Electron microscopy of frozen biological objects: a study using cryosectioning and cryosubstitution

Freezing of bulk biological objects was investigated by X-ray cryodiffraction. Freezing at atmospheric pressure of most microscopic biological samples gives rise to large hexagonal crystals and leads to poor structural preservation of these specimens. High-pressure freezing induces the formation of different ices (hexagonal, cubic and a high-pressure form) consisting of crystals having sizes smaller than those formed at atmospheric pressure. With both freezing methods, a cryoprotectant has to be added to the biological object to avoid the formation of ice crystals. However, special cases can be encountered: some biological objects contain large amounts of natural cryoprotectant or have a low water content. In these cases, vitrification can be achieved, especially using high-pressure freezing. Cryo-sectioning can be performed on vitrified samples, and the sections studied by electron cryomicroscopy. Images and electron diffraction patterns having a resolution better than 2 and 0.2 nm, respectively, can be obtained with such sections. Because samples containing crystalline ices cannot be cryosectioned, their structure has to be studied using cryosubstitution and resin embedding. We show that bacteria, yeast, and ciliate and marine worm elytrum have cellular compartments with an organization that has not been described by classical techniques relying on chemical fixation of the tissues. A high-pressure artefact affecting the Paramecium trichocysts is described. Such artefacts are not general; for example, we show that 70% of high-pressure frozen yeast cells survive successive high-pressure freezing and thawing steps.     

Vitrification depth can be increased more than 10-fold by high-pressure freezing

Biological specimens prepared for cryoelectron microscopy seem to suffer less damage when they are frozen under 2 kbar pressure rather than under normal conditions. The volume that can be well preserved is larger. This fact has been illustrated in a number of publications on a number of different samples. However, there is a lack of quantitative data concerning the depth of this good specimen preservation. Catalase crystals in various sugar solutions have been used as test objects and vitrification, as determined by electron diffraction, has been used as the criterion for good freezing. Keeping all other conditions equal, the depth of vitrification is approximately 10 times larger with freezing at high, rather than normal, pressure. The high-pressure vitrification depth in a 15–20% sugar solution averages 200 μm. Fully vitrified specimens up to 700 μm in thickness are obtained. When crystalline water is observed it is frequently in the form of high-density ice II, III or IX. These results are probably also relevant for typical biological specimens. The advantage of high-pressure freezing must be balanced by the possible consequences of a considerably increased cooling time and by the damage that may be induced by the pressure.     

Structure and function of frozen cells: freezing patterns and post-thaw survival.

Freezing patterns and post-thaw survival of cells varies with different cooling rates. The optimal cooling rates, indicating the highest percentage survival, were different in yeast and red blood cells. A difference of freezing patterns was also noticed in preparations frozen above and below the optimal cooling rate for each cell, namely, cell shrinkage at lower rates and intracellular ice formation at higher rates which showed similar trends in both the cells, even though there was some shifting of the optimum. Ultra-rapid freezing and addition of cryoprotectants are useful ways to minimize ice crystal formation and to cause such ice formations to approach the vitreous state. Ice crystals are hardly detectable in yeast cells as well as in erythrocytes, when these cells are frozen ultra-rapidly in the presence of cryoprotective agents in moderate concentration.     

The relative efficiency of cryogens used for plunge-cooling biological specimens

Coolants used for freezing biological specimens were tested for cooling performance in the continuous plunge mode. Results from bare thermocouples showed that ethane cooled faster than propane or a propane: pentane mixture, even when warmed to 25 K above its freezing point. Propane coolants were more efficient than Freon 22 and the slowest cooling occurred in boiling liquid nitrogen. Hydrated gelatin specimens showed similar results with ethane cooling about 33% faster than propane. Epoxy resin specimens cooled faster than hydrated gelatin specimens of similar size. Hydrated and resin specimens cooled over increasing distances as plunge velocity increased. A bare thermocouple, however, cooled over a constant distance when plunged above a critical velocity. This phenomenon may reflect vapour formation and its suppression at high plunge velocities. The rate of cooling in hydrated specimens is shown to have an absolute limit and cannot be modelled by bare thermocouples or resin specimens.     

