Hemophilic factor VIII C1- and C2-domain missense mutations and their modeling to the 1.5-angstrom human C2-domain crystal structure.

Factor VIII C domains contain key binding sites for von Willebrand factor (vWF) and phospholipid membranes. Hemophilic patients were screened for factor VIII C-domain mutations to provide a well-characterized series. Mutated residues were localized to the high-resolution C2 structure and to a homology model of C1. Of 30 families found with mutations in the C domains, there were 14 missense changes, and 9 of these were novel. Of the missense mutations, 10 were associated with reduced vWF binding and 8 were at residues with surface-exposed side chains. Six of the 10 mutants had nearly equivalent factor VIII clotting activity and antigen level, suggesting that reduced vWF binding could cause hemophilia by reducing factor VIII stability in circulation. When the present series was combined with previously described mutations from an online international database, 11 C1 and C2 mutations in patients with mild or moderately severe hemophilia A were associated with antibody-inhibitor development in at least one affected individual. Of these substitutions, 6 occurred at surface-exposed residues. As further details of the C1 structure and its interface with C2 become available, and as binding studies are performed on the plasma of more patients with hemophilic C-domain mutations, prediction of surface binding sites should improve, allowing confirmation by site-specific mutagenesis of surface-exposed residues.

Structure of the C2 domain of human factor VIII at 1.5 A resolution.

Human factor VIII is a plasma glycoprotein that has a critical role in blood coagulation. Factor VIII circulates as a complex with von Willebrand factor. After cleavage by thrombin, factor VIIIa associates with factor IXa at the surface of activated platelets or endothelial cells. This complex activates factor X (refs 6, 7), which in turn converts prothrombin to thrombin in the presence of factor Va (refs 8, 9). The carboxyl-terminal C2 domain of factor VIII contains sites that are essential for its binding to von Willebrand factor and to negatively charged phospholipid surfaces. Here we report the structure of human factor VIII C2 domain at 1.5 A resolution. The structure reveals a beta-sandwich core, from which two beta-turns and a loop display a group of solvent-exposed hydrophobic residues. Behind the hydrophobic surface lies a ring of positively charged residues. This motif suggests a mechanism for membrane binding involving both hydrophobic and electrostatic interactions. The structure explains, in part, mutations in the C2 region of factor VIII that lead to bleeding disorders in haemophilia A.

The crystal structure of a bacterial, bifunctional 5,10 methylene-tetrahydrofolate dehydrogenase/cyclohydrolase.

The structure of a bifunctional 5,10-methylene-tetrahydrofolate dehydrogenase/cyclohydrolase from Escherichia coli has been determined at 2.5 A resolution in the absence of bound substrates and compared to the NADP-bound structure of the homologous enzyme domains from a trifunctional human synthetase enzyme. Superposition of these structures allows the identification of a highly conserved cluster of basic residues that are appropriately positioned to serve as a binding site for the poly-gamma-glutamyl tail of the tetrahydrofolate substrate. Modeling studies and molecular dynamic simulations of bound methylene-tetrahydrofolate and NADP shows that this binding site would allow interaction of the nicotinamide and pterin rings in the dehydrogenase active site. Comparison of these enzymes also indicates differences between their active sites that might allow the development of inhibitors specific to the bacterial target.

Crystal structure of human recombinant ornithine aminotransferase.

