Fluid bilayer structure determination by the combined use of x-ray and neutron diffraction. II. "Composition-space" refinement method.

This is the second of two papers describing a method for the joint refinement of the structure of fluid bilayers using x-ray and neutron diffraction data. We showed in the first paper (Wiener, M. C., and S. H. White. 1990. Biophys. J. 59:162-173) that fluid bilayers generally consist of a nearly perfect lattice of thermally disordered unit cells and that the canonical resolution d/hmax is a measure of the widths of quasimolecular components represented by simple Gaussian functions. The thermal disorder makes possible a "composition space" representation in which the quasimolecular Gaussian distributions describe the number or probability of occupancy per unit length across the width of the bilayer of each component. This representation permits the joint refinement of neutron and x-ray lamellar diffraction data by means of a single quasimolecular structure that is fit simultaneously to both diffraction data sets. Scaling of each component by the appropriate neutron or x-ray scattering length maps the composition space profile to the appropriate scattering length space for comparison to experimental data. Other extensive properties, such as mass, can also be obtained by an appropriate scaling of the refined composition space structure. Based upon simple bilayer models involving crystal and liquid crystal structural information, we estimate that a fluid bilayer with hmax observed diffraction orders will be accurately represented by a structure with approximately hmax quasimolecular components. Strategies for assignment of quasimolecular components are demonstrated through detailed parsing of a phospholipid molecule based upon the one-dimensional projection of the crystal structure of dimyristoylphosphatidylcholine. Finally, we discuss in detail the number of experimental variables required for the composition space joint refinement. We find fluid bilayer structures to be marginally determined by the experimental data. The analysis of errors, which takes on particular importance under these circumstances, is also discussed.     

Purification of single-strand DNA binding protein from an Escherichia coli lysate using counter-current chromatography,partition and precipitation

Single-strand DNA binding protein (SSB) from Escherichia coli lysate was purified by counter-current chromatography (CCC) using the ammonium sulfate precipitation method in a coiled column. About 5 ml of E. coli lysate was separated by CCC using a polymer phase system composed of 16% (w/w) polyethylene glycol (PEG) 1000 and 17% (w/w) ammonium sulfate aqueous polymer two-phase solvent system. The precipitation of proteins in the lysate took place in the CCC column, and the SSB protein was eluted in the fraction 51-56. Many other impurities were either eluted immediately after the solvent front or precipitated in the column. The identities of the proteins in the fractions and in the precipitate were confirmed by SDS-polyacrylamide gel electrophoresis with Coomassie Brilliant Blue staining.     

Crystallization and preliminary x-ray investigation of soybean beta-amylase.

Beta-Amylase [1, 4-alpha-D-glucan maltohydrolase, EC 3.2.1.2] has been purified from defatted soybean meal by fractional precipation with ammonium sulfate, ion-exchange chromatography on CM- and DEAE-Sephadex and gel filtration chromatography on Sephadex G-100. Two different components of beta-amylase were crystallized from ammonium sulfate solutions, and the homogeneity of each preparation was confirmed by sedimentation and disc electrophoretic analyses. Both components of soybean beta-amylase formed large single crystals (trigonal crystal system) from 40--50 per cent saturated ammonium sulfate solution buffered at pH 5.4 on dialyzing concentrated protein solution in the apparatus of Zeppezauer et al. Preliminary X-ray diffraction data gave a hexagonal lattice with unit cell dimensions a=86.1 A and c=144.4 A. The space group corresponds to P3121 or P3221, and one asymmetric unit contains one molecule of beta-amylase, assuming a crystal density of 1.25 g/ml and a molecular weight of the enzyme of 60,000 daltons. In this case, the crystal has a volume of 2.53 A-3 per atomic mass unit, and the percentage of protein in the crystal is about 52.     

