The mechanism of regulation of hexokinase: new insights from the crystal structure of recombinant human brain hexokinase complexed with glucose and glucose-6-phosphate.

BACKGROUND: Hexokinase I is the pacemaker of glycolysis in brain tissue. The type I isozyme exhibits unique regulatory properties in that physiological levels of phosphate relieve potent inhibition by the product, glucose-6-phosphate (Gluc-6-P). The 100 kDa polypeptide chain of hexokinase I consists of a C-terminal (catalytic) domain and an N-terminal (regulatory) domain. Structures of ligated hexokinase I should provide a basis for understanding mechanisms of catalysis and regulation at an atomic level. RESULTS: The complex of human hexokinase I with glucose and Gluc-6-P (determined to 2.8 A resolution) is a dimer with twofold molecular symmetry. The N- and C-terminal domains of one monomer interact with the C- and N-terminal domains, respectively, of the symmetry-related monomer. The two domains of a monomer are connected by a single alpha helix and each have the fold of yeast hexokinase. Salt links between a possible cation-binding loop of the N-terminal domain and a loop of the C-terminal domain may be important to regulation. Each domain binds single glucose and Gluc-6-P molecules in proximity to each other. The 6-phosphoryl group of bound Gluc-6-P at the C-terminal domain occupies the putative binding site for ATP, whereas the 6-phosphoryl group at the N-terminal domain may overlap the binding site for phosphate. CONCLUSIONS: The binding synergism of glucose and Gluc-6-P probably arises out of the mutual stabilization of a common (glucose-bound) conformation of hexokinase I. Conformational changes in the N-terminal domain in response to glucose, phosphate, and/or Gluc-6-P may influence the binding of ATP to the C-terminal domain.

Refined crystal structures of glucoamylase from Aspergillus awamori var. X100.

The refined crystal structures of a proteolytic fragment of glucoamylase from Aspergillus awamori var. X100 have been determined at pH 6 and 4 to a resolution of 2.2 A and 2.4 A, respectively. The models include the equivalent of residues 1 to 471 of glucoamylase from Aspergillus niger and a complete interpretation of the solvent structure. The R-factors of the pH 6 and 4 structures are 0.14 and 0.12, respectively, with root-mean-square deviations of 0.014 A and 0.012 A from expected bondlengths. The enzyme has the general shape of a doughnut. The "hole" of the doughnut consists of a barrier of hydrophobic residues at the center, which separates two water-filled voids, one of which serves as the active site. Three clusters of water molecules extend laterally from the active site. One of the lateral clusters connects the deepest recess of the active site to the surface of the enzyme. The most significant difference in the pH 4 and 6 structures is the thermal parameter of water 500, the putative nucleophile in the hydrolysis of maltooligosaccharides. Water 500 is associated more tightly with the enzyme at pH 4 (the pH of optimum catalysis) than at pH 6. In contrast to water 500, Glu179, the putative catalytic acid of glucoamylase, retains the same conformation in both structures and is in an environment that would favor the ionized, rather than the acid form of the side-chain. Glycosyl chains of 5 and 8 sugar residues are linked to Asparagines 171 and 395, respectively. The conformations of the two glycosyl chains are similar, being superimposable on each other with a root-mean-square discrepancy of 1.9 A. The N-glycosyl chains hydrogen bond to the surface of the protein through their terminal sugars, but otherwise do not interact strongly with the enzyme. The structures have ten serine/threonine residues, to each of which is linked a single mannose sugar. The structure of the ten O-glycosylated residues taken together suggests a well-defined conformation for proteins that have extensive O-glycosylation of their polypeptide chain.

Crystal structures of adenylosuccinate synthetase from Escherichia coli complexed with GDP, IMP hadacidin, NO3-, and Mg2+.

