The crystal structure of human isopentenyl diphosphate isomerase at 1.7 A resolution reveals its catalytic mechanism in isoprenoid biosynthesis

The essential Saccharomyces cerevisiae pre-messenger RNA splicing protein 24 (Prp24) has four RNA recognition motifs (RRMs) and facilitates U6 RNA base-pairing with U4 RNA during spliceosome assembly. Prp24 is a component of the free U6 small nuclear ribonucleoprotein particle (snRNP) but not the U4/U6 bi-snRNP, and so is thought to be displaced from U6 by U4/U6 base-pairing. The interaction partners of each of the four RRMs of Prp24 and how these interactions direct U4/U6 pairing are not known. Here we report the crystal structure of the first three RRMs and the solution structure of the first two RRMs of Prp24. Strikingly, RRM 2 forms extensive inter-domain contacts with RRMs 1 and 3. These contacts occupy much of the canonical RNA-binding faces (beta-sheets) of RRMs 1 and 2, but leave the beta-sheet of RRM 3 exposed. Previously identified substitutions in Prp24 that suppress mutations in U4 and U6 spliceosomal RNAs cluster primarily in the beta-sheet of RRM 3, but also in a conserved loop of RRM 2. RNA binding assays and chemical shift mapping indicate that a large basic patch evident on the surface of RRMs 1 and 2 is part of a high affinity U6 RNA binding site. Our results suggest that Prp24 binds free U6 RNA primarily with RRMs 1 and 2, which may remodel the U6 secondary structure. The beta-sheet of RRM 3 then influences U4/U6 pairing through interaction with an unidentified ligand.     

The 2.2 A resolution structure of RpiB/AlsB from Escherichia coli illustrates a new approach to the ribose-5-phosphate isomerase reaction

Ribose-5-phosphate isomerases (EC 5.3.1.6) interconvert ribose 5-phosphate and ribulose 5-phosphate. This reaction permits the synthesis of ribose from other sugars, as well as the recycling of sugars from nucleotide breakdown. Two unrelated types of enzyme can catalyze the reaction. The most common, RpiA, is present in almost all organisms (including Escherichia coli), and is highly conserved. The second type, RpiB, is present in some bacterial and eukaryotic species and is well conserved. In E.coli, RpiB is sometimes referred to as AlsB, because it can take part in the metabolism of the rare sugar, allose, as well as the much more common ribose sugars. We report here the structure of RpiB/AlsB from E.coli, solved by multi-wavelength anomalous diffraction (MAD) phasing, and refined to 2.2A resolution. RpiB is the first structure to be solved from pfam02502 (the RpiB/LacAB family). It exhibits a Rossmann-type alphabetaalpha-sandwich fold that is common to many nucleotide-binding proteins, as well as other proteins with different functions. This structure is quite distinct from that of the previously solved RpiA; although both are, to some extent, based on the Rossmann fold, their tertiary and quaternary structures are very different. The four molecules in the RpiB asymmetric unit represent a dimer of dimers. Active-site residues were identified at the interface between the subunits, such that each active site has contributions from both subunits. Kinetic studies indicate that RpiB is nearly as efficient as RpiA, despite its completely different catalytic machinery. The sequence and structural results further suggest that the two homologous components of LacAB (galactose-6-phosphate isomerase) will compose a bi-functional enzyme; the second activity is unknown.     

