Crystal structure of 2,5-diketo-D-gluconic acid reductase A complexed with NADPH at 2.1-A resolution

The three-dimensional structure of Corynebacterium 2, 5-diketo-D-gluconic acid reductase A (2,5-DKGR A; EC 1.1.1.-), in complex with cofactor NADPH, has been solved by using x-ray crystallographic data to 2.1-A resolution. This enzyme catalyzes stereospecific reduction of 2,5-diketo-D-gluconate (2,5-DKG) to 2-keto-L-gulonate. Thus the three-dimensional structure has now been solved for a prokaryotic example of the aldo-keto reductase superfamily. The details of the binding of the NADPH cofactor help to explain why 2,5-DKGR exhibits lower binding affinity for cofactor than the related human aldose reductase does. Furthermore, changes in the local loop structure near the cofactor suggest that 2,5-DKGR will not exhibit the biphasic cofactor binding characteristics observed in aldose reductase. Although the crystal structure does not include substrate, the two ordered water molecules present within the substrate-binding pocket are postulated to provide positional landmarks for the substrate 5-keto and 4-hydroxyl groups. The structural basis for several previously described active-site mutants of 2,5-DKGR A is also proposed. Recent research efforts have described a novel approach to the synthesis of L-ascorbate (vitamin C) by using a genetically engineered microorganism that is capable of synthesizing 2,5-DKG from glucose and subsequently is transformed with the gene for 2,5-DKGR. These modifications create a microorganism capable of direct production of 2-keto-L-gulonate from D-glucose, and the gulonate can subsequently be converted into vitamin C. In economic terms, vitamin C is the single most important specialty chemical manufactured in the world. Understanding the structural determinants of specificity, catalysis, and stability for 2,5-DKGR A is of substantial commercial interest.     

Crystal structure of Escherichia coli cystathionine gamma-synthase at 1.5 A resolution

The transsulfuration enzyme cystathionine gamma-synthase (CGS) catalyses the pyridoxal 5'-phosphate (PLP)-dependent gamma-replacement of O-succinyl-L-homoserine and L-cysteine, yielding L-cystathionine. The crystal structure of the Escherichia coli enzyme has been solved by molecular replacement with the known structure of cystathionine beta-lyase (CBL), and refined at 1.5 A resolution to a crystallographic R-factor of 20.0%. The enzyme crystallizes as an alpha4 tetramer with the subunits related by non-crystallographic 222 symmetry. The spatial fold of the subunits, with three functionally distinct domains and their quaternary arrangement, is similar to that of CBL. Previously proposed reaction mechanisms for CGS can be checked against the structural model, allowing interpretation of the catalytic and substrate-binding functions of individual active site residues. Enzyme-substrate models pinpoint specific residues responsible for the substrate specificity, in agreement with structural comparisons with CBL. Both steric and electrostatic designs of the active site seem to achieve proper substrate selection and productive orientation. Amino acid sequence and structural alignments of CGS and CBL suggest that differences in the substrate-binding characteristics are responsible for the different reaction chemistries. Because CGS catalyses the only known PLP-dependent replacement reaction at Cgamma of certain amino acids, the results will help in our understanding of the chemical versatility of PLP.     

Three-dimensional structure of human tissue inhibitor of metalloproteinases-2 at 2.1 A resolution

The three-dimensional structure of human tissue inhibitor of metalloproteinases-2 (TIMP-2) was determined by X-ray crystallography to 2.1 A resolution. The structure of the inhibitor consists of two domains. The N-terminal domain (residues 1-110) is folded into a beta-barrel, similar to the oligonucleotide/oligosaccharide binding fold otherwise found in certain DNA-binding proteins. The C-terminal domain (residues 111-194) contains a parallel stranded beta-hairpin plus a beta-loop-beta motif. Comparison of the structure of uncomplexed human TIMP-2 with that of bovine TIMP-2 bound to the catalytic domain of human MMP-14 suggests an internal rotation between the two domains of approximately 13 degrees upon binding to the protease. Furthermore, local conformational differences in the two structures that might be induced by formation of the protease-inhibitor complex have been found. The most prominent of these involves residues 27-40 of the A-B beta-hairpin loop. Structure-based alignment of amino acid sequences of representatives of the TIMP family maps the sequence differences mainly to loop regions, and some of these differences are proposed to be responsible for the particular properties of the various TIMP species.     

