Current outcomes when optimizing 'standard' sample preparation for single-particle cryo-EM.

Although high-resolution single-particle cryo-electron microscopy (cryo-EM) is now producing a rapid stream of breakthroughs in structural biology, it nevertheless remains the case that the preparation of suitable frozen-hydrated samples on electron microscopy grids is often quite challenging. Purified samples that are intact and structurally homogeneous - while still in the test tube - may not necessarily survive the standard methods of making extremely thin, aqueous films on grids. As a result, it is often necessary to try a variety of experimental conditions before finally finding an approach that is optimal for the specimen at hand. Here, we summarize some of our collective experiences to date in optimizing sample preparation, in the hope that doing so will be useful to others, especially those new to the field. We also hope that an open discussion of these common challenges will encourage the development of more generally applicable methodology. Our collective experiences span a diverse range of biochemical samples and most of the commonly used variations in how grids are currently prepared. Unfortunately, none of the currently used optimization methods can be said, in advance, to be the one that ultimately will work when a project first begins. Nevertheless, there are some preferred first steps to explore when facing specific problems that can be more generally recommended, based on our experience and that of many others in the cryo-EM field.

Near-concentric Fabry-Pérot cavity for continuous-wave laser control of electron waves.

Manipulating free-space electron wave functions with laser fields can bring about new electron-optical elements for transmission electron microscopy (TEM). In particular, a Zernike phase plate would enable high-contrast TEM imaging of soft matter, leading to new opportunities in structural biology and materials science. A Zernike phase plate can be implemented using a tight, intense continuous laser focus that shifts the phase of the electron wave by the ponderomotive potential. Here, we use a near-concentric cavity to focus 7.5 kW of continuous-wave circulating laser power at 1064 nm into a 7 µm mode waist, achieving a record continuous laser intensity of 40 GW/cm2. Such parameters are sufficient to impart a phase shift of 1 rad to a 10 keV electron beam, or 0.16 rad to a 300 keV beam. Our numerical simulations confirm that the standing-wave phase shift profile imprinted on the electron wave by the intra-cavity field can serve as a nearly ideal Zernike phase plate.

Specimen Behavior in the Electron Beam.

It has long been known that cryo-EM specimens are severely damaged by a level of electron exposure that is much lower than what is needed to obtain high-resolution images from single macromolecules. Perhaps less well appreciated in the cryo-EM literature, the vitreous ice in which samples are suspended is equally sensitivity to radiation damage. This chapter provides a review of several fundamental topics such as inelastic scattering of electrons, radiation chemistry, and radiation biology, which-together-can help one to understand why radiation damage occurs so "easily." This chapter also addresses the issue of beam-induced motion that occurs at even lower levels of electron exposure. While specimen charging may be a contributor to this motion, it is argued that both radiation-induced relief of preexisting stress and damage-induced generation of additional stress may be the dominant causes of radiation-induced movement.

Accurate modeling of single-particle cryo-EM images quantitates the benefits expected from using Zernike phase contrast.

The use of a Zernike-type phase plate in biologic cryo-electron microscopy allows the imaging, without using defocus, of what are predominantly phase objects. It is thought that such phase-plate implementations might result in higher quality images, free from the problems of CTF correction that occur when images must be recorded at extremely high values of defocus. In single-particle cryo-electron microscopy it is hoped that these improvements in image quality will facilitate work on structures that have proved difficult to study, either because of their relatively small size or because the structures are not completely homogeneous. There is still a need, however, to quantitate how much improvement can be gained by using a phase plate for single-particle cryo-electron microscopy. We present a method for quantitatively modeling the images recorded with 200keV electrons, for single particles embedded in vitreous ice. We then investigate what difference the use of a phase-plate device could have on the processing of single-particle data. We confirm that using a phase plate results in single-particle datasets in which smaller molecules can be detected, particles can be more accurately aligned and problems of heterogeneity can be more easily addressed.

Images of paraffin monolayer crystals with perfect contrast: minimization of beam-induced specimen motion.