Crystal structures of adenylosuccinate synthetase from Esherichia coli complexed with Mg2+, IMP, GDP, NO3- and hadacidin at 298 and 100 K have been refined to R-factors of 0.188 and 0.206 against data to 2.8 A and 2.5 A resolution, respectively. Conformational changes of up to 9 A relative to the unligated enzyme occur in loops that bind to Mg2+, GDP, IMP and hadacidin. Mg2+ binds directly to GDP, NO3-, hadacidin and the protein, but is only five-coordinated. Asp13, which approaches, but does not occupy the sixth coordination site of Mg2+, hydrogen bonds to N1 of IMP. The nitrogen atom of NO3- is approximately 2.7 A from O6 of IMP, reflecting a strong electrostatic interaction between the electron-deficient nitrogen atom and the electron-rich O6. The spatial relationships between GDP, NO3- and Mg2+ suggest an interaction between the beta,gamma-bridging oxygen atom of GTP and Mg2+ in the enzyme-substrate complex. His41 hydrogen bonds to the beta-phosphate group of GDP and approaches bound NO3-. The aldehyde group of hadacidin coordinates to the Mg2+, while its carboxyl group interacts with backbone amide groups 299 to 303 and the side-chain of Arg303. The 5'-phosphate group of IMP interacts with Asn38, Thr129, Thr239 and Arg143 (from a monomer related by 2-fold symmetry). A mechanism is proposed for the two-step reaction governed by the synthetase, in which His41 and Asp13 are essential catalytic side-chains.

Refinement of a molecular model for lamprey hemoglobin from Petromyzon marinus.

A molecular model for the protein and ambient solvent of the complex of cyanide with methemoglobin V from the sea lamprey Petromyzon marinus yields an R-factor of 0.142 against X-ray diffraction data to 2.0 A resolution. The root-mean-square discrepancies from ideal bond length and angle are, respectively, 0.014 A and 1.5 degrees. Atoms that belong to planar groups deviate by 0.012 A from planes determined by a least-squares procedure. The average standard deviation for chiral volumes, peptide torsion angle and torsion angles of side-chains are 0.150 A3, 2.0 degrees and 19.4 degrees, respectively. The root-mean-square variation in the thermal parameters of bonded atoms of the polypeptide backbone is 1.21 A2; the variation in thermal parameters for side-chain atoms is 2.13 A2. The model includes multiple conformations for 11 side-chains of the 149 amino acid residues of the protein. We identify 231 locations as sites of water molecules in full or partial occupancy. The sum of occupancy factors for these sites is approximately 154, representing 28% of the 550 molecules of water within the crystallographic asymmetric unit. The environment of the heme in the cyanide complex of lamprey methemoglobin resembles the deoxy state of the mammalian tetramer. In particular, the bond between atom NE2 of the proximal histidine and the Fe lies 5.1 degrees from the normal of the heme plane. In deoxy- and carbonmonoxyhemoglobins, the deviations from the normal to the heme plane are 7 to 8 degrees and 1 degree, respectively. Furthermore, the inequality in the distance of atom CD2 of the proximal histidine from the pyrrole nitrogen of ring-C of the heme (distance = 3.29 A) and CE1 from the pyrrole nitrogen of ring-A (distance = 3.06 A) is characteristic of deoxyhemoglobin, not carbonmonoxyhemoglobin, where these distances are equal. Finally, a hydrogen bond exists between carbonyl 111 and the hydroxyl of tyrosine 149. The corresponding hydrogen link in the mammalian tetramer is central to the T to R state transition and is present in deoxyhemoglobin but absent in carbonmonoxyhemoglobin. We suggest that the low affinity of oxygen for lamprey hemoglobin may be a consequence of these T-state geometries.

Refined crystal structures of unligated adenylosuccinate synthetase from Escherichia coli.

