Single Amino Acid Substitution in Bacillus sphaericus Phenylalanine Dehydrogenase Dramatically Increases Its Discrimination between Phenylalanine and Tyrosine Substrates Stephen Y. K. Seah, ‡,⊥ K. Linda Britton, § David W. Rice, § Yasuhisa Asano, | and Paul C. Engel* ,‡ Department of Biochemistry and Conway Institute of Biomolecular and Biomedical Research, UniVersity College Dublin, Belfield, Dublin 4, Ireland, Krebs Institute for Biomolecular Research, Department of Molecular Biology and Biotechnology, UniVersity of Sheffield, P.O. Box 594, Sheffield S10 2UH, United Kingdom, and Biotechnology Research Center, Toyama Prefectural UniVersity, 5180 Kurokawa, Kosugi, Toyama 939-03, Japan ReceiVed March 12, 2002; ReVised Manuscript ReceiVed May 20, 2002 ABSTRACT: Homology-based modeling of phenylalanine dehydrogenases (PheDHs) from various sources, using the structures of homologous enzymes Clostridium symbiosum glutamate dehydrogenase and Bacillus sphaericus leucine dehydrogenase as a guide, revealed that an asparagine residue at position 145 of B. sphaericus PheDH was replaced by valine or alanine in PheDHs from other sources. This difference was proposed to be the basis for the poor discrimination by the B. sphaericus enzyme between the substrates L-phenylalanine and L-tyrosine. Residue 145 of this enzyme was altered, by site-specific mutagenesis, to hydrophobic residues alanine, valine, leucine, and isoleucine, respectively. The resultant mutants showed a high discrimination, above 50-fold, between L-phenylalanine and L-tyrosine. This higher specificity toward L-phenylalanine was due to K m values for L-phenylalanine lowered more than 20-fold compared to the values for L-tyrosine. The greater specificity for L-phenylalanine in the wild-type Bacillus badius enzyme, which has a valine residue in the corresponding position, was also found to be largely due to a lower K m for this substrate. Activities were also measured with a range of six amino acids with aliphatic, nonpolar side chains, and with the corresponding oxoacids, and in all cases the specificity constants for these substrates were increased in the mutant enzymes. As with phenylalanine, these increases are mainly attributable to large decreases in K m values. The label phenylalanine dehydrogenase (PheDH 1 ) (EC 1.4.1.20) denotes a class of enzymes that catalyze the reversible NAD + -dependent oxidative deamination of a broad range of hydrophobic amino acid substrates, with particular preference for aromatic side chains. The enzyme from Bacillus sphaericus is unusual within this class since its activities toward L-phenylalanine and L-tyrosine are similar (1, 2). In contrast, PheDH from Rhodococcus sp. M4 shows 34-fold discrimination in favor of L-phenylalanine (3). Similarly, with the enzymes from Bacillus badius and Sporosarcina ureae, tyrosine gives only 9 and 5.4%, respectively, of the specific activity with phenylalanine (1, 4). It is of interest to understand the structural basis for this difference in discrimination between aromatic amino acid substrates that differ by only a single hydroxyl substituent. Determining the molecular details of substrate recognition in the family of amino acid dehydrogenases has also practical relevance, as these enzymes are useful as biocatalysts for stereospecific synthesis of amino acids, and as diagnostic reagents for a variety of inborn errors of amino acid metabolism. Indeed, the thermostability of PheDH from B. sphaericus would recommend its use as a diagnostic reagent. However, major interference from tyrosine makes it unsuitable for measuring serum L-phenylalanine levels in the diagnosis of phenylketonuria (PKU) (2). The rational design of enzymes with altered substrate specificities for use in chemical synthesis or clinical diagnosis is a major goal of our protein engineering efforts. Several notable examples of successful redesign of substrate speci- ficities depend on prior knowledge of the high-resolution structure of the enzyme, or at least of the corresponding enzyme from another biological species. In the absence of such information, site-directed mutagenesis may be guided by patterns of amino acid conservation, by the results of chemical modification studies, or by molecular modeling if the structure of a related, homologous protein is known. Homology-based modeling has been fruitful, for example, in the redesign of coenzyme specificity in glutathione reductase and lipoamide dehydrogenase (5, 6). A critical question remains, however, as to the level of amino acid sequence similarity that might justify such an approach. Across the family of amino acid dehydrogenases, similarity in terms of amino acid sequence identity is not strikingly high. For instance, the overall sequence identity between * Corresponding author. Phone 00 353 1 716 1547; Fax 00 353 1 283 7211; E-mail paul.engel@ucd.ie. ‡ University College Dublin. § University of Sheffield. | Toyama Prefectural University. ⊥ Present address: Department of Microbiology, University of Guelph, Guelph, Ontario, Canada. 1 Abbreviations: PheDH, phenylalanine dehydrogenase; LeuDH, leucine dehydrogenase; GluDH, glutamate dehydrogenase; NAD + , oxidized nicotinamide adenine dinucleotide; NADH, reduced nicotin- amide adenine dinucleotide. SDS-PAGE, sodium dodecyl sulfate- polyacrylamide gel electrophoresis. 11390 Biochemistry 2002, 41, 11390-11397 10.1021/bi020196a CCC: $22.00 © 2002 American Chemical Society Published on Web 08/30/2002