Freezing of aqueous specimens: an X-ray diffraction study

The effects on water of two cooling methods, immersion in a liquid cryogen and high-pressure freezing, were studied by X-ray cryodiffraction on different sucrose solutions. The nature of the ice formed by each method depends on both the sucrose concentration and the specimen thickness. In order to compare the two methods, we mainly studied specimens having a thickness of 0.2 mm. Under these conditions, freezing by immersion gives rise to hexagonal (IH), cubic (IC) and amorphous (IV) ices when the sucrose concentration (weight/weight) has a value within the range 0–30%, 30–60%, 60% and higher, respectively. The temperature of the phase transitions IV–IC, IC–IH depends on the sucrose concentration. High-pressure freezing gives rise to two specific forms of ice: an amorphous and a crystalline ice (ice III). Ice III is observed when pure water samples are high-pressure frozen provided that the sample temperature does not rise above −150 °C. Above this temperature, ice III transforms into hexagonal ice. Amorphous ice is formed when the sucrose concentration is higher than 20%. The amorphous ice formed under high pressure has a similar, but not identical, X-ray diffraction pattern to that of amorphous ice formed at atmospheric pressure. While the X-ray diffraction pattern of amorphous ice formed at atmospheric pressure (IV) shows a broad ring at a position corresponding to 0.37 nm, that of high-pressure amorphous ice (IVHP) shows a broader ring, located at 0.35 nm. IVHP presents a phase transition (IVHP–IV) at temperatures that depend on the sucrose concentration. We also observed that some precautions have to be taken in order to minimize the alcohol contamination of high-pressure frozen samples. The ice-phase diagram presented in this paper should be taken into account in all methods dedicated to the structural study of frozen biological specimens.     

Chemical cryoprotection for structural studies

With the ascendency of techniques for ultrarapid cooling and the successful control of ice crystal damage by purely physical means, it has become necessary to make a case for the continued use of chemical antifreeze agents in any circumstances. These circumstances include the need to explore tissues deeper than those superficial layers, whose morphology can be preserved by ultrarapid cooling, the avoidance of superficial areas of damage inflicted by dissection and tissue slicing, and situations where the growth of ice crystals must be controlled throughout the specimen, as for example for the cutting of frozen sections. The control of ice crystal damage would ideally be by the avoidance of any freezing at all but in practice can be seen in terms of the control of ice crystal size, which in turn depends on the density of nuclei that develop in the specimen and the rate of accretion of water molecules to the crystallites that form on these nuclei. Chemical antifreeze agents act in a variety of ways to increase the density of nuclei and/or to reduce the rate of growth of ice crystals: by promoting nucleation, by enhancing subcooling and by increasing the viscosity of the extracellular medium. In this way, large numbers of small ice crystals are produced. The different types of cryofixative agent each have their own advocates, advantages, special applications and drawbacks and these must be taken into account when considering the range of methods available for the analysis of cryofixed material.     

Advances in ultrarapid freezing for the preservation of cellular ultrastructure

Most of our current knowledge of cellular ultrastructure is derived from studies of chemically fixed and chemically cryoprotected preparations. In the first part of this review, we document the many artifacts associated with chemical techniques that render them unsuitable for further refinement of our understanding of cellular ultrastructure. The best method currently available for the preservation of cellular ultrastructure is ultrarapid freezing. The second part of this review is a consideration of the physics of ice crystal formation in biological systems, which suggests that ice crystals will be present in any frozen, uncryoprotected specimen. We define an ultrarapidly frozen preparation as one in which the ice crystals are so small as to be invisible at the electron microscopic level. Improvements in the ease of application and reliability of ultrarapid freezing techniques have reached the point that these techniques can be used by anyone requiring the best achievable preservation of cellular ultrastructure. In the third part of this review, we describe and critique the five methods of ultrarapid freezing in current use.     

Cooling rate and ice-crystal measurement in biological specimens plunged into liquid ethane,propane, and Freon 22

Specimens sandwiched between copper planchettes were plunged up to a depth of 430 mm into coolants used for cryofixation. Hydrated gelatin containing a miniature thermocouple was used to mimic the behaviour of tissue during freezing. Gelatin and red blood cells were used for ice-crystal analysis. Ethane produced the fastest cooling rates and the smallest ice-crystal profiles, and Freon 22 produced the slowest cooling rates and the largest crystal profiles. Smaller crystal profiles were often seen in the centre of the specimens than in subsurface zones. The results show that ethane, rather than propane, should be used for freezing metal-sandwiched freeze-fracture specimens by the plunging method, and probably also in the jet-cooling method. They further suggest that good cryofixation could occur at the centre of thin specimens rather than only at their surfaces. Comparison between theoretical and experimental ice-crystal sizes was satisfactory, indicating that where the experimental parameters can be defined then realistic predictions can be made regarding cryofixation results.     