Ultrahigh resolution structure of a class A beta-lactamase: on the mechanism and specificity of the extended-spectrum SHV-2 enzyme

Bacterial beta-lactamases hydrolyze beta-lactam antibiotics such as penicillins and cephalosporins. The TEM-type class A beta-lactamase SHV-2 is a natural variant that exhibits activity against third-generation cephalosporins normally resistant to hydrolysis by class A enzymes. SHV-2 contains a single Gly238Ser change relative to the wild-type enzyme SHV-1. Crystallographic refinement of a model including hydrogen atoms gave R and R(free) of 12.4% and 15.0% for data to 0.91 A resolution. The hydrogen atom on the O(gamma) atom of the reactive Ser70 is clearly seen for the first time, bridging to the water molecule activated by Glu166. Though hydrogen atoms on the nearby Lys73 are not seen, this observation of the Ser70 hydrogen atom and the hydrogen bonding pattern around Lys73 indicate that Lys73 is protonated. These findings support a role for the Glu166-water couple, rather than Lys73, as the general base in the deprotonation of Ser70 in the acylation process of class A beta-lactamases. Overlay of SHV-2 with SHV-1 shows a significant 1-3 A displacement in the 238-242 beta-strand-turn segment, making the beta-lactam binding site more open to newer cephalosporins with large C7 substituents and thereby expanding the substrate spectrum of the variant enzyme. The OH group of the buried Ser238 side-chain hydrogen bonds to the main-chain CO of Asn170 on the Omega loop, that is unaltered in position relative to SHV-1. This structural role for Ser238 in protein-protein binding makes less likely its hydrogen bonding to oximino cephalosporins such as cefotaxime or ceftazidime.     

The crystal structure of the olfactory marker protein at 2.3 A resolution

Olfactory marker protein (OMP) is a highly expressed and phylogenetically conserved cytoplasmic protein of unknown function found almost exclusively in mature olfactory sensory neurons. Electrophysiological studies of olfactory epithelia in OMP knock-out mice show strongly retarded recovery following odorant stimulation leading to an impaired response to pulsed odor stimulation. Although these studies show that OMP is a modulator of the olfactory signal-transduction cascade, its biochemical role is not established. In order to facilitate further studies on the molecular function of OMP, its crystal structure has been determined at 2.3 A resolution using multiwavelength anomalous diffraction experiments on selenium-labeled protein. OMP is observed to form a modified beta-clamshell structure with eight antiparallel beta-strands. While OMP has no significant sequence homology to proteins of known structure, it has a similar fold to a domain found in a variety of existing structures, including in a large family of viral capsid proteins. The surface of OMP is mostly convex and lacking obvious small molecule binding sites, suggesting that it is more likely to be involved in modulating protein-protein interaction than in interacting with small molecule ligands. Three highly conserved regions have been identified as leading candidates for protein-protein interaction sites in OMP. One of these sites represents a loop known to mediate ligand interactions in the structurally homologous EphB2 receptor ligand-binding domain. This site is partially buried in the crystal structure but fully exposed in the NMR solution structure of OMP due to a change in the orientation of an alpha-helix that projects outward from the structurally invariant beta-clamshell core. Gating of this conformational change by molecular interactions in the signal-transduction cascade could be used to control access to OMP's equivalent of the EphB2 ligand-interaction loop, thereby allowing OMP to function as a molecular switch.     

The refined crystal structure of Pseudomonas putida lipoamide dehydrogenase complexed with NAD+ at 2.45 A resolution.

The three-dimensional structure of one of the three lipoamide dehydrogenases occurring in Pseudomonas putida, LipDH Val, has been determined at 2.45 A resolution. The orthorhombic crystals, grown in the presence of 20 mM NAD+, contain 458 residues per asymmetric unit. A crystallographic 2-fold axis generates the dimer which is observed in solution. The final crystallographic R-factor is 21.8% for 18,216 unique reflections and a model consisting of 3,452 protein atoms, 189 solvent molecules and 44 NAD+ atoms, while the overall B-factor is unusually high: 47 A2. The structure of LipDH Val reveals the conformation of the C-terminal residues which fold "back" into the putative lipoamide binding region. The C-terminus has been proven to be important for activity by site-directed mutagenesis. However, the distance of the C-terminus to the catalytically essential residues is surprisingly large, over 6 A, and the precise role of the C-terminus still needs to be elucidated. In this crystal form LipDH Val contains one NAD+ molecule per subunit. Its adenine-ribose moiety occupies an analogous position as in the structure of glutathione reductase. However, the nicotinamide-ribose moiety is far removed from its expected position near the isoalloxazine ring and points into solution. Comparison of LipDH Val with Azotobacter vinelandii lipoamide dehydrogenase yields an rms difference of 1.6 A for 440 well defined C alpha atoms per subunit. Comparing LipDH Val with glutathione reductase shows large differences in the tertiary and quaternary structure of the two enzymes. For instance, the two subunits in the dimer are shifted by 6 A with respect to each other. So, LipDH Val confirms the surprising differences in molecular architecture between glutathione reductase and lipoamide dehydrogenase, which were already observed in Azotobacter vinelandii LipDH. This is the more remarkable since the active sites are located at the subunit interface and are virtually identical in all three enzymes.     