Quantitative analysis of electron microscope images of organic and biological two-dimensional crystals has previously shown that the absolute contrast reached only a fraction of that expected theoretically from the electron diffraction amplitudes. The accepted explanation for this is that irradiation of the specimen causes beam-induced charging or movement, which in turn causes blurring of the image due to image or specimen movement. In this paper, we used three different approaches to try to overcome this image-blurring problem in monolayer crystals of paraffin. Our first approach was to use an extreme form of spotscan imaging, in which a single image was assembled on film by the successive illumination of up to 50,000 spots, each of a diameter of around 7 nm. The second approach was to use the Medipix II detector with its zero-noise readout to assemble a time-sliced series of images of the same area in which each frame from a movie with up to 400 frames had an exposure of only 500 electrons. In the third approach, we simply used a much thicker carbon support film to increase the physical strength and conductivity of the support. Surprisingly, the first two methods involving dose fractionation in space or time produced only partial improvements in contrast whereas the third approach produced many virtually perfect images, where the absolute contrast predicted from the electron diffraction amplitudes was observed in the images. We conclude that it is possible to obtain consistently almost perfect images of beam-sensitive specimens if they are attached to an appropriately strong and conductive support; however great care is needed in practice and the problem remains of how to best image ice-embedded biological structures in the absence of a strong, conductive support film.

Design of an electron microscope phase plate using a focused continuous-wave laser.

We propose a Zernike phase contrast electron microscope that uses an intense laser focus to convert a phase image into a visible image. We present the relativistic quantum theory of the phase shift caused by the laser-electron interaction, study resonant cavities for enhancing the laser intensity and discuss applications in biology, soft-materials science and atomic and molecular physics.

Three-dimensional diffractive imaging for crystalline monolayers with one-dimensional compact support.

The use of a compact support constraint along the beam direction is considered as a solution to the phase problem for diffraction by two-dimensional protein crystals. Specifically we apply the iterative Gerchberg-Saxton-Fienup algorithm to simulated three-dimensional transmission electron diffraction data from monolayer organic crystals. We find that oversampling along the reciprocal-lattice rods (relrods) normal to the monolayer alone does not solve the phase problem in this geometry in general. However, based on simulations for a crystalline protein monolayer (lysozyme), we find that convergence is obtained in three dimensions if phases are supplied from a few high resolution electron microscope images recorded at small tilts to the beam direction. In the absence of noise, amplitude-weighted phase residuals of around 5 degrees, and a cross-correlation coefficient of 0.96 between the true and estimated potential are obtained if phases are included from images at tilts of up to 15 degrees. The performance is almost as good in the presence of noise at a level that is comparable to that commonly observed in electron crystallography of proteins. The method should greatly reduce the time and labor needed for data acquisition and analysis in cryo-electron microscopy of organic thin crystals by avoiding the need to record images at high tilt angles.

Structure of an early intermediate in the M-state phase of the bacteriorhodopsin photocycle.

The structure of an early M-intermediate of the wild-type bacteriorhodopsin photocycle formed by actinic illumination at 230 K has been determined by x-ray crystallography to a resolution of 2.0 A. Three-dimensional crystals were trapped by illuminating with actinic light at 230 K, followed by quenching in liquid nitrogen. Amide I, amide II, and other infrared absorption bands, recorded from single bacteriorhodopsin crystals, confirm that the M-substate formed represents a structure that occurs early after deprotonation of the Schiff base. Rotation about the retinal C13-C14 double bond appears to be complete, but a relatively large torsion angle of 26 degrees is still seen for the C14-C15 bond. The intramolecular stress associated with the isomerization of retinal and the subsequent deprotonation of the Schiff base generates numerous small but experimentally measurable structural changes within the protein. Many of the residues that are displaced during the formation of the late M (M(N)) substate formed by three-dimensional crystals of the D96N mutant (Luecke et al., 1999b) are positioned, in early M, between their resting-state locations and the ones which they will adopt at the end of the M phase. The relatively small magnitude of atomic displacements observed in this intermediate, and the well-defined positions adopted by nearly all of the atoms in the structure, may make the formation of this structure favorable to model (simulate) by molecular dynamics.

