium cacodylate pH 7.5 at 4°C. The crystals of SaDltA were transferred to a cryoprotectant solution containing 20%(v/v) glycerol for a few seconds before being cooled in liquid nitrogen. A set of X-ray diffraction data was collected at 100 K using an ADSC Quantum 315r CCD detector system (Area Detector Systems Corporation, Poway, California, USA) at the BL-7A experimental station of Pohang Light Source, Republic of Korea. The SaDltA crystal belonged to the monoclinic space group P2₁, with unit-cell parameters a = 89.46, b = 88.51, c = 130.85 Å, α = γ = 90.00, β = 91.67°. The raw data were processed and scaled using HKL-2000 (Otwinowski & Minor, 1997). Table 1 summarizes the data-collection statistics.

Table 1
Data-collection and refinement statistics

The structure was solved with a data set from a single crystal. Values in parentheses are for the highest resolution shell.

Data collection
Wavelength (Å)	0.9000
Space group	P21
a, b, c (Å)	89.47, 88.51, 130.85
α, β, γ (°)	90.0, 91.67, 90.0
Resolution (Å)	50.00–2.55 (2.64–2.55)
Rmerge	0.103 (0.527)
〈I/σ(I)〉	10.3 (7.3)
No. of reflections	333087 (8768)
Completeness (%)	99.5 (98.2)
Multiplicity	5.0 (4.7)
Wilson B factor (Å^2)	40.1
CC1/2	0.96 (0.80)
Refinement
Resolution (Å)	39.91–2.55
No. of unique reflections	66484
Completeness (%)	99.5
Rwork/Rfree (%)	18.9/23.9
No. of atoms
Protein	15270
ATP	124
Mg^2+	4
Water	381
B factors (Å^2)
Protein	47.9
ATP	41.9
Mg^2+	46.7
Water	42.9
R.m.s. deviations
Bond lengths (Å)	0.012
Bond angles (°)	1.48
Ramachandran plot (%)
Favored	96.87
Outliers	0.31
PDB code	7vhv

2.3. Pyrophosphate-detection assay

The adenylation activity of recombinant SaDltA was determined using a pyrophosphate-detection assay. All of the reagents used in this assay were purchased from Merck. Assays were performed with 5 µM SaDltA, 5 units ml⁻¹ in­organic pyrophosphatase, 5 mM ATP, 100 mM KCl, 10 mM MgCl₂, 50 mM Tris–HCl buffer pH 7.4 and various concentrations of alanine at 37°C. The enzyme reaction was initiated by the addition of alanine, and 20 µl reaction solution was retrieved every 2 min. The retrieved reaction solution was immediately mixed with 380 µl dye solution consisting of 0.033%(w/v) malachite green and 1.3%(w/v) ammonium molybdate in 1.0 M HCl and incubated for 90 s. The absorbance at 620 nm was measured using a BioSpectrometer (Eppendorf, Hamburg, Germany) and the absorbance without 5 µM SaDltA was used as a blank. Determination of the adenylation activity of SaDltA in the presence of WT or S36A mutant SaDltC was performed essentially as described above. Specifically, the reaction was conducted in the presence of SaDltA (5 µM), 5 mM D-alanine and 300 µM WT or S36A mutant SaDltC. The initial rates of the enzyme reaction were derived from the time courses of phosphate accumulation. Rate–concentration curves against alanine were fitted to the Michaelis–Menten equation using Prism 6 (GraphPad, San Diego, California, USA) to obtain kinetic parameters.

2.4. Data deposition

Atomic coordinates and structure-factor amplitudes for the structure of SaDltA have been deposited in the Protein Data Bank (PDB) with accession code 7vhv.