The crystal structure of a 3D domain-swapped dimer of RNase A at a 2.1-A resolution

The dimer of bovine pancreatic ribonuclease A (RNase A) discovered by Crestfield, Stein, and Moore in 1962 has been crystallized and its structure determined and refined to a 2.1-A resolution. The dimer is 3D domain-swapped. The N-terminal helix (residues 1-15) of each subunit is swapped into the major domain (residues 23-124) of the other subunit. The dimer of bull seminal ribonuclease (BS-RNase) is also known to be domain-swapped, but the relationship of the subunits within the two dimers is strikingly different. In the RNase A dimer, the 3-stranded beta sheets of the two subunits are hydrogen-bonded at their edges to form a continuous 6-stranded sheet across the dimer interface; in the BS-RNase dimer, it is instead the two helices that abut. Whereas the BS-RNase dimer has 2-fold molecular symmetry, the two subunits of the RNase A dimer are related by a rotation of approximately 160 degrees. Taken together, these structures show that intersubunit adhesion comes mainly from the swapped helical domain binding to the other subunit in the "closed interface" but that the overall architecture of the domain-swapped oligomer depends on the interactions in the second type of interface, the "open interface." The RNase A dimer crystals take up the dye Congo Red, but the structure of a Congo Red-stained crystal reveals no bound dye molecule. Dimer formation is inhibited by excess amounts of the swapped helical domain. The possible implications for amyloid formation are discussed.     

Crystal structure of intact elongation factor EF-Tu from Escherichia coli in GDP conformation at 2.05 A resolution

The crystal structure of intact elongation factor Tu (EF-Tu) from Escherichia coli in GDP-bound conformation has been determined using a combination of multiple isomorphous replacement (MIR) and multiwavelength anomalous diffraction (MAD) methods. The current atomic model has been refined to a crystallographic R factor of 20.3 % and free R-factor of 26.8 % in the resolution range of 10-2.05 A. The protein consists of three domains: domain 1 has an alpha/beta structure; while domain 2 and domain 3 are beta-barrel structures. Although the global fold of the current model is similar to those of published structures, the secondary structural assignment has been improved due to the high quality of the current model. The switch I region (residues 40-62) is well ordered in this structure. Comparison with the structure of EF-Tu in GDP-bound form from Thermus aquaticus shows that although the individual domain structures are similar in these two structures, the orientation of domains changes significantly. Interactions between domains 1 and 3 in our E. coli EF-Tu-GDP complex are quite different from those of EF-Tu with bound GTP from T. aquaticus, due to the domain rearrangement upon GTP binding. The binding sites of the Mg2+ and guanine nucleotide are revealed in detail. Two water molecules that co-ordinate the Mg2+ have been identified to be well conserved in the GDP and GTP-bound forms of EF-Tu structures, as well as in the structure of Ras p21 with bound GDP. Comparisons of the Mg2+ binding site with other guanine nucleotide binding proteins in GDP-bound forms show that the Mg2+ co-ordination patterns are well preserved among these structures.     

The 2.0-A resolution structure of Escherichia coli histidine-containing phosphocarrier protein HPr. A redetermination

The x-ray structure of Escherichia coli HPr has been redetermined at 2.0-A resolution. In contrast to the previous study (El-Kabbani, O. A. L., Waygood, E. B., and Delbaere, L. T. J. (1987) J. Biol. Chem. 262, 12926-12929), the overall structure is, in general, similar to other reported NMR and x-ray HPr structures, although there are some important differences in detail. The overall folding topology of HPr is a classical open-faced beta-sandwich, consisting of four antiparallel beta-strands and three alpha-helices. The least square refinement produced an R index of 0.135 for all measured unique data between 8.0 and 2.0 A resolution. The active center consists of His15 which is hydrogen bonded to a sulfate anion, and Arg17 which has a fully open conformation. This corresponds to the first observed "semi-closed" conformation of the active center of HPr. The Streptococcus faecalis HPr structure (Jia, Z., Vandonselaar, M., Quail, J. W., and Delbaere, L. T. J. (1993) Nature 361, 94-97) has the "open" conformation in which the side chains of His15 and Arg17 are directed as far away from each other as possible. The Bacillus subtilis HPr (Herzberg, O., Reddy, P., Sutrina, S., Saier, M. H., Jr., Reizer, J., and Kapadia, G. (1992) Proc. Natl. Acad. Sci. U.S.A. 89, 2499-2503) has the "closed" conformation in which the side chains of His15 and Arg17 are close together with a sulfate anion located in the active center. The open conformation represents the unphosphorylated form of HPr whereas the closed conformation likely resembles the phosphorylated form of HPr. The semi-closed conformation observed in the E. coli HPr structure could represent a structural intermediate on the phosphorylation/dephosphorylation pathway of HPr.     

