Crystal structure of peptide deformylase from Staphylococcus aureus in complex with actinonin, a naturally occurring antibacterial agent



Bacterial peptide deformylase (PDF; EC represents an attractive target for developing new antibacterial agents because it is essential in prokaryotes and is conserved throughout eubacteria.1–3 It catalyzes the removal of the formyl group from the N-terminal methionine residue of newly synthesized polypeptides. Removal of the N-formyl group from the nascent polypeptide chain is required for proper folding and function of proteins in eubacteria, whereas in eukaryotes formylation and deformylation appear to occur only in some organelles.4 Bacterial PDFs fall into class I or class II, and structural data on both classes are available. Staphylococcus aureus, a Gram-positive pathogen, is a major cause of hospital- and community-acquired infections. Development of methicillin-resistance and, more recently, vancomycin-resistance in S. aureus poses a serious health problem. Structure-based discovery of new antibacterial agents against S. aureus would be aided by detailed structural information about the mode of inhibitor binding to its active site. Actinonin, a naturally occurring hydroxamic acid pseudopeptide, is a potent inhibitor of bacterial PDFs.5 Until now, no structure has been reported for complexes of S. aureus PDF (SaPDF; class II PDF) with any inhibitors. In order to provide the missing structural data, we have determined the crystal structure of SaPDF in complex with its inhibitor actinonin at 1.90 Å resolution.

Materials and Methods.

A recombinant SaPDF fused with a C-terminal eight-residue tag [LEHHHHHH] was overexpressed in E. coli C41(DE3) cells and purified by two chromatographic steps involving affinity and size exclusion columns. Its crystals were grown by the hanging drop vapor diffusion method at 297K in the presence of 6.8-fold molar excess actinonin. The protein solution was 30 mg/mL in 25 mM Tris-HCl buffer at pH 7.5 containing 200 mM sodium chloride, while the reservoir solution consisted of 23% (w/v) PEG 4000, 50 mM Tris-HCl (pH 8.5), 15% (v/v) glycerol, 100 mM MgCl2 and 20 mM CaCl2. These conditions are similar to but not identical to those reported previously by another group.6 X-ray diffraction data were collected at 100K on a Quantum 4R CCD detector at BL-38B1 of SPring-8.

The crystals belong to the orthorhombic space group C2221 with cell parameters a = 94.42 Å, b = 120.84 Å, c = 48.06 Å and α = β = γ = 90°. These cell parameters are similar to those reported previously.6 A monomer is present in each asymmetric unit, with a corresponding crystal volume per protein mass (VM) of 3.33 Å3/Da and a solvent content of 59.5%. By using the model of SaPDF [Protein Data Bank (PDB) code 1LQW; A chain]7 as the probe, the structure of SaPDF was solved by molecular replacement and refined using the CNS program. The refined model (PDB code 1Q1Y) consists of 186 amino acid residues of a monomer (residues 1–183 of SaPDF plus the first three residues of the C-terminal tag), a molecule of actinonin, a metal ion (Zn2+) and 288 water molecules. Cys111 in the active site is oxidized to sulfinic acid (Cys-SO2H). When we tried the model refinement with cysteine sulfonic acid (Cys-SO3H) as the oxidized Cys111, a large negative peak appeared around the SG atom of Cys111 in an |Fo| − |Fc| map, confirming that sulfinic acid is correct. Metal analysis by Inductively Coupled Plasma-Atomic Emission Spectrometry indicated that the recombinant enzyme is bound with mostly zinc ions. The crystallographic R- and Rfree-values were found to be 20.5% and 23.5%, respectively, for the resolution range 20.0–1.90 Å. All of the non-glycine residues are located in the most favored region of the Ramachandran plot. Refinement statistics are summarized in Table I.

Table I. Refinement Statistics
  • a

    Values in parentheses refer to the highest resolution shell (2.02–1.90 Å and 1.97–1.85 Å for SaPDF and PaPDF, respectively).