3. Results

3.1. Overall structure of S. aureus DltA

DltA from S. aureus (SaDltA) was crystallized in the presence of ATP and Mg²⁺. The structure was solved by the molecular-replacement method employing ATP-complexed B. cereus DltA (PDB entry 3fcc; Osman et al., 2009) as the search model (Table 1). The monoclinic crystal, belonging to space group P2₁, contains four DltA molecules (chains A, B, C and D) in the asymmetric unit. Two of the complexes (chains C and D) are relatively poorly defined in the electron-density map, yet fourfold noncrystallographic symmetry (NCS) averaging was used throughout the initial steps of the refinement, yielding well defined electron density for each crystallographic unit. Analysis of the protein interfaces in the asymmetric unit did not show specific interactions that would indicate the formation of oligomers; the complexation significance score (CSS), which indicates the significance of the interface for assembly formation, was 0.000. The atomic model of SaDltA has been refined to a crystallographic R factor of 0.189 (R_free = 0.239; Table 1) and features a characteristic `adenylation conformation' (Fig. 1a).

Figure 1 Overall structure of SaDltA. (a) Ribbon diagram and surface representation of two perpendicular views of SaDltA (the N-terminal lobes and C-terminal lobes are colored orange and blue, respectively). The ATP molecule bound in the active site is shown using a stick representation and is colored by atom type. (b) Sequence conservation of SaDltA (see also Fig. 2) mapped onto the surface of the structure and colored from low to high conservation using a red to green gradient. (c) Left: close-up view showing the ATP- and Mg^2+-bound active site of SaDltA. The 2Fo − Fc electron-density map (gray mesh) contoured at 1.0σ is shown around an all-atom representation of the P-loop. Center: close-up view showing the ATP- and Mg^2+-bound active site of B. cereus DltA (PDB entry 3fcc). Right: close-up view showing the D-alanine-adenylate-bound active site of B. cereus DltA (PDB entry 3dhv). (d) B-­factor distribution mapped onto the surface of SaDltA chains A and B (left) or chains C and D (right). The B-factor distribution is colored from low (20 Å^2) to high (90 Å^2) using a blue to red gradient. (e) Close-up view showing the interaction between the P-loop and residues residing in the hinge region of the Ppant blocking loop. Chain A of SaDltA is colored orange (N-lobe) and blue (C-lobe) and chain C of SaDltA is colored gray.

Similar to other adenylate-forming enzymes (Schmelz & Naismith, 2009), DltA utilizes several distinct structural conformations to catalyze two sequential reactions: (i) adenylation of D-Ala at the expense of ATP followed by (ii) thioesterification of the 4′-phosphopantetheinyl (Ppant) prosthetic group of the D-Ala carrier protein DltC (Supplementary Fig. S1a). Upon substrate binding by the substrate-free `open conformation' (Gulick, 2009; Weissman, 2015), binding of ATP and the D-Ala substrate facilitates the structural change of the enzyme towards the `adenylation conformation' by inducing electrostatic changes in the active-site pocket and mutual approach of the N-terminal and the C-terminal lobes, closing the active site from the bulk solvent and completing the active site capable of D-Ala adenylation. Subsequent D-Ala adenylate formation and breakage of the ATP phosphodiester bond result in a 140° rotation of the C-terminal subdomain towards the `thiolation conformation', in which the Ppant group of DltC can approach the active site of DltA for loading of D-Ala to complete the reaction (Du et al., 2008; Osman et al., 2009; Du & Luo, 2014). The SaDltA structure features the canonical adenylation conformation, allowing it to grasp the ATP substrate and Mg²⁺ ion in the active site that lies between the N- and C-terminal lobes and posing it for the D-alanine adenylation reaction with the Ppant substrate channel blocked by the loop (Ppant blocking loop; residues 463–476) in the C-terminal lobe (Figs. 1a and 1c).

