Crystal structure of the Escherichia coli shikimate kinase I (AroK) that confers sensitivity to mecillinam

Escherichia coli shikimate kinase I (SKI, AroK; EC 2.7.1.71; 173 residues; predicted molecular mass = 19,538 Da) catalyzes the fifth step in the seven-step chorismate biosynthesis pathway unique to microorganisms, plants, and parasites.1-4 AroK is a member of the nucleoside monophosphate (NMP) kinase family and has been shown to convert shikimic acid to shikimate-3-phosphate using adenosine triphosphate (ATP) as a cosubstrate. In E. coli, phosphorylation of shikimic acid is carried out by two isozymes: shikimate kinase I encoded by the aroK gene and shikimate kinase II (SKII, AroL) encoded by the aroL gene. Both genes are located in different parts of the bacterial chromosome and form parts of distinct operons.1, 3, 5, 6 Expression of the aroK gene, unlike that of aroL (controlled by the TyrR and TrpR repressors1), appears to be constitutive and the enzyme has approximately 100-fold lower affinity for shikimic acid than SKII with an estimated Km of 20 mM for AroK and 200 μM for SKII.7 AroK can functionally substitute for SKII, but double mutants in aroK and aroL are aromatic amino acid auxotrophs.8 These observations suggest that SKII is the dominant enzyme of the pathway, with AroK playing a secondary role and possibly participating in an as yet unidentified cellular process.9 It is remarkable that loss of wildtype AroK activity confers resistance to mecillinam [a bulky β-lactam antibiotic that binds to E. coli penicillin-binding protein 2 (PBP2)10], which is separable from the shikimate kinase activity of the enzyme.11 Although mecillinam sensitivity appears to be associated with AroK and not AroL, it may not be the primary function of the AroK protein. A number of other bacteria, for example, Haemophilus influenzae and Mycobacterium tuberculosis, encode only AroK but not AroL homologs. X-ray structure determination of AroK revealed a compact α/β protein (molecular dimensions 39 × 35 × 46 Å) that is structurally similar to Erwinia chrysanthemi SKII.12, 13 The linear arrangement of the secondary structural elements within the polypeptide chain is β1, αA, β2, αB, αC, αD, β3, 310, αE, β4, αF, αG, β5, and αH (Fig. 1). The central part of the molecule, composed of a five-stranded parallel β-sheet (β5–β4–β1–β3–β2), is sandwiched between five α-helices [Fig. 2(a)]. Alignment of proteins similar to E. coli AroK identified in an iterative PSI-BLAST30 search (14 iterations; December 2001). Secondary structural elements are shown above the aligned sequences, and the locations of linear motifs are labeled. Grey spheres indicate disordered residues. Color-coding denotes sequence conservation (white-to-green ramp, 30–100% identity). AK_ECOLI (GI:728898), E. coli shikimate kinase I; AK_SALTY (GI:16762800), Salmonella typhimurium shikimate kinase I (Z-score = 3e-57; 97% identity); AK_VIBCH (GI:15642624), Vibrio cholerae shikimate kinase I (Z-score = 6e-55; 81% identity); AK_MYCTU (GI:2492975), Mycobacterium tuberculosis shikimate kinase I (Z-score = 7e-41; 38% identity); AK_AQUAE (GI:7673950), Aquifex aeolicus shikimate kinase I (Z-score = 3e-41; 35% identity); AL_ECOLI (GI:114198), E. coli shikimate kinase II (Z-score = 4e-44; 31% identity); AL_ERWCH (GI:114199), E. chrysanthemi shikimate kinase II (Z-score = 2e-43; 33% identity); A1_EMENI (GI:3834343), amino acids 863–1053 of the pentafunctional arom polypeptide from Emericella nidulans encoding the shikimate kinase domain (28% identity; Z-score = 5e-35). (a). RIBBONS31 drawing of AroK with labeled N- and C-termini and secondary structural elements. Selected residue numbers are also provided. (b). PyMOL32-generated structure-based superposition of ADP from the structure of E. chrysanthemi SKII (PDB ID 2SHK) with residues important for metal coordination and nucleotide binding shown as atomic stick figures and labeled. A cocrystal structure of E. chrysanthemi SKII and shikimate did not reveal a recognizable electron density feature corresponding to shikimate, but made it possible to implicate several charged residues in shikimate binding. All residues identified in this manner are structurally equivalent