Improved specimen preparation for cryo-electron microscopy using a symmetric carbon sandwich technique

Image shift due to beam-induced specimen charging has become the most severe problem in electron microscopy for imaging two-dimensional (2D) crystals of biological macromolecules, especially in the case of highly tilted specimens. Image shift causes diffraction spots perpendicular to the tilt axis to disappear even at medium or low resolution. The yield of good images from tilted specimens prepared on a single layer of continuous carbon support film is therefore very low. In this paper, we have used 2D crystals of aquaporin-4 to investigate the effect of a carbon sandwich preparation method on specimen charging. We find that a larger number of images show sharp diffraction spots perpendicular to the tilt axis if crystals are placed in between two sheets of carbon film as compared to images taken from specimens prepared by the conventional single carbon support film technique. Our results demonstrate that the reproducible carbon sandwich preparation technique overcomes the severe specimen charging problem and thus has the potential to significantly speed up structure analysis by electron crystallography.     

Structure of CaATPase: electron microscopy of frozen-hydrated crystals at 6 A resolution in projection

Thin, three-dimensional crystals of CaATPase have been studied at high resolution by electron crystallography. These crystals were grown by adding purified CaATPase to appropriate concentrations of lipid, detergent and calcium. A thin film of crystals was then rapidly frozen and maintained in the frozen-hydrated state during electron microscopy. The resulting electron diffraction patterns extend to 4.1 A resolution and images contain phase data to 6 A resolution. By combining Fourier amplitudes from electron diffraction patterns with phases from images, a density map has been calculated in projection. Comparison of this map from unstained crystals with a previously determined map from negatively stained crystals reveals distinct contributions from intramembranous and extramembranous protein domains. On the basis of this distinction and of the packing of molecules in the crystal, we have proposed a specific arrangement for the ten alpha-helices that have been suggested as spanning the bilayer.     

Electron crystallography of macromolecular periodic arrays on phospholipid monolayers

Electron crystallography has the potential of yielding structural information equivalent to x-ray diffraction. The major difficulty has been preparing specimens with the required structural order and size for diffraction and imaging in the electron microscope. 2D crystallization on phospholipid monolayers is capable of fulfilling both of these requirements. Crystals can form as a result of specific interactions with a protein's ligand or an analog, suitably linked to a lipid tail; or on a surface of complementary head-group charge. With such choices, the availability of a suitable lipid is limited only by synthetic chemistry. Ultimately, it is the quality and regularity of the protein-protein interactions that determine the crystalline order, as it is with any protein crystal. In the case of streptavidin, the monolayer crystal diffracts beyond 2.5 Å. A 3 Å projection map reconstructed from electron diffraction amplitudes and phases from images shows density which can be interpreted as β-sheets and clusters of side chains. It remains to be shown that the monolayer crystals are flat and diffract as well at high tilt angle as untilted. Technological issues such as charging must be resolved. With parallel advances in data collection and processing, electron crystallography of monolayer macromolecular crystals will eventually take its place beside x-ray crystallography and NMR as a routine and efficient structural technique.     

7A projection map of the S-layer protein sbpA obtained with trehalose-embedded monolayer crystals

Two-dimensional crystallization on lipid monolayers is a versatile tool to obtain structural information of proteins by electron microscopy. An inherent problem with this approach is to prepare samples in a way that preserves the crystalline order of the protein array and produces specimens that are sufficiently flat for high-resolution data collection at high tilt angles. As a test specimen to optimize the preparation of lipid monolayer crystals for electron microscopy imaging, we used the S-layer protein sbpA, a protein with potential for designing arrays of both biological and inorganic materials with engineered properties for a variety of nanotechnology applications. Sugar embedding is currently considered the best method to prepare two-dimensional crystals of membrane proteins reconstituted into lipid bilayers. We found that using a loop to transfer lipid monolayer crystals to an electron microscopy grid followed by embedding in trehalose and quick-freezing in liquid ethane also yielded the highest resolution images for sbpA lipid monolayer crystals. Using images of specimens prepared in this way we could calculate a projection map of sbpA at 7A resolution, one of the highest resolution projection structures obtained with lipid monolayer crystals to date.     