Crystal structure of the D85S mutant of bacteriorhodopsin: model of an O-like photocycle intermediate.

Crystal structures are reported for the D85S and D85S/F219L mutants of the light-driven proton/hydroxyl-pump bacteriorhodopsin. These mutants crystallize in the orthorhombic C222(1) spacegroup, and provide the first demonstration that monoolein-based cubic lipid phase crystallization can support the growth of well-diffracting crystals in non-hexagonal spacegroups. Both structures exhibit similar and substantial differences relative to wild-type bacteriorhodopsin, suggesting that they represent inherent features resulting from neutralization of the Schiff base counterion Asp85. We argue that these structures provide a model for the last photocycle intermediate (O) of bacteriorhodopsin, in which Asp85 is protonated, the proton release group is deprotonated, and the retinal has reisomerized to all-trans. Unlike for the M and N photointermediates, where structural changes occur mainly on the cytoplasmic side, here the large-scale changes are confined to the extracellular side. As in the M intermediate, the side-chain of Arg82 is in a downward configuration, and in addition, a pi-cloud hydrogen bond forms between Trp189 NE1 and Trp138. On the cytoplasmic side, there is increased hydration near the surface, suggesting how Asp96 might communicate with the bulk during the rise of the O intermediate.

Review: automatic particle detection in electron microscopy.

Advances in cryoEM and single-particle reconstruction have led to results at increasingly high resolutions. However, to sustain continuing improvements in resolution it will be necessary to increase the number of particles included in performing the reconstructions. Manual selection of particles, even when assisted by computer preselection, is a bottleneck that will become significant as single-particle reconstructions are scaled up to achieve near-atomic resolutions. This review describes various approaches that have been developed to address the problem of automatic particle selection. The principal conclusions that have been drawn from the results so far are: (1) cross-correlation with a reference image ("matched filtering") is an effective way to identify candidate particles, but it is inherently unable to avoid also selecting false particles; (2) false positives can be eliminated efficiently on the basis of estimates of particle size, density, and texture; (3) successful application of edge detection (or contouring) to particle identification may require improvements over currently available methods; and (4) neural network techniques, while computationally expensive, must also be investigated as a technology for eliminating false particles.

Chemical and physical evidence for multiple functional steps comprising the M state of the bacteriorhodopsin photocycle.

In the photocycle of bacteriorhodopsin (bR), light-induced transfer of a proton from the Schiff base to an acceptor group located in the extracellular half of the protein, followed by reprotonation from the cytoplasmic side, are key steps in vectorial proton pumping. Between the deprotonation and reprotonation events, bR is in the M state. Diverse experiments undertaken to characterize the M state support a model in which the M state is not a static entity, but rather a progression of two or more functional substates. Structural changes occurring in the M state and in the entire photocycle of wild-type bR can be understood in the context of a model which reconciles the chloride ion-pumping phenotype of mutants D85S and D85T with the fact that bR creates a transmembrane proton-motive force.

Review: electron crystallography: present excitement, a nod to the past, anticipating the future.

From a modest beginning with negatively stained samples of the helical T4 bacteriophage tail, electron crystallography has emerged as a powerful tool in structural biology. High-resolution density maps, interpretable in terms of an atomic structure, can be obtained from specimens prepared as well-ordered, two-dimensional crystals, and the resolution achieved with helical specimens and icosahedral viruses is approaching the same goal. A hybrid approach to determining the molecular structure of complex biological assemblies is generating great interest, in which high-resolution structures that have been determined for individual protein components are fitted into a lower resolution envelope of the large complex. With this as background, how much more can be anticipated for the future? Considerable scope still remains to improve the quality of electron microscope images. Automation of data acquisition and data processing, together with the emergence of computational speeds of 10(12) floating point operations per second or higher, will make it possible to extend high-resolution structure determination into the realm of single-particle microscopy. As a result, computational alignment of single particles, i.e., the formation of "virtual crystals," can begin to replace biochemical crystallization. Since single-particle microscopy may remain limited to "large" structures of 200 to 300 kDa or more, however, smaller proteins will continue to be studied as helical assemblies or as two-dimensional crystals. The further development of electron crystallography is thus likely to turn increasingly to the use of single particles and small regions of ordered assemblies, emphasizing more and more the potential for faster, higher throughput.

