The examination of frozen-hydrated specimens by cryo-electron microscopy is a rapidly developing technique, which has much potential for the determination of biological structures. The ability of cryo-electron microscopy to determine conformational changes1 and transitionary states2 is an exciting area of science, and has already provided many insights into cellular processes. Although there are a few technical difficulties with the technique, it is believed by those in the scientific community that cryo-electron microscopy will develop into a powerful analytical tool.
The technique of electron microscopy allows electrons to be focused by electromagnetic lenses to form high-resolution images. These images give the positional information of atoms, or groups of atoms, in an object relative to one another. The equipment in the microscope allows specimens of interest to be tilted to angles of 80°. Thus, three-dimensional models are able to be created through the use of computer processing, and image reconstruction. This gives electron microscopy major advantages over other structural imaging techniques such as X-ray crystallography. As a result, electron microscopy is often an important step in the determination of the structure of biological macromolecules, their arrangement in cells and the way they interact with other molecules.
When a sample is placed in an electron microscope, the water present evaporates because of the presence of the high vacuum environment. As such, traditional methods of sample preparation have aimed to remove the water present in the specimen before studying it. This has involved a variety of techniques, including embedding the specimen in either plastic, salts of heavy-metals or glucose, or replicating the sample in platinum. Unfortunately, with all embedding methods, it was unknown how much water remained around the protein in the preparation. This could lead to incorrect structures.
Negative staining has been the conventional method of sample preparation of the specimen used by scientists for electron microscopy. This involves staining the protein in the sample with a heavy-metal salt such as uranyl acetate and then dehydrating the specimen. The heavy metal ions will surround the hydrophilic3 exterior of the protein. When viewed under the electron microscope, these give a larger signal4, and thus will provide greater contrast from the biological area of the sample. This enables the protein to be viewed easily. Unfortunately, this process can cause distortions in the shape of the protein.
However, using these techniques, scientists are only able to examine the dehydrated specimen. These methods are unsatisfactory, because when a protein crystal is dehydrated, it does not diffract X-ray or electron radiation suitably to produce a high-resolution image. As a result, scientists have continued to try to find a technique that enables the specimen to be examined in its natural state.
Traditionally catalase crystals have been used to test and verify new techniques. This is because the protein is easy to crystallise into thin plates, which produce excellent diffraction patterns in the microscope. The catalase protein has an added benefit, as its diffraction pattern changes according to its state of hydration. Thus, examination of the diffraction pattern produced enables scientists to identify how much of the native state of the catalase protein has been retained by the technique.
Research was initially undertaken to solve the hydration problem through the use of closed cell, thin window environmental chambers. However, in 1974, Parsons and co-workers achieved better results using differentially pumped hydration stages. Both of these techniques aim to keep the specimen in a saturated water vapour environment. Parsons managed to obtain electron diffraction patterns using the environmental cell. He was able to produce diffraction patterns to a resolution of 3Å5 from unstained and unfixed catalase crystals in the hydrated state. He was also able to localise the phase-separated lipid domains in hydrated human erythrocyte membranes6 , and thus, observe their movements.
Unfortunately, this method required physical separation between the specimen and the rest of the electron microscope column7. This was difficult to carry out without a severe loss in resolution, possibly due to water vapour scattering the electron beam. The only possible alternative to this technique was to examine hydrated specimens that had been frozen. This is cryo-electron microscopy.
In cryo-electron microscopy of frozen-hydrated specimens, the sample is first prepared as a thin film. It is then cooled rapidly, ensuring that the aqueous phase does not crystallise. Following these stages, the sample is transferred into the microscope, where a specimen-cooling holder maintains the low temperature. The specimen can then be photographed just like any other sample preparation. This results in a specimen that is unstained, and fully hydrated. The contrast of the image is produced due to the differences in scattering density of the buffer and the protein8. Contrast is increased by defocusing the object. As with all objects placed inside an electron microscope, the sample is subjected to high levels of ionising radiation. However, due to the absence of protection by chemical cross-links or the stabilisation of the stain molecules, the sample is more easily damaged than if other preparation techniques had been used. Fortunately, minimal exposure techniques enable excellent images to be produced that have little radiation damage.
