ALBERT

Notes: The development of high-intensity X-ray sources and the use of insertion devices will make it possible to collect data routinely from protein crystals at very short wavelengths (λ ≤ 0.5 Å). Possible benefits of using shorter wavelengths can be inferred from the improvement in the quality of the data when using a wavelength λ ∼ 0.9 Å instead of one close to the Cu Kα emission edge. In addition to fewer absorption errors, two factors might contribute to this improvement. These are an increase in the lifetime of the protein crystal and a better signal-to-background ratio. In this paper we address the second of these. In order to compare the quality of the data and the relative background level in the diffraction patterns at different wavelengths two data sets have been collected at λ = 0.92 and 0.55 Å. The results obtained from data processing and careful measurement of the background in the raw images suggest that, in the absence of absorption errors and radiation damage, data collection at very short wavelengths does not provide higher quality data. There is no improvement in the signal-to-background ratio in the short-wavelength data.

Notes: The crystal structure of metmyoglobin from yellowfin tuna (Thunnus albacares) has been determined by molecular replacement methods and refined to a conventional R factor of 0.177 for all observed reflections in the range of 6.0–1.70 Å resolution. Like other myoglobins for which a high-resolution structure is available, the polypeptide chain is organized into several helices that cooperate to form a hydrophobic pocket into which the heme prosthetic group is non-covalently bound; however, the D helix observed in other myoglobins is absent in myoglobin from yellowfin tuna and has been replaced with a random coil. As well, the A helix has a pronounced kink due to the presence of Pro16. The differences in structure between this and sperm whale myoglobin can be correlated with their reported dioxygen affinity and dissociation. The structure is in agreement with reported fluorescence data which show an increased Trp14...heme distance in yellowfin tuna compared to sperm whale myoglobin.

Notes: High-resolution single crystals of a catalytic RNA molecule derived from the sequence of the satellite RNA of tobacco ringspot virus have been obtained. The unit-cell volumes of the RNA crystals vary depending on the crystallization conditions and temperature. The best crystal form, when flash frozen, has space group P1 with unit-cell dimensions a = 53.08, b = 71.81, c = 28.03 Å, α = 98.43, β = 104.32 and γ = 74.54°. This form diffracts to a resolution of 2.4 Å. A heavy-atom derivative search is in progress.

Notes: The automated microbatch technique developed at Imperial College has been used to establish a phase diagram for crystallization. The concentrations of the protein (carboxypeptidase G2) and precipitant (PEG 4000) were varied, while pH and temperature were kept constant. The diagram consists of an undersaturation and a supersaturation zone, the latter being subdivided into the metastable, nucleation and precipitation zones. In the metastable zone, crystals may grow but nucleation of crystals does not occur. It is the best zone for growth of X-ray diffraction quality crystals because of the slower growth rate and the avoidance of uncontrolled nucleation, which uses up protein in the formation of tiny crystals. Nevertheless, in practice, it is rarely well defined or used because nuclei must be introduced artificially into the system. The new method used here consists of setting crystallization droplets at nucleation conditions and later diluting them to conditions where nucleation has not been observed. Single diffracting crystals of typical dimensions 0.3 × 0.3 × 0.2 mm were routinely obtained in the metastable zone, equivalent to the best (very rarely) obtained crystals in the nucleation zone.

Notes: The crystallization of a variant of Bacillus lentus subtilisin and the native enzyme was achieved using identical conditions. The variant B. lentus was found to grow in two crystal forms, form 1 and form 2, whereas the native B. lentus subtilisin enzyme crystallized in only one, form 1. Form 2 crystals, once obtained, were found to grow much more rapidly than form 1 crystals. The lattice contacts and structural changes giving both crystal forms have been examined. The results show that crystal form 2 has a more complex network of interactions. There is also a small surface conformational change in the form 2 structure relative to the native and variant form 1 crystals and at least two solvent molecules bound to the enzyme in crystal form 1 are displaced in crystal form 2. In addition, a site specific substitution in the variant at position 27 induces a `short' lattice contact which does not exist in the native B. lentus or the form 2 variant B. lentus. These results suggest that in some circumstances engineered variants could be designed to crystallize more rapidly than the native enzyme.

Notes: X-ray or neutron diffraction studies have shown, at the atomic level, that water molecules occupy well determined sites inside or at the surface of biological macromolecules. These water molecules are constitutive of biomolecules and play a crucial role in their structural and functional properties. Upon crystallization some water molecules are either desolvated or involved in crystal packing. The role of water in determining crystal packing has been experimentally confirmed by several X-ray analyses.

Notes: Large single crystals of the dodecylmaltoside (DDM) complex of a polytopic integral membrane transport protein, the Neurospora plasma membrane H+-ATPase, have been obtained using an approach that attempts to take into account the possibly radically different physicochemical properties of the protein surfaces and the detergent micellar collar. The overall goal of the crystallization strategy employed was to identify conditions in which the protein surfaces of the DDM–ATPase complex are moderately insoluble and in which the DDM micellar collar is also near its solubility limit. The first step was to screen a variety of commonly used protein precipitants for those that were able to induce the aggregation of pure DDM micelles. The concentration at which any precipitant induced DDM micellar aggregation was hoped to be close to the concentration at which it might induce insolubility of the detergent micellar collar of the DDM–ATPase complex. Of the nine precipitants tried, seven, all polyethylene glycols (PEGs), were able to induce DDM micelle insolubility. The seven PEGs were then tested for their effect on the solubility of the DDM–ATPase complex at a concentration slightly below that necessary to induce DDM micellar aggregation. Three of the PEGs caused extensive precipitation of the ATPase at this concentration and were, therefore, shelved. The other four PEGs did not induce precipitation at the concentration employed and were subsequently used at this concentration for crystallization trials in which the protein concentration was varied. Encouragingly, crystalline plates of the ATPase were obtained for each of the four PEGs tried, indicating that the overall approach may be valid. Unfortunately, the crystals obtained were visibly flawed, suggesting that the correct balance of protein surface and DDM micelle insolubility had not yet been reached. The ionic strength of the crystallization trials was then raised, which was known from other experiments to render the protein surfaces of the ATPase less soluble while having no effect on the DDM micellar aggregation point. For one of the PEGs, PEG 4000, this brought on a new, well formed hexagonal crystal habit. Subsequent optimization of the initial conditions has yielded large single hexagonal crystals of the H+-ATPase roughly 0.4 × 0.4 × 0.15 mm in size, holding promise for exploration of the structure of the ATPase by X-ray diffraction analysis.

Notes: Protein crystal growth often depends on the combination of many different factors. Some affect protein solubility directly; others may act indirectly by causing conformational changes. Systematic characterization of these factors can be important for generating good crystals. It can also provide useful insight into the biochemical behavior of the protein being crystallized. Here we focus on statistical methods to achieve these two objectives. (1) Characterization of a protein system by analyzing patterns of crystal polymorphism under different levels of biochemical parameters, such as ligands and pH. Tests of the reproducibility of crystal growth experiments indicate that quantitative scales of crystal quality can be statistically significant. Analysis of variance for a replicated, full-factorial design in which four factors were tested at two levels has been used to demonstrate highly significant, biochemically relevant, two-factor interactions strongly implicating pH and ligand-dependent conformational changes. (2) Optimization of crystal growth via response-surface methods. `Minimum predicted variance' designs provide for efficient response-surface experiments aimed at constructing quadratic models in several dimensions. We have used such models to improve crystal size and quality significantly for three forms of Bacillus stearothermophilus tryptophanyl-tRNA synthetase. In one case we can now avoid having to increase the size by repeated seeding, a difficult procedure that also produces unwanted growth of satellite crystals. Graphs of two-dimensional level surfaces reveal a number of ridges, where the same result is obtained for many combinations of the factors usually varied when trying to improve crystals. An important inference is that it may be better to sample simultaneously for the effects of protein concentration and supersaturation. For a system involving only one crystallizing agent, supersaturation can be approximated as the product of protein and precipitant concentrations. Use of this search direction significantly improves the performance of response-surface experiments. Advantages of growing crystals at stationary points of their response surfaces include better crystals and higher reproducibility, since crystal growth at stationary points is insulated from the deleterious effects of experimental fluctuations. This arises because the derivatives of the response are by definition zero with respect to the experimental variables. Quantitative analysis of appropriately designed crystal growth experiments can thus be a powerful way to characterize complex and interacting biochemical dependencies in macromolecular systems and optimize parameters important to the crystallography.

Notes: The Escherichia coli molecular chaperone cpn60 oligomer, [cpn60]14, also called GroEL, has been crystallized and examined by X-ray crystallography and self-rotation function calculations. The crystals show unit-cell dimensions a = 143.3, b = 154.6 and c = 265 Å, with α = 82, β = 95 and γ = 107°. The space group is P1 and crystals diffract to 7 Å. X-ray analysis shows that the oligomer has one sevenfold symmetry axis and seven twofold axes that are all perpendicular to the sevenfold. The symmetry suggests that [cpn60]24 consists of two heptamers, [cpn60]7, stacked on top of each other. The orientations of the symmetry axes of the two independent [cpn60]14 oligomers in the triclinic unit cell have been determined relative to the crystallographic axes. The two oligomers in the unit cell are arranged side-by- side, but the second oligomer is rotated 26° around the sevenfold axis relative to the first oligomer.

Notes: The crystals of most proteins are poorly ordered and diffract to lower resolutions than other crystals of simple and inorganic compounds. The use of two novel methods for gel protein crystal growth, utilizing liquid diffusion and vapor diffusion, are described for the growth of lysozyme and canavalin. Crystallization using gels has been demonstrated to improve crystal quality by reducing convective flow, sedimentation, nucleation and twinning. Preliminary X-ray diffraction data are also presented.

Notes: A procedure which allows an investigator to supply a crystal with fresh mother material without inducing significant growth defects is described. This technique requires that the crystal is grown in a gelled hanging or sitting drop. An example concerning a model macromolecule, hen egg-white lysozyme, is given. Extension of this procedure to other macromolecules is discussed.

Notes: Two populations of aggregates are generally indentified in supersaturated solutions of biological macromolecules: small aggregates of a size which is less than 5 nm and large aggregates, the largest of which are at least one order of magnitude bigger. In order to understand the role played by the microporous network of a gel in the growth and behaviour of these different species in the prenucleation period, an in situ observation of nucleation has been carried out using either free solutions or solutions trapped in agarose gels. In a previous study, free solutions were investigated by small-angle neutron scattering (SANS) to identify the small aggregates. Optical observations, made under the same conditions, revealed the formation of an amorphous precipitate which disappeared at the end of the experiment. The sedimentation of this phase, which occurs in free solution but never occurs in gelled solution, depletes the solution bulk and this could explain why the nucleation density is higher in agarose gel than in free solution. The case of silica gel, the behaviour of which is completely different with respect to nucleation, will be discussed.