Chemical bonding effects in the determination of protein structures by electron crystallography.

Scattering of electrons is affected by the distribution of valence electrons that participate in chemical bonding and thus change the electrostatic shielding of the nucleus. This effect is particularly significant for low-angle scattering. Thus, while chemical bonding effects are difficult to measure with small-unit cell materials, they can be substantial in the study of proteins by electron crystallography. This work investigates the magnitude of chemical bonding effects for a representative collection of protein fragments and a model ligand for nucleotide-binding proteins within the resolution range generally used in determining protein structures by electron crystallography. Electrostatic potentials were calculated by ab initio methods for both the test molecules and for superpositions of their free atoms. Differences in scattering amplitudes can be well over 10% in the resolution range below 5 A and are especially large in the case of ionized side chains and ligands. We conclude that the use of molecule-based scattering factors can provide a much more accurate representation of the low-resolution data obtained in electron crystallographic studies. The comparison of neutral and ionic structure factors at resolutions below 5 A can also provide a sensitive determination of charge states, important for biological function, that is not accessible from X-ray crystallographic measurements.

Structural characterization of the L-to-M transition of the bacteriorhodopsin photocycle.

Structural intermediates occurring in the photocycle of wild-type bacteriorhodopsin are trapped by illuminating hydrated, glucose-embedded purple membrane at 170 K, 220 K, 230 K, and 240 K. We characterize light-induced changes in protein conformation by electron diffraction difference Fourier maps, and relate these to previous work on photocycle intermediates by infrared (FTIR) spectroscopy. Samples illuminated at 170 K are confirmed by FTIR spectroscopy to be in the L state; a difference Fourier projection map shows no structural change within the 0.35-nm resolution limit of our data. Difference maps obtained with samples illuminated at 220 K, 230 K, and 240 K, respectively, reveal a progressively larger structural response in helix F when the protein is still in the M state, as judged by the FTIR spectra. Consistent with previous structural studies, an adjustment in the position or in the degree of ordering of helix G accompanies this motion. The model of the photocycle emerging from this and previous studies is that bacteriorhodopsin experiences minimal change in protein structure until a proton is transferred from the Schiff base to Asp85. The M intermediate then undergoes a conformational evolution that opens a hydrated "half-channel," allowing the subsequent reprotonation of the Schiff base by Asp96.

Differences in hydration structure near hydrophobic and hydrophilic amino acids.

We use molecular dynamics to simulate recent neutron scattering experiments on aqueous solutions of N-acetyl-leucine-amide and N-acetyl-glutamine-amide, and break down the total scattering function into contributions from solute-solute, solute-water, water-water, and intramolecular correlations. We show that the shift of the main diffraction peak to smaller angle that is observed for leucine, but not for glutamine, is attributable primarily to alterations in water-water correlations relative to bulk. The perturbation of the water hydrogen-bonded network extends roughly two solvation layers from the hydrophobic side chain surface, and is characterized by a distribution of hydrogen bonded ring sizes that are more planar and are dominated by pentagons in particular than those near the hydrophilic side chain. The different structural organization of water near the hydrophobic solute that gives rise to the inward shift in the main neutron diffraction peak under ambient conditions may also provide insight into the same directional shift for pure liquid water as it is cooled and supercooled.

Direct evidence for modified solvent structure within the hydration shell of a hydrophobic amino acid.

Neutron scattering experiments are used to determine scattering profiles for aqueous solutions of hydrophobic and hydrophilic amino acid analogs. Solutions of hydrophobic solutes show a shift in the main diffraction peak to smaller angle as compared with pure water, whereas solutions of hydrophilic solutes do not. The same difference for solutions of hydrophobic and hydrophilic side chains is also predicted by molecular dynamics simulations. The neutron scattering curves of aqueous solutions of hydrophobic amino acids at room temperature are qualitatively similar to differences between the liquid molecular structure functions measured for ambient and supercooled water. The nonpolar solute-induced expansion of water structure reported here is also complementary to recent neutron experiments where compression of aqueous solvent structure has been observed at high salt concentration.

