Lipid A is the hydrophobic anchor of lipopolysaccharide in Gram-negative bacteria and is required for growth of most Gram-negative bacteria.1, 2 It is also necessary for maintaining the integrity of the outer membrane as a barrier to toxic chemicals.3, 4 Therefore, the study of the enzymes involved in lipid A biosynthesis would be useful for the development of new antibacterial drugs against Gram-negative bacteria.5 UDP-N-acetylglucosamine acyltransferase (LpxA) is the first enzyme of the lipid A biosynthetic pathway. It catalyzes the transfer of an R-3-hydroxyacyl chain from R-3-hydroxy-acyl carrier protein (ACP) to UDP-N-acetylglucosamine (UDP-GlcNAc) at the glucosamine 3-OH position. Among nine enzymes of lipid A biosynthetic pathway, information on the 3D structure is available on LpxA from Escherichia coli only.6 It is a trimer composed of three identical subunits of 262 residues and contains a left-handed parallel β-helix motif. However, the crystal structure of E. coli LpxA determined at 2.6-Å resolution did not contain any bound ligand and provided little information on the active site. The interaction site of LpxA involved in binding ACP is also unknown. Therefore, further structural data on LpxA will be valuable for a better understanding of the active site and structure-based inhibitor design. Here, we present the crystal structure of LpxA from Heliobacter pylori refined using 2.1-Å data. The sequence identity between H. pylori LpxA (270 residues, 29,855 Da) and that from E. coli is 39.3% over the entire polypeptide chain. Thanks to higher resolution, we could assign solvent molecules as well as bound ions. Further, an extra electron density is present in the putative active site and we tentatively interpret this unknown ligand as a detergent molecule, which seems to mimic the acyl chain of the substrate or the product. On the basis of this observation, together with the location of strictly conserved residues and the highly positively charged surface of the C-terminal helical domain, we propose a model for the complex between LpxA and ACP.
Materials and Methods.
Overexpression of H. pylori LpxA as a fusion with a C-terminal eight-residue tag, its crystallization, and X-ray data collection have been reported elsewhere.7 The structure was solved by the molecular replacement method using the 2.6-Å structure of E. coli LpxA6 [Protein Data Bank (PDB) ID 1LXA] as a search model. The model of H. pylori LpxA (PDB ID 1J2Z) has been refined to crystallographic Rwork and Rfree values of 22.2 and 26.4%, respectively, for reflections with F > 2σ in the resolution range 20–2.1 Å. It consists of 2001 nonhydrogen protein atoms from 259 amino acid residues (residues 2–260) in a monomer, 1 tartrate ion, 2 sulfate ions, 1 detergent molecule (1-s-octyl-β-D-thioglucoside), and 148 water molecules in the asymmetrical unit. One-hundred ninety residues (85.6%) of 222 nonglycine and nonproline residues are in the most favored regions of the Ramachandran plot and 32 residues (14.4%) in the additionally allowed regions. Refinement statistics are shown in Table I. Compared with most other LpxAs, H. pylori LpxA is longer at its C-terminus by about 10 residues. This C-terminal extension (residues 261–270) and eight residues from the C-terminal tag are not visible in the electron density map, presumably because they are disordered in the crystal.
Table 1. Refinement Statistics
Values in parentheses refer to the highest-resolution shell (2.23–2.1 Å).
Rfree is calculated from the randomly selected 10% set of reflections not included in the calculation of Rwork.
H. pylori LpxA is a homo-trimer in its quaternary structure [Fig. 1(A)] and its subunit possesses a highly similar fold as that of E. coli LpxA.6 The root mean square (RMS) difference between H. pylori and E. coli enzymes is 1.22 Å for 229 Cα atom pairs. Each subunit can be divided into two distinct domains. The N-terminal domain (residues 2–186) is folded into the left-handed parallel β-helix motif (LβH), which comprises 28 β-strands. The first 27 β-strands are wound in 9 helical turns and the last β-strand is only one third of a helical turn [Fig. 1(B)]. The fourth turn has a 12-residue insertion between strands β11 and β12, and the fifth turn has another 7-residue insertion between β14 and β15, respectively. Theses insertions cover the strictly conserved residues (His118, His121, His140), which could possibly play important catalytic roles. Other turns except the ninth turn comprise exactly 18 amino acid residues in three β-strands. Each turn of the β-helix is approximately triangular and three β-helix motifs pack nicely into a trimer. The trimer formation is critical for the catalytic function of LpxA, as discussed below.