Notes: Crystals of the binary complex of alcohol dehydrogenase from Sulfolobus solfataricus with NADH were shown to be twinned and not suitable for automated data collection. Several crystallization trials, performed with the aim of eliminating twinning, are described. Interestingly, crystals grown from agarose gel have been demonstrated to have a unique reciprocal lattice. These crystals are monoclinic, space group C2, with cell dimensions a = 134.47 (9), b = 85.26 (5), c = 71.76 (8) Å, β = 97.53 (4)°, and showed significant diffraction beyond 3.0 Å resolution.

Notes: The parameters affecting the crystal quality of complexes between p21H-ras and caged GTP have been investigated. The use of pure diastereomers of caged GTP complexed to the more stable p21(G12P)′ mutant of p21 and the addition of n-octyl-β-D-glucopyranoside improved the reproducibility and decreased the mosaicity of the crystals significantly. Furthermore, the crystallization technique was changed from the batch method to the sitting-drop technique. With the availability of a larger yield of well ordered crystals, it was possible to extend the time-resolved crystallographic investigations on p21H-ras. A structure of p21(G12P)′:GTP could be obtained 2 min after photolytic removal of the cage group and led to the identification of a previously unidentified conformation for the so-called catalytically active loop L4. The refinement of five data sets collected within 2 min at different times (2–4, 11–13, 20–22, 30–32 and 90–92 min) after the initiation of the intrinsic GTPase reaction of the protein indicates that the synchrotron Laue method can be used to detect small structural changes and alternative conformations, but is presently limited in the analysis of larger rearrangements since these produce diffuse and broken electron density.

Notes: The tetrameric flavoenzyme 2,4-pentadienoyl-CoA reductase has been crystallized from solutions containing polyethylene glycol as precipitant. The crystals grow in the monoclinic space group C2 with unit-cell dimensions a = 160.2, b = 120.2, c = 95.3 Å, β = 99.0°. The packing parameter VM is 2.3 Å3 Da−1 (Matthews parameter) for four monomers per asymmetric unit. Complete data sets to about 2.9 Å resolution have been collected.

Notes: Ribosome-inactivating protein from barley seeds has been crystallized using polyethylene glycol as precipitant. The crystal belongs to the monoclinic space group C2, with unit-cell parameters a = 88.36, b = 62.59, c = 53.18 Å and β = 108.62°. The asymmetric unit contains one molecule of ribosome-inactivating protein with a corresponding crystal volume per protein mass (Vm) of 2.32 Å3 Da−1 and a solvent content of 47% by volume. The crystal diffracts to about 2.3 Å with X-rays from a rotating-anode source and is very stable in the X-ray beam. X-ray data (nearly complete to 2.4 Å Bragg spacing) have been collected from a native crystal.

Notes: A platinum chromophore, chloro(2,2′:6′,2′′-terpyridine)platinum(II) chloride, previously used in labelling active-site histidines of serine proteases, proves to be a useful reagent in heavy-atom derivatization of protein crystals for X-ray crystallographic phase determination.

Notes: Galactose-1-phosphate uridylyltransferase catalyzes the formation of UDP-galactose during normal cellular metabolism, making it an essential enzyme in all cells. The enzyme from Escherichia coli has been crystallized at pH 5.9 in the presence of phenyl-UDP (P1-5′-uridyl-P2-phenyl diphosphate), a substrate analog, using PEG 10 000 in combination with Li2SO4 and NaCl. Crystals belong to space group P21212 with unit-cell dimensions a = 58.6, b = 217.6 and c = 69.6 Å. There is one dimer or two subunits in the asymmetric unit. Crystals are relatively insensitive to X-ray radiation and diffract beyond 2.5 Å resolution. A low-resolution native data set has been recorded.

Notes: The three-dimensional structure of human dicupric monooxalate lactoferrin, Cu2oxLf, has been determined to 2.0 Å resolution, using X-ray diffraction data collected by diffractometry to 2.5 Å resolution, and oscillation photography on a synchrotron source to 2.0 Å resolution. Difference electron-density maps calculated between Cu2oxLf and both dicupric lactoferrin, Cu2Lf, and diferric lactoferrin, Fe2Lf, showed that the oxalate had replaced a carbonate in the C-terminal binding site, and that, relative to Cu2Lf, there were no significant differences in the N-terminal site. The structure was then refined crystallographically by restrained least-squares methods. The final model, in which the r.m.s. deviation in bond distances is 0.017 Å, contains 5314 protein atoms (691 residues), two Cu2+ ions, one bicarbonate ion, one oxalate ion, 325 solvent molecules and one sugar residue. The crystallographic R factor of 0.193 is for 46 134 reflections in the range 8.0 to 2.0 Å resolution. The oxalate ion is coordinated to copper in a 1,2-bidentate fashion, and the added bulk of the anion results in the rearrangement of the side chains of nearby arginine and tyrosine residues. No other major alterations in the molecule can be observed, the overall protein structure being the same as that for Cu2Lf and Fe2Lf.

Notes: This paper gives an overview of the science of crystals of biological macromolecules. The historical background of the field is outlined and the main achievements and open problems are discussed from both biological and physical–chemical viewpoints. Selected results, including data from the authors, illustrate this overview. The perspectives of crystallogenesis for structural biology, but also more general trends, are presented.

Notes: A dilute solution parameter obtained from static light-scattering measurements is proposed as a predictor for protein crystallization experiments. The osmotic second virial coefficients, B22, have been measured for a variety of proteins in solvents that are known to promote crystallization and the values for B22 were found to lie within a fairly narrow range which we refer to as a crystallization slot. Solution conditions which were known not to favor crystallization of the proteins resulted in B22 values well outside the crystallization slot.

Notes: The early stages of the crystallization process of porcine pancreatic α-amylase were investigated by quasi-elastic light scattering. It is shown that at 288 and 293 K the diffusion coefficient does not monotonically change with increasing protein concentration but passes through a maximum at 10 mg ml−1. In supersaturated solutions, prior to nucleation, the protein is strictly monodisperse. Nucleation induces the formation of aggregates and a polydispersity of, for example, 18% for an initial supersaturation C/Ce = 5.8. Monodispersity is restored after the nuclei have grown and partially consumed the solute. On the other hand, polydispersity increases up to 20% at 298 K if the protein concentration decreases to 3–4 mg ml−1, values at which the solutions are under-saturated. When the protein concentration exceeds 5–6 mg ml−1 the protein becomes monodisperse again. These results, confirmed by those of another system we are studying (bovine pancreatic trypsin inhibitor), are at variance with the statements that supersaturation is always at the origin of aggregation and polydispersity, and that in undersaturated solutions the diffusion coefficient should remain constant for obtaining crystals once the solutions are supersaturated.

Notes: Quasi-elastic light scattering (QELS) was used to investigate quantitatively the mechanisms of nucleation, postnucleation growth, and dissolution in ensembles of both crystalline and amorphous aggregates of satellite tobacco mosaic virus (STMV), ferritin, apoferritin and pumpkin seed globulin. At low supersaturation conditions, as described previously for small molecule crystallization, the metastable region was obtained. Under these conditions aggregation took place, but crystallization did not proceed and critical nuclei did not form over a long period of time. The critical solution supersaturation necessary to obtain crystals, σ = ln(c/s) where c and s are concentration and solubility of protein, varied from ∼0.1 for pumpkin seed globulin to ∼0.9 for STMV. For higher supersaturation conditions when aggregation processes leading to formation of crystals are not established immediately but after a certain induction period, the supersaturation-dependent critical nuclear size, Rc, for different macromolecular systems was estimated from time-dependent size-distribution analyses to be in the range of ∼103 for proteins such as pumpkin globulin to approximately 10 for virus particles. From the same data, the molar interfacial free energy was deduced to be 3.3–9.2 kJ mol−1. These are believed to be among the first estimates for macromolecular crystals. Under conditions of moderate supersaturation where induction periods preceded the appearance of critical nuclei, the potential barriers for formation were estimated to be in the range 8.3–50 kJ mol−1. Growth and dissolution kinetics for pumpkin seed globulin were investigated. These experiments allowed determination of protein solubility versus solution temperature, protein and precipitant concentrations. Aggregation patterns which lead to crystal formation are distinctly different to those which produce an amorphous precipitate. The results provide additional evidence that QELS can be used to find general criteria that allow one to discriminate between conditions for a given protein system leading to crystalline or amorphous states at early stages of the aggregation process.

Notes: This laboratory has explored the potential of a combination of three analytical techniques to study the nucleation of chicken egg-white lysozyme. Collisional quenching of the fluorescent molecule SPQ [6-methoxy-N-(3-sulfopropyl)quinolinium] by chloride ions was used to determine the binding of the crystallizing agent to the protein at equilibrium and kinetically. Calorimetric measurements show that this binding generates an exothermic peak larger than the energy released during the early stages of nucleation. Light scattering intensity measurements were used to follow the aggregation kinetics.

Notes: Chloromuconate cycloisomerase (E.C. 5.5.1.7) is an enzyme involved in the 2,4-dichlorophenoxyacetate degradation pathway of Alcaligenes eutrophus JMP134 (pJP4). The crystal structure of this protein was determined at 3 Å resolution by molecular-replacement techniques using atomic coordinates from the reported crystal structure of the homologous muconate cycloisomerase (E.C. 5.5.1.1) from Pseudomonas putida as the search model (42% identical positions in the sequences). Structure refinement by simulated-annealing and restrained least-squares techniques converged at R = 0.195. In the crystals studied, space group I4, the protein is present as two octamers per unit cell with two subunits per asymmetric unit. Each subunit consists of two globular domains, one of which forms an α/β-barrel. Comparison of this structure with that of muconate cycloisomerase reveals the reasons for the altered substrate specificity of chloromuconate cycloisomerase. Marked differences are observed in polarity, accessibility and hydrogen-bonding potential in the channel leading into the active site as well as in the active center itself.