Display and interpretation of solvent electron density distributions in insulin crystals

In macromolecular crystallography, three-dimensional contour surfaces are useful for interactive computer graphics displays of the protein electron density but are less effective for presenting static images of large volumes of solvent density. A raster-based computer graphics program which displays depth-cued projections of continuous density distributions has been developed to analyze the distribution of solvent atoms in macromolecular crystals. Maps of the water distribution in the cubic insulin crystal show some well-ordered waters, which are bound to surrounding protein atoms by multiple hydrogen bonds, and an ill-defined solvent structure at a greater distance from the protein surface. Molecular dynamics calculations were used to assist in the interpretation of the time-varying solvent structure within two enclosed cavities in the crystal. Two water molecules that ligate a sodium ion were almost immobile during the simulation but the majority of water molecules were found to move rapidly between the density maxima identified from the crystallographic refinement.     

The hydration shell of myoglobin

The space in the unit cell of a metmyoglobin crystal not occupied by myoglobin atoms was filled with water using Monte Carlo calculations. Independent calculations with different amounts of water have been performed. Structure factors were calculated using the water coordinates thus obtained and the known coordinates of the myoglobin atoms. A comparison with experimental structure factors showed that both the low and the high resolution regime could be well reproduced with 814 Monte Carlo water molecules per unit cell with a B-value of 50 Ų. The Monte Carlo water molecules yield a smaller standard R-value (0.166) than using a homogeneous electron density for the simulation of the crystal water (R = 0.212). A reciprocal space refinement of the water and the protein coordinates has been performed. Monte Carlo calculations can be used to obtain information for crystallographically invisible parts of the unit cell and yield better coordinates for the visible part in the refinement. Correspondence to: F. Parak     

Analysis of solvent structure in proteins using neutron D2O-H2O solvent maps: pattern of primary and secondary hydration of trypsin.

A method of determining the water structure in protein crystals is described using neutron solvent difference maps. These maps are obtained by comparing the changes in diffracted intensities between two data sets, one in which H2O is the major solvent constituent, and a second in which D2O is the solvent medium. To a good first approximation, the protein atom contributions to the scattering intensities in both data sets are equal and cancel, but since H2O and D2O have very different neutron-scattering properties, their differences are accentuated to reveal an accurate representation of the solvent structure. The method also employs a series of density modification steps that impose known physical constraints on the density distribution function in the unit cell by making real space modifications directly to the density maps. Important attributes of the method are that (1) it is less subjective in the assignment of water positions than X-ray analysis; (2) there is threefold improvement in the signal-to-noise ratio for the solvent density; and (3) the iterative density modification produces a low-biased representation of the solvent density. Tests showed that water molecules with as low as 10% occupancy could be confidently assigned. About 300 water sites were assigned for trypsin from the refined solvent density; 140 of these sites were defined in the maps as discrete peaks, while the remaining were found within less-ordered channels of density. There is a very good correspondence between the sites in the primary hydration layer and waters found in the X-ray structure. Most water sites are clustered into H-bonding networks, many of which are found along intermolecular contact zones. The bound water is equally distributed between contacting apolar and polar atoms at the protein interface. A common occurrence at hydrophobic surfaces is that apolar atoms are circumvented by one or more waters that are part of a larger water network. When the effects on surface accessibility by neighboring molecules in the crystal lattice are taken into consideration, only about 29% of the surface does not interface ordered water. About 25% of the ordered water is found in the second hydration sphere. In many instances these waters bridge larger clusters of primary layer waters. It is apparent that, in certain regions of the crystal, the organization of ordered water reflects the characteristics of the crystal environment more than those of trypsin's surface alone.     

Structure of ferricytochrome c' from Rhodospirillum molischianum at 1.67 A resolution

The structure of ferricytochrome c' from Rhodospirillum molischianum has been crystallographically refined to 1.67 A resolution using a combination of reciprocal space and restrained least-squares refinement methods. The final crystallographic R-factor for 30,533 reflections measured with I greater than sigma (I) between infinity and 1.67 A is 0.188. The final model incorporates 1944 unique protein atoms (of a total of 1972) together with 194 bound solvent molecules. The structure has been analysed with respect to its detailed conformational properties, secondary structural features, temperature factor behavior, bound solvent sites, and heme geometry. The asymmetric unit of the cytochrome c' crystal contains a dimer composed of chemically identical 128-residue polypeptide chains. Although the refined structure shows the monomers to be very similar, examination of the differences that do occur allows an evaluation of how different lattice contacts affect protein conformation and solvent binding. In particular, comparison of solvent binding sites in the two subunits allows identification of a common set that are not altered by lattice interactions. The preservation of these solvent interactions in different lattice environments suggests that they play a structural role in protein stabilization in solution. The refined structure additionally reveals some new features that relate to the ligand binding properties and unusual mixed-spin state character of cytochrome c'. Finally, comparison of the heme binding geometry in cytochrome c' and other structurally unrelated c-type cytochromes shows that two alternative, but sterically favorable, conformational variants occur among the seven examples examined.     