There are many advantages to using frozen-hydrated techniques. It removes many of the limitations that traditional methods could not overcome, and as such offers many new possibilities. It enables structural features to be studied in different chemical states, increasing the possibility for the identification of conformational changes within the sample. There is also the opportunity to study interactions between macromolecules in their native environment. Before the concept of cryo-electron microscopy became a reality, these were not able to be studied at high-resolution by electron microscopy. As a result, many laboratories have adopted the cryo-electron microscopy technique, producing a large and extremely diverse range of publications.
This series of Guide Entries will review the origins of cryo-electron microscopy, detail processes in the technique, give an insight into various studies that have been performed using cryo-electron microscopy and suggest a few possibilities for cryo-electron microscopy in the future.
The Origins and Development of Cryo-Electron Microscopy
The use of frozen-hydrated techniques to study biological specimens under the electron microscope was first proposed in 1952 by Fernandez-Moran. Unfortunately, his results were not of a high-enough resolution to be beneficial to the scientific community. It was not until the 1970s that scientists were able to produce sufficiently adequate results. Chanzy used the technique of quench-freezing and cryo-observation, and was able to produce complexed and hydrated polysaccharide crystals and hydrated protein. These were of a sufficiently high resolution that they could be used to perform electron diffraction experiments.
Despite Chanzy's work, it was the studies performed separately by Unwin and Henderson and Glaeser that began the revolution in cryo-electron microscopy. Unwin and Henderson published work on both catalase and purple membrane, whilst Glaeser was able to actually quantify the amount of beam damage produced. However, cryo-electron microscopy still had a large number of major technical difficulties that needed to be addressed. The most difficult problem that needed to be overcome was the fact that water crystallises upon cooling. Thus, not only are the cells and organisms destroyed, the ice crystals that were produced damaged the biological structures that were to be studied.
The real break-through for cryo-electron microscopy was a study by Taylor and Glaeser in 1974, published only a few months after Parsons published his environmental cell data. Using a more stable cold stage, Taylor and Glaeser were able to study the structure of frozen bovine liver catalase crystals, which had been prepared using the frozen-hydrated technique. They managed to obtain resolutions of 3.4Å. Subsequent to this initial study, Taylor and Glaeser were able to produce many more micrographs of catalase that had both good resolution and contrast. Unfortunately, the images of this material only contained data to a resolution of 11.5Å. This was due to the movement of the stage on which the specimen was mounted.
The images showed the production of contrast as a result of the density differences. Thus, specimens did not require staining. As protein is denser than ice, it appeared as a dark area, whereas in a negative stain, the contrast is reversed and the protein would appear as a light area. However, in comparison to negative stain images, there was a severe decrease in the amount of contrast. This is because the density difference between protein and ice is far less than the density difference between heavy-metal ions and biological molecules. This problem with contrast was thought to be a major limitation to the cryo-electron microscopy technique. However, if phase contrast is used the difference is minimal. Phase contrast results when the specimen being observed is taken out of focus by a few micrometers.
Taylor and Glaeser also studied the periodic array (distribution) of the surface-layer protein on the outer membrane of Spirillum serpens. This was prepared in the same way as the catalase protein. The contrast that resulted was of such a high definition that the bilayer profile of the membrane could easily be seen. However, the disadvantage of this technique was that the sample had to be placed in distilled water. This was to prevent evaporation, which would cause thinning to the layer of film in which the specimen was held. Evaporation can cause structures within the specimen to be displaced. If the layer of film is too thin, evaporation may even cause particles to emerge from the layer of water, which will lead to them becoming freeze-dried. This is explained further later. Despite this drawback, the study highlighted several benefits of cryo-electron microscopy. It also eliminated suggestions that the freezing process could cause severe disorder to the macromolecular structure.
The next major development for cryo-electron microscopy was the discovery that vitrified ice could be used to freeze the sample. The concept of vitrification was proposed in 1940 by Luyet and Gehenio. It relies on the principle that a liquid is cooled so quickly that the individual constituent molecules are immobilised before they have time to crystallise. However, this method was not considered. This was due to the discovery that glycerol acts as a cryo-protectant, which made the method of growing ice crystals under controlled conditions more successful.