SaDltA consists of two structurally distinct lobes: an N-terminal lobe (residues 1–377), a C-terminal lobe (residues 382–485) and a flexible interdomain hinge region (residues 378–381) connecting the two lobes (Figs. 1a and 2). The flexible nature of the hinge region, consisting of four highly conserved residues (Gly-Arg-Ile-Asp), allows the proper relative orientation of the N- and C- lobes for the adenylation conformation and thiolation conformation (Yonus et al., 2008). Clear electron density for ATP and magnesium in the active-site pocket is observed in all chains in the asymmetric unit (Supplementary Fig. S1b). The active-site pocket is positioned in a deep cleft between the N- and C- terminal lobes of DltA formed by conserved amino-acid residues mostly from the large N-terminal domain, including the highly conserved P-loop (Figs. 1a, 1b and 2). Interestingly, we first observed a clear density for the P-loop of DltA in the adenylation conformation, which has not been observed in other DltA structures (Figs. 1c and 1d), as discussed in detail below.

Figure 2 Alignment of DltA from S. aureus, B. cereus and B. subtilis. Orange backgrounds indicate 100% sequence identity and yellow backgrounds indicate amino acids with similar physicochemical properties. The P-loop, hinge region and unique short loop of S. aureus discussed in the text are marked with blue boxes. Sequence-conservation scores were calculated from the alignment of 100 DltA sequences from different species using Scorecons (Valdar, 2002). The resulting per-residue scores were then plotted against the S. aureus DltA sequence.

The characteristic structural features between chains were identified when we superimposed the four SaDltA monomers in the asymmetric unit (Fig. 1e and Supplementary Fig. S2). The most prominent difference between chains is the structural stability of the aforementioned P-loop (S. aureus residues ¹⁴⁴Thr-Ser-Gly-Ser-Thr-Gly-Glu-Pro-Lys¹⁵²) and interactions involving the P-loop that covers the one side of the active site. The P-loop of DltA has an amino-acid composition similar to that of the P-loop or Walker A motif found in other adenylate-forming enzymes (Schmelz & Naismith, 2009); thus, the loop has been suggested to bind ATP or pyrophosphate, despite the lack of direct structural data to support this assumption. In chains A and B of SaDltA we observed clear electron density allowing us to unambiguously assign the residues of the P-loop (Fig. 1c), whereas the electron density of the corresponding residues in chains C and D was very weak and we could not model four residues (residues 148–151), indicating significant flexibility of the region (Fig. 1d). In chains A and B the P-loop directly interacts with the hinge-region residue Asp381, as well as the C-terminal domain residues Asn471 and Lys473 that reside in the Ppant blocking loop, while the corresponding interactions are absent in chains C and D (Figs. 1d and 1e). Conceivably, the observed interactions involving P-loop residues are likely to stabilize the overall conformation of the protein by generating a more compressed and stable structure in which the C-terminal lobe moves towards the N-terminal lobe (Supplementary Fig. S2). Indeed, the C-terminal lobes of chains A and B have nearly identical structures [root-mean-square (r.m.s.) deviations of 0.38 Å between chains A and B] with an average B factor of 45–50 Ų, whereas the C-terminal lobes of chains C and D, in which the P-loops are disordered, reveal a variety of conformations (r.m.s. deviations of 1.56 Å between chains C and D) with an average B factor of 64–73 Ų (Fig. 1d). Interestingly, the previously reported substrate-free structure of B. cereus DltA (Du & Luo, 2014) shows a noticeably more flexible structure than the other adenylation-conformation structure of DltA, and the disordered region observed in the apo structure (B. cereus residues Arg397–Glu413 and Lys433–Tyr440) coincides well with the region that shows structural variation in SaDltA chains C and D (S. aureus DltA β22–β23 and β24–β25), in which the P-loops are disordered (Supplementary Fig. S2). These observations indicate that the binding of the P-loop to the β- and γ-phosphates of the ATP molecule, as well as binding to the Ppant blocking loop and hinge region, may be a prerequisite for the adenylation reaction by promoting the structural transition from the apo state to the adenylation state (Supplementary Fig. S2b; Du & Luo, 2014; Chen et al., 2015).