Crystal structures of unligated adenylosuccinate synthetase from Escherichia coli in space groups P2(1) and P2(1)2(1)2(1) have been refined to R-factors of 0.199 and 0.206 against data to 2.0 and 2.5 A, respectively. Bond lengths and angles deviate from expected values by 0.011 A and 1.7 degrees for the P2(1) crystal form and by 0.015 A and 1.7 degrees for the P2(1)2(1)2(1) crystal form. The fold of the polypeptide chain is dominated by a central beta-sheet, which is composed of nine parallel strands and a tenth antiparallel strand. Extending off from this central beta-sheet are four subdomains. The four subdomains contribute loops of residues that are disordered or have high thermal parameters. At least three of these loops (residues 42 to 52, 120 to 131 and 298 to 304) contribute essential residues to the putative active site of the synthetase. In the absence of ligands, much of the active site of the synthetase exists in an ill-defined conformational state. Two, nearly independent regions contribute residues to the interface between polypeptide chains of the synthetase dimer. A pair of helices (H4 and H5) interact with their symmetry-equivalent mates by way of residues that are not conserved amongst the known sequences of the synthetase. The second interface region involves conserved residues belonging to structural elements that connect strands of the central beta-sheet. Residues putatively involved in the binding of IMP lie at or near the interface between polypeptide chains of the dimer. Of the four sequence elements putatively common to all GTP hydrolases, the synthetase has only the guanine recognition element and a glycine-rich loop (P-loop). Although the base recognition element is essentially identical with those of the p21 ras and G alpha proteins, the P-loop of the synthetase is extended in size relative to the P-loops of other GTP hydrolases. The P-loop has two acid residues (Asp13 and Glu14), which are found in the P-loops of only the synthetase family. Glu14 may be involved in the stabilization of the enlarged P-loop of the synthetase, whereas Asp13 may play a role in catalysis and in the coordination of Mg2+. The structural elements of the p21 ras and G alpha proteins responsible for binding Mg2+ are either absent from the synthetase or unavailable for the coordination of metal cations.

Structural asymmetry in the CTP-liganded form of aspartate carbamoyltransferase from Escherichia coli.

The protein and solvent structure of the CTP-liganded form of aspartate carbamoyltransferase from Escherichia coli yields an R-factor of 0.155 for data to a resolution of 2.6 A. The model has 7353 protein atoms, 945 sites for solvent, and two molecules of CTP. A total of 25 of the 912 residues of the model exist in more than one conformation. The root-mean-square deviation of bond lengths and angles from their ideal values is 0.013 A and 2.1 degrees, respectively. The model reported here reflects a correction in the trace of the regulatory chain. One molecule of CTP binds to each of the two regulatory chains of the asymmetric unit of the crystal. The interactions between the pyrimidine of each CTP molecule and the protein are similar. The 4-amino group of CTP binds to the carbonyl groups of residues 89 (tyrosine) and 12 (isoleucine) of the regulatory chain. The nitrogen of position 3 of the pyrimidine binds to the amide group of residue 12; the 2-keto group binds to lysine 60. The 2'-OH group of the ribose forms hydrogen bonds with lysine 60 and the carbonyl group of residue 9 (valine). The binding of the phosphate groups of CTP to the regulatory chain probably reflects an incomplete association of CTP with the enzyme at pH 5.8. A lattice contact influences the interaction between the triphosphate group of one CTP molecule and the protein. For the other CTP molecule, only lysine 94 binds to the phosphate groups of CTP. Of the two regulatory and two catalytic chains of the asymmetric unit of the crystal, there are only two significant violations of non-crystallographic symmetry. The active site in the vicinity of arginine 54 of one catalytic chain is larger than the active site of its non-crystallographic mate. The "expanded" cavity accommodates four solvent molecules in the vicinity of arginine 54 as opposed to two molecules of water for the "contracted" cavity. Furthermore, arginine 54 in the "expanded" pocket adopts two conformations, either hydrogen-bonding to glutamate 86 or to the phenolic oxygen atom of tyrosine 98; residues 86 and 98 are in a catalytic chain related by 3-fold symmetry to the catalytic chain of arginine 54. In the "contracted" pocket, arginine 54 binds only to glutamate 86.(ABSTRACT TRUNCATED AT 400 WORDS)

Preliminary X-ray crystallographic study of cyanase from Escherichia coli.

Cyanase, an oligomeric enzyme of Escherichia coli that catalyzes the decomposition of cyanate to ammonia and bicarbonate, crystallizes in the space group P1 with unit cell parameters a = 85.96 A, b = 83.17 A, c = 83.28 A, alpha = 110.29 degrees, beta = 118.29 degrees and gamma = 72.40 degrees. Crystals diffract to a resolution of at least 2.5 A. The crystal data, in conjunction with a subunit molecular weight of 17,008, suggest that two oligomers are in the asymmetric unit of the crystal and that eight subunits comprise a single oligomer.