Ornithine aminotransferase (OAT), a pyridoxal-5'-phosphate dependent enzyme, catalyses the transfer of the delta-amino group of L-ornithine to 2-oxoglutarate, producing L-glutamate-gamma-semialdehyde, which spontaneously cyclizes to pyrroline-5-carboxylate, and L-glutamate. The crystal structure determination of human recombinant OAT is described in this paper. As a first step, the structure was determined at low resolution (6 A) by molecular replacement using the refined structure of dialkylglycine decarboxylase as a search model. Crystallographic phases were then refined and extended in a step-wise fashion to 2.5 A by cyclic averaging of the electron density corresponding to the three monomers within the asymmetric unit. Interpretation of the resulting map was straightforward and refinement of the model resulted in an R-factor of 17.1% (Rfree=24.3%). The success of the procedure demonstrates the power of real-space molecular averaging even with only threefold redundancy. The alpha6-hexameric molecule is a trimer of intimate dimers with a monomer-monomer interface of 5500 A2 per subunit. The three dimers are related by an approximate 3-fold screw axis with a translational component of 18 A. The monomer fold is that of a typical representative of subgroup 2 aminotransferases and very similar to those described for dialkylglycine decarboxylase from Pseudomonas cepacia and glutamate-1-semialdehyde aminomutase from Synechococcus. It consists of a large domain that contributes most to the subunit interface, a C-terminal small domain most distant to the 2-fold axis and an N-terminal region that contains a helix, a loop and a three stranded beta-meander embracing a protrusion in the large domain of the second subunit of the dimer. The large domain contains the characteristic central seven-stranded beta-sheet (agfedbc) covered by eight helices in a typical alpha/beta fold. The cofactor pyridoxal-5'-phosphate is bound through a Schiff base to Lys292, located in the loop between strands f and g. The C-terminal domain includes a four-stranded antiparallel beta-sheet in contact with the large domain and three further helices at the far end of the subunit. The active sites of the dimer lie, about 25 A apart, at the subunit and domain interfaces. The conical entrances are on opposite sides of the dimer. In the active site, R180, E235 and R413 are probable substrate binding residues. Structure-based sequence comparisons with related transaminases in this work support that view. In patients suffering from gyrate atrophy, a recessive hereditary genetic disorder that can cause blindness in humans, ornithine aminotransferase activity is lacking. A large number of frameshift and point mutations in the ornithine aminotransferase gene have been identified in such patients. Possible effects of the various point mutations on the structural stability or the catalytic competence of the enzyme are discussed in light of the three-dimensional structure.

Human ornithine aminotransferase complexed with L-canaline and gabaculine: structural basis for substrate recognition.

BACKGROUND: Ornithine aminotransferase (OAT) is a 45 kDa pyridoxal-5'-phosphate (PLP)-dependent enzyme that catalyzes the conversion of L-ornithine and 2-oxoglutarate to glutamate-delta-semialdehyde and glutamic acid, respectively. In humans, loss of OAT function causes an accumulation of ornithine that results in gyrate atrophy of the choroid and retina, a disease that progressively leads to blindness. In an effort to learn more about the structural basis of this enzyme's function, we have determined the X-ray structures of OAT in complex with two enzyme-activated suicide substrates: L-canaline, an ornithine analog, and gabaculine, an irreversible inhibitor of several related aminotransferases. RESULTS: The structures of human OAT bound to the inhibitors gabaculine and L-canaline were solved to 2.3 A at 110K by difference Fourier techniques. Both inhibitors coordinate similarly in the active site, binding covalently to the PLP cofactor and causing a 20 degrees rotation in the cofactor tilt relative to the ligand-free form. Aromatic-aromatic interactions occur between the bound gabaculine molecule and active-site residues Tyr85 and Phe177, whereas Tyr55 and Arg180 provide specific contacts to the alpha-amino and carboxyl groups of L-canaline. CONCLUSIONS: The OAT-L-canaline complex structure implicates Tyr55 and Arg180 as the residues involved in coordinating with the natural substrate ornithine during normal enzyme turnover. This correlates well with two enzyme-inactivating point mutations associated with gyrate atrophy, Tyr55-->His and Arg180-->Thr. The OAT-gabaculine complex provides the first structural evidence that the potency of the inhibitor is due to energetically favourable aromatic interactions with residues in the active site. This aromatic-binding mode may be relevant to structure-based drug design efforts against other omega-aminotransferase targets, such as GABA aminotransferase.

Crystallization and preliminary X-ray diffraction studies of recombinant human ornithine aminotransferase.

Human liver ornithine aminotransferase was expressed in Escherichia coli and purified by ammonium sulfate fractionation and anion exchange column chromatography. The purified recombinant enzyme is fully active and crystallized readily over a wide range of polyethylene glycol concentrations. The crystals belong to the trigonal space group P3(1)21 (or its enantiomorph P3(2)21) with unit cell parameters a = b = 116.3 A, and c = 190.0 A, alpha = beta = 90 degrees, gamma = 120 degrees. There are three monomers per asymmetric unit. Self-rotation function studies revealed both 2-fold and 3-fold non-crystallographic symmetry, with the local 3-fold axis being tilted 15 degrees from the c axis and perpendicular to a crystallographic dyad. A complete native data set to 2.3 A resolution was collected using synchrotron radiation.