Structure of light-harvesting chlorophyll a/b protein complex from plant photosynthetic membranes at 7 A resolution in projection

The structure of thin three-dimensional crystals of the light-harvesting chlorophyll a/b protein complex, an integral membrane protein from the photosynthetic membrane of chloroplasts, has been determined at 7 A (1 A = 0.1 nm) resolution in projection. The structure analysis was carried out by image processing of low-dose electron micrographs, and electron diffraction of thin three-dimensional crystals preserved in tannin. The three-dimensional crystals appeared to be stacks of two-dimensional crystals having p321 symmetry. Results of the image analysis indicated that the crystals were disordered, due to random translational displacement of stacked layers. This was established by a translation search routine that used the low-resolution projection of a single layer as a reference. The reference map was derived from the symmetrized average of two images that showed features consistent with the projected structure of negatively stained two-dimensional crystals. The phase shift resulting from the displacement of each layer was corrected. Phase shifts were then refined by minimizing the phase residual, bringing all layers to the same phase origin. Refined phases from different images were in agreement and reliable to 7 A resolution. A projection map was generated from the averaged phases and electron diffraction amplitudes. The map showed that the complex was a trimer composed of three protein monomers related by 3-fold symmetry. The projected density within the protein monomer suggested membrane-spanning alpha-helices roughly perpendicular to the crystal plane. The density in the centre and on the periphery of the trimeric complex was lower than that of the protein, indicating that this region contained low-density matter, such as lipids and antenna chlorophylls.     

7 Å projection map of the S-layer protein sbpA obtained with trehalose-embedded monolayer crystals

Two-dimensional crystallization on lipid monolayers is a versatile tool to obtain structural information of proteins by electron microscopy. An inherent problem with this approach is to prepare samples in a way that preserves the crystalline order of the protein array and produces specimens that are sufficiently flat for high-resolution data collection at high tilt angles. As a test specimen to optimize the preparation of lipid monolayer crystals for electron microscopy imaging, we used the S-layer protein sbpA, a protein with potential for designing arrays of both biological and inorganic materials with engineered properties for a variety of nanotechnology applications. Sugar embedding is currently considered the best method to prepare two-dimensional crystals of membrane proteins reconstituted into lipid bilayers. We found that using a loop to transfer lipid monolayer crystals to an electron microscopy grid followed by embedding in trehalose and quick-freezing in liquid ethane also yielded the highest resolution images for sbpA lipid monolayer crystals. Using images of specimens prepared in this way we could calculate a projection map of sbpA at 7 Å resolution, one of the highest resolution projection structures obtained with lipid monolayer crystals to date.     

Fragment-based phase extension for three-dimensional structure determination of membrane proteins by electron crystallography

In electron crystallography, membrane protein structure is determined from two-dimensional crystals where the protein is embedded in a membrane. Once large and well-ordered 2D crystals are grown, one of the bottlenecks in electron crystallography is the collection of image data to directly provide experimental phases to high resolution. Here, we describe an approach to bypass this bottleneck, eliminating the need for high-resolution imaging. We use the strengths of electron crystallography in rapidly obtaining accurate experimental phase information from low-resolution images and accurate high-resolution amplitude information from electron diffraction. The low-resolution experimental phases were used for the placement of α helix fragments and extended to high resolution using phases from the fragments. Phases were further improved by density modifications followed by fragment expansion and structure refinement against the high-resolution diffraction data. Using this approach, structures of three membrane proteins were determined rapidly and accurately to atomic resolution without high-resolution image data.     

Rational design of lipid for membrane protein crystallization

The lipidic cubic phase has been used to grow crystals of membrane proteins for high-resolution structure determination. However, the original, so-called, in meso method does not work reliably at low temperatures, where proteins are generally more stable, because the hosting lipid turns solid. The need existed therefore for a lipid that forms the cubic phase and that supports crystal growth at low temperatures. We created a database of phase diagrams and used it to design such a lipid. X-ray diffraction showed that the new lipid exhibits designed phase behavior. Further, it produces diffraction quality membrane protein crystals by the in meso method at 6 degrees C. This demonstrates that lipidic materials, like their protein counterparts are amenable to rational design. The same approach as used in this study should find application in extending the range of membrane proteins crystallizable by the in meso method and in tailoring transport of cubic phases for controlled delivery and uptake.     