Streptomyces griseus aminopeptidase: X-ray crystallographic structure at 1.75 A resolution

The X-ray crystal structure of the enzyme Streptomyces griseus aminopeptidase (SGAP) has been determined in its double zinc form to 1.75 A resolution, in its apo-enzyme from (zinc removed) to 2.1 A resolution, and as a mercury replaced derivative to 2.1 A resolution. The structure solution was achieved by single isomorphous replacement with phasing from anomalous scattering (SIRAS), followed by density modification with histogram matching. The protein consists of a central beta-sheet made up of eight parallel and antiparallel strands, surrounded by helices on either side. The active site is located at the carbonyl ends of two middle strands of the beta-sheet region. Two sections of the chain that could not be traced were Glu196 to Arg202, which borders the active site, and the final seven C-terminal residues starting with Gly278. The active site contains two zinc cations, each with similar ligands, at a distance of 3.6 A from each other. An unknown molecule appears to be bound to both zinc ions in the active site at partial occupancy and has been modelled as a phosphate ion. A calcium binding site has also been identified, consistent with the observations that calcium modulates the activity of the enzyme, and increases its heat stability. The mechanism by which the calcium cation modulates enzyme activity is not apparent, since the location of the calcium binding site is approximately 25 A distant from the active site zinc ions. Comparison of the structure of SGAP to other known aminopeptidases shows that the enzyme is most similar to Aeromonas proteolytica aminopeptidase (AAP). Both enzymes share a similar topology, although the overall sequence identity is very low (24% in aligned regions). The coordination of the two active site zinc cations in SGAP resembles that of AAP. These two microbial enzymes differ from bovine lens leucine aminopeptidase (LAP) in both overall structure and in coordination of the two zinc ions.     

X-ray structure at 1.76 A resolution of a polypeptide phospholipase A2 inhibitor

The high resolution crystal structure of a natural PLA2 inhibitor has been determined by Patterson search methods. In the heterodimeric, neurotoxic complex, vipoxin, isolated from the venom of Bulgarian viper, PLA2 inhibitor represents the non-toxic subunit. The model was refined to a crystallographic R-factor of 15.5% for data between 6 and 1.76 A resolution. The packing of the inhibitor in the crystal reveals close contacts between the molecules, which are symmetry-related by the 2-fold axes of the lattice. These pairs associate as a crystallographic dimer, stabilized by a set of interactions, including van der Waals contacts between residues from symmetry-related pairs, denoted as the recognition site and the recognition surface. Residues Ph3, Trp31 and Tyr119 represent the recognition site of inhibitor which possibly fits to the hydrophobic wall of the target PLA2. The topology of the inhibitor represents the PLA2 type of folding: three long helices and a beta-hairpin. Superposition of the structure of the inhibitor shows an almost complete overlap with different mammalian and viper PLA2 in the backbone and in the position of the sidechains of the residues that belong to the active centre and the hydrophobic wall. A "lock and key" mechanism of recognition of its native PLA2 in gland cells and other toxic PLA2 in vitro has been suggested. The mechanism includes complementary "head to tail" interactions between the recognition site of the inhibitor and a recognition surface located on the hydrophobic wall of the target PLA2. Having a high spatial homology with the PLA2 family of enzymes but opposing their action, the inhibitor from vipoxin presents an example of a divergent evolution of an ancient PLA2. The presence of a space for binding calcium in the inhibitor is believed to be a rudiment and proof of a common origin with PLA2.     