In 1980, Brüggeller and Mayer produced vitrified water by projecting water microdroplets violently into a cryogen. In the following year, Dubochet and McDowall produced a thin layer of vitrified water by spreading a layer of water on a support and immersing it into liquid ethane. This opened the way to producing vitrified samples that could then be examined in the electron microscope. Films 100-200nm thin that contained biological specimens were studied. They appeared to have no crystalline ice present and the structural integrity was maintained to an extremely high resolution.
Within a year of the Dubochet study, a number of key observations were made. It was seen that a thin, uniform layer of solution could be prepared on a carbon support pre-treated by glow discharge in amyl amine. A glow discharge is a low pressure plasma maintained between two electrodes. It is particularly effective at sputtering and ionising materials from solid surfaces. Furthermore, in 1984, it was discovered that a supporting film need not be used at all. With time, many of the initial difficulties of the technique had either been removed or taken into consideration. For example, the problem of beam damage, which is explained further later, had been found to be significantly reduced at low temperatures.
An additional setback to the progress of cryo-electron microscopy was that it was believed that water would alter the resistance of the specimen to the electron beam. However, in 1978, Glaeser and Taylor and disproved this theory. This was supported by Lepault, in 1983.
Today, image processing techniques have developed to such an extent that the contrast can be accentuated, and the three-dimensional structure can be determined by combining the projected densities in the individual micrographs. All these many findings have been combined through time to produce a simple yet extremely powerful method for the production of high-resolution diffraction patterns and images of hydrated solutions. There have also been many developments and advancements in microscope technology, leading to the sophisticated technique that is used today. This has resulted in cryo-electron microscopy becoming the chosen technique to study biological samples in their nearly native state.
Procedures Involved in Cryo-Electron Microscopy
Cryo-electron microscopy is a technique that requires patience, thought, good preparation and familiarity with the equipment used. Any mistake made in the procedure can cause a long delay whilst waiting for the vacuum to recover or waiting for the cryo-holder to be re-warmed and dried.
All electron microscopes must have an additional cryo-specimen holder in order to enable the viewing of the frozen-hydrated specimen. It is normal for manufacturers to supply electron microscopes that contain its own cooling holder. However, it is possible to purchase a standard cryo-holder and place it within an existing microscope. Fortunately, most cryo-holders now are able to perform to the same standard as room-temperature holders. The temperature at which the specimen should be maintained should be below the temperature required for devitrification, usually between -125°C and -150°C. This will prevent the ice from recrystallising.
Anticontaminators reduce the amount of amorphous ice on the specimen that can arise from condensation of water vapour from the air. These can be beneficial if the specimen in the cryo-holder is not protected by cooled shields during transfer to the microscope and the time required to stabilise the sample within the microscope.
There are many different methods in which a sample can be prepared for cryo-electron microscopy. However, each method should ensure that the specimen is embedded in a layer of ice, which is slightly thicker than the specimen itself. After preparation, the specimens are able to be stored for long periods of time, many years if necessary, in a grid box that is kept under liquid nitrogen. This can be beneficial as all specimens can be prepared in a job lot, and subsequently viewed under the microscope when required.
A small amount of the specimen in suspension is placed on to a grid at a suitable concentration to produce good results. This is then thinned into a film approximately 1000Å thick by simultaneously blotting both sides with filter paper. Immediately, the grid must be plunged into the cryogen. The excess cryogen is subsequently removed with cooled filter paper in the freezing chamber, and the grid is transferred on to the microscope by passing it rapidly between baths of liquid nitrogen. It is mounted on a liquid-nitrogen cooled specimen holder and quickly inserted into the microscope. After the vacuum has been restored, the protective shields can be retracted and the specimen examined. Alternative methods include placing the sample between two support films.
Unfortunately, despite the efforts of even the most experienced scientist, it is extremely rare to find a preparation with more than 50% of the grid filled with ideal ice thickness. Thus, a more reproducible method needs to be investigated.
The supporting grid on which the sample is placed is generally made from carbon. There are two distinct types of film on which the specimen can be placed: continuous carbon films and holey films. The higher the quality of carbon support film that is used, the better the results that are obtained. When the films are fresh, they are very slightly hydrophilic, which produces a thin film of liquid. However, films become hydrophobic as they age. This results in grid squares with ice that is too thick being found adjacent to holes that are dry. If films are too hydrophilic, the sample often adheres to the carbon instead of being evenly spread over the holes.