Notes: Transforming growth factor-β is a multifunctional cell-growth regulator and is a member of the TGF-β superfamily of cytokines. Each monomer is 112 amino acids long and the mature active form is a 25 kDa homodimer. Recently, the crystal structure of TGF-β2 has been determined independently in two laboratories [Daopin, Piez, Ogawa & Davies (1992). Science, 257, 369–373; Schlunegger & Grütter (1992). Nature (London), 358, 430–434] and subsequently refined to higher resolutions [Daopin, Li & Davies (1993). Proteins Struct. Funct. Genet. In the press; Schlunegger & Grütter (1993). J. Mol. Biol. In the press]. A detailed structural comparison shows that the two structures are nearly identical with the differences mostly located on the mobile regions of the molecule. The r.m.s. differences between the two structures are 0.10 Å for 104 pairs of Cα atoms, 0.15 Å for 434 pairs of main-chain atoms, 0.33 Å for 860 out of 890 pairs of protein atoms and a correlation of 90% between the temperature B factors of all protein atoms. Based on a comparison of the water molecules, a B value of 60.0 Å2 is recommended as the cut off for modeling new waters. The structural identity is striking because in one case the material was expressed in vivo in CHO cells whereas in the other case it was expressed in E. coli and had to be refolded in vitro. The overall coordinate errors are estimated to be 0.21 Å from the Luzzati plot, 0.18 Å from the σA plot, 0.24 Å with Cruickshank's equations and 0.25 Å using the empirical method of Perry & Stroud. These estimates are comparable to the r.m.s. structure superposition. The r.m.s. differences correlate very well with the crystallographic B values and the relation is best described with the Cruickshank formula. In addition to the estimation of an overall error, a new application of the Cruickshank formula is presented here to estimate the local errors.

Notes: The X-ray diffraction pattern of the crystals of the non-selfcomplementary hexadeoxyribonucleotide d(CGCACG)·d(CGTGCG) can be indexed in four different space groups: (i) P65 and P21, with cell parameters a = 17.75 (1), b = 17.76 (1), c = 42.77 (3) Å, α = 90, β = 90, γ = 120°, and (ii) P212121 and C2, with cell parameters a = 17.75 (1), b = 30.74 (2), c = 42.77 (3) Å, α = 90, β = 90, γ = 90°. While the Rmerge for the equivalent reflections in the different space groups indicates that P21 is the correct choice in the present case, it is demonstrated that the near degeneracy of the space groups arises out of the fact that the DNA molecule is nearly cylindrical. A perfect cylinder would show perfect degeneracy.

Notes: We report the crystallization of samples of a recombinant preparation of human interleukin-1 receptor antagonist protein (IRAP) and solution of the crystal structure by isomorphous replacement methods. Crystals were obtained by the hanging-drop vapor-diffusion method at 277 K from solutions of PEG 4000 containing sodium chloride, dithiothreitol and PIPES [sodium piperazione-N,N′-bis(2-ethanesulfonate)] buffer at pH 7.0. Crystals appear within about a week and grow as truncated tetragonal bipyramids to 0.3–0.6 mm on an edge. X-ray diffraction data from these crystals specify space group P43212 and unit-cell dimensions of a = b = 72.35(26), c = 114.7(8) Å and Z = 16 (two molecules per asymmetric unit). Fresh crystals diffract to about 2.3 Å resolution. The search for heavy-atom derivatives has produced two, potassium gold cyanide and trimethyl lead chloride, as same-site, single-site derivatives. Inspection of an electron-density map at 4 Å resolution calculated with these derivatives confirms that the IRAP molecule is a member of the interleukin-1 structural family.

Notes: The structure of the ADP complex of the enzyme 3-phosphoglycerate kinase (PGK, E.C. 2.7.2.3) from Bacillus stearothermophilus NCA-1503 has been determined by the method of molecular replacement. The structure has been refined to an R factor of 0.16 for all data between 10.0 and 1.65 Å resolution, using data collected on the Hendrix–Lentfer imaging plate at the EMBL outstation in Hamburg. The r.m.s. deviations from stereochemical ideality are 0.010 and 0.011 Å for bonds and planes, respectively. Although crystallized in the presence of the nucleotide product MgATP, the high-resolution structure reveals the bound nucleotide to be MgADP reflecting the low intrinsic ATPase activity of PGK. Although the two domains of this enzyme are found to be some 4.5° closer together than is found in the yeast and horse-muscle apo-enzyme structures, this structure represents the `open' rather than the `closed', catalytically competent form, of the enzyme.

Notes: Yeast initiator tRNA crystals exhibit strong X-ray diffuse scattering. This scattering can be used to extract information about lattice-coupled and intramolecular motions in the crystals. The amplitudes and correlation distances of these motions can be estimated by calculating the diffuse scattering and comparing the results with the observed scattering. Results indicate that both anisotropic, lattice-coupled motions as well as short-range correlated local disorder in the anticodon arm contribute to the overall disorder in the crystals. These types of motions can be correlated with aspects of tRNA function. This additional information complements the results from analysis of crystallographic data and provides a more detailed picture of the structure and dynamics of the molecule. The degree to which the methodology presented here can account for the observed diffuse scattering from tRNA represents a significant step forward in the ability to use this conventionally discarded information, and encourages the ultimate extension of these ideas to a wide variety of macromolecular systems.

Notes: Two new crystal forms of isoenzyme 3-3 of rat liver glutathione S-transferase (GST 3-3) have been obtained. They were grown under essentially the same crystallization conditions as those reported for the C2 crystal form [Fu, Rose, Chung, Tam & Wang (1991). Acta Cryst. B47, 813–814]. The new crystals belong to space group P21 with one form having cell dimensions a = 101.6, b = 69.5, c = 81.4 Å, and β = 113.6°, and the other form having cell parameters a = 97.4, b = 81.1, c = 69.4 Å and β = 109.2°. These new crystals diffract to at least 2.5 Å, resolution. The molecular packing arrangements in these P21 crystals have been found by molecular replacement studies.

Notes: Lipase from Chromobacterium viscosum has been purified to homogeneity and crystallized in a form suitable for X-ray diffraction analysis from 10-14% polyethylene glycol 4000 and 10-14% 2-methyl-2,4-pentane diol at pH 6.4 in the presence of 0.25%(w/v) n-octyl-β-D-glucopyranoside. These crystals belong to space group P21212 with refined lattice constants a = 41.1 Å, b= 156.8, c = 43.6 Å, indicating a cell content of one monomer per asymmetric unit of the crystal. The crystals diffract to a resolution of 2.2 Å.

Notes: Xylose isomerase from Bacillus coagulans has been crystallized in two different crystal forms. One crystal form is in space group P21212, cell dimensions a = 462, b = 165, c = 82 Å. The other is in space group I422, cell dimensions a = b = 113, c = 153 Å.

Notes: The packing of glutathione reductase from Escherichia coli in crystal form T showed a place where two molecules are at a distance of only 6 Å between the closest atoms, i.e. where a contact is almost made. In order to form this contact with hydrogen bonds, two amino-acid residues were exchanged. This mutation had no effect on molecular packing or the resolution limit of the X-ray diffraction, but facilitated crystal nucleation dramatically and possibly increased the crystal growth rate and shortened the crystallization time.

Notes: Crystals of the tetra-heme cytochrome c3 (Mr = 13 kDa, 107 residues, four heme groups) from sulfate- and nitrate-reducing Desulfovibrio desulfuricans ATCC 27774 have been obtained and crystallographically characterized. They belong to space group P6122 with cell dimensions a = b = 61.84 (4) and c = 109.7 (2) Å, and Z = 12. Intensity data were initially collected on a FAST system with a rotating-anode X-ray source leading to a total of 22 592 observations, from which only 4930 were unique, in the resolution range 20.0–2.4 Å with an Rmerge(I) of 7.0%. Higher resolution data were measured on a FAST system at station 9.6 of the SRS (Daresbury, England), leading to 19 328 intensities, of which 11 179 were unique, in the resolution range 20.0–1.75 Å and an Rmerge(I) of 5.5%. Cross-rotation and translation functions were performed with ALMN and TFSGEN programs from the CCP4 suite. The packing of the molecules in the unit cell was checked with TOM/FRODO. Rigid-body refinement of the model and subsequent refinement using molecular dynamics were performed with X-PLOR, leading to a current R factor of 25.9%, for data up to 2.3 Å.

Notes: An error in the paper by Naismith, Habash, Harrop, Helliwell, Hunter, Wan, Weisgerber, Kalb & Yariv [Acta Cryst. (1993), D49, 561–571] is corrected. The first sentence of the caption for Fig. 6 on p. 568 should read: The S1 (Cd2+) and S2 (Ca2+) metal sites.

Notes: The speed of electron-density fitting during X-ray structure solution and refinement, and the quality of the protein model resulting, can both be enhanced by the use of databases of main- and side-chain conformations. Three structures are compared in this report, one refined at high resolution (1.7 Å), and two at lower resolutions using either the database method (2.4 Å resolution) or more traditional empirical electron-density fitting (1.9 Å resolution). An analysis of peptide orientation was used as an aid in finding unusual portions of main-chain structure. The fit of side chains to known rotamer conformations was used to help determine the accuracy of these atomic positions. In addition, the use of an objective measure of the fit of structures to electron-density maps was evaluated, both alone and in combination with side-chain conformational information.

Notes: Ferredoxin I (Fd I) from Equisetum arvense is an iron–sulfur protein composed of 95 amino-acid residues and one [2Fe–2S] cluster. It crystallized in the space group P21, a = 30.4, b = 57.4, c = 47.5 Å and β = 78.7° with two molecules per asymmetric unit. X-ray diffraction data up to 1.8 Å resolution were collected by using a Rigaku four-circle diffractometer. The initial model of Fd I, which was derived by the molecular replacement method using a structure of the Fd I from the blue–green alga Aphanothece sacrum, was refined by molecular dynamics simulation and a least-squares minimization with stereochemical restraints. Positional parameters and isotropic temperature factors for 1420 non-H protein atoms and 183 water molecules were refined on 13 838 observed structure factors (Fo 〉 σFo) between 10.0 and 1.8 Å resolution. The final Rfactor was 17.0%, and the standard deviation of atomic position estimated by Luzzati plot [Luzzati (1952). Acta Cryst. 5, 802–810] was 0.2 Å. The electron-density map was well defined for the two independent molecules except for the N-terminal residue and the three C-terminal residues. Equivalent Cα atoms of two independent molecules in the asymmetric unit were superposed by the least-squares method with root-mean-square deviations of 0.26 Å. Reasonable structural differences were observed at a polypeptide segment having few intramolecular interactions. Highly flexible regions of the molecule were assigned from the structural differences between the two independent molecules in the crystal and the distribution of temperature factors along the polypeptide chain.

Notes: A computer program, VOIDOO, is described which can be employed in the study of cavities such as they occur in macromolecular structures (in particular, in proteins). The program can be used to detect unknown cavities or to delineate known cavities, either of which may be connected to the outside of the molecule or molecular assembly under study. Optionally, output files can be requested that contain a description of the shape of the cavity which can be displayed by the crystallographic modelling program O. Additionally, VOIDOO can be used to calculate the volume of a molecule and to create a file containing data pertaining to the surface of the molecule which can also be displayed using O. Examples of the use of VOIDOO are given for P2 myelin protein, cellular retinol-binding protein and cellobiohydrolase II. Finally, operational definitions to discern different types of cavity are introduced and guidelines for assessing the accuracy and improving the comparability of cavity calculations are given.