The C-terminal helical domain (residues 187–260) contains four α-helices. Three C-terminal domains in a trimer do not contact each other and they provide a highly positively charged surface at the bottom side of the trimeric enzyme [Fig. 1(C)]. Because H. pylori and E. coli ACPs are highly acidic with calculated pI values of 3.85 and 3.98, respectively, we suggest that this positively charged surface at the bottom of the trimer is likely to be the binding site for ACP. In a crude docking attempt, three molecules of E. coli ACP8 can be positioned nicely into three identical pockets, each of which is formed between the two subunits. The proposed binding site of ACP is further supported by the distribution of strictly conserved residues. Two strictly conserved residues, Asn194 and Arg199, are located at N-terminal and C-terminal ends of the helix α1 in the C-terminal domain, respectively. Their side-chains point toward the proposed ACP binding site, which is adjacent to the putative catalytic site, as defined by other strictly conserved residues (discussed below). In the proposed binding model, Arg199 is in proximity of a highly negatively charged surface patch of ACP. The residues of the C-terminal domain on the other side of the proposed ACP binding pocket are not highly conserved and it appears that ACP interacts mainly through the face of the binding pocket lined with strictly conserved Asn194 and Arg199.
Among 26 LpxA sequences, 11 residues are strictly conserved: 9 residues (Gly53, Gly83, Glu90, Gly108, His118, His121, Gly139, His140, Gln157) in the N-terminal β-helix domain and 2 (Asn194, Arg199) in the C-terminal helical domain. Four conserved glycine residues are most likely to play a structural role. The role of the strictly conserved Glu90 is unclear from the structure; it is located at the subunit interface and is a little separated from residues His118, His121, His140, and Gln157, which are clustered underneath the insertion loops of the β-helix motif. Three histidine residues, His118(His122), His140(His144), and His156(His160), were implicated for the substrate binding through chemical modification and site-directed mutagenesis of E. coli LpxA.9 The corresponding residues of E. coli LpxA are given in parentheses. In addition, Lys72(Lys76) and Arg199*(Arg204*) have also been suggested to play some roles in substrate binding. An asterisk after the residue number denotes that the residue comes from a neighboring subunit. Lys72(Lys76) and His156(His160) are less strictly conserved and are likely to play a minor role in substrate binding. His121(His125) and Gln157(Gln161) are likely to play an important role in either catalysis or substrate binding. His121(His125) was proposed to act as a general base to increase the nucleophilicity of the glucosamine 3-OH of UDP-GlcNAc during its attack on the thioester carbonyl of the acyl-ACP donor.9 As mentioned above, Asn194*(Asn199*) and Arg199*(Arg*204) from the C-terminal α-helical domain of the neighboring subunit are likely to be important in recognizing the holo-ACP, in particular its conserved features such as 4′-phosphopantetheine group and a negatively charged surface patch.
An extra electron density for an unknown ligand is present in the putative active site of H. pylori LpxA. We tentatively interpret it as a molecule of the detergent 1-s-octyl-β-D-thioglucoside, which was added to the hanging drop and was found to be necessary for crystallization.7 The strictly conserved residues His118, His121, His140, and Gln157 point toward the extra electron density and they clearly define the location of the catalytic site. The last carbon atom at the end of the octyl group in the detergent is ≈10 Å from the Cα atom of Gly169*. When the octyl chain is extended by six more carbon atoms, the distance becomes ≈2.5 Å. Gly173 of E. coli LpxA and Met169 of Pseudomonas aeruginosa LpxA, corresponding to Gly169 of H. pylori LpxA, were found to determine the specificity for the acyl chain length.10E. coli LpxA is selective for the 14-carbon myristoyl chain, whereas P. aeruginosa LpxA is selective to the 10-carbon decanoyl chain.10 The side-chain oxygen atom of Gln157 lies ≈4 Å from the C3 atom of the octyl group. This indicates that Gln157 may be important in recognizing the hydroxyl group of the R-3-hydroxymyristoyl chain. And, H. pylori LpxA is most likely to prefer the myristoyl chain like E. coli LpxA.
The authors thank Prof. N. Sakabe and his staff for assistance during data collection at beamline BL-18B of Photon Factory, Japan. This work was supported by the Center for Functional Analysis of Human Genome (21st Century Frontier Program) of the Korea Ministry of Science and Technology. B.I.L. is supported by the BK21 Fellowship.