Continuous carbon films are conventionally prepared on plastic-coated grids. In the final step of preparation, the plastic is washed away by placing the grids on a water-cooled surface in a device that refluxes 9 acetone. This has the added benefit that it cleans the surface and makes it hydrophilic, removing the need to perform glow-discharge. This enables the sample to cover the surface in a regular, thin layer. However, this type of preparation has the disadvantage that the carbon support can give background noise to the image.
Holey films are the most common form of preparation that is used today. They remove all traces of interference that could be caused by continuous carbon film. This is required for high resolution studies. There are many ways to produce holey films, but each of them leaves a network of holes of a desired size in the carbon. Those with a fine, strong, lacy structure are most suited to cryo-electron microscopy. A carbon coating of around 200Å enables the grid to be washed as a continuous film, resulting in similar surface properties. Here the sample is spread, unsupported, across holes in a carbon film. Surface tension and evaporation cause the layer to break. Fortunately, there is about 1 second when the layer is around 0.1?m, just before the layer breaks. This gives ample time to form and vitrify the thin layer. The excess water is removed with blotting paper.
A problem can arise when holey grids are used in conjunction with proteins that are positively charged, such as DNA binding proteins. Because the carbon film has a slight negative charge, especially after glow discharging, it readily attracts positively charged proteins. Therefore, such a protein will preferentially land on the carbon as opposed to the holes. To prevent this, a technique has been developed which uses a thinner, secondary carbon layer. The holey grid is made in the same way as previously. However, before depositing the sample, an extremely thin layer of carbon is deposited above the original film. As the carbon is now continuous, there is no preference as to where the protein will be deposited. Furthermore, the layer is thin enough to not have a large impact on the signal to noise ratio 10 of the sample.
It is estimated that a freezing rate of 107 K/sec is required to obtain vitreous ice from pure water. Thus, the rate to vitrify a biological sample must also be high. The type of cryogen used affects the rate of freezing due to differences in the thermal properties of coolants. Liquid nitrogen, ethane or propane have all been used in the frozen-hydrated technique. However, ethane slush held in a container of liquid nitrogen is the most commonly used cryogen today.
Liquid propane and ethane appear to have similar cooling properties. However, they differ in the effect of the residual amount left on the grid. Ethane, with a boiling point of -89°C rapidly sublimes during transfer from the specimen holder to the microscope. Propane on the other hand has a boiling point of -42°C, and may not evaporate completely. This can affect the diffraction pattern produced. Nitrogen is not used directly. This is because the gas formation around the specimen that occurs when the sample is introduced into liquid nitrogen slows the cooling and leads to the formation of ice crystals.
Other Methods of Freezing
There are other techniques to freeze samples other than the one described above. Jet freezing involves the specimen being kept out of direct contact with the cryogen by a thin metal foil. However, this makes the preparation method more difficult to perform. High-pressure cooling is possible. Pressures of 2100atm produce the best cooling. This enables crystallisation to be delayed until required. However, the technique that is most used today is Freeze-Slamming. This involves projecting the specimen onto a metal block that has been cooled by liquid helium or nitrogen. This produces good cryo-protection of the surface.
The analysis of transient structures is possible if the sample is subjected to a change in conditions immediately prior to freezing. This can be performed by spraying droplets of one of the reactants onto the sample. As freezing is almost instantaneous, the transitory states are easily captured. By altering the delay with which the droplets are sprayed can result in a series of micrographs which detail how the conformational change occurs.
Droplets around 1?m in diameter are applied with the use of an atomiser spray. They spread rapidly over the surface due to surface tension, and only minimally disturb the film beneath. The droplets only manage to partially move the particles away from the point of impact. Mixing occurs rapidly by the process of diffusion. Fortunately, there are no detectable changes in the concentration of the sample, and highly resolved images are still possible to produce.
Crystalline specimens are usually handled in a slightly different way. These are often placed on a continuous carbon film in glucose or tannic acid and cooled directly in liquid nitrogen. This is possible because very high concentrations of these sugars act as a cryo-protectant and inhibit ice formation.