Notes: An automatic molecular-replacement procedure has been applied to solve the crystal structure of cytochrome c2 from Rhodopseudomonas viridis. The structure was solved on the basis of the structure of tuna cytochrome c as a search model using an automatic processing program system, AUTOMR. The refinements by molecular dynamics and restrained least-squares methods result in a current crystallographic R factor of 0.219 for diffraction data at 3 Å resolution.

Notes: Crystals of cadmium-substituted azurin have been prepared by diffusing CdII into crystals of apo-azurin grown previously and their structure has been determined at high resolution by X-ray crystallography. Data to 1.8 Å resolution were collected by Weissenberg photography (with image plates) using synchrotron radiation. These data were combined with a 2.2 Å diffractometer data set to give 90% coverage to 1.8 Å. An initial model was derived from the isomorphous CuII-azurin structure, and the cadmium and ligand positions added from `omit' maps. Refinement was by restrained least squares (program PROLSQ), to a final R value of 0.168 for all data in the range 10.0–1.8 Å (23 349 reflections). The final model of 1954 protein atoms, two CdII ions (occupancy 0.75), four SO{_4^{2-}} ions and 239 water molecules has r.m.s. deviations of 0.015, 0.045 and 0.013 Å from standard bond lengths, angle distances and planar groups. The protein structure is essentially the same as that of CuII-azurin, with an r.m.s. deviation of 0.18 Å for 97% of main-chain atoms after superposition of the two structures. The Cd atom is within 0.2 Å of the equivalent copper position, displaced slightly away from the axial Met ligand towards the carbonyl O atom of Gly45. The latter has also moved slightly towards the metal, by a rotation of the peptide unit, to give a Cd—O bond of 2.76 Å. The Cd—S(Cys) bond is lengthened to 2.39 Å. The coordination geometry is slightly more tetrahedral than for CuII, and the cadmium–oxygen interaction is consistent with the presence of an oxygen ligand in the coordination sphere of stellacyanin.

Notes: Monoclonal antibody 4B7 is a neutralizing antibody that binds the protein Pfs25 in the sexual stages of the malaria parasite Plasmodium falciparum and completely blocks transmission of the parasite from human serum to the mosquito host. Here we report the identification of the epitope on Pfs25 recognized by 4B7 and the crystallization of the intact murine monoclonal antibody with peptides corresponding to that epitope. This study highlights the importance of ligands in the crystallization of proteins. In this case peptides have been used to modulate the solubility of the peptide–IgG complex and may have provided different or additional crystal contacts to create or enhance a crystalline reticulum. Multiple crystal forms characterize this crystallization and the various peptides, differing both in length and sequence, have been used to investigate how such changes affect nucleation and crystal growth.

Notes: The diffraction resolution of crystals of the guanine nucleotide-exchange factor complex, EF–Tu–Ts, has been extended from 5.0 to 2.5 Å by lowering the solvent content in the crystals as well as the temperature of data collection. The common form of EF–Tu–Ts crystal belongs to space group P212121 with a = 81.1, b = 109.9, c = 207.5 Å and has a solvent content of 61%. The crystals diffract to a resolution of 5.0 Å at 293 K and 4.0 Å at 273 K. When cryoprotective agents are slowly diffused into the crystals, the cell constants shrink to a = 74.4, b = 109.9, c = 198.7 Å and the solvent content falls to 55%. After the cryoprotective agent has been added, the crystals diffract to 2.7 Å resolution at 293 or 273 K and 2.5 Å at 250 K. X-ray diffraction data, collected before and after the transformation of individual EF–Tu–Ts crystals, demonstrate that a large percentage of the improvement in diffraction resolution is due solely to the addition of cryoprotective agents. The transfer procedures for the successful introduction of cryoprotective agents into EF–Tu–Ts crystals as well as the general applicability to other crystal systems will be discussed.

Notes: A computer-simulation method is proposed for studying the hydrodynamic interactions of rigid protein molecules. It is a combination of Stokes dynamics and continuum hydrodynamics. The Stokes equations of motion for the protein molecules, the creeping-flow equation for the solvent together with the no-slip boundary conditions give a complete representation of the system. The resulting three-dimensional boundary-value problem can be rewritten in a two-dimensional form (without any loss of information) considering the surfaces of the particles only. Then, by solving the equations on discrete surface elements, the so-called mobility matrix is determined in which all hydrodynamic interactions are included. Finally, after calculation of the conservative forces and the stochastic force, the new velocities of the protein molecules can be determined. The simulation method can be applied to arbitrary particle shapes. It can also handle arbitrary flow fields, and the effects of applying a flow field to the system can be studied. From analysis of the trajectories, information can be gained on the kinetics and thermodynamics in the early stages of the crystallization process.

Notes: All hitherto solved crystal structures of the catalytic (C) subunit of cAMP-dependent protein kinase can be classified into two groups, those with a closed and those with an open conformation of the ATP-binding lobe. The molecules with the closed conformation are all related by a crystallographic 21 axis that connects them into an infinite-chain motif. The motif has only one large contact region that involves many residues, several of them in the ATP-binding lobe, embedded in an extensive network of water molecules. The dominant feature of this region is the hydrophobic interaction between Trp196 and Arg133, Arg134. This motif has been found so far in three different crystal forms, two correspond to ternary enzyme–inhibitor–ATP complexes with mammalian and recombinant C, and one to a binary enzyme–inhibitor complex with recombinant C. The open conformation has been found in two closely related crystal structures, both of cubic symmetry, of the apoenzyme and a binary complex of the mammalian catalytic subunit. In this cubic structure of the binary complex, the hydrogen-bonded intramolecular contacts between Arg18 of the inhibitor and the ATP-binding lobe of the binary and ternary complexes of the recombinant enzyme are missing due to a strong hydrophobic intermolecular contact involving the diiodinated Tyr7. In solution, no crystal contacts prevent these hydrogen bonds involving Arg18 from forming so that it is likely that the binary complex with Tyr7 of the peptide inhibitor iodinated or not, can assume the closed conformation in solution. While the closed structure very likely represents a stable conformation in solution, there is no evidence to suggest that the open conformation represents a unique stable conformational state of the enzyme in solution.

Notes: A novel procedure has been developed for locating heavy-atom positions in crystals of macromolecules. This method used genetic algorithms (GA's) to search for heavy-atom sites that are consistent with an observed difference Patterson function. The procedure is straightforward to apply, space-group independent, and particularly powerful for cases involving non-crystallographic symmetry of multiple heavy atoms in the asymmetric unit. In this paper, we introduce how GA's are used for determining the heavy-atom positions and show how this method is more efficient than a sequential search.

Notes: Ferricytochromes c were crystallized at low ionic strength by macroseeding techniques. Large crystals were grown by seed-induced self-nucleation which occurred anywhere in the drop, regardless of the location of the seed crystal. This unusual crystal-seeding method worked reproducibly in our hands, and X-ray quality crystals have been prepared of several ferricytochromes c: horse, rat (recombinant wild type), and two site-directed mutants of the latter, tyrosine 67 to phenylalanine (Y67F) and asparagine 52 to isoleucine (N52I). Crystals of any one of these four proteins could be used as seeds for the crystallization of any one of the others. All the crystals are of the same crystal form, with space group P212121. There are two protein molecules per asymmetric unit. The crystals are stable in the X-ray beam and diffract to at least 2.0 Å, resolution. Full crystallographic data sets have been collected from single crystals of all four proteins.

Notes: Crystals of macromolecules often have two or more molecules per asymmetric unit, or contain domains of a macromolecule or a macromolecular complex that are structurally independent. In such cases the conventional molecular-replacement method attempts to determine the position of each structural unit independently. Typically, some parts of the structure can be determined more easily or more reliably than other parts. Methods are proposed whereby information from a part of a crystal structure that has been determined can be used to help determine the structure of the remainder. Two different strategies are discussed, `subtraction' and `addition'. With `subtraction' strategy the Patterson function of the known part of the structure is subtracted from the `observed' Patterson. This approach is found to be most effective in the context of the rotation function in that it eliminates peaks that are irrelevant to the desired solution. With `addition' strategy the structure factors of the known component are added to those of the search model. This procedure is most effective in the context of the translation function because it brings the structure factors calculated from the search model closer to those observed. Methods of applying the fast Fourier transform to facilitate these calculations are described. A number of examples are provided including structures of mutants of T4 lysozyme that might not have been solved without recourse to the proposed methods. A method of including information from a heavy-atom derivative in a translation function is also developed and shown to be superior in some situations to the conventional translation function.

Notes: Core tracing is a threshold-independent method of determining connectivity (long chains of high-density values) in electron-density maps. It gives visually sparse pictures of large volumes which are useful for initial fitting and for molecular-boundary determination. New methods for visual presentation of the traces are suggested by the way that the connectivity is parameterized in terms of local connections between maxima and the saddle (lowest) points along the connecting paths. The algorithm also partitions the density into small compact volumes containing the maxima. These volumes are useful for localization and statistical analysis.

Notes: The crystal structures of two anionic inhibitor complexes of human carbonic anhydrase I (HCAI), namely, HCAI–iodide and HCAI–Au(CN)2−, have been refined by the restrained least-squares method at 2.2 and 2 Å nominal resolution, respectively, with good stereochemistry for the final models. The R values have improved from 30.3 to 16.6% for HCAI–iodide and from 28.8 to 17.1% for HCAI–Au(CN)2−. The sites of inhibitor binding as elucidated are totally different in the two structures. The iodide anion replaces the zinc-bound H2O/OH− ligand and renders the enzyme inactive. This result confirms that the zinc-bound H2O/OH− is the activity-linked group in carbonic anhydrase enzymes. Au(CN)2− binds at a different and new site near the zinc ion, without liganding to the metal. The N atom of Au(CN)2− is within hydrogen-bonding distance of the zinc-bound H2O/OH− group which shifts by about 0.4 Å away from the zinc ion in relation to its position in the native HCAI. It is proposed that the presence of the inhibitor Au(CN)2− results in a conformational reorientation of the activity-linked group, due to hydrogen-bond formation with the inhibitor, which in turn sterically hinders the binding of the substrate CO2 molecule in the active site, leading to the inhibition of HCAI enzyme.