to those in AroK (Asp34, Arg58, Glu61, Gly78–80, and Arg139 in E. chrysanthemi SKII correspond to Asp36, Arg60, Glu63, Gly81–83, and Arg140 in E. coli AroK). The putative shikimate-binding pocket is lined with additional hydrophobic side-chains that are likely to stabilize the binding of shikimate. One of these substrate-binding site residues (Leu83 in E. chrysanthemi) is replaced with Lys86 in E. coli AroK [Fig. 3(a)]. We propose that loss of a stabilizing hydrophobic residue at this position may explain the significantly lower affinity of E. coli AroK for shikimic acid as compared to that of SKII. (a). A cut-away view of the putative AroK shikimate-binding pocket. For clarity, only residues 7–87 and 121–148 of AroK are displayed. (b). GRASP33 representation of the cut-away view calculated using a water probe radius of 1.4Å. The surface electrostatic potential is colored red and blue, representing electrostatic potentials from −20 to 20 KBT, where KB is the Boltzmann constant and T is temperature in Kelvin. Calculations were performed with an ionic strength of 0 and dielectric constants of 80 and 2 for solvent and protein, respectively. E. coli AroK contains motifs found in E. chrysanthemi SKII and many nucleotide-binding proteins (Fig. 1): Walker A- (11–18: Gly–Pro–Met–Gly–Ala–Gly–Lys–Ser; consensus Gly–XXXX–Gly–Lys–Thr/Ser, where X is any residue) and B-motifs [77–82: Val–Leu–Ala–Thr–Gly–Gly; consensus ZZ–Asp–XX–Gly, where Z is a hydrophobic residue; in E. coli AroK the Asp residue is replaced with Ala; the function of the consensus Asp residue to coordinate Mg2+ is taken over by Asp34 and Asp36; see Fig. 2(b)].12, 14. The adenine-binding motif found in the C-terminal portion of AroK differs significantly from the consensus sequence (Val/Ile–Asp–Ala–X–Gln/Asn–X–Pro) found in bacterial SKIIs and adenylyl kinase type-II isoenzymes. Although in AroK the corresponding polypeptide chain segment (152–159: Ile–Arg–Thr–Asp–Asp–Gln–Ser–Ala), which was not visible in our electron density maps, lacks several consensus residues, it is possible that this loop participates in adenine binding because nucleotide recognition depends solely on backbone interactions.15 Finally, the N-terminus of αG in AroK contains a stable proline box motif (Z–Pro–XXZZ)16 encompassing residues 128–133 in AroK (Pro–Pro–Arg–Glu–Val–Leu) that is absent in the structure of E. chrysanthemi SKII. The NMP kinases contain “lid” domains that are thought to undergo large conformational changes during catalysis. In AroK, this region spans residues 115–126 (Figs. 1 and 3). A single-point mutation in AroK11 [Leu133Pro in the proline box at the junction between αG and the “lid”; Fig. 3(a)] abolishes mecillinam sensitivity without affecting shikimate kinase activity. The Leu133Pro substitution may introduce a kink in the helix17 and in combination with Pro128 and Pro129 increase the rigidity of the “lid” domain. We propose that the phenotype of this mutation results from perturbation of the hydrophobic environment necessary for mecillinam binding or by impairing opening of the “lid” to accommodate mecillinam but not the smaller shikimate substrate. A DALI18 search of the Protein Data Bank (http://www.rcsb.org/, November 2001) with the atomic coordinates of AroK identified 254 structurally similar proteins with Z scores > 2.0. Six of these proteins are closely related to AroK (Z-score > 9.0): E. coli shikimate kinase II [PDB ID 1SHK; Z-score = 23.9; root mean square deviation (RMSD). = 1.4 Å for 149 α-carbon pairs; 34% identity), Bacillus stearothermophilus adenylate kinase (PDB ID 1ZIN; Z-score = 10.8; RMSD = 3.0 Å for 134 α-carbon pairs; 17% identity), Streptomyces venezuelae chloramphenicol phosphotransferase (PDB ID 1QHS; Z-score = 10.4; RMSD = 3.1 Å for 137 α-carbon pairs; 21% identity), Sulfolobus acidocaldarius adenylate kinase (PDB ID 1NKS; Z-score = 10.4; RMSD = 3.5 Å for 143 α-carbon pairs; 15% identity), Penicillium chrysogenum adenosine-5'-phosphosulfate kinase (PDB ID 1D6J; Z-score = 9.9; RMSD = 3.1 Å for 122 α-carbon pairs; 23% identity), and E. coli cytidine monophosphate kinase (PDB ID 1CKE; Z-score = 9.1; RMSD = 3.7 Å for 127 α-carbon pairs; 17% identity). Automated homology