Formation of Ice
At low temperature and pressure, water can be frozen into three predominant forms, vitreous, cubic and hexagonal.
Hexagonal ice is obtained from cooled liquid water or by warming vitreous or cubic ice. Hexagonal crystals are relatively large and are usually striated by bend contours if they originate from the freezing of liquid water. They appear as grains with regular polygonal contours, which can be as small as 20-30nm if they originate from the condensation of atmospheric water vapour on a cold specimen or in the cryogen. Hexagonal ice is the form found in ice cubes and snowflakes.
Cubic ice can be obtained in three ways: slow deposition of water vapour in vacuum at a temperature between approximately –135 and 100°C, warming vitreous water, or by very rapid cooling of liquid water. Cubic ice appears as a mosaic of small crystals with dimensions around 0.1mm. It produces an electron diffraction pattern that is characteristic for a powder.
Vitreous ice is obtained by slow deposition of water vapour on a cold substrate or by rapid cooling of liquid water. It has the same density as cubic or hexagonal ice. Vitreous water appears as a smooth layer without internal structure. Observations suggest that vitreous water is polymorphous and that a second, low-temperature high-density form may have been observed by electron microscopy.
It has been suggested that biological molecules are cryo-protectants, affecting the way that water freezes at a radial distance of 20-30Å. This means that the area in which the specimen is found can remain vitreous even if other areas of ice are crystalline. This could explain why specimens can still produce good results even if cooled in a less effective cryogen.
Unfortunately, whole cells and tissues are often too thick to enable them to be spread into a thin layer to be examined by cryo-electron microscopy. This problem can be overcome by reducing the thickness of the specimen before it enters the electron microscope. This can be performed by vitrifying the specimen and then cutting it into the thin sections required for observation. Knives are currently available which can section at either a 35° or 45° angle, and cut thicknesses from about 30nm to 1 micron.
Several biological specimens have been examined using the technique of cryo-sectioning. These include bacteria, insect flight muscle and culture cells. Unfortunately, sectioning is an extremely difficult task, and cutting artefacts are common. Distortions are also unable to be easily relaxed after sectioning. These can be left, or can be crushed into ripples that are known as 'crevasses'. These are a network of lines that are roughly parallel to the knife edge. They are found on most vitrified sections, with pure water and lipid droplets being the most susceptible for crevassing. The thickest regions of the sections are where the crevassed regions are found. Zierold has performed many studies into the phenomenon. These distortions cause a loss in long-range order of the structure. In addition, it makes it difficult for images to increase the signal-to-noise ratio.
Another problem can be 'chatter';. This is usually caused by variation in the section thickness due to vibration of the specimen with respect to the knife. It has a period range of less than 0.1mm to several micrometers. Chatter can be prevented by fixing the specimen more tightly. However, chatter can also be caused by an irregular cut made by the knife. This can cause cutting dependent beam damage.
Observation of the Specimen
The specimen is usually found in only a few regions on the grid. These can be identified with low magnification, minimising the extent of the irradiation on the sample before studying it more closely. The contrast of the specimen depends on the specimen itself, the defocus value of the objective lens and the thickness of the ice. The ability of the specimen to provide contrast cannot be altered. The defocus value can be set to optimise the contrast and will enhance the visibility of low-resolution structural detail. Finally, the thicker the film of ice, the greater the background noise. Thus, the thinner the film of ice, the better the contrast produced.
There are three common methods for observing and recording the image produced by electron microscopy. Visualisation on a fluorescent screen is useful for aligning the microscope. It is based on the fact that when electrons hit the screen, visible light is produced.
The most common method for recording data is photographic film. These are extremely sensitive and can record high resolution data. Once developed, the film is scanned into a computer for analysis. The resolution is then limited by available computer memory and scanning resolution, which allows the best resolution of about 2Å per pixel.
Charged-coupled device (CCD) chips remove the need for film by transferring the image directly into a computer. The CCD camera allows the quick recording of images, which enables the quality of sample to be judged. The high contrast of the CCD images, as compared with film images, also makes them very suitable for publication.