The crystal structure of free human hypoxanthine-guanine phosphoribosyltransferase reveals extensive conformational plasticity throughout the catalytic cycle

Human hypoxanthine-guanine phosphoribosyltransferase (HGPRT) catalyses the synthesis of the purine nucleoside monophosphates, IMP and GMP, by the addition of a 6-oxopurine base, either hypoxanthine or guanine, to the 1-beta-position of 5-phospho-alpha-d-ribosyl-1-pyrophosphate (PRib-PP). The mechanism is sequential, with PRib-PP binding to the free enzyme prior to the base. After the covalent reaction, pyrophosphate is released followed by the nucleoside monophosphate. A number of snapshots of the structure of this enzyme along the reaction pathway have been captured. These include the structure in the presence of the inactive purine base analogue, 7-hydroxy [4,3-d] pyrazolo pyrimidine (HPP) and PRib-PP.Mg2+, and in complex with IMP or GMP. The third structure is that of the immucillinHP.Mg(2+).PP(i) complex, a transition-state analogue. Here, the first crystal structure of free human HGPRT is reported to 1.9A resolution, showing that significant conformational changes have to occur for the substrate(s) to bind and for catalysis to proceed. Included in these changes are relative movement of subunits within the tetramer, rotation and extension of an active-site alpha-helix (D137-D153), reorientation of key active-site residues K68, D137 and K165, and the rearrangement of three active-site loops (100-128, 165-173 and 186-196). Toxoplasma gondii HGXPRT is the only other 6-oxopurine phosphoribosyltransferase structure solved in the absence of ligands. Comparison of this structure with human HGPRT reveals significant differences in the two active sites, including the structure of the flexible loop containing K68 (human) or K79 (T.gondii).     

Atomic (0.94 A) resolution structure of an inverting glycosidase in complex with substrate

The crystal structure of Clostridium thermocellum endoglucanase CelA in complex with cellopentaose has been determined at 0.94 A resolution. The oligosaccharide occupies six D-glucosyl-binding subsites, three on either side of the scissile glycosidic linkage. The substrate and product of the reaction occupy different positions at the reducing end of the cleft, where an extended array of hydrogen-bonding interactions with water molecules fosters the departure of the leaving group. Severe torsional strain upon the bound substrate forces a distorted boat(2,5) B conformation for the glucosyl residue bound at subsite -1, which facilitates the formation of an oxocarbenium ion intermediate and might favor the breakage of the sugar ring concomitant with catalysis.     