3.2. SaDltA shows distinct structural features compared with other DltA structures

There are several previously published DltA structures in the literature from B. cereus and B. subtilis. The structures from B. cereus are in the adenylation conformation [PDB entries 3fcc (complexed with ATP and Mg²⁺), 3fce (complexed with ATP and Ca²⁺), 3dhv (complexed with D-Ala-adenylate) and 4pzp (apo)] (Du & Luo, 2014; Osman et al., 2009), while a structure from B. subtilis features the thiolation conformation (PDB entry 3e7w; complexed with AMP; Yonus et al., 2008).

While the overall architecture of SaDltA is similar to those of previously reported B. cereus DltA structures in the adenylation conformation, several discrete structural dissimilarities were identified in SaDltA, especially in the interface between the large and small lobes, in addition to the aforementioned distinct structural features of the P-loop.

Interestingly, in SaDltA we observed many fewer inter­actions between the large N-lobe and the small C-lobe compared with DltA from B. cereus (Fig. 3). In SaDltA a very short loop (Thr325–Gly328) links the two β-strands β16 and β17, in comparison to 12 residues that loop out in other DltAs (Figs. 2 and 3a). The observed differences in the length of the loop region lead to substantial differences in the interdomain interactions formed between this structural element and the C-terminal lobe of the protein. In B. cereus DltA the positively charged Lys345 positioned in the loop makes hydrogen-bond interactions with the highly conserved Glu401 and Glu410 in an antiparallel hairpin right next to the hinge residues. However, these interactions are missing in SaDltA (Fig. 3a). Although the linker is inherently flexible, a degree of rigidity is introduced and may serve to keep the antiparallel hairpin region tethered, therefore stabilizing the overall adenylation conformation of the protein.

Figure 3 Several key interdomain interactions are absent in SaDltA. (a) Close-up view of the short loop (Thr325–Gly328) linking β-strands β16 and β17 of SaDltA (orange and blue, chain A) and the equivalent region in B. cereus (gray). Hydrogen-bond interactions are shown as dotted lines. Note that the interaction of Lys345 with Glu410 and Gln401 is absent in the structure of S. aureus DltA. (b) Left: close-up view of the hydrogen-bond interactions between Arg268 in the N-lobe and Ser403 and Val405 in the C-lobe of SaDltA (orange and blue, chain A). Right: close-up view of the hydrogen-bond interactions observed in the equivalent region of B. cereus DltA (gray; PDB entry 3fcc). Hydrogen-bond interactions are shown as dotted lines.

In addition, several other key hydrophilic interdomain interactions are missing in SaDltA. In structures of B. cereus DltA in the adenylation conformation, interactions between the N- and C-lobes are predominantly mediated by hydrogen bonds/salt bridges. In B. cereus DltA, the Arg419, Glu425 and His416 (B. cereus numbering) residues in the C-terminal lobe highly stabilize the interface between the N- and C-lobes by forming salt bridges/hydrophilic interactions with Glu247, Lys280 and Lys317 on the N-terminal lobe, respectively (Fig. 3b). However, these interactions are absent in S. aureus DltA, and the only interactions observed in the region are the hydrogen bonds between the backbone of Ser403/Val406 and the side chain of Arg268 (Fig. 3b). The absence of this interaction is likely to affect the adenylation–substrate-free conformational equilibrium of DltA and could reduce the catalytic efficiency of the enzyme, as the DltA–substrate interactions may be used to compensate for the energetic cost of stabilizing the adenylation state of the enzyme. Indeed, the catalytic efficiency of the DltA enzyme for the adenylation reaction is lower than that of previously reported DltAs (as discussed in detail below), but the substrate specificity is higher.