Preliminary X-ray crystallographic study of adenylosuccinate synthetase from Escherichia coli.

Adenylosuccinate synthetase, a dimeric enzyme of 96,000 Mr, catalyzes the first committed step toward the de novo biosynthesis of AMP. Large, single crystals of adenylosuccinate synthetase from Escherichia coli grow from solutions of polyethylene glycol and ammonium sulfate. Crystals from ammonium sulfate belong to the orthorhombic space group P212121 with unit cell parameters a = 79.0 A, b = 70.2 A and c = 152.6 A. Crystals from polyethylene glycol belong to the space group P21, having unit cell parameters of a = 71.16 A, b = 71.99 A, c = 82.95 A, and beta = 71.52 degrees. The asymmetric units of both crystal forms probably contain the entire dimeric enzyme.

Complex of N-phosphonacetyl-L-aspartate with aspartate carbamoyltransferase. X-ray refinement, analysis of conformational changes and catalytic and allosteric mechanisms.

The allosteric enzyme aspartate carbamoyltransferase of Escherichia coli consists of six regulatory chains (R) and six catalytic chains (C) in D3 symmetry. The less active T conformation, complexed to the allosteric inhibitor CTP has been refined to 2.6 A (R-factor of 0.155). We now report refinement of the more active R conformation, complexed to the bisubstrate analog N-phosphonacetyl-L-aspartate (PALA) to 2.4 A (R-factor of 0.165, root-mean-square deviations from ideal bond distances and angles of 0.013 A and 2.2 degrees, respectively). The antiparallel beta-sheet in the revised segment 8-65 of the regulatory chain of the T conformation is confirmed in the R conformation, as is also the interchange of alanine 1 with the side-chain of asparagine 2 in the catalytic chain. The crystallographic asymmetric unit containing one-third of the molecule (C2R2) includes 925 sites for water molecules, and seven side-chains in alternative conformations. The gross conformational changes of the T to R transition are confirmed, including the elongation of the molecule along its threefold axis by 12 A, the relative reorientation of the catalytic trimers C3 by 10 degrees, and the rotation of the regulatory dimers R2 about the molecular twofold axis by 15 degrees. No changes occur in secondary structure. Essentially rigid-body transformations account for the movement of the four domains of each catalytic-regulatory unit; these include the allosteric effector domain, the equatorial (aspartate) domain, and the combination of the polar (carbamyl phosphate) and zinc domain, which moves as a rigid unit. However, interfaces change, for example the interface between the zinc domain of the R chain and the equatorial domain of the C chain, is nearly absent in the T state, but becomes extensive in the R state of the enzyme; also one catalytic-regulatory interface (C1-R4) of the T state disappears in the more active R state of the enzyme. Segments 50-55, 77-86 and 231-246 of the catalytic chain and segments 51-55, 67-72 and 150-153 of the regulatory chain show conformational changes that go beyond the rigid-body movement of their corresponding domains. The localized conformational changes in the catalytic chain all derive from the interactions of the enzyme with the inhibitor PALA; these changes may be important for the catalytic mechanism. The conformation changes in segments 67-72 and 150-153 of the regulatory chain may be important for the allosteric control of substrate binding. On the basis of the conformational differences of the T and R states of the enzyme, we present a plausible scheme for catalysis that assumes the ordered binding of substrates and the ordered release o

Regulation of hexokinase I: crystal structure of recombinant human brain hexokinase complexed with glucose and phosphate.

