D-alanine:D-alanine ligase (Ddl) catalyses the dimerization of D-alanine before its incorporation in peptidoglycan precursors. The synthesis of D-alanine:D-alanine begins with an attack on the first D-alanine by the γ-phosphate of adenosine triphosphate (ATP) to yield an acylphosphate. That is followed by attack by the amino group of the second D-alanine, which eliminates the phosphate and produces the D-alanine:D-alanine dipeptide.1, 2
Peptidoglycan biosynthesis has long been an attractive target for antibacterial drugs, such as D-cycloserine, glycopeptide antibiotics (vancomycin and teicoplanins), and β-lactams (penicillin and cephalosporins).3–5 Vancomycin-type antibiotics, for example, bind directly to the D-alanine:D-alanine terminus, thereby inhibiting crosslinking by the transpeptidase. Notably, bacteria that show vancomycin resistance, which develops after prolonged clinical treatment with vancomycin, possess an inactive Ddl and rely on another ligase, D-alanine:D-lactate ligase (Van), which produces D-alanine:D-lactate rather than D-alanine:D-alanine for cell wall synthesis. The switch from D-alanine:D-alanine peptidoglycan termini to D-alanine:D-lactate results in the loss of crucial hydrogen bonding interactions that causes a 1000-fold reduction in vancomycin binding affinity.6–8
X-ray crystallographic studies of Ddl and Van have contributed significantly to our understanding of the ligand specificity these two enzymes and suggest that a His residue in Van plays a critical role. A positive charge on the side chain imidazole nitrogen of His would attract the negatively charged lactate to the second substrate binding site at pH values less than 7, but at higher pH values Van would predominantly synthesize D-alanine:D-alanine. In Ddl, a Tyr residue [Tyr216 in Escherichia coli (Eco) DdlB, Tyr232 in Thermus caldophilus (Tca) Ddl] occupies the same spatial position as the His residue, and the hydroxyl group of the Tyr interacts with the COOH-terminal of the second D-alanine substrate.9–11
The structure of Eco DdlB complexed with ADP/phosphorylated phosphinate (PDB ID: 2DLN)12 or with ADP/phosphorylated phosphonate (PDB ID: 1IOV)13 has been determined, as have the structures of Leuconostoc mesenteroides (Lme) D-Alanine:D-Lactate ligase complexed with ADP and phosphinophosphate (PDB ID:1EHI)14 and Enterococcus faecium (Efa) VanA complexed with ADP and phosphinophosphate (PDB ID:1E4E).15 However, to analyze the reaction mechanisms of these enzymes and their associated conformational changes, it is necessary to know the structures of both the substrate-bound and substrate-free forms of these enzymes. Our aim in the present study, therefore, was to grow crystals of Ddl that diffracted to high resolution in the absence of substrates. Here we report the X-ray structure of Tca Ddl resolved to a resolution of 1.9 Å and describe the conformational differences of the apo structure, comparing it with the structures of the previously described transition state analogue complex.
Materials and Methods.
DNA encoding Tca Ddl was subcloned between the EcoRI and PstI sites of the expression vector pKK223-3 (Pharmacia Biotech), which was then used to transform E. coli strain JM109. Expression of Tca Ddl was induced by treating the cells with 1 mM isopropyl β-D-thiogalactopyranoside (IPTG) for 6 h at 310 K. The bacteria were then resuspended in lysis buffer (50 mM potassium phosphate, pH 7.5) and lysed by sonication, after which the lysate was centrifuged at 14,000 × g for 30 min. Tca Ddl was then purified by heating the supernatant at 353 K for 30 min with gentle agitation. The denatured contaminating proteins were removed by centrifugation at 14,000 × g for 20 min, after which the supernatant was applied to a DEAE-Sepharose column (Pharmacia Biotech), and the matrix was washed with 10 bed volumes of lysis buffer. The protein was eluted with a linear KCl gradient (0–1 M KCl) and further purified using size exclusion chromatography on a Superdex 200 column (Pharmacia) preequilibrated with 20 mM HEPES-NaOH (pH 7.5) and 150 mM KCl. The fractions containing Tca Ddl were collected and concentrated by ultrafiltration (Amicon Centricon 10). The molecular weight of the Tca Ddl obtained was approximately 60 kDa, which corresponds to the dimeric state, as estimated by size exclusion chromatography.
Tca Ddl was crystallized at room temperature (294 ± 1 K) using the hanging-drop vapor-diffusion method. Crystals were grown on a siliconized cover slip by equilibrating a mixture containing 2 μl of protein solution [12 mg mL−1 protein in 20 mM HEPES-NaOH (pH 7.5), 150 mM KCl] and 1 μL of reservoir solution [29% (w/v) PEG MME 2000, 0.1 M Tris-HCl (pH 8.5), 0.2 M Sodium Acetate, 0.2 M KSCN] against 0.5 mL of reservoir solution. Crystals formed from the precipitate after 2 days and grew to a largest dimension of 0.3 mm.