X-ray crystal structure of a pea lectin-trimannoside complex at 2.6 A resolution

The x-ray crystal structure of pea lectin, in complex with a methyl glycoside of the N-linked-type oligosaccharide trimannosyl core, methyl 3,6-di-O-(alpha-D-mannopyranosyl)-alpha-D-mannopyranoside, has been solved by molecular replacement and refined at 2.6-A resolution. The R factor is 0.183 for all data in the 8.0 to 2.6 A resolution range with an average atomic temperature factor of 26.1 A2. Strong electron density for a single mannose residue is found in the monosaccharide-binding site suggesting that the trisaccharide binds primarily through one of the terminal alpha-linked mannose residues. The complex is stabilized by hydrogen bonds involving the protein residues Asp-81, Gly-99, Asn-125, Ala-217, and Glu-218, and the carbohydrate oxygen atoms O3, O4, O5, and O6. In addition, the carbohydrate makes van der Waals contacts with the protein, involving Phe-123 in particular. These interactions are very similar to those found in the monosaccharide complexes with concanavalin A and isolectin 1 of Lathyrus ochrus, confirming the structural relatedness of this family of proteins. Comparison of the pea lectin complex with the unliganded pea lectin and concanavalin A structures indicates differences in the conformation and water structure of the unliganded binding sites of these two proteins. Furthermore, a correlation between the position of the carbohydrate oxygen atoms in the complex and the bound water molecules in the unliganded binding sites is found. Binding of the trimannose core through a single terminal monosaccharide residue strongly argues that an additional fucose-binding site is responsible for the high affinity pea lectin-oligosaccharide interactions.     

X-ray structure of nucleoside diphosphate kinase complexed with thymidine diphosphate and Mg2+ at 2-A resolution

We report the crystal structure of nucleoside diphosphate kinase (NDP kinase) from Dictyostelium discoideum with thymidine diphosphate (dTDP) and Mg2+ bound at the active site. The structure has been refined to an R-factor of 18.3% at 2-A resolution. The base stacks on the aromatic ring of Phe 64 near the protein surface and is wedged between the side chains of Phe 64 and Val 116. The sugar and the pyrophosphate are deeper inside the protein and make numerous H-bonds with protein side chains. There is no backbone interaction with the nucleotide. A Mg2+ ion bridges the alpha- and beta-phosphates and interacts with the protein via water molecules. NDP kinase shows little specificity toward ribonucleotides and deoxyribonucleotides. This property, required by the enzyme biological function, can now be analyzed by comparing the crystal structures of free, ADP-ligated, and dTDP-ligated enzymes. The most significant differences are located in residues 60-64, which adapt their conformation to allow Phe 64 to stack on both types of bases. Nonspecific binding is achieved by the absence of polar interaction between the base and protein atoms. The ribose of ADP and the deoxyribose of dTDP occupy similar positions, their hydroxyl groups interacting with Lys 16 and Asn 119. The H-bond between Lys 16 and the O2' hydroxyl of ADP is replaced by a similar interaction with a water molecule in the dTDP complex. The beta-phosphate position is the same for ADP and dTDP, suggesting that the mechanism of phosphate transfer is the same for all substrates ofNDP kinase.(ABSTRACT TRUNCATED AT 250 WORDS)     

X-ray crystal structure of glycinamide ribonucleotide synthetase from Escherichia coli

Glycinamide ribonucleotide synthetase (GAR-syn) catalyzes the second step of the de novo purine biosynthetic pathway; the conversion of phosphoribosylamine, glycine, and ATP to glycinamide ribonucleotide (GAR), ADP, and Pi. GAR-syn containing an N-terminal polyhistidine tag was expressed as the SeMet incorporated protein for crystallographic studies. In addition, the protein as isolated contains a Pro294Leu mutation. This protein was crystallized, and the structure solved using multiple-wavelength anomalous diffraction (MAD) phase determination and refined to 1.6 A resolution. GAR-syn adopts an alpha/beta structure that consists of four domains labeled N, A, B, and C. The N, A, and C domains are clustered to form a large central core structure whereas the smaller B domain is extended outward. Two hinge regions, which might readily facilitate interdomain movement, connect the B domain and the main core. A search of structural databases showed that the structure of GAR-syn is similar to D-alanine:D-alanine ligase, biotin carboxylase, and glutathione synthetase, despite low sequence similarity. These four enzymes all utilize similar ATP-dependent catalytic mechanisms even though they catalyze different chemical reactions. Another ATP-binding enzyme with low sequence similarity but unknown function, synapsin Ia, was also found to share high structural similarity with GAR-syn. Interestingly, the GAR-syn N domain shows similarity to the N-terminal region of glycinamide ribonucleotide transformylase and several dinucleotide-dependent dehydrogenases. Models of ADP and GAR binding were generated based on structure and sequence homology. On the basis of these models, the active site lies in a cleft between the large domain and the extended B domain. Most of the residues that facilitate ATP binding belong to the A or B domains. The N and C domains appear to be largely responsible for substrate specificity. The structure of GAR-syn allows modeling studies of possible channeling complexes with PPRP amidotransferase.     