Hexokinase I, the pacemaker of glycolysis in brain tissue and red blood cells, is comprised of two similar domains fused into a single polypeptide chain. The C-terminal half of hexokinase I is catalytically active, whereas the N-terminal half is necessary for the relief of product inhibition by phosphate. A crystalline complex of recombinant human hexokinase I with glucose and phosphate (2.8 A resolution) reveals a single binding site for phosphate and glucose at the N-terminal half of the enzyme. Glucose and phosphate stabilize the N-terminal half in a closed conformation. Unexpectedly, glucose binds weakly to the C-terminal half of the enzyme and does not by itself stabilize a closed conformation. Evidently a stable, closed C-terminal half requires either ATP or glucose 6-phosphate along with glucose. The crystal structure here, in conjunction with other studies in crystallography and directed mutation, puts the phosphate regulatory site at the N-terminal half, the site of potent product inhibition at the C-terminal half, and a secondary site for the weak interaction of glucose 6-phosphate at the N-terminal half of the enzyme. The relevance of crystal structures of hexokinase I to the properties of monomeric hexokinase I and oligomers of hexokinase I bound to the surface of mitochondria is discussed.

Crystal structures of mutant monomeric hexokinase I reveal multiple ADP binding sites and conformational changes relevant to allosteric regulation.

Hexokinase I, the pacemaker of glycolysis in brain tissue, is composed of two structurally similar halves connected by an alpha-helix. The enzyme dimerizes at elevated protein concentrations in solution and in crystal structures; however, almost all published data reflect the properties of a hexokinase I monomer in solution. Crystal structures of mutant forms of recombinant human hexokinase I, presented here, reveal the enzyme monomer for the first time. The mutant hexokinases bind both glucose 6-phosphate and glucose with high affinity to their N and C-terminal halves, and ADP, also with high affinity, to a site near the N terminus of the polypeptide chain. Exposure of the monomer crystals to ADP in the complete absence of glucose 6-phosphate reveals a second binding site for adenine nucleotides at the putative active site (C-half), with conformational changes extending 15 A to the contact interface between the N and C-halves. The structures reveal distinct conformational states for the C-half and a rigid-body rotation of the N-half, as possible elements of a structure-based mechanism for allosteric regulation of catalysis.

Spontaneous subunit exchange in porcine liver fructose-1,6-bisphosphatase.

No evidence to date suggests the possibility of subunit exchange between tetramers of mammalian fructose-1,6-bisphosphatase. An engineered fructose-1,6-bisphosphatase, with subunits of altered electrostatic charge, exhibits spontaneous subunit exchange with wild-type enzyme in the absence of ligands. The exchange process reaches equilibrium in approximately 5 h at 4 degrees C, as monitored by non-denaturing gel electrophoresis and anion exchange chromatography. Active site ligands, such as fructose 6-phosphate, abolish subunit exchange at the level of the monomer, but permit dimer-dimer exchanges. AMP, alone or in the presence of active site ligands, abolishes all exchange processes. Exchange phenomena may play a role in the kinetic mechanism of allosteric regulation of fructose-1,6-bisphosphatase.

Crystallization and preliminary X-ray analysis of human brain hexokinase.

Human brain hexokinase type I, expressed in Escherichia coli, has been crystallized from polyethylene glycol 8000 in the presence of inorganic phosphate. The crystals are hexagonal needles of diameter 0.25 mm, diffracting to a resolution of 3.5 A on a rotating-anode/area-detector system. The crystals belong to the space group P3(1)21/P3(2)21 with cell dimensions a = b = 171.5 A and c = 99.4 A. The specific volume of the crystal is 4.2 A3/Da, suggesting an asymmetric unit with a single 100 kDa molecule and a solvent content of 71% by volume or two molecules of hexokinase with a solvent content of 41%. The complex of hexokinase with glucose crystallizes under similar conditions, giving crystals of the same morphology.

Multiple crystal forms of hexokinase I: new insights regarding conformational dynamics, subunit interactions, and membrane association.

Hexokinase I is comprised of homologous N- and C-terminal domains, and binds to the outer membrane of mitochondria. Reported here is the structure of a new crystal form of recombinant human hexokinase I, which complements existing crystal structures. Evidently, in some packing environments and even in the presence of glucose and glucose 6-phosphate the N-terminal domain (but not the C-terminal domain) can undergo oscillations between closed and partially opened conformations. Subunit interfaces, present in all known crystal forms of hexokinase I, promote the formation of linear chains of hexokinase I dimers. Presented is a model for membrane-associated hexokinase I, in which linear chains of hexokinase I dimers are stabilized by interactions with mitochondrial porin.