modeling with MODPIPE19 using AroK as a template yielded 99 models of both prokaryotic and eukaryotic proteins (model score > 0.7, model length > 89 residues), which can be subdivided into two categories. One group contains models comparable in length to AroK that represent other shikimate, adenylate, cytidylate, and adenylylsulfate kinases, plus some small ABC transporter proteins. The second group contains 22 models of multidomain proteins with length > 300 residues known to bind and process nucleotides. Further biochemical and biophysical studies of the AroK protein will be required to establish its precise functional role in bacteria, which should be facilitated by the availability of an X-ray structure and a detailed analysis of results from structural studies of other AroK proteins already in progress.20, 21 The aroK open reading frame (ORF) was amplified by polymerase chain reaction (PCR) using the forward primer containing a BamHI restriction site (GTCGTGGATCCATGGCAGAGAAACGCAATATCTTTCTG), a reverse primer containing an XhoI site (GTCGTCTCGAGTCAGTTGCTTTCCAGCATGTGAATAATCTG), and E. coli UT5600 DNA as a template using standard protocols.22 The amplified insert was cloned into the corresponding sites of the pGEX-6P-1 plasmid. S–Met and Se–Met glutathione S-transferase (GST)-tagged proteins were expressed in E. coli BL21 (overnight induction at 18°C). Proteins were purified on glutathione and Sepharose Q resins following established procedures.23 Proteins for crystallization were dialyzed against 20 mM HEPES pH 7.5, 100 mM potassium chloride, and 3 mM dithiothreitol, concentrated to 10 mg/mL, and filtered through a 0.1-μm filter. Gel filtration and dynamic light scattering experiments documented that AroK is monomeric in solution (data not shown). MALDI-MS confirmed the identity of the purified recombinant AroK (measured mass = 19,982.52 ± 15 Da, predicted mass = 19,949.5 Da; the measured peak is likely to represent a sodium adduct of the protein). Diffraction-quality Se–Met and S–Met AroK crystals (rectangular blocks) were obtained by hanging-drop vapor diffusion at room temperature against a reservoir containing 0.1 M MES pH 6.5, 0.2 M ammonium sulfate, and 30% monomethyl ether of polyethylene glycol 5000. Crystals were cryoprotected by transfer to the mother liquor supplemented with 20% glycerol for 15–30 s and immersion in liquid propane (see Table I for space group and unit cell dimensions). Diffraction data were collected under standard cryogenic conditions with a MARCCD detector on Beamline X9A (National Synchrotron Light Source, Brookhaven National Laboratory), processed and scaled using Denzo/Scalepack.24 The structure of the Se-Met protein was determined with data from a two-wavelength anomalous diffraction experiment.25 Four selenium positions were located with SnB 2.126 and refined using MLPHARE24 (figure of merit = 0.434). Density modification of the MLPHARE-refined phases yielded a high-quality map suitable for automated model building. ARP/wARP27 built 169 residues distributed into 17 fragments that encompassed both molecules comprising the asymmetrical unit. Following partial refinement in CNS 1.0,28 missing residues were added manually with O.29 There was poor or no electron density for residues 1–2, 115–120, 153–157, and 172–173 in molecule A and for residues 1–2, 113–127, 153–157, and 172–173 in molecule B, and they were therefore omitted from refinement. The final model, consisting of 305 of 346 residues and 258 water molecules, was refined against 2.05-Å resolution data obtained from an S–Met crystal to an R factor of 21.7% and an Rfree value of 27.1% (see Table I for a summary of X-ray data and refinement statistics). Refined atomic coordinates and structure factors have been deposited in the Protein Data Bank (PDB ID 1KAG). The authors thank Dr. K. Rajashankar from BNL for help with data collection and Drs. J.B. Bonanno, D. Jeruzalmi, and S.A. Gibney for useful discussions. This work was supported by NIGMS Grants P50-GM62529 (S.K.B.) and GM20276 (M.J.R.). S.K.B. is an Investigator in the Howard Hughes Medical Institute. The E. coli AroK protein represents target T535 from the New York Structural Genomics Research Consortium.