The relevance of dose-fractionation in tomography of radiation-sensitive specimens.

It is commonly assumed that the number of projections required for single-axis tomography precludes its application to most beam-labile specimens. However, Hegerl and Hoppe have pointed out that the total dose required to achieve statistical significance for each voxel of a computed 3D reconstruction is the same as that required to obtain a single 2D image of that isolated voxel, at the same level of statistical significance. Thus a statistically significant 3D image can be computed from statistically insignificant projections, as long as the total dose that is distributed among these projections is high enough that it would have resulted in a statistically significant projection, if applied to only one image. We have tested this critical theorem by simulating the tomographic reconstruction of a realistic 3D model created from an electron micrograph. The simulations verify the basic conclusions of the theorem and extend its validity to the experimentally more realistic conditions of high absorption, signal-dependent noise, varying specimen contrast and missing angular range. Individual projections in the series of fractionated-dose images could be aligned by cross-correlation because they contained significant information derived from the summation of features from different depths in the structure. This latter information is generally not useful for structural interpretation prior to 3D reconstruction, owing to the complexity of most specimens investigated by single-axis tomography. These results demonstrate that it is feasible to use single-axis tomography with soft X-ray and electron microscopy of frozen-hydrated specimens.

Crystallographic extraction and averaging of data from small image areas.

The accuracy of structure factor phases determined from electron microscope images is determined mainly by the level of statistical significance, which is limited by the low level of allowed electron exposure and by the number of identical unit cells that can be averaged. It is shown here that Fourier transforms of small image fields of purple membrane (a two-dimensional crystal consisting of bacteriorhodopsin and endogenous lipids) can be combined to provide the same quality of phases as are obtained from Fourier transforms of large image fields of the same total area. Although Fourier transforms of such small image fields are statistically significant only at lower resolution, the data from many such image fields can be averaged at the calculated positions of high-resolution reciprocal lattice points to give accurate phases. More specifically, when images of a size that can be recorded with CCD cameras are processed individually, key parameters including lattice vectors, defocus, crystal and beam tilts, and common phase origin can be accurately determined.

Two progressive substrates of the M-intermediate can be identified in glucose-embedded, wild-type bacteriorhodopsin.

Glucose-embedded bacteriorhodopsin shows M-intermediates with different Amide I infrared bands when samples are illuminated at 240 or 260 K, in contrast with fully hydrated samples where a single M-intermediate is formed at all temperatures. In hydrated, but not in glucose-embedded specimens, the N intermediate is formed together with M at 260 K. Both Fourier transform infrared and electron diffraction data from glucose-embedded bacteriorhodopsin suggest that at 260 K a mixture is formed of the M-state that is trapped at 240 K, and a different M-intermediate (MN) that is also formed by mutant forms of bacteriorhodopsin that lack a carboxyl group at the 96 position, necessary for the M to N transition. The fact that an MN species is trapped in glucose-embedded, wild-type bacteriorhodopsin suggests that the glucose samples lack functionally important water molecules that are needed for the proton transfer aspartate 96 to the Schiff base (and, thus, to form the N-intermediate); thus, aspartate 96 is rendered ineffective as a proton donor.

The bacteriorhodopsin photocycle: direct structural study of two substrates of the M-intermediate.

Changes in protein structure that occur during the formation of the M photointermediate of bacteriorhodopsin can be directly visualized by electron diffraction techniques. A modified preparation technique for glucose-embedded crystals was employed to ensure sufficient hydration of the crystals, which was needed for the formation of the M intermediate at low temperature. Samples containing a high percentage of the M intermediate were trapped by rapidly cooling the crystals with liquid nitrogen after illumination with filtered green light at 240 and 260 K, respectively. Difference Fourier projection maps are presented for the M intermediates formed at these two temperatures. The diffraction data clearly show that statistically significant structural changes occur upon formation of the M intermediate at 240 K and then further upon formation of the second specimen that is produced at 260 K.