Conducting,semiconducting and insulating objects observed by low-energy electron holography

To demonstrate the potential of low energy in line projection holography, we study the reconstruction of the experimental holograms of three electrically different objects: a conducting, a semiconducting and an insulating object. The reconstructions of these holograms provide meaningful results for a large range of magnification of the object. The comparison between the reconstructed images and the scanning electron microscopy (SEM) micrographs of the same objects shows that the shapes and the dimensions of the reconstructed objects are identical to those obtained by conventional SEM. So, the simple assumptions needed to the reconstruction are justified. The reconstructions show a 2 nm resolution and appear superior than the best obtained SEM micrographs, when available. We also show some limitations of the reconstruction process. We point out that both numerical artifacts and experimental conditions are responsible for these limitations.     

Self-pressurized rapid freezing (SPRF): a novel cryofixation method for specimen preparation in electron microscopy

A method is described for the cryofixation of biological specimens for ultrastructural analysis and immunocytochemical detection studies. The method employs plunge freezing of specimens in a sealed capillary tube into a cryogen such as liquid propane or liquid nitrogen. Using this method a number of single-cell test specimens were well preserved. Also multicellular organisms, such as Caenorhabditis elegans , could be frozen adequately in low ionic strength media or even in water. The preservation of these unprotected specimens is comparable to that achieved with high-pressure freezing in the presence of cryoprotectant. The results are explained by the fact that cooling of water in a confined space below the melting point gives rise to pressure build-up, which may originate from the conversion of a fraction of the water content into low-density hexagonal ice and/or expansion of water during supercooling. Calculations indicate the pressure may be similar in magnitude to that applied in high-pressure freezing. Because the specimens are plunge cooled, suitable cryogens are not limited to liquid nitrogen. It is shown that a range of cryogens and cryogen temperatures can be used successfully. Because the pressure is generated inside the specimen holders as a result of the cooling rather than applied from an external source as in high-pressure freezing, the technique has been referred to as self-pressurized rapid freezing.     

Time-resolved cryo-electron microscopy of vitrified muscular components

Biological objects may be arrested in defined stages of their activity by fast freezing and may then be structurally examined. If the time between the start of activity and freezing is controlled, structural rearrangements due to biological function can be determined. Cryo-electron microscopy shows great potential for the study of such time-dependent phenomena. This study examines the actin polymerization process using cryo-electron microscopy of vitrified specimens. Actin filaments are shown to undergo a structural change during polymerization. In the early stages of the polymerization process (t < 2 min), filaments exhibit a pronounced structural variation and frequently show a central low-density area. In the later stages of the polymerization, F-actin-ADP filaments have a more uniform appearance and rarely display a central low-density area. These findings, analysed on the basis of a previously proposed polymerization model, suggest that polymerization intermediates (F-actin-ATP and more probably F-actin-ADP-P_i) and filaments at steady state (F-actin-ADP) have different structures. To investigate the physiological relevance of these results at the cellular level, the potential of cryo-substitution in preserving the structure of muscular fibre was assessed. Optical diffraction patterns of relaxed and contracted frog cutaneous muscle are similar to the corresponding X-ray diffraction patterns. The resolution of the images extends to about 7 nm. These results show that dynamic study of muscle contraction is possible using cryo-substitution.     

Structure determination of tubular crystals of membrane proteins. III. Solvent flattening

Solvent flattening is considered to be a principal means for improving the data quality in X-ray crystallography. It could be equally effective for tubular crystals of membrane proteins imaged by electron microscopy because of the large empty space inside the tubes. However, tubular crystals are difficult objects for solvent flattening due to lack of electron diffraction amplitudes. Therefore, solvent flattening was used to align images more accurately and to improve the completeness of the data by reducing contributions of noise in the solvent (+ lipid) region. The methods developed were tested with the tubular crystals of Ca2+-ATPase embedded in amorphous ice. The improvement of the data quality was remarkable when solvent flattening was applied to many individual images before averaging. In this way, noises contaminated in the protein region by contrast transfer function were removed effectively. Solvent flattening was far more powerful than simple averaging described in Part II of this series (K. Yonekura, C. Toyoshima, Ultramicroscopy 84 (2000) 15).     

Diffraction effect of x-rays on the accuracy of quantitative analysis of crystals

In order to understand an apparent discrepancy of concentration of constituent elements in the quantitative analysis of thin KCl crystals with an energy dispersive X-ray analyzer, which has been partly explained by fluorescent X-ray excitation, more detailed experiments were carried out on binary alkali halides and tricomponent amorphous material by changing the azimuthal angle. Intensity ratios of X-rays from the constituent elements in KCl and KBr single crystals varied considerably according to the rotation of the specimens. The observed differences between maximum and minimum intensity ratios were 15% for a thin KCl crystal about 100 nm thick and 20% for a bulk KCl crystal, which corresponded to 4.1% and 4.5% differences in atomic concentration, respectively. It was ascertained that for amorphous materials such as Co₈₇Zr₅Nb₈, such a variation of the intensity ratio of X-rays was not observed. It is thus proved that the variations of the intensity ratios of X-rays from the constituent elements with azimuthal angle for KCl and KBr are attributed to the diffraction effect of characteristic X-rays generated in crystalline specimens. This effect, which is the same as that in the production of Kossel patterns, is one of the essential factors limiting the accuracy of the analysis.     