The crystal structure of the "open" and the "closed" conformation of the flexible loop of trypanosomal triosephosphate isomerase

Triosephosphate isomerase has an important loop near the active site which can exist in a "closed" and in an "open" conformation. Here we describe the structural properties of this "flexible" loop observed in two different structures of trypanosomal triosephosphate isomerase. Trypanosomal triosephosphate isomerase, crystallized in the presence of 2.4 M ammonium sulfate, packs as an asymmetric dimer of 54,000 Da in the crystallographic asymmetric unit. Due to different crystal contacts, peptide 167-180 (the flexible loop of subunit-1) is an open conformation, whereas in subunit-2, this peptide (residues 467-480) is in a closed conformation. In the closed conformation, a hydrogen bond exists between the tip of the loop and a well-defined sulfate ion which is bound to the active site of subunit-2. Such an active site sulfate is not present in subunit-1 due to crystal contacts. When the native (2.4 M ammonium sulfate) crystals are transferred to a sulfate-free mother liquor, the flexible loop of subunit-2 adopts the open conformation. From a closed starting model, this open conformation was discovered through molecular dynamics refinement without manual intervention, despite involving C alpha shifts of up to 7 A. The tip of the loop, residues 472, 473, 474, and 475, moves as a rigid body. Our analysis shows that in this crystal form the flexible loop of subunit-2 faces a solvent channel. Therefore the open and the closed conformations of this flexible loop are virtually unaffected by crystal contacts. The actual observed conformation depends only on the absence or presence of a suitable ligand in the active site.     

Structural analysis of T4 DNA helix destabilizing protein (gp32 I) crystal by electron microscopy

Low dose electron diffraction and imaging techniques have been applied to the study of the crystalline structure of gp32*I, a DNA helix destabilizing protein derived from bacteriophage T4 gene 32 protein. A quantitative analysis of intensities from electron diffraction patterns from tilted, multilayered gp32*I crystal has provided the unit cell thickness of the crystal. The three-dimensional phases indicate that the space group P2(1)2(1)2. By taking into account the unit cell volume and the solvent content in the crystal, it was deduced that there is one gp32*I molecule in each asymmetric unit. A projected density map of unstained, glucose-embedded gp32*I crystal was synthesized with amplitudes from electron diffraction intensities and phases from electron images with reflections out to 7.6 A. Because of the similarity in the scattering density between glucose and protein, this projected map cannot be interpreted with certainty. A low resolution three-dimensional reconstruction shows that the protein molecule is about 90 A long and about 20 A in diameter. Because the dimer is formed around a dyad axis, the protein molecules comprising it must be arranged head-to-head. This dimeric arrangement of the proteins in the unit cell may be implicated as one of the conformational states of this protein in solution.     

Crystallization and preliminary X-ray crystallographic analysis of Mirabilis antiviral protein.

Mirabilis antiviral protein is a single-chain ribosome-inactivating protein purified from the tuberous root of Mirabilis jalapa L. We obtained several forms of crystals of the protein by the hanging drop vapor diffusion method, but most of these crystals were not suitable for X-ray crystallography. After refining the growth conditions, crystals of crystallographic quality were grown in 20-microliters droplets of an equi-volume mixture of 1.5% (w/v) protein solution and a reservoir solution containing 49 to 50% (w/v) ammonium sulfate and 50 mM-ammonium citrate (pH 5.4) at room temperature. Addition of 2 mM-adenine sulfate reduced twinning and "crystal shower". The resulting trigonal crystals diffract beyond 2.5 A resolution using a rotating anode X-ray generator. The space group was determined to be P3(1)21 or P3(2)21 (a = b = 103.9.A, c = 134.6 A, alpha = beta = 90 degrees, gamma = 120 degrees) based on their precession photography of h0l and hk0 zones. There seems to be three monomers in an asymmetric unit for VM = 2.51 A3/Da.     