The digital structural analysis of cadmium selenide crystals by a method of ion beam thinning for high resolution electron microscopy

A digital processing method using a scanning densitometer system for structural analysis of electron micrographs was successfully applied to a study of cadmium selenide crystals, which were prepared by an argon-ion beam thinning method.Based on Fourier techniques for structural analysis from a computer-generated diffractogram, it was demonstrated that when cadmium selenide crystals were sufficiently thin to display the higher order diffraction spots at a high resolution approaching the atomic level, they constitute an alternative hexagonal lattice of imperfect wurtzite phase from a superposition of individual harmonic images by the enhanced scattering amplitude and corrected phase.From the structural analysis data, a Fourier synthetic lattice image was reconstructed, representing the precise location and three-dimensional arrangement of each of the atoms in the unit cell. Extensively enhanced lattice defect images of dislocations and stacking faults were also derived and shown graphically.     

Effect of sodium polyacrylate on the hydrolysis of octacalcium phosphate

Octacalcium phosphate (OCP) hydrolysis into hydroxyapatite (HA) has been investigated in aqueous solutions at different concentrations of sodium polyacrylate (NaPA). In the absence of the polyelectrolyte, OCP undergoes a complete transformation into HA in 48 h. The hydrolysis is inhibited by the polymer, which is significantly adsorbed on the crystals, up to about 22 wt.%. A polymer concentration of 10(-2) mM is sufficient to cause a partial inhibition of OCP to HA transformation, which is completely hindered at higher concentrations. The small platelet-like crystals in the TEM images of partially converted OCP can display electron diffraction patterns characteristic either of OCP single crystals or of polycrystalline HA, whereas the much bigger plate-like crystals exhibit diffraction patterns characteristic of OCP single crystals. The polyelectrolyte adsorption on OCP crystals is accompanied by an increase of their mean length and by a significant reduction of the coherence length of the perfect crystalline domains along the c-axis direction. It is suggested that the carboxylate-rich polyelectrolyte is adsorbed on the hydrated layer of the OCP (100) face, thus inhibiting its in situ hydrolysis into HA.     

High-resolution spot-scan electron microscopy of microcrystals of an alpha-helical coiled-coil protein

We describe the electron microscopy of a crystalline assembly of an alpha-helical coiled-coil protein extracted from the ootheca of the praying mantis. Electron diffraction patterns of unstained crystals show crystal lattice sampling of the coiled-coil molecular transform to a resolution beyond 1.5 A. Using a "spot-scan" method of electron imaging, micrographs of unstained crystals have been obtained that visibly diffract laser light from crystal spacings as small as 4.3 A. A projection map was calculated to 4 A using electron diffraction amplitudes and phases from computer-processed images. The projection map clearly shows modulations in density arising from the 5.1 A alpha-helical repeat, the first time this type of modulation has been revealed by electron microscopy. The crystals have p2 plane group symmetry with a = 92.4 A, b = 150.7 A, y = 92.4 degrees. Examination of tilted specimens shows that c is approximately 18 A, indicating that the unit cell is only one molecule thick. A preliminary interpretation shows tightly packed molecules some 400 A long lying with their long axes in the plane of the projection. The molecules have a coiled-coil configuration for most of their length. The possible modes of packing of the molecules in three dimensions are discussed.     

Three-dimensional electron diffraction of plant light-harvesting complex

Electron diffraction patterns of two-dimensional crystals of light-harvesting chlorophyll a/b-protein complex (LHC-II) from photosynthetic membranes of pea chloroplasts, tilted at different angles up to 60°, were collected to 3.2 Å resolution at -125°C. The reflection intensities were merged into a three-dimensional data set. The Friedel R-factor and the merging R-factor were 21.8 and 27.6%, respectively. Specimen flatness and crystal size were critical for recording electron diffraction patterns from crystals at high tilts. The principal sources of experimental error were attributed to limitations of the number of unit cells contributing to an electron diffraction pattern, and to the critical electron dose. The distribution of strong diffraction spots indicated that the three-dimensional structure of LHC-II is less regular than that of other known membrane proteins and is not dominated by a particular feature of secondary structure.     