High-resolution crystal structure of the Fab-fragments of a family of mouse catalytic antibodies with esterase activity

The crystal structures of four related Fab fragments of a family of catalytic antibodies displaying differential levels of esterase activity have been solved in the presence and in the absence of the transition-state analogue (TSA) that was used to elicit the immune response. The electron density maps show that the TSA conformation is essentially identical, with limited changes on hapten binding. Interactions with the TSA explain the specificity for the D rather than the L-isomer of the substrate. Differences in the residues in the hapten-binding pocket, which increase hydrophobicity, appear to correlate with an increase in the affinity of the antibodies for their substrate. Analysis of the structures at the active site reveals a network of conserved hydrogen bond contacts between the TSA and the antibodies, and points to a critical role of two conserved residues, HisL91 and LysH95, in catalysis. However, these two key residues are set into very different contexts in their respective structures, with an apparent direct correlation between the catalytic power of the antibodies and the complexity of their interactions with the rest of the protein. This suggests that the catalytic efficiency may be controlled by contacts arising from a second sphere of residues at the periphery of the active site.     

Amine free crystal structure: the crystal structure of d(CGCGCG)2 and methylamine complex crystal

We succeeded in the crystallization of d(CGCGCG)2 and methylamine Complex. The crystal was clear and of sufficient size to collect the X-ray crystallographic data up to 1.0 A resolution using synchrotron radiation. As a result of X-ray crystallographic analysis of 2Fo-Fc map was much clear and easily traced. It is the first time monoamine co-crystallizes with d(CGCGCG)2. However, methylamine was not found from the complex crystal of d(CGCGCG)2 and methylamine. Five Mg ions were found around d(CGCGCG)2 molecules. These Mg ions neutralized the anion of 10 values of the phosphate group of DNA with five Mg2+. DNA stabilized only by a metallic ion and there is no example of analyzing the X-ray crystal structure like this. Mg ion stabilizes the conformation of Z-DNA. To use monoamine for crystallization of DNA, we found that we can get only d(CGCGCG)2 and Mg cation crystal. Only Mg cation can stabilize the conformation of Z-DNA. The method of using the monoamine for the crystallization of DNA can be applied to the crystallization of DNA of long chain of length in the future like this.     

A metallo-beta-lactamase enzyme in action: crystal structures of the monozinc carbapenemase CphA and its complex with biapenem

One strategy developed by bacteria to resist the action of beta-lactam antibiotics is the expression of metallo-beta-lactamases. CphA from Aeromonas hydrophila is a member of a clinically important subclass of metallo-beta-lactamases that have only one zinc ion in their active site and for which no structure is available. The crystal structures of wild-type CphA and its N220G mutant show the structural features of the active site of this enzyme, which is modeled specifically for carbapenem hydrolysis. The structure of CphA after reaction with a carbapenem substrate, biapenem, reveals that the enzyme traps a reaction intermediate in the active site. These three X-ray structures have allowed us to propose how the enzyme recognizes carbapenems and suggest a mechanistic pathway for hydrolysis of the beta-lactam. This will be relevant for the design of metallo-beta-lactamase inhibitors as well as of antibiotics that escape their hydrolytic activity.     

Crystal structure of decameric fructose-6-phosphate aldolase from Escherichia coli reveals inter-subunit helix swapping as a structural basis for assembly differences in the transaldolase family

Fructose-6-phosphate aldolase from Escherichia coli is a member of a small enzyme subfamily (MipB/TalC family) that belongs to the class I aldolases. The three-dimensional structure of this enzyme has been determined at 1.93 A resolution by single isomorphous replacement and tenfold non-crystallographic symmetry averaging and refined to an R-factor of 19.9% (R(free) 21.3%). The subunit folds into an alpha/beta barrel, with the catalytic lysine residue on barrel strand beta 4. It is very similar in overall structure to that of bacterial and mammalian transaldolases, although more compact due to extensive deletions of additional secondary structural elements. The enzyme forms a decamer of identical subunits with point group symmetry 52. Five subunits are arranged as a pentamer, and two ring-like pentamers pack like a doughnut to form the decamer. A major interaction within the pentamer is through the C-terminal helix from one monomer, which runs across the active site of the neighbouring subunit. In classical transaldolases, this helix folds back and covers the active site of the same subunit and is involved in dimer formation. The inter-subunit helix swapping appears to be a major determinant for the formation of pentamers rather than dimers while at the same time preserving importing interactions of this helix with the active site of the enzyme. The active site lysine residue is covalently modified, by forming a carbinolamine with glyceraldehyde from the crystallisation mixture. The catalytic machinery is very similar to that of transaldolase, which together with the overall structural similarity suggests that enzymes of the MipB/TALC subfamily are evolutionary related to the transaldolase family.     