Phosphocholine binding immunoglobulin Fab McPC603. An X-ray diffraction study at 2.7 A

The crystal structure of the Fab of McPC603, a phosphocholine-binding mouse myeloma protein, has been refined at 2.7 A resolution by a combination of restrained least-squares refinement and molecular modeling. The overall structure remains as previously reported, with an elbow bend angle between the variable and constant modules of 133 degrees. Some adjustments have been made in the structure of the loops as a result of the refinement. The hypervariable loops are all visible in the electron density map with the exception of three residues in the first hypervariable loop of the light chain. A sulfate ion occupies the site of binding of the phosphate moiety of phosphocholine.     

Solvent structure in crystals of trypsin determined by X-ray and neutron diffraction.

The solvent structure in orthorhombic crystals of bovine trypsin has been independently determined by X-ray diffraction to 1.35 A resolution and by neutron diffraction to 2.1 A resolution. A consensus model of the water molecule positions was obtained using oxygen positions identified in the electron density map determined by X-ray diffraction, which were verified by comparison to D2O-H2O difference neutron scattering density. Six of 184 water molecules in the X-ray structure, all with B-factors greater than 50 A2, were found to be spurious after comparison with neutron results. Roughly two-thirds of the water of hydration expected from thermodynamic data for proteins was localized by neutron diffraction; approximately one-half of the water of hydration was located by X-ray diffraction. Polar regions of the protein are well hydrated, and significant D2O-H2O difference density is seen for a small number of water molecules in a second shell of hydration. Hydrogen bond lengths and angles calculated from unconstrained refinement of water positions are distributed about values typically seen in small molecule structures. Solvent models found in seven other bovine trypsin and trypsinogen and rat trypsin structures determined by X-ray diffraction were compared. Internal water molecules are well conserved in all trypsin structures including anionic rat trypsin, which is 65% homologous to bovine trypsin. Of the 22 conserved waters in trypsin, 19 were also found in trypsinogen, suggesting that they are located in regions of the apoprotein that are structurally conserved in the transition to the mature protein. Seven waters were displaced upon activation of trypsinogen. Water structure at crystal contacts is not generally conserved in different crystal forms. Three groups of integral structural water molecules are highly conserved in all solvent structures, including a spline of water molecules inserted between two beta-strands, which may resemble an intermediate in the formation of beta sheets during the folding of a protein.     

Crystallization studies of glycosylated and unglycosylated human recombinant interleukin-2.

Glycosylated interleukin-2 (glyIL-2) has been crystallized in two crystal forms, and unglycosylated interleukin-2 (uIL-2) has been crystallized in three forms. The glycosylated form of the human recombinant IL-2 has been crystallized from 1.9 M ammonium sulfate, pH 6.5 to 7.0 in the hexagonal space group P6(2)22 or its enantiomorph. The crystals diffract to 2.8 A and contain two or three molecules per asymmetric unit. A second crystal form grows from 1.4 to 1.5 M ammonium sulfate in 0.2 M ammonium acetate, pH 5.0-5.5, as polycrystalline rosettes which are not suitable for even a preliminary crystallographic analysis. The uIL-2 crystallizes from 1.0 to 1.7 M ammonium sulfate, 0.2 M ammonium acetate, pH 4.5-5.6 in the monoclinic space group P2(1), and less frequently in the orthorhombic space group P2(1)2(1)2(1) from 2.5 M ammonium sulfate, pH 4.5 to 5.7. Cross-seeding uIL-2 with seeds from hexagonal crystals of glyIL-2 promotes nucleation of trigonal crystals of unglycosylated IL-2. These trigonal crystals belong to the space group P3(1)21 or its enantiomorph, with similar cell dimensions to the glyIL-2 hexagonal crystals.     