Temperature-controlled structure and kinetics of ripple phases in one- and two-component supported lipid bilayers

Temperature-controlled atomic force microscopy (AFM) has been used to visualize and study the structure and kinetics of ripple phases in one-component dipalmitoylphosphatidylcholine (DPPC) and two-component dimyristoylphosphatidylcholine-distearoylphosphatidylcholine (DMPC-DSPC) lipid bilayers. The lipid bilayers are mica-supported double bilayers in which ripple-phase formation occurs in the top bilayer. In one-component DPPC lipid bilayers, the stable and metastable ripple phases were observed. In addition, a third ripple structure with approximately twice the wavelength of the metastable ripples was seen. From height profiles of the AFM images, estimates of the amplitudes of the different ripple phases are reported. To elucidate the processes of ripple formation and disappearance, a ripple-phase DPPC lipid bilayer was taken through the pretransition in the cooling and the heating direction and the disappearance and formation of ripples was visualized. It was found that both the disappearance and formation of ripples take place virtually one ripple at a time, thereby demonstrating the highly anisotropic nature of the ripple phase. Furthermore, when a two-component DMPC-DSPC mixture was heated from the ripple phase and into the ripple-phase/fluid-phase coexistence temperature region, the AFM images revealed that several dynamic properties of the ripple phase are important for the melting behavior of the lipid mixture. Onset of melting is observed at grain boundaries between different ripple types and different ripple orientations, and the longer-wavelength metastable ripple phase melts before the shorter-wavelength stable ripple phase. Moreover, it was observed that the ripple phase favors domain growth along the ripple direction and is responsible for creating straight-edged domains with 60 degrees and 120 degrees angles, as reported previously.     

Low dose electron microscopy of the crotoxin complex thin crystal

The crotoxin complex from Crotalus d. terrificus rattlesnake venom was crystallized in the form of thin platelets. These crystals were prepared by the glucose embedding technique and examined by low dose electron microscopy. Electron diffraction patterns and images have been recorded to 2.2 and 4.5 A, respectively. By a combination of electron and X-ray diffraction techniques, the space group of this crystal was determined to be P4(2)22 with eight crotoxin complex molecules in one unit cell with dimensions of 38.8 A x 38.8 A x 256.8 A. The Patterson maps and the symmetry reliability factors calculated from the electron diffraction intensities clearly showed the existence of three types of electron diffraction patterns in different crystals. The phases in the computer-calculated transform of the low dose images also show the variation in symmetry among crystals. These phenomena are explained by the presence of crystals consisting of one-half, three-quarter and one unit cell in thickness. The interpretation of the computer reconstructed two-dimensional density map was limited, partly because of the similarity in density between the protein and the embedding glucose and partly because of the non-uniqueness in relating projected structure to the three-dimensional structure.     

Two-dimensional structure of plant light-harvesting complex at 3.7 A [corrected] resolution by electron crystallography

The structure of the light-harvesting chlorophyll a/b-protein complex has been determined at 3.7 A resolution in projection by electron diffraction, electron microscopy and image analysis. Diffraction patterns and high-resolution spotscan images of two-dimensional crystals stabilized with tannin were recorded at low temperature. Phases of structure factors were obtained directly by image processing, after correction of the images for lattice distortions, defocus and beam tilt. Amplitudes were measured by electron diffraction. The projection map shows the detailed structure of the trimeric complex, suggesting the positions of two domains of potential structural and functional homology, of one membrane-spanning alpha-helix approximately perpendicular to the membrane plane and of several tightly bound lipid molecules.     

A comparison of negatively stained electron micrographs and projections obtained from single crystal X-ray studies: The lipovitellin complex from lamprey

The analysis of single crystals of the lipovitellin complex from lamprey has made it possible to compare electron density projections derived from X-ray diffraction and density modification methods with previously published electron micrographs. The close correlation between the images obtained by the two methods demonstrates that the fidelity of images obtained by electron microscopy is excellent despite the loss of specimen order. These correlations also attest to the ability of density modification methods, currently used in macromolecular crystallography, to estimate the phases of small angle X-ray reflections.     