Ultrastructure of the intact skeleton of the human erythrocyte membrane.

Filamentous skeletons were liberated from isolated human erythrocyte membranes in Triton X-100, spread on fenestrated carbon films, negatively stained, and viewed intact and unfixed in the transmission electron microscope. Two forms of the skeleton were examined: (a) basic skeletons, stripped of accessory proteins with 1.5 M NaCl so that they contain predominantly polypeptide bands 1, 2, 4.1, and 5; and (b) unstripped skeletons, which also bore accessory proteins such as ankyrin and band 3 and small plaques of residual lipid. Freshly prepared skeletons were highly condensed. Incubation at low ionic strength and in the presence of dithiothreitol for an hour or more caused an expansion of the skeletons, which greatly increased the visibility of their elements. The expansion may reflect the opening of spectrin from a compact to an elongated disposition. Expanded skeletons appeared to be organized as networks of short actin filaments joined by multiple (5-8) spectrin tetramers. In unstripped preparations, globular masses were observed near the centers of the spectrin filaments, probably corresponding to complexes of ankyrin with band 3 oligomers. Some of these globules linked pairs of spectrin filaments. Skeletons prepared with a minimum of perturbation had thickened actin protofilaments, presumably reflecting the presence of accessory proteins. The length of these actin filaments was highly uniform, averaging 33 +/- 5 nm. This is the length of nonmuscle tropomyosin. Since there is almost enough tropomyosin present to saturate the F-actin, our data support the hypothesis that tropomyosin may determine the length of actin protofilaments in the red cell membrane.

Lipid-protein interaction at solid-water interface. Adsorption of human apo-high density lipoprotein to amphiphilic interfaces.

Hydrophobic glass beads with well characterized physical properties were used as a model system to study at the amphiphilic interface the properties of apolipoproteins A-I and A-II from human serum high density lipoproteins. In this study, spherical glass beads with known diameter were coated covalently with a film of silicone to varying surface density. The decrease in surface tension induced by coating was directly related to the increase in silicone film density and likely to the hydrophobicity of the glass surface. The adsorption of apo-A-I and apo-A-II to the hydrophobic glass bead surface was determined by following the decrease of 1) the radioactivity of preparations of 125I-iodinated proteins from the solution, 2) the UV absorbance of the solution at 206 nm, and 3) the fluorescence emitted by the complex formed between free protein and Fluram II in solution. All of the three measurements gave identical results. Both proteins adsorbed rapidly and reversibly to the hydrophobic glass surface. The adsorption isotherms followed the Langmuir equation with apo-A-II showing a higher surface affinity; delta Gaff = RT ln Kd has a value of -9.1 kcal/mol and -10.5 kcal/mol for apo A-I and apo A-II, respectively. The addition of canine serum high density lipoprotein (HDL) to the above system caused a rapid desorption of apolipoproteins from the beads into the aqueous phase and adsorption onto the HDL surface with no detectable structural changes of this lipoprotein. The results indicate that apo-A-I and apo-A-II can reversibly be adsorbed at a solid hydrophobic surface and that these apoproteins are capable of moving into a HDL particle if added to the system via a solution phase. The data suggest that the rate limiting aspect of the desorption-adsorption processes is the concentration of the apoproteins in solution.

Ultrastructure of unit fragments of the skeleton of the human erythrocyte membrane.