Protonation status of metal-bound ligands can be determined by quantum refinement

The protonation status of key residues and bound ligands are often important for the function of a protein. Unfortunately, protons are not discerned in normal protein crystal structures, so their positions have to be determined by more indirect methods. We show that the recently developed quantum refinement method can be used to determine the position of protons in crystal structures. By replacing the molecular-mechanics potential, normally used in crystallographic refinement, by more accurate quantum chemical calculations, we get information about the ideal structure of a certain protonation state. By comparing the refined structures of different protonation states, the one that fits the crystallographic raw data best can be decided using four criteria: the R factors, electron density maps, strain energy, and divergence from the unrestrained quantum chemical structure. We test this method on alcohol dehydrogenase, for which the pK(a) of the zinc-bound solvent molecule is experimentally known. We show that we can predict the correct protonation state for both a deprotonated alcohol and a neutral water molecule.     

Crystal structure and electron transfer properties of cytochrome c3

The crystal structure of cytochrome c3 from the sulfate-reducing bacteria Desulfovibrio desulfuricans, Norway strain, has been determined through the fitting of the recently completed primary structure to a 2.5 A resolution electron density map. The phase calculations were based on three mercurial derivatives; anomalous scattering data were used to refine the four heme iron positions. A preliminary refinement of the molecular model has led to a conventional crystallographic R factor of 34%. Cytochrome c3 is folded in two structural domains with one heme in each, the two other heme moieties lying in a large groove dividing the molecule. The core of the protein is the compact four-heme cluster which presents a relatively high degree of solvent exposure. The structural pattern of redox centers suggests that electron transfer might occur through direct contacts between some of the heme groups, via the overlapping system of pi oribitals or via intervening amino acid side chains or both.     

Three-dimensional structure of a light chain dimer crystallized in water. Conformational flexibility of a molecule in two crystal forms

The three-dimensional structure of an immunoglobulin light chain dimer (Mcg) crystallized in deionized water (orthorhombic form) was determined at 2.0 A resolution by phase extension and crystallographic refinement. This structure was refined side-by-side with that of the same molecule crystallized in ammonium sulfate (trigonal form). The dimer adopted markedly different structures in the two solvents. "Elbow bend" angles between pseudo 2-fold axes of rotation relating pairs of "variable" (V) and "constant" (C) domains were found to be 132 degrees in the orthorhombic form and 115 degrees in the trigonal form. Modes of association of the V domains and, to a lesser extent, the pairing interactions of the C domains were different in the two structures. Alterations in the V domain pairing were reflected in the shapes of the binding regions and in the orientations of the side-chains lining the walls of the binding sites. In the trigonal form, for instance, the V domain interface was compartmentalized into a main binding cavity and a deep pocket, whereas these spaces were continuous in the orthorhombic structure. Patterns of ordered water molecules were quite distinct in the two crystal types. In some cases, the solvent structures could be correlated with conformational changes in the proteins. For example, close contacts between V and C domains of monomer 1 of the trigonal form were not retained in orthorhombic crystals. Ordered water molecules filled the space created when the two domains moved apart.     

Correlation for the partition behavior of proteins in aqueous two-phase systems: effect of surface hydrophobicity and charge

Correlations to describe the effect of surface hydrophobicity and charge of proteins with their partition coefficient in aqueous two-phase systems were investigated. Polyethylene glycol (PEG) 4000/phosphate, sulfate, citrate, and dextran systems in the presence of low (0.6% w/w) and high (8.8% w/w) levels of NaCl were selected for a systematic study of 12 proteins. The surface hydrophobicity of the proteins was measured by ammonium sulfate precipitation as the inverse of their solubility. The hydrophobicity values measured correlated well with the partition coefficients, K, obtained in the PEG/salt systems at high concentration of NaCl (r = 0.92-0.93). In PEG/citrate systems the partition coefficient correlated well with protein hydrophobicity at low and high concentrations of NaCl (r = 0.81 and 0.93, respectively). The PEG/citrate system also had a higher hydrophobic resolution than other systems to exploit differences in the protein's hydrophobicity. The surface charge and charge density of the proteins was determined over a range of pH (3-9) by electrophoretic titration curves; PEG/salt systems did not discriminate well between proteins of different charge or charge density. In the absence of NaCl, K decreased slightly with increased positive charge. At high NaCl concentration, K increased as a function of positive charge. This suggested that the PEG-rich top phase became more negative as the concentration of NaCl in the systems increased and, therefore, attracted the positively charged proteins. The effect of charge was more important in PEG/dextran systems at low concentrations of NaCl. In the PEG/dextran systems at lower concentration of NaCl, molecular weight appeared to be the prime determinant of partition, whereas no clear effect of molecular weight could be found in PEG/salt systems.     