A zipper-like duplex in DNA: the crystal structure of d(GCGAAAGCT) at 2.1 A resolution

BACKGROUND: The replication origin of the single-stranded (ss)DNA bacteriophage G4 has been proposed to fold into a hairpin loop containing the sequence GCGAAAGC. This sequence comprises a purine-rich motif (GAAA), which also occurs in conserved repetitive sequences of centromeric DNA. ssDNA analogues of these sequences often show exceptional stability which is associated with hairpin loops or unusual duplexes, and may be important in DNA replication and centromere function. Nuclear magnetic resonance (NMR) studies indicate that the GCGAAAGC sequence forms a hairpin loop in solution, while centromere-like repeats dimerise into unusual duplexes. The factors stabilising these unusual secondary structure elements in ssDNA, however, are poorly understood. RESULTS: The nonamer d(GCGAAAGCT) was crystallised as a bromocytosine derivative in the presence of cobalt hexammine. The crystal structure, solved by the multiple wavelength anomalous dispersion (MAD) method at the bromine K-edge, reveals an unexpected zipper-like motif in the middle of a standard B-DNA duplex. Four central adenines, flanked by two sheared G.A mismatches, are intercalated and stacked on top of each other without any interstrand Watson-Crick base pairing. The cobalt hexammine cation appears to participate only in crystal cohesion. CONCLUSIONS: The GAAA consensus sequence can dimerise into a stable zipper-like duplex as well as forming a hairpin loop. The arrangement closes the minor groove and exposes the intercalated, unpaired, adenines to the solvent and DNA-binding proteins. Such a motif, which can transform into a hairpin, should be considered as a structural option in modelling DNA and as a potential binding site, where it could have a role in DNA replication, nuclease resistance, ssDNA genome packaging and centromere function.     

Crystal structure of oxidized trimethylamine N-oxide reductase from Shewanella massilia at 2.5 A resolution

The periplasmic trimethylamine N-oxide (TMAO) reductase from the marine bacteria Shewanella massilia is involved in a respiratory chain, having trimethylamine N-oxide as terminal electron acceptor. This molybdoenzyme belongs to the dimethyl sulfoxide (DMSO) reductase family, but has a different substrate specificity than its homologous enzyme. While the DMSO reductases reduce a broad spectra of organic S-oxide and N-oxide compounds, TMAO reductase from Shewanella massilia reduces only TMAO as the natural compound. The crystal structure was solved by molecular replacement with the coordinates of the DMSO reductase from Rhodobacter sphaeroides. The overall fold of the protein structure is essentially the same as the DMSO reductase structures, organized into four domains. The molybdenum coordination sphere is closest to that described in the DMSO reductase of Rhodobacter capsulatus. The structural differences found in the protein environment of the active site could be related to the differences in substrate specificity of these enzymes. In close vicinity of the molybdenum ion a tyrosine residue is missing in the TMAO reductase, leaving a greater space accessible to the solvent. This tyrosine residue has contacts to the oxo groups in the DMSO reductase structures. The arrangement and number of charged residues lining the inner surface of the funnel-like entrance to the active site, is different in the TMAO reductase than in the DMSO reductases from Rhodobacter species. Furthermore a surface loop at the top of the active-site funnel, for which no density was present in the DMSO reductase structures, is well defined in the oxidized form of the TMAO reductase structure, and is located on the border of the funnel-like entrance of the active center.     

Refined X-ray structure of Dictyostelium discoideum nucleoside diphosphate kinase at 1.8 A resolution