Notes: Ferritin, the iron-storage protein, binds porphyrins, metalloporphyrins and the fluorescent dyes ANS (8-anilino-1-naphthalenesulfonic acid) and TNS (2-p-toluidinyl-6-naphthalenesulfonic acid), similarly to apo-myoglobin. Octahedral crystals of horse-spleen apo-ferritin (HSF; 174 amino acids) complexes prepared by the addition of haem, hematoporphyrin or Sn-protoporphyrin IX to a solution of apo-ferritin crystallize in space group F432 with cell parameter a = 184.0 Å. X-ray crystallographic analysis of single crystals prepared from a mixture containing haem or Sn-protoporphyrin IX shows that the haem-binding sites in these crystals are occupied by protoporphyrin IX, which is free of metal, rather than by the original metalloporphyrin. The present paper describes the structure of horse-spleen apo-ferritin cocrystallized with Sn-protoporphyrin IX. The 6797 reflections up to 2.6 Å resolution used in the refinement were obtained from a data set recorded on a Nicolet/Xentronics area detector with Cu Kα radiation from a Rigaku RU 200 rotating anode. The final structure comprises 1613 non-H atoms, two Cd atoms and 170 solvent molecules. Four residues are described as disordered. The root-mean-square deviations from ideal bond lengths and angles are 0.013 A and 2.88°, respectively. Protoporphyrins are observed in special positions on the twofold axes of the ferritin molecule with a stoichiometry of 0.4 per subunit.

Notes: The structure of the copper protein plastocyanin from poplar leaves (Populus nigra var. italica) at 173 K has been subjected to two independent refinements, using a single set of synchrotron X-ray data at 1.6 A resolution. Energy-restrained refinement using the program EREF resulted in lower root-mean-square deviations from ideal geometry (e.g. 0.011 Å for bond lengths) but a higher residual R (0.153) than restrained least-squares refinement using the program PROLSQ (0.014 Å, 0.132). Electron-density difference maps in both refinements provided evidence for disorder at some side chains and solvent atoms, and the PROLSQ refinement made allowance for this disorder. The number of solvent sites identified at the 4σ(ρ) level was 171 in the EREF refinement and 189 in the PROLSQ refinement; 159 of the solvent sites are common to both refinements within 1 Å. The root-mean-square differences between the atomic positions produced by the two refinements are 0.08 Å for Cα atoms, 0.08 Å for backbone atoms and 0.12 Å for all non-H atoms (excluding six obvious outliers) of the protein molecule. The two sets of Cu–ligand bond lengths differ by up to 0.07 Å, and the ligand–Cu–ligand angles by up to 7°. At 173 K the volume of the unit cell is 4.2% smaller than at 295 K. Greater order in the solvent region is indicated by the location of 79 more solvent sites, the identification of extensive networks of hydrogen-bonded rings of solvent molecules, and a general decrease in the thermal parameters. Within the unit cell, the protein molecules are significantly translated and rotated from their positions at ambient temperature. An important structural change at low temperature is a 180° flip of the peptide group at Ser48-Gly49. Nearly all other significant differences between the structures of the protein at 173 and 295 K occur at exposed side chains. If the backbone atoms in the 173 and 295 K structures are superposed, excluding atoms involved in the peptide flip, the root-mean- square difference between the positions of 393 atoms is 0.25 Å. Two internal water molecules, not included in previous descriptions of poplar plastocyanin, have been located. The plastocyanin Cu-site geometry at 173 K is not significantly different from that at 295 K. If plastocyanin undergoes a change in Cu-site geometry at low temperature, as has been suggested on the basis of resonance Raman spectroscopic evidence, then the change is not detected within the limits of precision of the present results.

Notes: Crystals have been obtained of glyceraldehyde 3-phosphate dehydrogenase from the extreme thermophile, Thermus aquaticus. This enzyme is stable and active at 363 K, thus its three-dimensional structure should add insight into the structural basis of protein thermostability. Large high-quality crystals were grown using isopropanol and polyethylene glycol at pH 8.4. They crystallize in the orthorhombic space group P212121 with cell dimensions a = 144.77 (6), b = 148.77 (5), c = 149.50 (7) Å, and diffract to beyond 2.8 Å. The volume of the unit cell and the packing observed in other GAPDH structures suggest that there are two tetramers per asymmetric unit. With 300 kDa/asymmetric unit expected in this form, its solution represents a challenging molecular replacement problem. A low-resolution data set has been recorded and used to carry out self-rotation, cross-rotation and Patterson-correlation refinement calculations. We found that the Q molecular axes of both tetramers are approximately coincident with the crystallographic a axis, and the non-crystallographic symmetry relating the two tetramers is approximately a rotation of 90° about the a axis.

Notes: Diffraction data to 2.7 Å resolution were measured on crystals of the homotetramers of components II and III of the cytoplasmic hemoglobin of the symbiont-harboring clam Lucina pectinata. Even though the crystallization conditions are different and the sequence homology of the two hemoglobins is only 63%, the crystals are isomorphous to each other and to the heterotetramer Hb II/III, implying that the residues primarily involved in the intermolecular interactions and responsible for crystal cohesion may be invariant.

Notes: The molecular structures of cobalt- and nickel-substituted concanavalin A have been refined at 1.6 and 2.0 Å resolution, respectively. Both metal derivatives crystallize in space group I222 with approximate cell dimensions a = 89, b = 87 and c = 63 Å and one monomer in the asymmetric unit. The final R factor for Co-substituted concanavalin A is 17.8% for 29 211 reflections with F 〉 1.0σ(F) between 8.0 and 1.6 Å. For Ni-substituted concanavalin A the final R factor is 15.9% for 16 128 reflections with F 〉 1.0σ(F) between 8.0 and 2.0 Å resolution. Both structures contain a transition-metal binding site and a calcium-binding site but, unlike Cd-substituted concanavalin A, do not have a third metal-binding site. The Co-substituted concanavalin A structure diffracts to the highest resolution of any concanavalin A structure reported to date. A comparison of the structures of Ni-, Co-, Cd-substituted and native concanavalin A gives an indication of coordinate errors, which is a useful baseline for comparisons with saccharide complexes of concanavalin A described in other work. We also give a detailed account of multiple conformations which were found for five side-chain residues.

Notes: The CCP4 (Collaborative Computational Project, number 4) program suite is a collection of programs and associated data and subroutine libraries which can be used for macromolecular structure determination by X-ray crystallography. The suite is designed to be flexible, allowing users a number of methods of achieving their aims and so there may be more than one program to cover each function. The programs are written mainly in standard Fortran77. They are from a wide variety of sources but are connected by standard data file formats. The package has been ported to all the major platforms under both Unix and VMS. The suite is distributed by anonymous ftp from Daresbury Laboratory and is widely used throughout the world.

Notes: We have crystallized a variety of RNA oligonucleotides in a form suitable for X-ray diffraction studies using polyethylene glycol with a low-molecular-weight distribution (PEG 400) as the precipitant. Crystallization experiments on a set of 26 RNA oligomers ranging from eight to 12 nucleotides in length resulted in eight diffraction-quality crystals. Of these eight RNA crystals, six utilized PEG 400 as the precipitating agent. We have also been able to obtain large single crystals of a DNA–RNA hybrid, transfer RNA (two different conditions) and a catalytic RNA from PEG 400 solutions. These results suggest that PEG 400 may be a generally useful alternative to 2-methyl-2,4-pentanediol (MPD) which has, thus far, been the most successful precipitant for DNA oligomers.

Notes: It is of considerable interest to separate the processes of viral infectivity and virion assembly. Until recently this has only been possible with viruses that could be disassembled and reassembled in vitro. Even in these cases it was difficult to establish the authenticity of reassembled capsid protein because of possible irreversible damage that may have occurred to the protein during disassembly. An ideal method for the study of virus assembly is a protein expression system in which conditions are appropriate for spontaneous particle formation from freshly synthesized polypeptides. The baculovirus expression system has proven to be an excellent means to this end. Recently, this approach has been used to study the T = 3 Flock House insect virus and it has been demonstrated that subunits with the wild-type protein sequence, and with site-specific mutations that prevent particle maturation, will assemble and crystallize. This same approach has now been used at Purdue to study the T = 4 Nudaurelia ω capensis insect virus. There is no cell culture system currently available for the study of NωV, thus the expression system provides the first opportunity to study assembly under controlled conditions.

Notes: Lysozyme, which is known to crystallize readily in the presence of many salts, has never been crystallized by salting out with ammonium sulfate. In the present study, lysozyme was first completely desalted by treatment with strong cation- (H+ form) and anion- (OH− form) exchange resins. This leads to a protein solution with only H+ and OH− as counterions, corresponding to its isoionic point. Addition of 2.5–3 molar equivalents of H2SO4 to isoionic lysozyme decreases the pH value to 9–8 and allows crystallization to take place. The space group was found to be P43212, similar to the classical lysozyme crystals grown in the presence of NaCl at pH 4.5, with unit-cell dimensions a = b = 78.9, c = 38.5 Å. Tentative explanation of the sulfate/lysozyme interaction was addressed by mass spectrometry, and shows non-covalent binding of the ions on the protein.

Notes: Early steps in the crystallization process of pancreatic ribonuclease have been investigated by time-dependent fluorescence anisotropy, using a labeled protein as a fluorescent probe. Previous experiments have shown that steady-state fluorescence anisotropy is sensitive to protein–protein interactions and can be used to find new crystallization conditions. The present work describes an attempt, by means of time-resolved experiments, to detect and characterize species appearing in the early stages of the crystallization pathway. Fluorescence anisotropy decay was measured with synchrotron radiation as a light source under a variety of conditions where it is known that the solutions tend towards crystallization; the decay was analyzed by a maximum-entropy method that calculates a rotational correlation-time distribution. Fluorescence anisotropy originates in the Brownian rotatory motion of macromolecules and the values of the correlation times are related to the size and shape of different species present in the solution. In the presence of high salt concentrations, a bimodal distribution is always observed. Whereas a peak of protein monomer is still present, a second peak appears as a stable intermediate in the crystallization pathway. The correlation time of this new species varies between two and three times the correlation time of the monomer. The second peak is possibly the symmetrical dimer of the ribonuclease molecules commonly observed in all the high-salt crystal forms.

Notes: Mirabilis anti-viral protein (MAP) is a ribosome-inactivating protein from Mirabilis jalapa L. Since MAP is effective over a broad spectrum of species, the protein is difficult to express in heterologous hosts such as Escherichia coli. Recently, we obtained a MAP mutant, Y72F which exhibits a lower (1/100) activity against E. coli ribosomes while retaining almost full activity against mammalian cells [Habuka, Miyano, Kataoka, Tsuge & Noma (1992). J. Biol. Chem. 267, 7758–7760]. For the crystallographic studies, the Y72F MAP expression vector with an OmpA leading sequence was constructed and expressed in E. coli. The Y72F MAP mutant was then isolated and purified from the cell culture medium. Crystals were grown using the crystallization conditions for the native MAP crystals [Miyano et al. (1992). J. Mol. Biol. 226, 281–283]: 50% ammonium sulfate containing 50 mM ammonium citrate and 2 mM adenine sulfate, pH 5.4. The crystals belong to space group P3121 (or P3221) with a = b = 104.1 and c = 134.3 Å. The crystals are isomorphous with the wild-type crystals but diffract to higher resolution. Imaging-plate photographs of the Y72F mutant showed sharp intense spots without the streaking observed in the native crystals.