Crystal structure of bovine beta-trypsin at 1.5 A resolution in a crystal form with low molecular packing density. Active site geometry, ion pairs and solvent structure

The crystal structure of bovine pancreatic beta-trypsin (BPT) has been determined from a novel orthorhombic crystal form which contains substantially more solvent (filling 57% of the volume of the unit cell) than previously determined orthorhombic (44%) and trigonal (37%) BPT structures. The native and benzamidine-inhibited crystal structures of BPT in ammonium sulphate at pH 5.3 have been determined for the new form by molecular replacement techniques. The structures have been refined at 1.5 A resolution with final R-values of 16.7% and 16.9%, respectively. Comparison with the previously refined old orthorhombic forms shows that the overall conformation of the protein backbone is highly conserved. A great number of previously undefined side-chains have been located in density. At the C terminus an extra ion pair involving lysines 87 and 107 has been revealed. A far more detailed picture of the ordered solvent structure has been derived. Thirty water clusters have been identified. A large water network extends from the calcium binding site to the activation area and the autolysis loop. There is evidence for a water channel reaching from the depth of the specificity pocket to the nearby protein surface which might be involved in the displacement of water molecules upon substrate binding. A sulphate anion which forms hydrogen bonds to the active site residues His57, Ser195 and Gly193 was for the first time positioned in clearly defined electron density. Interaction with the sulphate ion may explain the increase in the pKa value of His57 at high sulphate concentrations which was observed by nuclear magnetic resonance studies of a bacterial serine protease both in crystalline form and in solution. Thus, a His-Ser hydrogen bond will not exist in solvents containing sulphate at low pH (up to at least 6.8) where the imidazole of His57 is protonated. The new crystal form is of considerable interest for substrate binding studies. Wide solvent channels should allow diffusion of large substrates (comparable in size to, e.g. pancreatic trypsin inhibitor) into the enzyme crystal. The active site is accessible; intermolecular contact areas are further remote from the active site than in the old orthorhombic form.     

Multiple hydration layers in cubic insulin crystals.

Cubic insulin crystals contain approximately 30-A-diameter channels filled with aqueous solvent, providing a useful system in which to analyze hydration structure at a variety of distances from protein surfaces. Beginning with an atomic model for the protein and ordered water molecules, the density distribution in the solvent volume of the phasing model was iteratively refined to improve the fit of calculated structure factors with x-ray diffraction data. The free R value, which compares calculated structure factors with a subset of observed structure factors deliberately omitted from the refinement, was used to provide an objective confirmation of the effectiveness of the refinement procedure. Electron density maps of the solvent, computed using the solvent-refined phases and complete low-resolution diffraction data, reveal multiple hydration layers around the protein.     

Fractionation and recovery of whey proteins by hydrophobic interaction chromatography

A method for the recovery and fractionation of whey proteins from a whey protein concentrate (80%, w/w) by hydrophobic interaction chromatography is proposed. Standard proteins and WPC 80 dissolved in phosphate buffer with ammonium sulfate 1 M were loaded in a HiPrep Octyl Sepharose FF column coupled to a fast protein liquid chromatography (FPLC) system and eluted by decreasing the ionic strength of the buffer using a salt gradient. The results showed that the most hydrophobic protein from whey is α-lactalbumin and the less hydrophobic is lactoferrin. It was possible to recover 45.2% of β-lactoglobulin using the HiPrep Octyl Sepharose FF column from the whey protein concentrate mixture with 99.6% purity on total protein basis.