A combined freeze chamber and low temperature stage for an electron microscope

A low-temperature freeze-drying or freezing preparation chamber and specimen stage has been designed and constructed for use with an electron microscope. The system allows biological specimens to be prepared under conditions of temperature and pressure and their subsequent direct transfer to a cold stage maintained at — 175°C within the microscope. Attachment of both the chamber and stage is a simple procedure and does not interfere with the normal electrical and mechanical operation of the microscope. The stage is cooled with the aid of a liquid nitrogen reservoir system and can be maintained at — 175°C for about 80 min without the use of any continuous cooling system. Test specimens have shown that a resolution of about 1.0–1.5 nm can be maintained at — 175°C for 80 min once thermal equilibrium has been established.     

Methods for specimen thickness determination in electron microscopy

A new method is described for specimen thickness determination in transmission electron microscopy. This is carried out by marking specimens with gold particles and analysing the images of a tilt series by computer. The method makes it possible to distinguish populations of particles on different planes and calculate the distance between the planes with statistical variation. We have applied it to carbon films as test objects and compared the results with those obtained by transverse sectioning, STEM mass measurement, optical density and frequency change of a quartz crystal oscillator. We have then used the method for thickness measurement of multilayered protein crystals and thin sectioned cells.     

Electron microscopy of frozen water and aqueous solutions

Thin layers of pure water or aqueous solutions are frozen in the vitreous state or with the water phase in the form of hexagonal or cubic crystals, either by using a spray-freezing method or by spreading the liquid on alkylamine treated films. The specimens are observed in a conventional and in a scanning transmission electron microscope at temperatures down to 25 K. In general, the formation of crystals and segregation of solutes during freezing, devitrification and evaporation upon warming, take place as foreseen by previous X-ray, thermal, optical and electron microscopical studies. Electron beam damage appears in three forms. The devitrification of vitreous ice. The slow loss of material for the specimen at a rate of about one molecule of pure water for every sixty electrons. The bubbling in solutions of organic material for doses in the range of thousands of e nm^?2. We propose a possible model for the mechanism of beam damage in aqueous solutions. The structural and thermal properties of pure frozen water important for electron microscopy are summarized in an appendix.     

Recrystallization and ice-crystal growth in a biological specimen,as shown by a simple freeze substitution method

The sensory hairs of the silkmoth, Bombyx mori, are suitable test objects to check for recrystallization and secondary freezing damage in a biological object, because cryofixation by immersion into propane (90 K) routinely yields well-preserved specimens without noticeable freezing damage. After rewarming the frozen specimens for 10 min to 230 K (boiling propane), the tissue preservation has not deteriorated, and even after 45 min at 230 K, ice-crystal ghosts rarely exceed 50 nm. Two minutes at 250 K (in deep freezer) produced moderate freezing damage with ice-crystal ghosts of 30–75 nm, whereas 90 min at 250 K resulted in severe damage with ice-crystal ghosts well over 100 nm. Secondary freezing damage by ice-crystal growth upon rewarming well-frozen biological specimens, therefore, is a relatively slow process, depending not only on the temperature, but also on the exposure time. Moreover, with some biological specimens, secondary ice-crystal growth starts at much higher temperatures than previously guessed, and with short exposure times rarely should become a hazard in fine structure work.     

Electron diffraction of synthetic polymers: The model compound approach to polymer structure

X-ray fiber diffraction is a technique most often used to establish the structure of chemically regular crystalline polymers. However, this technique has many shortcomings. Some of them are: ambiguities regarding the choice of the space group, limited intensity data, often large diffraction spots, and overlapping intensities. Microsingle crystals of synthetic polymers, when they can be produced, happen to have dimensions that are well suited to being studied by electron diffraction. Electron diffraction data from microsingle crystals often complement the X-ray fiber data. Strictly speaking, one does not solve the crystal structure of a polymer in the conventional way but rather one chooses from among many potentially acceptable models the one which fits the X-ray and/or the electron diffraction data best. The various steps of model building, its placing within the unit cell, and structure refinement will be discussed.