Notes: Ruminant haemoglobin (Hb) extracted from river buffalo (Bubalus bubalis) has been purified and crystallized. Two different Hb forms of the phenotype BB gave isomorphous crystals which diffracted to 2.8 Å resolution and were not sensitive to radiation damage. Crystals of CO Hb have space group P212121 with unit-cell parameters a = 54.8, b = 64.0, c = 158.6 Å, and contain one Hb molecule per asymmetric unit.

Notes: We report the cDNA sequence determination and the crystal structure of the Fab fragment of a murine IgG1,λ antibody (HC19), specific for an influenza virus hemagglutinin. The HC19 Fab-fragment structure has been refined; the crystallographic R-factor is 19.5% at 2.3 Å resolution. We have compared the conformation of HC19 complementarity determining regions (CDRs) with those of CDR loops of Fab structures available from the Protein Data Bank. These loops were chosen based on the identity of key residues, following the canonical-structure approach; four CDRs have a main-chain conformation very similar to the canonical structure that had been identified. HC19 L1 CDR adopts a conformation clearly distinct from all L1 CDRs that belong to a chain of a different class or origin; this is determined by the nature of a few residues at positions in the sequence different from those of key residues in other light chains. This canonical structure should be representative of most murine λ-class light chains, as inferred from the very high sequence homologies of these polypeptides.

Notes: The active centers of NiFe hydrogenase from Desulfovibrio vulgaris Miyazaki F have been located in the electron-density map calculated at 4 Å resolution. The electron-density map based on five heavy-atom derivatives showed four strong peaks which were clearly distinguished from the protein region. These strong densities have been successfully assigned to three iron–sulfur clusters and one Ni atom by a difference Fourier technique with coefficients of the best phases from the multiple isomorphous replacement (MIR) method and structure factors obtained at five wavelengths (1.040, 1.487, 1.730, 1.743 and 1.750 Å) with the use of a synchrotron radiation source. Four active centers are approximately lined up at a distance of ca 13 Å, which seems reasonable if they are connected with the electron-transfer chain.

Notes: Human methylamine-treated complement C3 (C3-MA) and C3b (C3b-MA) have been crystallized using ammonium sulfate as precipitant. The crystals of the two compounds are morphologically indistinguishable though they belong to different space groups. We show that only minor alterations in packing are responsible for the change in space group. Crystals of C3-MA are tetragonal [P41(3)22, a = b = 135, c = 610 Å] with two molecules per asymmetric unit. Crystals of C3b-MA are also tetragonal [P41(3)212, a = b = 191, c = 610 Å] with four molecules per asymmetric unit. The maximum diffraction observed is 7.7 Å at cryogenic temperature using synchrotron radiation.

Notes: To assess the accuracy of refined structures, a comparison was made using independently determined structures of the same protein in the same crystal form. The models were re-refined against a common data set to minimize the effects of different data and different refinement protocols. The process did not converge to a single model. Rather the structures differed from each other by 0.84 Å which was roughly three times that predicted by a Luzzati analysis [Luzzati (1952). Acta Cryst. 5, 802–810]. The individual structures are equally valid and at least partially independent as evidenced by a reduction of the R factor by 0.013 when a simple linear combination is used. Only 29 solvent molecules were common to all four models.

Notes: Three crystal forms of the sweet-tasting protein thaumatin from the African berry Thaumatococcus daniellii have been grown. These include two naturally occurring isoforms, A and B, that differ by a single amino acid, and a recombinant form of isoform B expressed in yeast. The crystals are of space groups C2 with a = 117.7, b = 44.9, c = 38.0 Å, and β = 94.0°, P212121 with a = 44.3, b = 63.7 and c = 72.7 Å, and a tetragonal form P41212 with a = b = 58.6 and c = 151.8 Å. The structures of all three crystals have been solved by molecular replacement and subsequently refined to R factors of 0.184 for the monoclinic at 2.6 Å, 0.165 for the orthorhombic at 1.75 Å, and 0.181 for the tetragonal, also at 1.75 Å resolution. No solvent was included in the monoclinic crystal while 123 and 105 water molecules were included in the higher resolution orthorhombic and tetragonal structures, respectively. A bound tartrate molecule was also clearly visible in the tetragonal structure. The r.m.s. deviations between molecular structures in the three crystals range from 0.6 to 0.7 Å for Cα atoms, and 1.1 to 1.3 Å for all atoms. This is comparable to the r.m.s. deviation between the three structures and the starting model. Nevertheless, several peptide loops show particularly large variations from the initial model.

Notes: The crystal structure of the ternary complex of horse liver alcohol dehydrogenase (LADH) with the coenzyme NADH and inhibitor dimethyl sulfoxide (DMSO) has been refined by simulated annealing with molecular dynamics and restrained positional refinement using the program X-PLOR. The two subunits of the enzyme were refined independently. The space group was P1 with cell dimensions a = 51.8, b = 44.5, c = 94.6 Å, α = 104.8, β = 102.3 and γ = 70.6°. The resulting crystallographic R factor is 17.3% for 62 440 unique reflections in the resolution range 10.0–1.8 Å. A total of 472 ordered solvent molecules were localized in the structure. An analysis of secondary-structure elements, solvent content and NADH binding is presented.

Notes: Based on amino-acid sequence homology, it is predicted that ribosomal protein L14 is a member of a recently identified family of structurally related RNA-binding proteins. To verify this, the gene for Bacillus stearothermophilus L14 has been cloned, and the protein has been purified and crystallized. The crystals are in space group C2 with cell dimensions a = 67.0, b = 32.7, c = 49.4 Å, and β = 101.8°, and there is one molecule in the asymmetric unit (Vm = 2.0 Å3 Da−1). They are of high quality, and a native data set has been collected to a resolution of 1.6 Å, with an Rmerge of 5.3%.

Notes: The direct methods program SAYTAN was applied to simulated data at various resolutions from three oligonucleotides. Success in solving the structures was found to depend more upon the resolution of the data than upon errors in the data or the complexity of the structure. Collecting the data at a reduced temperature has little effect, unless it alters the mosaicity of the crystal or changes the resolution of the data. The presence of a heavy atom dramatically improved the phase refinement, particularly at low resolution.

Notes: The crystal structure of glutaminase–asparaginase from Acinetobacter glutaminasificans has been reinterpreted and refined to an R factor of 0.171 at 2.9 Å resolution, using the same X-ray diffraction data that were used to build a preliminary model of this enzyme [Ammon, Weber, Wlodawer, Harrison, Gilliland, Murphy, Sjölin & Roberts (1988). J. Biol. Chem. 263, 150–156]. The current model, which does not include solvent, is based in part on the related structure of Escherichia coli asparaginase and is significantly different from the structure of the enzyme from A. glutaminasificans described previously. The reason for the discrepancies has been traced to insufficient phasing power of the original heavy-atom derivative data, which could not be compensated for fully by electron-density modification techniques. The corrected structure of A. glutaminasificans glutaminase–asparaginase is presented and compared with the preliminary model and with the structure of E. coli asparaginase.

Notes: In recent years it has been shown that direct methods are capable of solving the structures of small proteins. Mukherjee & Woolfson [Acta Cryst. (1993), D49, 9–12] have shown that useful phase sets can be produced even at 3 Å resolution but that the standard figures of merit could not distinguish the better phase sets from others. They found modified forms of the standard figures of merit that could pick out better phase sets for 2 Å resolution or higher. Gilmore, Henderson & Bricogne [Acta Cryst. (1991), A47, 842–846] have shown that evaluation of the log-likelihood gain, coming from entropy-maximization procedures, is also very successful in picking out good protein phases sets. A new figure of merit is described, based on the expected charactistics of an electron-density map for a protein, and comparisons are made with the other figures of merit mentioned above.

Notes: Raster3D Version 2.0 is a program suite for the production of photorealistic molecular graphics images. The code is hardware independent, and is particularly suited for use in producing large raster images of macromolecules for output to a film recorder or high-quality color printer. The Raster3D suite contains programs for composing illustrations of space-filling models, ball-and-stick models and ribbon-and-cylinder representations. It may also be used to render figures composed using other graphics tools, notably the widely used program Molscript [Kraulis (1991). J. Appl. Cryst. 24, 946–950].

Notes: The crystal structures of porcine and human aldehyde reductase, an enzyme implicated in complications of diabetes, have been determined by X-ray diffraction methods. The crystallographic R factor for the refined porcine aldehyde reductase model is 0.19 at 2.8 Å resolution. There are two molecules in the asymmetric unit related by a local non-crystallographic twofold axis. The human aldehyde reductase model has been refined to an R factor of 0.21 at 2.48 Å resolution. The amino-acid sequence of porcine aldehyde reductase revealed a remarkable homology with human aldehyde reductase. The coenzyme-binding site residues are conserved and adopt similar conformations in human and porcine aldehyde reductase apo-enzymes. The tertiary structures of aldhyde reductase and aldose reductase are similar and consist of a β/α-barrel, with the coenzyme-binding site located at the carboxy-terminus end of the strands of the barrel. The crystal structure of porcine and human aldehyde reductase should allow in vitro mutagenesis to elucidate the mechanism of action for this enzyme and facilitate the effective design of specific inhibitors.