The X-ray structure of the nucleoside diphosphate kinase (NDP kinase) from Dictyostelium discoideum has been refined at 1.8 A resolution from a hexagonal crystal form with a 17 kDa monomer in its asymmetric unit. The atomic model was derived from the previously determined structure of a point mutant of the protein. It contains 150 amino acid residues out of 155, and 95 solvent molecules. The R-factor is 0.196 and the estimated accuracy of the average atomic position, 0.25 A. The Dictyostelium structure is described in detail and compared to those of Drosophila and Myxococcus xanthus NDP kinases. The protein is a hexamer with D3 symmetry. Residues 8 to 138 of each subunit form a globular alpha/beta domain. The four-stranded beta-sheet is antiparallel; its topology is different from other phosphate transfer enzymes, and also from the HPr protein which, like NDP kinase, carries a phosphorylated histidine. The same topology is nevertheless found in several other proteins that bind mononucleotides, RNA or DNA. Strand connections in NDP kinase involve alpha-helices and a 20-residue segment called the Kpn loop. The beta-sheet is regular except for a beta-bulge in edge strand beta 2 and a gamma-turn at residue Ile120 just preceding strand beta 4. The latter may induce strain in the main chain near the active site His122. The alpha 1 beta 2 motif participates in forming dimers within the hexamer, helices alpha 1 and alpha 3, the Kpn loop and C terminus, in forming trimers. The subunit fold and dimer interactions found in Dictyostelium are conserved in other NDP kinases. Trimer interactions probably occur in all eukaryotic enzymes. They are absent in the bacterial Myxococcus xanthus enzyme which is a tetramer, even though the subunit structure is very similar. In Dictyostelium, contacts between Kpn loops near the 3-fold axis block access to a central cavity lined with polar residues and filled with well-defined solvent molecules. Biochemical data on point mutants highlight the contribution of the Kpn loop to protein stability. In Myxococcus, the Kpn loops are on the tetramer surface and their sequence is poorly conserved. Yet, their conformation is maintained and they make a similar contribution to the substrate binding site.     

A flavodoxin that is required for enzyme activation: the structure of oxidized flavodoxin from Escherichia coli at 1.8 A resolution

In Escherichia coli, flavodoxin is the physiological electron donor for the reductive activation of the enzymes pyruvate formate-lyase, anaerobic ribonucleotide reductase, and B12-dependent methionine synthase. As a basis for studies of the interactions of flavodoxin with methionine synthase, crystal structures of orthorhombic and trigonal forms of oxidized recombinant flavodoxin from E. coli have been determined. The orthorhombic form (space group P2(1)2(1)2(1), a = 126.4, b = 41.10, c = 69.15 A, with two molecules per asymmetric unit) was solved initially by molecular replacement at a resolution of 3.0 A, using coordinates from the structure of the flavodoxin from Synechococcus PCC 7942 (Anacystis nidulans). Data extending to 1.8-A resolution were collected at 140 K and the structure was refined to an Rwork of 0.196 and an Rfree of 0.250 for reflections with I > 0. The final model contains 3,224 non-hydrogen atoms per asymmetric unit, including 62 flavin mononucleotide (FMN) atoms, 354 water molecules, four calcium ions, four sodium ions, two chloride ions, and two Bis-Tris buffer molecules. The structure of the protein in the trigonal form (space group P312, a = 78.83, c = 52.07 A) was solved by molecular replacement using the coordinates from the orthorhombic structure, and was refined with all data from 10.0 to 2.6 A (R = 0.191; Rfree = 0.249). The sequence Tyr 58-Tyr 59, in a bend near the FMN, has so far been found only in the flavodoxins from E. coli and Haemophilus influenzae, and may be important in interactions of flavodoxin with its partners in activation reactions. The tyrosine residues in this bend are influenced by intermolecular contacts and adopt different orientations in the two crystal forms. Structural comparisons with flavodoxins from Synechococcus PCC 7942 and Anaebaena PCC 7120 suggest other residues that may also be critical for recognition by methionine synthase.     

Crystal structure of Pseudomonas mevalonii HMG-CoA reductase at 3.0 angstrom resolution

The rate-limiting step in cholesterol biosynthesis in mammals is catalyzed by 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase, a four-electron oxidoreductase that converts HMG-CoA to mevalonate. The crystal structure of HMG-CoA reductase from Pseudomonas mevalonii was determined at 3.0 angstrom resolution by multiple isomorphous replacement. The structure reveals a tightly bound dimer that brings together at the subunit interface the conserved residues implicated in substrate binding and catalysis. These dimers are packed about a threefold crystallographic axis, forming a hexamer with 23 point group symmetry. Difference Fourier studies reveal the binding sites for the substrates HMG-CoA and reduced or oxidized nicotinamide adenine dinucleotide [NAD(H)] and demonstrate that the active sites are at the dimer interfaces. The HMG-CoA is bound by a domain with an unusual fold, consisting of a central alpha helix surrounded by a triangular set of walls of beta sheets and alpha helices. The NAD(H) is bound by a domain characterized by an antiparallel beta structure that defines a class of dinucleotide-binding domains.