3.3. ATP-binding pocket of SaDltA

The active-site pocket of SaDltA is positioned in a deep cleft between the N- and C-terminal lobes formed by conserved amino-acid residues mostly from the large N-terminal domain, including the highly conserved P-loop (Fig. 4a). As mentioned earlier, we observed electron density for the P-loop in chains A and B of SaDltA and we are limiting our discussion to chain A of the molecule in the asymmetric unit, as chain A shows the clearest density map. The overall architecture of the SaDltA active site resembles that of DltA from B. cereus in the adenylation state, and characteristic key conserved residues essential for substrate binding and the adenylation reaction are present in the active site of SaDltA (Figs. 4a and 4b). The planar adenosine ring of ATP forms extensive hydrophilic interactions with highly conserved residues including Gly287, Tyr286, Thr285 and Gly262 (Fig. 4a). In SaDltA the adenosine ring makes π-stacking interactions with the aromatic side chain of Tyr286. The ribose moiety of ATP is held in place by hydrogen bonds to the side chains of Asp365 and Lys473. While the residues involved in the interaction between the adenosine ring and ribose parts of the bound ATP in SaDltA are similar to those of B. cereus DltA in the adenylation conformation (Fig. 4b), we observed unique features around the β- and γ-phosphate parts of the bound ATP. In SaDltA the P-loop folds over the phosphate parts of the nucleotide and wraps around the phosphate moiety to form an extensive hydrogen-bonding network, presumably positioning the phosphates of ATP for catalysis (Fig. 4a). The hydrogen-bonding network between the loop and phosphate includes hydrophilic interaction clusters consisting of Mg²⁺, Thr144, Ser145, Gly146, Ser147, Thr148 and Lys152. The hydrogen-bonding network between the β- and γ-phosphate parts of the bound ATP and the P-loop may help to maintain structural rigidity and stabilize ionization states to modulate the turnover rate and control the local charge balance during the adenylation reactions.

Figure 4 Detailed active-site structures of S. aureus DltA and B. cereus DltA. (a) Close-up view of the active site of S. aureus DltA (chain A). The bound Mg^2+ ion is shown as a purple sphere. The ATP molecule and the key residues engaged in the hydrogen-bonding network are shown as stick models. Residues from the N- and C-lobes are colored orange and blue, respectively. (b) Left: close-up view of the active site of ATP- and Mg^2+-complexed B. cereus DltA (PDB entry 3fcc). Residues engaged in the hydrogen-bonding network are shown as stick models and colored white. Right: close-up view of the active site of D-­alanine-adenylate-complexed B. cereus DltA (PDB entry 3dhv). Residues engaged in the hydrogen-bonding network are shown as stick models and colored cyan.

The D-alanine-binding pocket of SaDltA closely resembles that of DltA from B. cereus. In the D-Ala-AMP-bound B. cereus DltA structure (Du et al., 2008), the D-alanyl amino group makes hydrogen bonds to Thr297, Asp197 and Val301 and the D-alanyl methyl group points towards the small hydrophobic cavity formed by the side chains of Leu197, Met200 and Cys268. In particular, Cys268 has been shown to play a critical role in enantiomer selection, as its thiol group causes a steric clash with L-alanine but not with D-alanine (Du et al., 2008; Yonus et al., 2008). The corresponding residues involved in D-alanine binding are highly structurally conserved in SaDltA (Supplementary Fig. S3), suggesting that SaDltA would also prefer D-alanine as a substrate, as experimentally shown in the next section.