Refined structure for the complex of D-gluco-dihydroacarbose with glucoamylase from Aspergillus awamori var. X100 to 2.2 A resolution: dual conformations for extended inhibitors bound to the active site of glucoamylase.

The crystal structure at pH 4 of the complex of glucoamylase II(471) from Aspergillus awamori var. X100 with the pseudotetrasaccharide D-gluco-dihydroacarbose has been refined to an R-factor of 0.125 against data to 2.2 A resolution. The first two residues of the inhibitor bind at a position nearly identical to those of the closely related inhibitor acarbose in its complex with glucoamylase at pH 6. However, the electron density bifurcates beyond the second residue of the D-gluco-dihydroacarbose molecule, placing the third and fourth residues together at two positions in the active site. The position of relatively low density (estimated occupancy of 35%) corresponds to the location of the third and fourth residues of acarbose in its complex with glucoamylase at pH 6. The position of high density (65% occupancy) corresponds to a new binding mode of an extended inhibitor to the active site of glucoamylase. Presented are possible causes for the binding of D-gluco-dihydroacarbose in two conformations at the active site of glucoamylase at pH 4.

Crystal structure of S-glutathiolated carbonic anhydrase III.

S-Glutathiolation of carbonic anhydrase III (CAIII) occurs rapidly in hepatocytes under oxidative stress. The crystal structure of the S-glutathiolated CAIII from rat liver reveals covalent adducts on cysteines 183 and 188. Electrostatic charge and steric contacts at each modification site inversely correlate with the relative rates of reactivity of these cysteines toward glutathione (GSH). Diffuse electron density associated with the GSH adducts suggests a lack of preferred bonding interactions between CAIII and the glutathionyl moieties. Hence, the GSH adducts are available for binding by a protein capable of reducing this mixed disulfide. These properties are consistent with the participation of CAIII in the protection/recovery from the damaging effects of oxidative agents.

Environment of tryptophan 57 in porcine fructose-1,6-bisphosphatase studied by time-resolved fluorescence and site-directed mutagenesis.

The environment of Trp57, introduced by the mutation of a tyrosine in the dynamic loop of porcine liver fructose-1,6-bisphosphatase (FBPase), was examined using time-resolved fluorescence and directed mutation. The Trp57 enzyme was studied previously by X-ray crystallography and steady-state fluorescence, the latter revealing an unexpected redshift in the wavelength of maximum fluorescence emission for the R-state conformer. The redshift was attributed to the negative charge of Asp127 in contact with the indole side chain of Trp57. Time-resolved fluorescence experiments here reveal an indole side chain less solvent exposed and more rigid in the R-state, than in the T-state of the enzyme, consistent with X-ray crystal structures. Replacement of Asp127 with an asparagine causes a 6 nm blueshift in the wavelength of maximum fluorescence emission for the R-state conformer, with little effect on the emission maximum of the T-state enzyme. The data here support the direct correspondence between X-ray crystal structures of FBPase and conformational states of the enzyme in solution, and provide a clear example of the influence of microenvironment on the fluorescence properties of tryptophan.

Raman spectroscopy of avidin: secondary structure, disulfide conformation, and the environment of tyrosine.

An analysis of the amide I region of Raman spectra indicates that avidin has 10 +/- 5% and 55 +/- 4% of its residues in helical and beta-strand conformations, respectively. Predictions of secondary structure on the basis of the sequence of avidin are consistent with the high percentage of residues in the beta conformation. We observe no differences between the spectra of avidin in solution and in crystals nor is there a significant difference between the secondary structures of avidin and the complex of avidin with biotin. In addition, the ratio of the intensities of the tyrosine doublet at 826 and 855 cm-1 indicates the lone tyrosine side chain of an avidin subunit is in a strong hydrogen bond as a proton acceptor. The Raman data also indicate the single disulfide of an avidin subunit has dihedral angles of 0-50 degrees for each of its two C beta-S bonds and a dihedral angle of 85 +/- 20 degrees for its disulfide bond. We discuss the significance of these results in relation to findings of earlier work on avidin.