Notes: The three-dimensional structure of the complex between methyl α-D-mannopyranoside and concanavalin A has been refined at 2.0 Å resolution. Diffraction data were recorded from a single crystal (space group P212121, a = 123.7, b = 128.6, c = 67.2 Å) using synchrotron radiation at a wavelength of 1.488 Å. The final model has good geometry and an R factor of 19.9% for 58 871 reflections (82% complete), within the resolution limits of 8 to 2 Å, with F 〉 1.0σ(F). The asymmetric unit contains four protein subunits arranged as a dimer of dimers with approximate 222 point symmetry. Each monomer binds one saccharide molecule. Each sugar is bound to the protein by hydrogen bonds and van der Waals contacts. Although the four subunits are not crystallographically equivalent, the protein–saccharide interactions are nearly identical in each of the four binding sites. The differences that do occur between the four sites are in the structure of the water network which surrounds each saccharide; these networks are involved in crystal packing. The structure of the complex is compared with a refined saccharide-free concanavalin A structure. The saccharide-free structure is composed of crystallographically identical subunits, again assembled as a dimer of dimers, but with exact 222 symmetry. In the saccharide complex the tetramer association is different in that the monomers tend to separate resulting in fewer intersubunit interactions. The average temperature factor of the mannoside complex is considerably higher than that of the saccharide-free protein. The binding site in the saccharide-free structure is occupied by three ordered water molecules and the side chain of Asp71 from a neighbouring molecule in the crystal. These occupy positions similar to those of the four saccharide hydroxyls which are hydrogen bonded to the site. Superposition of the saccharide-binding site from each structure shows that the major changes on binding involve expulsion of these ordered solvents and the reorientation of the side chain of Tyrl00. Overall the surface accessibility of the saccharide decreases from 370 to 100 Å2 when it binds to the protein. This work builds upon the earlier studies of Derewenda et al. [Derewenda, Yariv, Helliwell, Kalb (Gilboa), Dodson, Papiz, Wan & Campbell (1989). EMBO J. 8, 2198–2193] at 2.9 Å resolution, which was the first detailed study of lectin–saccharide interactions.

Notes: X-ray diffraction intensities for tobacco necrosis virus crystals were collected at 5 Å resolution using a Weissenberg camera with a large cassette of radius 430 mm. The synchrotron radiation source at the Photon Factory was used. The crystal structure of the virus was obtained by 91 cycles of the non-crystallographic symmetry averaging. Secondary structures such as α-helices and β-structures were clearly identified in the electron-density map at 5 Å resolution. This virus resembles southern bean mosaic virus both in orientation of coat protein subunits and in their folding. Ordered and disordered parts of each subunit of tobacco necrosis virus are shorter and longer than the corresponding parts of the southern bean mosaic virus by 12 and 27 residues, respectively.

Notes: This paper describes the study of the effects of radiation damage on the quality of data collected from a protein crystal at 100 K. It is shown that radiation damage causes measurable effects in the diffraction pattern. This implies that, even at liquid nitrogen temperatures, there is a limit to the size of a crystal from which a complete data set can be collected.

Notes: A binary complex of dihydropteridine reductase and NADH crystallizes in the space group C2, with a = 222.2, b = 46.5, c = 95.3 Å and β = 101.1°. There are two dimers in the asymmetric unit. The structure was solved by molecular-replacement techniques and refined with 2.6 Å data to a crystallographic R factor of 16.8%. Each dimer has twofold non-crystallographic symmetry and the four individual monomers in the asymmetric unit have the same overall molecular conformation.

Notes: A model structure of the human complement enzyme factor D was built based on homology with related serine proteases. A molecular-replacement solution of the factor D crystal structure employing the homology model refined without manual intervention to an R factor of 0.249 with 2.4 Å native diffraction data. A multiple isomorphous replacement (MIR) electron-density map was subsequently produced, leading to a model refined at 2.0 Å resolution to an R factor of 0.188. A homology model built with commercial modeling software was subjected to the same procedure. Comparisons of the homology models with the final refined MIR structure are presented. Major discrepancies were found in critical active-site regions.

Notes: A variety of criteria were tested for identifying errors in protein crystal coordinates. Statistical analysis was based on comparisons of a highly refined crystal structure and several preliminary models derived from molecular replacement. A protocol employing temperature factors, real-space fit residuals, geometric strains, dihedral angles and shifts from the previous refinement cycle is developed. These results are generally applicable to the detection of errors in partially refined protein crystal structures.

Notes: Affinity-purified amaryllis lectin was used to grow single crystals using the hanging-drop method. The space group was found to be C2 with unit-cell dimensions a = 73.4 (1), b = 100.3 (1), c = 62.2 (1) Å and β = 137.3 (2)°. Data to 2.25 Å resolution have been recorded and solution of the structure is currently underway by means of molecular-replacement techniques.

Notes: An essential first step in most techniques for using anomalous-scattering data for phase determination is to determine the positions of the anomalous scatterers. This is usually done by use of the anomalous differences, either as input to a direct-methods procedure or to produce a Patterson map. If the arrangement of anomalous scatterers is noncentrosymmetric then it is also necessary to find their absolute configuration and a process is described for doing this based on the properties of the Ps function [Okaya, Saito & Pepinsky (1955). Phys. Rev. 98, 1857–1858]. If the arrangement of anomalous scatterers is centrosymmetric then the problem does not occur.

Notes: A Bayesian approach is applied to the calculation of Patterson functions and cross-Fourier maps in the analysis of multi-wavelength anomalous-diffraction (MAD) data. This procedure explicitly incorporates information available a priori on the likely magnitudes of partial structure factors (FA) corresponding to the anomalously scattering atoms, uses weighted-average estimates of FA, and incorporates estimates of errors in the data that are not represented in the instrumental uncertainties. The method is demonstrated by application to MAD data collected on selenomethionine-containing gene V protein.

Notes: A recently reported method – the improved condensing protocol [Subbiah (1991). Science, 252, 128–133; (1993). Acta Cryst. D49, 108–119] – for obtaining low-resolution macromolecular envelopes is applied to five varied test cases. These examples were chosen to illustrate the general applicability of the method to the wide range that typical macromolecular crystals adopt. The cases include small and large asymmetric unit volumes (4.7 × 104 to 1.17 × 106 Å3), low and high symmetry (2 to 12 symmetry elements), small and large proteins (1570 to 12 216 non-H atoms), orthogonal and non-orthogonal unit cells, a wide variety of space groups (P21 to P6322), small and large solvent contents (33–80%), and a case of non-crystallographic symmetry (threefold). In all five cases the inherent ambiguity of the condensing protocol in differentiating between bulk matter and bulk solvent is then resolved by use of the recently reported sign-fixing method (Subbiah, 1993).

Notes: Trypanothione reductase is an FAD-dependent disulfide oxidoreductase which catalyses the reduction of trypanothione using NADPH as co-factor. The enzyme is unique to protozoan parasites from the genera Trypanosoma and Leishmania and is an important target for the design of improved antitrypanocidal drugs. We present details of the structure of trypanothione reductase from Crithidia fasciculata solved by molecular replacement, using human glutathione reductase as a search model, and refined to an R factor of 16.1% with data between 8.0 and 2.6 Å resolution. The model comprises two subunits (one containing 487 residues, the other 486), an FAD prosthetic group, plus 392 solvent molecules. The last four C-terminal residues are not seen in either subunit and the density is poor for the N-terminal residue of subunit B. The model has a root-mean-square deviation from ideality of 0.016 Å for bond lengths and 3.2° for bond angles. Each subunit was independently refined in the latter stages of the analysis but the subunits remain similar as indicated by the root-mean-square deviation of 0.35 Å for Cα atoms. Trypanothione reductase has 36% sequence identity with human glutathione reductase and the root-mean-square deviation between the 462 Cα atoms in the secondary structural units common to the two proteins is 1.1 Å. However, there are large differences in the loop regions and significant shifts in the orientation of the four domains within each subunit. Domain II, which binds the dinucleotide co-factor, and domain IV, which forms the interface between the two subunits, are both rotated by approximately 5° with respect to domain I, which binds the FAD moiety, when compared with glutathione reductase. Crystals of trypanothione reductase have been soaked in the dinucleotide co-factor NADPH and N1-glutathionylspermidine disulfide substrate and the structure of the resulting complex determined at 2.8 Å resolution. Strong density is observed for the adenosine end of the co-factor which forms many charged interactions with the protein though the density for the nicotinamide moiety is more diffuse. The mode of binding indicates that NADP is bound to the enzyme in a similar conformation to that observed with human glutathione reductase.

Notes: The structure of chicken skeletal muscle troponin C (TnC) has been refined to an R value of 0.168, using 14 788 reflections, in the resolution range 8.0–1.78 Å. Our earlier 2 Å resolution structure [Satyshur, Rao, Pyzalska, Drendel, Greaser & Sundaralingam (1988). J. Biol. Chem. 263, 1628–1647] served as the starting model. The refined model includes atoms for all protein residues (1–162), 2 Ca2+ ions, 169 water molecules and one sulfate ion. The high-resolution refinement shows more clearly the details of the protein and water structure. The side chains Glu63, Cysl01, Arg123, Aspl40 and Asp152 adopt two discretely ordered conformations. The long central helix is only slightly curved/bent (7.9°) and all the central helix NH...O=C hydrogen bonds are intact. Seven of the nine carbonyl O atoms of the mid segment of this helix, including the D/E linker region, are hydrogen bonded to water molecules which weakens the helix hydrogen bonds. In contrast, in each of the protected upper and lower thirds of the long central helix, only two carbonyl O atoms are hydrogen bonded to water molecules. The hydrogen-bonding patterns displayed by some of the carbonyl O atoms of NT and A helices of the N-terminal domain and the F and H helices of the C-terminal domain, which are on the exposed surface of the protein, are similar. The B helix of the calcium-free site I is kinked, with the local helix axes at either end making an angle of 39°, by two inserted water molecules between N—H and O=C groups, breaking the adjacent helix hydrogen bonds. A sulfate ion from the crystallization buffer is also trapped in the B helix between the guanidinium group of Arg47 and these two inserted water molecules. The C helix of site II is devoid of similar hydration and is probably responsible for the different interhelical angles A/B at site I (134°) and C/D at site II (149°). Extensive interhelix hydrogen bonds occur between the side chains of the C and D helices of the `apo' site II: Gln51–Asp89, Asn52–Asp89, Glu57–Gln85, Glu57–Glu88 and Glu64–Arg84, which apparently are disrupted upon Ca uptake and the resulting rearrangement of the helices expose the side chains, lining the palm of the N-(and C-) terminal domains, for interaction with specific peptide fragment of troponin I (Tnl) during muscle contraction. The dominant crystal packing motif involves a head-to-tail interaction between the N-terminal domain A helix of one molecule and the palm of the C-terminal domain of the 32-related molecule, in a manner similar to that which can be expected for the TnC–TnI complex. Similar interactions may also be responsible for the dimerization of TnC at low pH.