3.4. S. aureus DltA prefers D-alanine over L-alanine

B. cereus DltA prefers D-alanine over L-alanine as a substrate (Du et al., 2008). To investigate whether SaDltA also prefers D-alanine, we measured the initial rates of the adenylation reaction catalyzed by SaDltA via pyrophosphate-detection assays. The initial rates were measured in the presence of various concentrations of D- or L-alanine, and the rate–concentration curves towards alanine were fitted to the Michaelis–Menten equation (Figs. 5a and 5b). The calculated kinetic parameters are listed in Fig. 5(c). The K_m of S. aureus DltA against D-alanine was lower than the K_m against L-alanine, and the catalytic efficiency (k_cat/K_m) against D-alanine was higher than the k_cat/K_m against L-alanine. These results show that SaDltA prefers D-alanine over L-alanine. For SaDltA, the change in K_m (for D-alanine over L-alanine) was approximately 106.71-fold, while according to previous studies it was about 13.1-fold for B. cereus DltA (Du et al., 2008). This indicates that the substrate specificity of SaDltA is much higher than that of B. cereus DltA. Although the catalytic efficiency (k_cat/K_m) of B. cereus DltA was higher than that of SaDltA, the change in k_cat/K_m (for D-alanine over L-alanine) for SaDltA was higher than that of B. cereus DltA.

Figure 5 Rate–concentration curves of the adenylation catalyzed by S. aureus DltA in the presence of D- or L-alanine. (a) Adenylation reaction with various concentrations of D-alanine. (b) Adenylation reaction with various concentrations of L-alanine. (c) Kinetic parameters for SaDltA. The average values of triplicate measurements and standard deviations are shown.

3.5. The Ppant group of S. aureus DltC is necessary for the activation of S. aureus DltA

The Ppant group of DltC is required for the second reaction step catalyzed by DltA (Heaton & Neuhaus, 1994), and its mimic coenzyme A has been shown to increase the adenylation activity of B. cereus DltA (Osman et al., 2009; Du & Luo, 2014). In order to examine whether S. aureus DltC (SaDltC) increases the adenylation activity of SaDltA and whether the Ppant group of DltC is necessary for activation, we measured the initial rates of the adenylation reaction catalyzed by SaDltA in the presence of wild-type (WT) SaDltC and the S36A DltC mutant. The S36A DltC mutant is incapable of being post-translationally modified with a Ppant group at the conserved Ser36 residue (May et al., 2005). To produce uniformly Ppant-group-modified WT SaDltC, we co-expressed WT DltC with acyl carrier protein synthase (AcpS), which catalyses the Ppant-modification reaction at Ser36 of SaDltC. To confirm that the residue responsible for the Ppant modification is Ser36, we performed an in vitro phosphopantetheinylation reaction using purified AcpS and WT SaDltC or S36A SaDltC, and analyzed it using native PAGE. WT SaDltC purified from E. coli migrated as two bands on conformationally sensitive native PAGE, as reported previously (Wood et al., 2018), and shifted to a single band upon AcpS-catalyzed phosphopantetheinylation (Supplementary Fig. S4a). In contrast, S36A SaDltC purified from E. coli migrated as a single band and did not show a band shift upon AcpS-catalyzed phosphopantetheinylation, confirming that the residue responsible for the Ppant modification of SaDltC is Ser36. The uniformity of modification was verified by matrix-assisted laser desorption/ionization (MALDI) time-of-flight (TOF) mass spectrometry (Supplementary Fig. S4b).

The adenylation activity of SaDltA was increased by approximately 7.2-fold in the presence of wild-type S. aureus DltC, but the S36A mutant did not significantly affect the activity (Fig. 6). These data indicate that SaDltC is able to activate the adenylation activitiy of SaDltA and that the Ppant group covalently attached to DltC is necessary for activation.

Figure 6 Ppant-modified SaDltC activates SaDltA. Initial rates of adenylation catalyzed by SaDltA (5 µM) in the presence of wild-type SaDltC (300 µM) or the S36A mutant (300 µM). Error bars represent the standard deviation (n = 3). Statistical significance was evaluated by one-way ANOVA with Tukey's multiple comparison test. Statistical values of p < 0.05 were considered to be statistically significant. ***, p < 0.0001; ns, not significant.

