STRUCTURE NOTE Crystal Structure of Escherichia coli DkgA, a Broad- Specificity Aldo-Keto Reductase Sandra Jeudy, Vincent Monchois, † Caroline Maza, Jean-Michel Claverie, and Chantal Abergel * Information Ge ´nomique et Structurale, CNRS, Marseille, France Key words: structural genomics; YqhE; NADPH oxy- doreductase; methylglyoxal detoxifica- tion; antibacterial target Introduction. The Structural and Genomics Informa- tion Laboratory is involved in a Structural and Functional Genomics program (BIGS; http://www.igs.cnrs-mrs.fr/ str_gen/) aiming at the discovery of new antibacterial targets among proteins that are ubiquitous in bacterial pathogens, exhibiting good sequence conservation, but whose precise biochemical or cellular functions remain unknown. Comprehensive bioinformatics and comparative genomics analyses were performed according to these criteria, resulting in the selection of 110 Escherichia coli candidate genes submitted to a structural genomics pipe- line, the details of which have been previously described. 1 Here we report the crystal structure of E. coli protein DkgA, product of the gene formerly referred to as yqhE (EcoGene accession number: EG13015). This 275-residue protein belongs to the aldo-keto reductase (AKR) family [Enzyme Commission (EC): 1.1.1.-]. AKRs constitute a group of diverse cytosolic, nicotinamide adenine dinucle- otide phosphate [NAD(P)H]-dependent oxidoreductases, that catalyze the reduction of a wide range of substrates, such as aldoses, aldehydes, steroids, or monosaccharides. The precise physiological role of most of these enzymes remains unknown, although they have been collectively associated with detoxification of a broad range of sub- strates, 2 including the toxic metabolite methylglyoxal. 3 The AKR family is represented in most living cells as monomeric (/) 8 -barrel proteins about 300 amino acids in length, exhibiting a catalytic tetrad composed of a ty- rosine, a lysine, an aspartate, and a histidine. Despite their common /-barrel fold, a conserved cofactor binding site, and the catalytic tetrad, members of the AKR family often exhibit little sequence similarity, in some cases sharing less than 10% identical residues. It has been proposed that the difference in substrate specificity exhib- ited by AKR family members is linked to the length and sequence of several highly variable loops in their three- dimensional (3D) structures. 4 The E. coli DkgA protein was previously shown to catalyze the reduction of 2,5-diketo-D-gluconic acid to 2-keto-L-gulonic acid, a key step in the biosynthesis of ascorbic acid. 5 Because of its industrial interest for the production of vitamin C, the homologous enzyme (EC: 1.1.1.274) in Corynebacterium has been the target of detailed biochemical and structural studies. 6–8 Subse- quently, DkgA was also found to be capable of stereoselec- tive -keto ester reductions on ethyl acetoacetate and other 2-substituted derivatives. 9 The 3D structure of the E. coli DkgA protein presented here was solved at 2.16 Å resolution by molecular replace- ment with the structure of the Corynebacterium homolog sharing 51% identical residues. Using NADPH as a cofac- tor, the recombinant DkgA protein was found to be active against a wide range of substrates, including methylg- lyoxal. Materials and Methods. Cloning, expression and purifi- cation. The gene encoding DkgA was amplified from E. coli K12 genomic DNA and directional cloning was per- formed using the Gateway system (Invitrogen). The poly- merase chain reaction (PCR) product was inserted by homologous recombination in the pDEST17 expression plasmid in phase with a N-terminal His 6 -tag, under the control of a T7 promoter. After transformation into DH5 cells, the purified plasmids were used for the overexpres- sion of the recombinant proteins using our soluble expres- sion screening protocol previously described. 1 The best result was obtained with E. coli strain BL21(DE3)pLysS. Cells were grown on 2  yeast-tryptone (2YT) medium containing ampicillin and chloramphenicol at 37°C and the pellet was resuspended in a 50 mM sodium phosphate, 300 mM NaCl buffer, pH 8.0 (buffer A), containing 0.1% Triton X-100 and 5% glycerol, before sonication. The cleared lysate was then applied to a 5 mL HiTrap Chelat- ing Column (Pharmacia) charged with Ni 2+ and equili- brated with buffer A. After sample injection, the column Grant sponsor: French Ministry of Industry; Grant number: 4906088. *Correspondence to: Chantal Abergel, Structural and Genomic Information Laboratory, CNRS-UPR2589, IBSM, 31 chemin Joseph Aiguier, Marseille, 13402 cedex 20, France. E-mail: chantal.abergel@igs.cnrs-mrs.fr † Present address: Protein’eXpert SA, 15 Rue des Martyrs, 38000 Grenoble, France Received 24 March 2005; Accepted 10 July 2005 Published online 11 November 2005 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/prot.20710 PROTEINS: Structure, Function, and Bioinformatics 62:302–307 (2006) © 2005 WILEY-LISS, INC.