  • b

    Rfree is calculated from the randomly selected 10% set of reflections not included in the calculation of Rwork.

Resolution range (Å)20.0–1.9020.0–1.85
Completeness (%)96.9 (85.6)a90.7 (83.6)a
No. of reflections used21,219 (2797)a59,330 (8287)a
No. of residues186338
No. of water molecules288304
No. of actinonin/Zn2+ ions1/12/2
Rwork/Rfreeb (%)20.5 (22.1)a/23.5 (27.1)a18.5 (19.0)a/21.4 (21.9)a
RMSD from ideal geometry  
 Bond lengths (Å)/bond angles (°)0.005/1.30.005/1.4
Average B-factor (Å2)  
 Protein atoms18.119.4
 Water molecules32.131.3
 Actinonin/Zn2+ ions34.0/29.324.2/17.1

Crystallization of PaPDF in the presence of actinonin and X-ray data collection have been reported previously.8 The structure of PaPDF was solved independently by molecular replacement using the model of E. coli PDF (EcPDF; PDB code 1BS4) as the probe, since no structure of PaPDF was available at the time when this work was completed. The refined model (PDB code 1IX1) consists of 338 amino acid residues of the two PDF chains (residues 2–168 of PaPDF plus the first two residues of the C-terminal tag), two molecules of actinonin, two Zn2+ ions and 304 water molecules in the asymmetric unit. The crystallographic R- and Rfree-values are 18.5% and 21.4%, respectively, for the resolution range 20.0–1.85 Å (Table I). All of the non-glycine residues are in the most favored region of the Ramachandran plot.

Results and Discussion.

The two independent monomers of PaPDF in the asymmetric unit are essentially identical to each other; they superimpose with a root mean square deviation (RMSD) of 0.26 Å for 167 Cα atoms. When we compared our PaPDF structure (PDB code 1IX1; A and B chains) to the PaPDF–actinonin complex model (1LRY),7 we obtained RMSD values of 0.38 and 0.37 Å for 165 Cα atom pairs, respectively. Our PaPDF structure refined at a 1.85 Å resolution represents a significant improvement in the accuracy of the actinonin-bound structure over the previously reported structure refined at 2.6 Å (1LRY).7 Superposition of our PaPDF model with A and B chains of the inhibitor-unbound PaPDF (containing the Asp84Glu mutation) determined at 1.8 Å (1N5N)9 gives RMSD values of 0.22-0.40 Å for 164 common Cα atoms. Our study largely confirms a previous report on the actinonin complex of PaPDF,7 with the modes of actinonin binding in the active site highly similar to each other. The metal ion is coordinated by Cys92, His134, His138 and the hydroxamate group of actinonin, and the sulfhydryl group of Cys92 is not oxidized in the crystal structure.