For the cryogenic experiments, a suitable cryoprotectant was determined to be reservoir solution plus 20% (v/v) glycerol. Successful flash freezing was achieved when the crystals were transferred directly from the drop to the cryoprotection solution and allowed to equilibrate for 1 min. A set of native data was collected at beam line NW-AR12 at the Photon Factory, Japan using the X-ray beam at a single wavelength (1.0000 Å) and diffracted up to 1.9 Å. Data sets were indexed and processed using HKL2000.16
The structure was determined by molecular replacement using the program MOLREP.17 For the crossrotation search, the structure of Eco DdlB (PDB ID:2DLN) was used as the search model, and the highest peak of the rotation function was used for the translation function. This model gave a strong single peak in the translation function. Rigid body refinements, simulated annealing, overall anisotropic B-factor and individual restrained B-factor refinements were performed using the program CNS.18 After these refinement steps, an interpretable electron density map was calculated. Many cycles of manual rebuilding, using the program O19 and refinement using CNS yielded a final crystallographic R-value of 22.8% (Rfree = 26.8%). Atomic coordinates have been deposited at the Protein Data Bank under accession code 2FB9.
Results and Discussion.
Overall structure of Tca Ddl: The crystal structure of Tca Ddl was solved by molecular replacement and refined with 1.9 Å resolution diffraction data (Rcryst = 22.8%, Rfree = 26.8%). The Ramachandran plot calculated with the program PROCHECK20 showed no residues with angular values in disallowed areas: 91% of residues were in the most favored regions, while 9% were in allowed regions. The crystals belong to the C2221 space group with one molecule per asymmetric unit. The final model of Tca Ddl includes one protein molecule (amino acid residues 1–322) and 115 water molecules. Data collection and refinement statistics are summarized in Table I.
Table I. Data Collection and Refinement Statistics
Rsym = Σ|〈I〉 − I|/Σ〈I〉.
Rcryst = Σ||Fo| − |Fc||/Σ|Fo|.
Rfree calculated with 10% of all reflections excluded from refinement stages using high resolution data.
Values in parentheses refer to the highest resolution shells.
The structure of the Tca Ddl monomer is composed of 13 β-strands (β1–β13) and 9 α-helices (α1–α9), and can be divided into three domains: an N-terminal domain (residues 1–107), a central domain (residues 108–195) and a C-terminal domain (residues 196–322) [Fig. 1(A) and (B)]. The ATP binding site is situated between the central and C-terminal domains, while the D-alanine:D-alanine binding site is between the N-terminal and central domains. The biologically relevant dimeric structure of Tca Ddl is formed by crystallographic twofold symmetry. The dimer interface involves the interactions between three α-helices (α3, α4, and α5) with a buried surface area of 1371 Å2 (69% nonpolar atoms and 31% polar atoms) per monomer. The abundance of hydrophobic residues (Phe97, Leu100, Val110, Ala111, Ala114, Leu115, Leu121, Val125, and Ala127) in this region suggests the involvement of favorable hydrophobic interactions in the dimerization [Fig. 1(C)].
Comparison of the apo and complexed structures: When the structure of Tca Ddl was superimposed onto those of Eco DdlB (PDB ID: 2DLN) and two D-Alanine:D-Lactate ligase structures (PDB ID:1EHI and 1E4E), major differences were observed around the lid loop region comprised of residues 222–235 [Fig. 2(A)]. The crystal structures of Ddl and the two D-Alanine:D-Lactate ligases revealed that the lid loop adopts a closed conformation (ω-loop conformation) within inhibitor complexes and contains catalytically important residues.12–15 In our Tca Ddl structure (apo-form), by contrast, the lid loop has an open conformation and extends away from the superimposed structures.
Structural superposition for the Tca Ddl apo structure (open form) and Eco DdlB structure (closed form) in complex with ADP/phosphorylated phosphinate (PDB ID:2DLN) gave a root-mean-square deviation (RMSD) of 1.4 Å; over 215 Cα atoms. The most dramatic movement occurred in the lid loop region, where the Cα atom of Tyr232 showed the maximum displacement between the open and closed states (about 16 Å). Notably, by interacting with the transferred phosphoryl group of ATP, this Tyr residue serves a key function during Ddl enzyme catalysis. In the closed Eco DdlB structure, Lys215 and Tyr216 (Lys231 and Tyr232 are the spatially equivalent residues in Tca Ddl) on the lid loop form hydrogen bonds with Ser150 and Ser151 (Ser162 and Ser163 are the spatially equivalent residues in Tca Ddl), respectively. Moreover, the central domain, which contains these two Ser residues, shows rotational movement between the open and closed forms with a total rotation of 17° (determined using the program DynDom21). It thus appears that it is with this rotation of the central domain and the swing motion of the lid loop that Ddl shifts to the closed conformation upon substrate binding [Fig. 2(B)].
In summary, we have resolved the structure of Tca Ddl in the absence of substrates (open conformation). Comparison of the open and the closed states reveals substrate-induced conformational changes, which we have divided into two compartments: the rotational movement of central domain, and the swing motion of the lid loop region. This movement induced by substrate binding causes appreciable conformation change in Ddl and brings the catalytic residues correctly into position to participate in catalysis.
We thank Professor N. Sakabe and Drs. N. Igarashi and N. Matsugaki for their generous support in X-ray data collection at NW-AR12 of Photon Factory (Tsukuba, Japan).