4. Discussion

Bacterial resistance to antibiotics is a global health emergency that is not only increasing rapidly but is also severe. By 2050, it has been estimated that the total number of deaths caused by antimicrobial resistance will have increased by ten million per year worldwide (O'Neill, 2014). Among antibiotic-resistant pathogens, S. aureus is especially notorious for its unique ability to rapidly become resistant to a broad range of antibiotics, making the pathogen extremely difficult to treat (Chambers & DeLeo, 2009). Therefore, S. aureus resistant to methicillin was classified by the World Health Organization (WHO) as one of 12 `priority pathogens' that pose the greatest threat to human health and for which new antibiotics are urgently needed (Willyard, 2017).

S. aureus has become resistant to antibiotics through numerous strategies, including the inactivation of antibiotics (for example the production of penicillinase, a penicillin-binding protein), cell-wall thickening, spontaneous mutations of topoisomerase IV and DNA gyrase, and overexpression of an efflux pump (for example NorA) (Chambers & DeLeo, 2009; Lowy, 2003; Schmitz et al., 1998; Kaatz & Seo, 1995; Fishovitz et al., 2014). In this study, we focused on the D-alanylation process mediated by four proteins (DltA, DltB, DltC and DltD) encoded by the dlt operon that have drawn much attention as novel antibiotic drug targets due to their indispensable role in the virulence and survival of S. aureus (Weidenmaier et al., 2003, 2004; Collins et al., 2002; Kristian et al., 2003).

Our crystal structure of SaDltA complexed with ATP and Mg²⁺ reveals detailed interactions involving the P-loop. We observed that the P-loop not only binds to the β- and γ-phosphate moieties of ATP, but also directly binds to residues in the hinge region and Ppant blocking loop in the C-lobe (Figs. 1c and 1e). We interpret these observations to indicate that the interactions involving the P-loop may be required for DltA to adopt a properly positioned structure for the adenylation reaction. In the absence of such interactions, as shown in chains C and D, the overall B factor of the C-lobe is significantly higher, supporting our assumption that inter­actions involving the P-loop stabilize the overall adenylation conformation of the protein (Figs. 1d and 1e).

In the SaDltA structure, several key interdomain inter­actions (between the N- and C-lobes) that are observed in the adenylation conformation of B. cereus DltA are missing (Fig. 3). This might affect the overall catalytic efficiency of the enzyme in the adenylation reaction, as additional substrate-binding energy is required to compensate for this entropic penalty. Indeed, the catalytic efficiency (k_cat/K_m) of SaDltA for D-alanine is lower than that of B. cereus DltA by an order of magnitude (Fig. 5c; Du et al., 2008). Interestingly, the cytoplasmic concentration of D-alanine in S. aureus reaches approximately 1.5 mM in the stationary phase, which is similar to the K_m value of SaDltA for D-alanine (Fig. 5c), while B. subtilis produces only marginal amounts of D-alanine (Lam et al., 2009). In addition to this, the enantioselectivity of SaDltA is much higher than that of B. cereus DltA (Fig. 5), suggesting that the rate of the D-alanine-adenylate-forming reaction of S. aureus in vivo may be higher than that of B. cereus.

The single enzyme DltA catalyzes two sequential reactions: (i) adenylation of D-alanine and (ii) thioesterification of the Ppant moiety attached to DltC. In the presence of Ppant-loaded SaDltC, which is a substrate of the second reaction, the adenylation reaction rate of SaDltA was increased by approximately 7.2-fold, which is in agreement with previous work (Fig. 6; Wood et al., 2018). Conceivably, the Ppant-binding pocket in DltA could be exploited for the development of selective inhibitors, although further structural work on Ppant–DltA complexes will be necessary.

Undoubtedly, a better understanding of the enzymes involved in the pathogenicity and virulence of S. aureus could be of great help in the design of new antibacterial agents. As DltA is a protein that is essential for the pathogenicity of S. aureus, the observed unique structural features as well as the kinetic parameters reported here may be useful in the design of effective antibiotic peptides or small molecules and may accelerate a rational approach for the development of effective antibiotics specific for S. aureus.

Footnotes

‡These authors contributed equally.