Despite a low level (31.4%) of sequence identity between SaPDF and PaPDF [Fig. 1(a)], their overall structures are similar, except at their C-termini. A long C-terminal helix in class I PaPDF is replaced by a loop in class II SaPDF [Fig. 1(b)]. Here we provide the first structural data on the actinonin complex of SaPDF. Until now, three crystal structures of SaPDF have been reported, but none of them are complexed with inhibitors such as actinonin. The first structure (1LQW; A and B chains)7 contains a Zn2+ ion in the active site but is inhibitor free. The active site cysteine Cys111 is not oxidized. The second structure (1LMH)10 is that of the selenomethionine-substituted enzyme carrying the R127K mutation, fused with a C-terminal tag (containing the H186Q mutation). It is also inhibitor free and the metal ion was modeled as Zn2+, with Cys111 not oxidized. The third structure (1LM4; A and B chains)9 was solved for the enzyme containing 12 extra residues (MGSDKIHHHHHH) at the N-terminus, and the bound metal ion was modeled as Fe3+. Part of the N-terminal expression tag (GSDKIH) of chain B is visible in the active site of chain A, mimicking the binding of a peptide substrate. One glycerol molecule was also identified in close vicinity to the active site of chain B. In both chains A and B, Cys111 was oxidized to sulfinic acid (Cys-SO2H). Superposition of our refined SaPDF model with A and B chains of 1LQW7 gave RMSD values of 0.34 and 0.48 Å for 183 common Cα atoms, respectively. Superposition against 1LMH10 gave a RMSD 0.15 Å for 184 common Cα atoms. Superposition against the A and B chains of 1LM49 gave RMSD values of 0.36 and 0.40 Å for 183 common Cα atoms, respectively. The results of these structural comparisons indicate that SaPDF undergoes little conformational change upon binding of actinonin. In the case of 1LMH,10 our structure provides evidence that the R127K mutation that was introduced inadvertently during Polymerase Chain Reaction (PCR) did not alter the structure significantly. Interestingly, this model with essentially the same C-terminal tag as our SaPDF construct shows the smallest deviation from our model; in the case of 1LM4,9 the small but detectable structural deviation may be caused by the N-terminal 12-residue tag introduced into the recombinant protein. This study shows that the binding modes of actinonin in SaPDF and PaPDF are highly similar to each other, despite the difference in oxidation states of the metal-coordinating cysteine residues [Fig. 1(c)]. Cys111 (OD1: 3.27 Å, OD2: 2.77 Å), His154 (NE2: 2.10 Å), His158 (NE2: 2.22 Å) and the hydroxamate group (O1: 2.36 Å, O2: 2.47 Å) of actinonin coordinate the Zn2+ ion in SaPDF. The electron density clearly indicates that Cys111 is oxidized to sulfinic acid (Cys-SO2H). Furthermore, the OD1 (3.27 Å) and OD2 (2.77 Å) of Cys111 sulfinic acid are close to the metal ion in SaPDF. It is not clear why the active site cysteine residue of SaPDF are more susceptible to oxidation than those of PaPDF.

Figure 1.

(a) Amino acid sequence alignment of seven peptide deformylases. Strictly conserved residues and semi-conserved residues are colored pink and yellow. Blue boxes represent three conserved motifs. The residue numbers are for SaPDF. SA is for peptide deformylase from S. aureus (PDB codes 1Q1Y, 1LM4), PA for P. aeruginosa (1IX1, 1LRY, 1N5N), EC for E. coli (1BSZ), TM for T. martima (1LME), PF for P. falciparum (1JYM), BS for B. stearothermophilus (1LQY) and SP for S. pneumoniae (1LM6). (b) Stereo view of the superimposed Cα backbones of SaPDF (blue) and PaPDF (orange). The spheres indicate Zn2+ ions (SaPDF, cyan; PaPDF, magenta). (c) Stereo view of the superposition of the active sites of SaPDF and PaPDF with bound actinonin. Cα backbones of SaPDF and PaPDF are shown in blue and orange lines, respectively. The inhibitor actinonin (SaPDF, cyan; PaPDF, magenta) and the Zn2+ ion (SaPDF, cyan; PaPDF, magenta) are drawn in ball-and-stick representation.


We report the crystal structure of SaPDF in complex with its inhibitor actinonin. The metal-coordinating Cys111 of SaPDF is oxidized to sulfinic acid. Despite the oxidation of this cysteine residue, actinonin is bound to the active site in a manner highly similar to that for PaPDF, in which the equivalent Cys92 is not oxidized. This study provides new structural data about the inhibitor-complex of SaPDF, which will be useful for antibacterial drug discovery efforts against S. aureus.


This work was supported by a grant from the Korea Science and Engineering Foundation (R03-2003-000-10031-0) to HJY and the Korea Ministry of Science and Technology (NRL-2001, grant no. M10318000132) to SWS. HJY, HLK, SKL, HWK, and HWK are recipients of BK21 fellowships.