We have examined fragments of the filamentous network underlying the human erythrocyte membrane by high-resolution electron microscopy. Networks were released from ghosts by extraction with Triton X-100, freed of extraneous proteins in 1.5 M NaCl, and collected by centrifugation onto a sucrose cushion. These preparations contained primarily protein bands 1 + 2 (spectrin), band 4.1 and band 5 (actin). The networks were partially disassembled by incubation at 37 degrees C in 2 mM NaPi (pH 7), which caused the preferential dissociation of spectrin tetramers to dimers. The fragments so generated were fractionated by gel filtration chromatography and visualized by negative staining with uranyl acetate on fenestrated carbon films. Unit complexes, which sedimented at approximately 40S, contained linear filaments approximately 7-8 nm diam from which several slender and convoluted filaments projected. The linear filaments had a mean length of 52 +/- 17 nm and a serrated profile reminiscent of F-actin. They could be decorated in an arrowhead pattern with S1 fragments of muscle heavy meromyosin which, incidentally, displaced the convoluted filaments. Furthermore, the linear filaments nucleated the polymerization of rabbit muscle G-actin, predominantly but not exclusively from the fast-growing ends. On this basis, we have identified the linear filaments as F-actin; we infer that the convoluted filaments are spectrin. Spectrin molecules were usually attached to actin filaments in clusters that showed a preference for the ends of the F-actin. We also observed free globules up to 15 nm diam, usually associated with three spectrin molecules, which also nucleated actin polymerization; these may be simple junctional complexes of spectrin, actin, and band 4.1. In larger ensembles, spectrin tetramers linked actin filaments and/or globules into irregular arrays. Intact networks were an elaboration of the basic pattern manifested by the fragments. Thus, we have provided ultrastructural evidence that the submembrane skeleton is organized, as widely inferred from less direct information, into short actin filaments linked by multiple tetramers of spectrin clustered at sites of association with band 4.1.

Glycosphingolipid-high density lipoprotein-3 interactions. I. Transfer of glycosphingolipid from phosphatidylcholine vesicles to high density lipoprotein-3.

Single bilayer vesicles (d less than 1.02 g/ml) of 3H-glycosphingolipids and [14C]phosphatidylcholine in the molar ratio of 1:7 were prepared by ethanolic injection of the lipid mixture into buffer, concentrated, and incubated with human serum high density lipoprotein-3 (HDL3; d = .14 g/ml) at 37 degrees C. Equilibrium ultracentrifugation of the incubation mixtures on a 0-22% NaBr gradient revealed the presence of three discrete lipid-protein complexes of density 1.03, 1.06, and 1.12 g/ml (Peaks I, II, and III, respectively). Each peak was homogeneous upon reultracentrifugation and the protein and radioactivity eluted as a single peak upon Sepharose CL-6B chromatography. Compositional analysis showed peak I to contain 2.6% protein (apo-A-I peptide) and 4.3% cholesterol, peak II to contain 17.6% protein (apo-A-I peptide) and 6.3% cholesterol, and peak III to have a composition similar to HDL3. Electron microscopy of negatively stained samples confirmed the homogeneity of the peaks and the similarity between peak III and HDL3. Peak II particles were larger than HDL3; peak I particles resembled fused or aggregated vesicles which could be removed by ultracentrifugation; disc-shaped particles were not seen in any of the fractions. Direct incubation of HDL3 or human serum with 3H-glycosphingolipid dispersions did not yield a glycolipid . HDL3 complex as judged by density gradient ultracentrifugation and Sepharose CL-6B chromatography. However, incubation of 3H-glycolipid/phosphatidylcholine vesicles with serum did result in transfer of 3H-glycolipid to the HDL fraction. It was concluded that glycolipids incorporated into a lipid membrane structure can interact with, and become incorporated into, high density lipoprotein.

Structure of human serum lipoproteins inferred from compositional analysis.

Analysis of the correlations between size and chemical composition of lipoproteins of normolipidemic human plasma shows that the structure of all circulating lipoproteins is consistent with a spherical model of radius r in which a spherical liquid core of cholesterol esters and triglycerides of radius = r --20.2 A is surrounded by a monolayer of cholesterol and phospholipids with closely hydrophobic ends on the surface of the core. The average molecular areas at this inner surface are Spl = 68.5 A2/molecule for phospholipids and Sc= 39.1 A2/molecule for cholesterol. The proteins are closely packed with the hydrophilic head groups of phospholipids at the outer surface of the particle, with S' pl = 62.7 A2/molecule for phospholipids and Saa = 15.6 A2/amino acid for proteins. The polar head group of free cholesterol does not participate in the packing of the outer layer and thus must be masked by proteins. Free cholesterol is distributed among the circulating lipoproteins--with the exception of very high density lipoprotein and perhaps chylomicrons--according to a thermodynamic equilibrium governed by the curvature of the surface of the particle.