Crystal structure of the polyketide cyclase AknH with bound substrate and product analogue: implications for catalytic mechanism and product stereoselectivity

AknH is a small polyketide cyclase that catalyses the closure of the fourth carbon ring in aclacinomycin biosynthesis in Streptomyces galilaeus, converting aklanonic acid methyl ester to aklaviketone. The crystal structure analysis of this enzyme, in complex with substrate and product analogue, showed that it is closely related in fold and mechanism to the polyketide cyclase SnoaL that catalyses the corresponding reaction in the biosynthesis of nogalamycin. Similarity is also apparent at a functional level as AknH can convert nogalonic acid methyl ester, the natural substrate of SnoaL, to auraviketone in vitro and in constructs in vivo. Despite the conserved structural and mechanistic features between these enzymes, the reaction products of AknH and SnoaL are stereochemically distinct. Supplied with the same substrate, AknH yields a C9-R product, like most members of this family of polyketide cyclases, whereas the product of SnoaL has the opposite C9-S stereochemistry. A comparison of high-resolution crystal structures of the two enzymes combined with in vitro mutagenesis studies revealed two critical amino acid substitutions in the active sites, which contribute to product stereoselectivity in AknH. Replacement of residues Tyr15 and Asn51 of AknH, located in the vicinity of the main catalytic residue Asp121, by their SnoaL counter-parts phenylalanine and leucine, respectively, results in a complete loss of product stereoselectivity.     

Trimeric crystal structure of the glycoside hydrolase family 42 beta-galactosidase from Thermus thermophilus A4 and the structure of its complex with galactose

The beta-galactosidase from an extreme thermophile, Thermus thermophilus A4 (A4-beta-Gal), is thermostable and belongs to the glycoside hydrolase family 42 (GH-42). As the first known structures of a GH-42 enzyme, we determined the crystal structures of free and galactose-bound A4-beta-Gal at 1.6A and 2.2A resolution, respectively. A4-beta-Gal forms a homotrimeric structure resembling a flowerpot. Each monomer has an active site located inside a large central tunnel. The N-terminal domain of A4-beta-Gal has a TIM barrel fold, as predicted from hydrophobic cluster analysis. The putative catalytic residues of A4-beta-Gal (Glu141 and Glu312) superimpose well with the catalytic residues of Escherichia coli beta-galactosidase. The environment around the catalytic nucleophile (Glu312) is similar to that in the case of E.coli beta-galactosidase, but the recognition mechanism for a substrate is different. Trp182 of the next subunit of the trimer constitutes a part of the active-site pocket, indicating that the trimeric structure is essential for the enzyme activity. Structural comparison with other glycoside hydrolases revealed that many features of the 4/7 superfamily are conserved in the A4-beta-Gal structure. On the basis of the results of 1H NMR spectroscopy, A4-beta-Gal was determined to be a retaining enzyme. Interestingly, the active site was similar with those of retaining enzymes, but the overall fold of the TIM barrel domain was very similar to that of an inverting enzyme, beta-amylase.     