The molecular and crystal structure of dextrans: a combined electron and X-ray diffraction study. II. A low temperature, hydrated polymorph

The crystal and molecular structure of a dextran hydrate has been determined through combined electron and X-ray diffraction analysis, aided by stereochemical model refinement. A total of 65 hk0 electron diffraction intensities were measured on frozen single crystals held at the temperature of liquid nitrogen, to a resolution limit of 1.6 A. The X-ray intensities were measured from powder patterns recorded from collections of the single crystals. The structure crystallizes in a monoclinic unit cell with parameters a = 25.71 A, b = 10.21 A, c (chain axis) = 7.76 A and beta = 91.3 degrees. The space group is P2(1) with b axis unique. The unit cell contains six chains and eight water molecules, with three chains of the same polarity and four water molecules constituting the asymmetric unit. Along the chain direction the asymmetric unit is a dimer residue; however, the individual glucopyranose residues are very nearly related by a molecular 2-fold screw axis. The conformation of the chain is very similar to that in the anhydrous structure, but the chain packing differs in the two structures in that the rotational positions of the chains about the helix axes (the chain setting angles) are considerably different. The chains still pack in the form of sheets that are separated by water molecules. The difference in the chain setting angles between the anhydrous and hydrate structures corresponds to the angle between like unit cell axes observed in the diffraction diagrams recorded from hybrid crystals containing both polymorphs. Despite some beam damage effects, the structure was determined to a satisfactory degree of agreement, with the residuals R'(electron diffraction) = 0.258 and R(X-ray) = 0.127.     

A maximum likelihood approach to two-dimensional crystals

Maximum likelihood (ML) processing of transmission electron microscopy images of protein particles can produce reconstructions of superior resolution due to a reduced reference bias. We have investigated a ML processing approach to images centered on the unit cells of two-dimensional (2D) crystal images. The implemented software makes use of the predictive lattice node tracking in the MRC software, which is used to window particle stacks. These are then noise-whitened and subjected to ML processing. Resulting ML maps are translated into amplitudes and phases for further processing within the 2dx software package. Compared with ML processing for randomly oriented single particles, the required computational costs are greatly reduced as the 2D crystals restrict the parameter search space. The software was applied to images of negatively stained or frozen hydrated 2D crystals of different crystal order. We find that the ML algorithm is not free from reference bias, even though its sensitivity to noise correlation is lower than for pure cross-correlation alignment. Compared with crystallographic processing, the newly developed software yields better resolution for 2D crystal images of lower crystal quality, and it performs equally well for well-ordered crystal images.     

Ab initio phasing of high-symmetry macromolecular complexes: successful phasing of authentic poliovirus data to 3.0 A resolution

A genetic algorithm-based computational method for the ab initio phasing of diffraction data from crystals of symmetric macromolecular structures, such as icosahedral viruses, has been implemented and applied to authentic data from the P1/Mahoney strain of poliovirus. Using only single-wavelength native diffraction data, the method is shown to be able to generate correct phases, and thus electron density, to 3.0 A resolution. Beginning with no advance knowledge of the shape of the virus and only approximate knowledge of its size, the method uses a genetic algorithm to determine coarse, low-resolution (here, 20.5 A) models of the virus that obey the known non-crystallographic symmetry (NCS) constraints. The best scoring of these models are subjected to refinement and NCS-averaging, with subsequent phase extension to high resolution (3.0 A). Initial difficulties in phase extension were overcome by measuring and including all low-resolution terms in the transform. With the low-resolution data included, the method was successful in generating essentially correct phases and electron density to 6.0 A in every one of ten trials from different models identified by the genetic algorithm. Retrospective analysis revealed that these correct high-resolution solutions converged from a range of significantly different low-resolution phase sets (average differences of 59.7 degrees below 24 A). This method represents an efficient way to determine phases for icosahedral viruses, and has the advantage of producing phases free from model bias. It is expected that the method can be extended to other protein systems with high NCS.     

A Strategy for Electron Tomographic Data Collection and Crystallographic Reconstruction of Biological Bundles

Structures of highly ordered biological bundles have unique features which call for special experimental and computational methods in electron cryomicroscopy. They can be considered as three-dimensional quasi-crystals and reconstructed using a crystallographic approach. However, they are neither “infinitely” large with respect to the borders of the bundle, nor are they a single unit cell in thickness along the viewing direction. Also, because of their shape, bundles do not generally have a preferred azimuthal orientation, which poses challenges for orientation estimation and refinement. We developed a strategy for recording and processing electron cryomicroscopic images that differs from classical two-dimensional crystalline reconstruction techniques. These developments allowed us to merge data from tomographic tilt series of ice-embedded acrosomal bundles. The goal is to determine accurately amplitudes and phases at the diffraction maxima in terms ofhklindices, and compute a three-dimensional map from the diffraction data.