Notes: The crystal structure of the recombinant calmodulin from Paramecium tetraurelia (rPCaM, Mr = 16 700, 148 residues) has been determined at 1.68 Å resolution. X-ray intensity data were collected at 263 K using a Siemens–Nicolet area detector and Cu Kα radiation from a rotating-anode source. A total of 35 936 observations were processed with XENGEN1.3 and scaled to yield 16 255 unique reflections with Rsymm(I) of 4.1%. The crystals are triclinic, with unit-cell dimensions a = 29.89, b = 53.42, c = 25.35 Å, α = 93.67, β = 96.88, γ = 89.24°, space group P1, with one molecule in the unit cell. The atomic coordinates of the wild-type Paramecium calmodulin (PCaM) studied in our laboratory provided the starting model. Refinement of the structure by X-PLOR and refitting it into omit maps yielded an R value of 0.194 for 15 965 reflections greater than 3σ(F) in the 6.0–1.68 Å resolution range. The final model contained 1165 protein atoms for all of the 148 residues, four Ca2+ ions, and 172 water molecules. The dumbbell structure has seven α-helices including a long 7.8 turn central helix connecting the two terminal domains each containing two EF-hand (helix–loop–helix motif) calcium-binding sites. The loops within each pair of EF-hand motifs in the N- and C-terminal domains are brought into juxtaposition to form a pair of hydrogen-bonded antiparallel β-sheets which are extended at either ends by water bridges. The four calcium-binding EF-hands are superposable with r.m.s. deviations of 0.31–0.79 Å. The best agreement is between site 1 and site 3 and the worst agreement is between site 1 and 4. The largest differences are in the ninth and tenth residues of the calcium-binding loops probably because of their involvement in the mini β-sheets. The calcium coordination distances vary between 2.04 and 2.69 Å, average 2.34 Å. The rPCaM and wild-type PCaM have an r.m.s. deviation of 0.36 Å for equivalent Cα atoms. The side chains of Lys13 and Lys115 are more extended in rPCaM compared to the wild type where the post-translational modified di- and tri-methylated lysine residues are more folded. The sequence of PCaM differs from those of mammalian (MCaM) and Drosophila calmodulin (DCaM), but the overall structures are very similar, with r.m.s,. deviations of 0.44 and 1.68 Å for equivalent Cα atoms, respectively. However, in rPCaM, the first four N-terminal residues stretch out and make intermolecular crystal contacts, in contrast to those in recombinant Drosophila calmodulin (rDCaM), they stretch out in the opposite direction and towards the second calcium-binding site (see note below), while in MCaM and wild-type PCaM, the N-terminal residues are not visible. The central helix in rPCaM has all its backbone hydrogen bonds intact with no unusually long separation between the carbonyl and amide groups as found in MCaM and rDCaM.

Notes: Today the determination of successful crystallization conditions for a particular macromolecule remains a highly empirical process. Sparse-matrix and grid-screening procedures are rapid and economical means to determine preliminary crystallization conditions. During optimization the variable set (pH, precipitant type and precipitant concentration) utilized in these procedures is screened in an attempt to determine appropriate conditions for the nucleation and growth of single crystals suitable for X-ray diffraction analysis. Unfortunately, in many cases this strategy will not produce single crystals suitable for X-ray diffraction analysis. We have explored, in an empirical sense, other tools for use during optimization. First, a new screening protocol is evaluated which employs less classical precipitating agents. Second, a set of 24 electrostatic crosslinking agents are evaluated for their ability to promote crystallization. Third, a panel of more than 30 detergents are evaluated for their ability to prevent sample aggregation and influence crystal growth.

Notes: Version 3.0 of the NIST/NASA/CARB Biological Macromolecule Crystallization Database (BMCD) includes crystal and crystallization data on all forms of biological macromolecules which have produced crystals suitable for X-ray diffraction studies. The data include summary information on each of the macromolecules, crystal data, crystallization conditions and comments about the crystallization procedure if it varies from the traditional methods employed for crystal growth. The database-management software maintains continuity with previous versions providing similar search procedures and displays. Version 3.0 of the BMCD includes protocols and results of crystallization experiments undertaken in space. These new data are comprised of both the NASA Protein Crystal Growth Archive, which includes information on all NASA-sponsored protein crystal growth experiments, and data describing other internationally sponsored microgravity macromolecule crystallization studies. The entries for the space growth crystallization experiments contain the crystallization protocols, apparatus descriptions, flight summary data, indication of success or failure of the experiments, references, etc. Other new features of the BMCD include the addition of crystallization procedures for small peptides and cross references to other structural biology databases.

Notes: The orthorhombic, or high-temperature, form of chicken egg-white lysozyme typically appears at temperatures ≥298 K. Solubility diagrams have been determined for this form of lysozyme from pH 4.0 to 5.4 in 0.2 pH increments using the micro-column technique. Data were collected in the 297–317 K temperature range which resulted in phase diagrams similar in overall shape to those obtained for the lower temperature tetragonal form. Specifically, the solubility increased with increasing temperature and decreased with increasing precipitant concentration. However, the solubility of the orthorhombic form is considerably less sensitive to temperature than the tetragonal form, resulting in a more flattened slope. On the other hand, pH effects on the high-temperature form were opposite to those on the low-temperature form. When holding the precipitant concentration constant, the solubility decreased with increasing pH for the orthorhombic form. Previous tetragonal data were incorporated with these orthorhombic data to produce intercept values. These values varied with both pH and precipitant concentration, but the general tendency of the slope was to decrease with increasing pH.

Notes: In protein crystallography, the initial experimental problem is the identification of physical and chemical conditions that will support nucleation and crystal growth. Ideally, experiments to search for such conditions would be based on a full-factorial structure, with variation in the temperature and solution composition. However, consideration of even a moderate number of possibilities for the composition of the system will result in factorial experiments which may be prohibitively large. In this paper it is proposed that search experiments for protein crystallization might be based on orthogonal arrays. These are subsets of full-factorial experiments which possess a great deal of symmetry, such that a uniform distribution of points throughout the experimental region is preserved. Such experiments have reasonable size, explore the proposed experimental region in a systematic fashion, and form a logical basis for a sequential approach to the search for crystallization conditions. Examples of such initial search experiments are given, and their application to some recent protein crystallization problems in this laboratory is described briefly. The relationship of this approach to other protein crystallization search procedures is also discussed.

Notes: Predispensed gradient matrices allow the boundary between precipitate and clear solution to be located very rapidly for a particular protein and precipitant. In many cases crystals grow in the trials which were used to identify this boundary. The method involves dispensing a series of between 10 and 72 microbatch trials in which some parameter, such as precipitant concentration, is gradually changed. (Protein is not dispensed at this stage.) Protein is then added to selected trials using a predetermined algorithm, which takes into account the level of precipitation caused by previous additions. Thirteen crystal forms were obtained using the method with eight proteins and eight precipitants. Six forms were prisms or plates with maximum dimensions above 400 μm.

Notes: A convenient method for screening crystallization conditions using an automated fast-screen protocol has been implemented and tested on an enoyl-CoA hydratase. The crystallization solutions for the initial screening and subsequent optimizations are prepared using a crystallization robot. Enoyl-CoA hydratase (E.C. 4.2.1.17), purified from rat-liver mitochondria, is one of the enzymes from the β-oxidation pathway of fatty-acid metabolism; it catalyzes the reversible hydration of 2-trans-enoyl-CoA's to L-3-hydroxy-acyl-CoA's. Different crystal forms, diffracting to 3.0 Å, were obtained.

Notes: Recent developments of the IMPAX system for automated crystallization are presented. A five-channel microtip has been introduced into the system thereby providing an extra degree of freedom for carrying out experiments. A new mouse-driven program for screening has been introduced, which creates a much wider scope for designing and executing screens covering new conditions of crystallization. The hardware has been adapted so that the system can also be used to set up vapour-diffusion trials. A simple design of a vapour-diffusion vessel, suitable for sitting drops of 2–15 μl, using smaller reservoir volumes (up to 100 μl), facilitates large-scale systematic trials.

Notes: Two different systems for recording crystallization results are described. Both programs are written in Microsoft Excel. Results can be entered either as text or numerical scores. In each system, one can search or query on the results, macromolecule name, experimental parameters, etc. and correlate these to specific crystallization conditions or trends in the crystallizations. Such information is valuable when optimizing crystal growth conditions and determining new crystallization parameters to test. XTAL RES is a Macintosh program that is linked to PCRYSTAL (software used to enter details of the crystallization experiments) through an Oracle database on a VAX network. A second system, the Electronic Notebook, was designed as a `stand-alone' Excel application running under Microsoft Windows.

Notes: Poly(ethylene) glycol monomethyl ethers (Peg-mmes) are a series of methyl substituted poly(ethylene) glycols that have been used with some success in the crystallization of a number of hydrophobic proteins. Crystallization of a lipase from Humicola lanuginosa complexed with the C12 substrate analogue from Peg-mme 5000, an endoglucanase 1 and a 59 kDa fragment of human topoisomerase IIα crystallized from Peg-mme are described. The use of Peg-mme for improving the quality of crystals previously grown from normal poly(ethylene) glycol 8000 is also described. We suggest that these modified Peg-mmes should be regularly used in screening for crystallization.

Notes: Major emphasis has been placed in recent years on kits for screening crystallization conditions of macromolecules. Such approaches have undoubtedly speeded up the initial screening and, to a certain extent, helped in reducing the amount of protein required for the initial survey. Factorial screening techniques, either full-factorial or sparse-matrix approaches, have proved successful in the crystallization of many proteins. However, in cases where the amount of protein is limited a systematic approach based on an a priori choice of precipitants may be preferable to an extensive search. The approach described here targets such situations. The approach consists of the determination of the solubility characteristics of the macromolecule under study as a function of precipitant and macromolecule concentrations to define a working range for these parameters. Conditions under which the protein is highly supersaturated, and hence more conducive to nucleation, are established so as to favor the formation of an initial stable nucleus which can be one of the dominant problems that hinders successful crystallization of proteins. Later, changes in solubility as a function of pH and as a result of the introduction of additives are evaluated. In addition, when ligands are available for the formation of macromolecular complexes, screening of different complexes is used as a means to increase the probability of obtaining crystals. Solubility information derived from one, or more, complexes that have been screened can be used for comparison and to aid in the crystallization of other complexes. Cross-seeding between complexes is an intrinsic part of the method and provides an efficient way of obtaining crystals when spontaneous nucleation is hard to achieve. In the example presented here, reverse screening has enabled the production of crystals of several peptide complexes with an anti-malaria antibody.

Notes: The advances in recombinant DNA technology in recent years have had a dramatic effect on the area of protein crystallization. Large amounts of pure protein produced in various expression systems have made it possible to conduct experiments that would have been impossible with material from natural sources. With many more laboratories becoming involved in crystallizing proteins a great deal of new information has been generated on techniques to eliminate the so called `bottleneck of crystallization' in determining a three-dimensional protein structure. More and more new and interesting proteins are being submitted to this laboratory for crystallization. Certain criteria may be set before crystallization trials are started, such as solubility, purity and aggregation tendencies. The introduction of robots now facilitates the screening of crystallization conditions. In cases where no crystals have been obtained after initial screening it can now be decided which possible modifications can be made to the protein itself to improve the chances of obtaining crystals.