A novel binding of GTP stabilizes the structure and modulates the activities of human phosphoglucose isomerase/autocrine motility factor

Phosphoglucose isomerase (PGI) catalyzes the interconversion between glucose 6-phosphate and fructose 6-phosphate in the glycolysis pathway. In mammals, the enzyme is also identical to the extracellular proteins neuroleukin, tumor-secreted autocrine motility factor (AMF) and differentiation and maturation mediator for myeloid leukemia. Hereditary deficiency of the enzyme causes non-spherocytic hemolytic anemia in human. In the present study, a novel interaction between GTP and human PGI was corroborated by UV-induced crosslinking, affinity purification and kinetic study. GTP not only inhibits the isomerization activity but also compromises the AMF function of the enzyme. Kinetic studies, including the Yonetani-Theorell method, suggest that GTP is a competitive inhibitor with a K_i value of 63 μM and the GTP-binding site partially overlaps with the catalytic site. In addition, GTP stabilizes the structure of human PGI against heat- and detergent-induced denaturation. Molecular modelling and dynamic simulation suggest that GTP is bound in a syn-conformation with the γ-phosphate group located near the phosphate-binding loop and the ribose moiety positioned away from the active-site residues.     

Structures of the “open” and “closed” state of trypanosomal triosephosphate isomerase,as observed in a new crystal form: Implications for the reaction mechanism

The structure of trypanosomal triosephosphate isomerase (TIM)has been solved at a resolution of 2.1Å in a new crystal form grown at pH 8.8 from PEG6000. In this new crystal form (space group C2, cell dimensions 94.8 Å, 48.3 Å, 131.0 Å, 90.0°, 100.3°, 90.0°), TIM is present in a ligand-free state. The asymmetric unit consists of two TIM subunits. Each of these subunits is part of a dimer which is sitting on a crystallographic twofold axis, such that the crystal packing is formed from two TIM dimers in two distinct environments. The two constituent monomers of a given dimer are, therefore, crystallographically equivalent. In the ligand-free state of TIM in this crystal form, the two types of dimer are very similar in structure, with the flexible loops in the “Open” conformation. For one dimer (termed molecule-1), the flexible loop (loop-6) is involved in crystal contacts. Crystals of this type have been used in soaking experiments with 0.4 M ammonium sulphate (studied at 2.4 Å resolution), and with 40 μM phosphoglycolohydroxamate (studied at 2.5 Å resolution). It is found that transfer to 0.4 M ammonuum sulphate (equal to 80 times the Ki of sulphate for TIM), gives rise to significant sulphate binding at the active site of one dimer (termed molecule-2), and less significant binding at the active site of the other. In neither dimer does sulphate induce a “closed” conformation. In a mother liquor containing 40 μM phosphoglycolohydroxamate (equal to 10 times the Ki of phosphoglycolohydroxamate for TIM), an inhibitor molecule binds at the active site of only that dimer of which the flexible loop is free from crystal contacts (molecule-2). In this dimer, it induces a closed conformation. These three structures are compared and discussed with respect to the mode of binding of ligand in the active site as well as with respect to the conformational changes resulting from ligand binding. © 1993 Wiley-Liss, Inc.     

Analysis of the class I aldolase binding site architecture based on the crystal structure of 2-deoxyribose-5-phosphate aldolase at 0.99A resolution

The crystal structure of the bacterial (Escherichia coli) class I 2-deoxyribose-5-phosphate aldolase (DERA) has been determined by Se-Met multiple anomalous dispersion (MAD) methods at 0.99A resolution. This structure represents the highest-resolution X-ray structure of an aldolase determined to date and enables a true atomic view of the enzyme. The crystal structure shows the ubiquitous TIM alpha/beta barrel fold. The enzyme contains two lysine residues in the active site. Lys167 forms the Schiff base intermediate, whereas Lys201, which is in close vicinity to the reactive lysine residue, is responsible for the perturbed pK(a) of Lys167 and, hence, also a key residue in the reaction mechanism. DERA is the only known aldolase that is able to use aldehydes as both aldol donor and acceptor molecules in the aldol reaction and is, therefore, of particular interest as a biocatalyst in synthetic organic chemistry. The uncomplexed DERA structure enables a detailed comparison with the substrate complexes and highlights a conformational change in the phosphate-binding site. Knowledge of the enzyme active-site environment has been the basis for exploration of catalysis of non-natural substrates and of mutagenesis of the phosphate-binding site to expand substrate specificity. Detailed comparison with other class I aldolase enzymes and DERA enzymes from different organisms reveals a similar geometric arrangement of key residues and implies a potential role for water as a general base in the catalytic mechanism.     

Structure determination and refinement of the Al complex of the D254,256E mutant of Arthrobacter -xylose isomerase at 2.40 Å resolution. Further evidence for inhibitor-induced metal ion movement

The structure of the D254,256E double mutant of Arthrobacter xylose isomerase with Al³⁺ at both metal-binding sites was determined by the molecular replacement method at a conventional R-factor of 0.179. Binding of the two Al³⁺ does not alter the overall structure significantly. However, there are local rearrangements in the octahedral co-ordination sphere of the Al³⁺. The inhibitor molecule moves somewhat away from the active site. Furthermore, evidence was revealed for metal ion movement from site 2₁ to site 2₂ upon double mutation. Xylose isomerase requires two divalent metal cations for activation. The catalytic metal ion is translocated 1.8 Å away from its initial position during the catalytic reaction. The fact that both activating and inactivating metals (including Al³⁺) were found exclusively at a single location in the double mutant was an indication that the consequently missing shuttle may account for the crippled catalytic efficiency.     

Lipidic cubic phase crystal structure of the photosynthetic reaction centre from Rhodobacter sphaeroides at 2.35A resolution

Well-ordered crystals of the bacterial photosynthetic reaction centre from Rhodobacter sphaeroides were grown from a lipidic cubic phase. Here, we report the type I crystal packing that results from this crystallisation medium, for which 3D crystals grow as stacked 2D crystals, and the reaction centre X-ray structure is refined to 2.35A resolution. In this crystal form, the location of the membrane bilayer could be assigned with confidence. A cardiolipin-binding site is found at the protein-protein interface within the membrane-spanning region, shedding light on the formation of crystal contacts within the membrane. A chloride-binding site was identified in the membrane-spanning region, which suggests a putative site for interaction with the light-harvesting complex I, the cytochrome bc(1) complex or PufX. Comparisons with the X-ray structures of this reaction centre deriving from detergent-based crystals are drawn, indicating that a slight compression occurs in this lipid-rich environment.     

The crystal structure of phenylpyruvate decarboxylase from Azospirillum brasilense at 1.5 A resolution. Implications for its catalytic and regulatory mechanism

Phenylpyruvate decarboxylase (PPDC) of Azospirillum brasilense, involved in the biosynthesis of the plant hormone indole-3-acetic acid and the antimicrobial compound phenylacetic acid, is a thiamine diphosphate-dependent enzyme that catalyses the nonoxidative decarboxylation of indole- and phenylpyruvate. Analogous to yeast pyruvate decarboxylases, PPDC is subject to allosteric substrate activation, showing sigmoidal v versus [S] plots. The present paper reports the crystal structure of this enzyme determined at 1.5 A resolution. The subunit architecture of PPDC is characteristic for other members of the pyruvate oxidase family, with each subunit consisting of three domains with an open alpha/beta topology. An active site loop, bearing the catalytic residues His112 and His113, could not be modelled due to flexibility. The biological tetramer is best described as an asymmetric dimer of dimers. A cysteine residue that has been suggested as the site for regulatory substrate binding in yeast pyruvate decarboxylase is not conserved, requiring a different mechanism for allosteric substrate activation in PPDC. Only minor changes occur in the interactions with the cofactors, thiamine diphosphate and Mg2+, compared to pyruvate decarboxylase. A greater diversity is observed in the substrate binding pocket accounting for the difference in substrate specificity. Moreover, a catalytically important glutamate residue conserved in nearly all decarboxylases is replaced by a leucine in PPDC. The consequences of these differences in terms of the catalytic and regulatory mechanism of PPDC are discussed.