A layer of particles, such as viruses, can be viewed either as images of separate particles or as images of the same structure, but taken from a variety of orientations. If the second perspective is taken, these views can potentially be re-combined to determine the three-dimensional structure of the original virus.
The process of three-dimensional reconstruction involves identifying the orientations of each of the views represented by the particle images and then combining these images to determine the structure. For asymmetrical specimens, each angle must be represented. However, for symmetrical macromolecules, fewer projections are required. Working with algorithms formulated by Crowther, a number of groups have developed methods for determining the structure of icosahedral particles.
The use of Euler angles assists in the determination of the orientation. They are directional angles, which indicate a position and orientation in space around a common centre. The concept of common lines and the use of Euler angles were introduced by Crowther in the 1970’s85.
? (Theta) - Defines the elevation above or below the equator.
f (Phi) - Defines the rotation (azimuth) around the equator.
? (Psi) - Defines the rotation around the centre of the position defined by ? and f.
These are shown in Figure 4.
To generate a reconstruction, ? and f need to be calculated in order to define the position of each particle. This is done by using the random conical tilt mechanism.
The use of a highly symmetrical particle allows the symmetry to serve as a guide in determining the orientation. In addition, it makes the process of computing the structure itself much more efficient.
Obviously, the reliability of the assignation of the orientation and the resolution of the information contained in the image limit the final resolution of the structure. A process of refinement allows the determination of the orientations to a higher degree of accuracy and reliability. Unfortunately, the orientations have to be determined with high precision, or the process remains extremely ineffective. This can be seen in Figure 5. This ultimately is determined by the resolution of the data. Sample variation cannot occur, as the process relies on the fact that differences between particles are caused by a change in orientation, not a contaminant. Tests for symmetry in the particle images allows identification of particles in the sample that have been distorted.
This picture demonstrates the necessity to take more than one image to create a three-dimensional reconstruction. Although the object being visualised is a Tiger, the image that is formed on the screen could be misinterpreted as a worm emerging from an apple.
The resolution and quality of the image can be improved by looking at more copies of the same projection and obtaining an average for the results. This will reduce the level of background noise. It is not unusual for a high-resolution structure, around 10-12Å, to require about 50,000 projections in total.
The recent advances in computer processing have been the key to the development of three-dimensional reconstruction. As greater and greater numbers of pictures can be processed increasingly quickly, the signal-to-noise ratio increases steadily. In addition, as the use of higher voltage and field emission gun sources has increased, the strength of the high resolution information transferred from the specimen to the film has increased. This is due to the use of phase contrast information. A more coherent electron source produces a better phase image. This increases the ability to determine the orientation of the particles. This further enhances the signal-to-noise ratio.
A review regarding high-resolution icosahedral reconstruction was published by Erika Mancini, Felix de Haas and Stephen D Fuller in 199786.
Advantages and Disadvantages to the Technique of Cryo-Electron Microscopy
There are many advantages to performing cryo-electron microscopy. However, the main advantage of using the technique is that the structure remains native and no dehydration is required at all as a result of the quickness of the cooling. Although this may have unknown disadvantages, comparisons with X-ray crystallographic data shows there to be little difference in the structure that is determined. As staining and fixing is not required, this removes drying, fixation, stain artefacts and reduces the level of interference on the sample by the researcher.
The freezing process is able to freeze a sample of up to 1?m in 10-4s or less. This short time enables transient states to be captured and studied readily. In addition, as the temperature is maintained at these very low levels, thermal vibrations are minimal. This removes a certain amount of noise from the images.
The choice of specimen buffer is made at the discretion of the researcher, not dictated by the technique. This allows the conditions in which the sample is to be placed to be set precisely. Spray freezing can be used to great advantage here, as it enables the environment in which the sample is placed to be changed, moments before freezing. However, it should be remembered that the extent of electron scattering, and as such the contrast obtained, depends on the concentration of the sample. Pure water will provide the highest degree of contrast.
The environment in which the sample is placed prior to freezing can be easily controlled. This can be used to a great advantage to initiate conformational changes. Temperature-jump devices have been used to control temperature changes more precisely just before the sample enters the cryogen. Controlling the environment is also advantageous as it allows humidity control. This prevents the invagination of lipid vesicles and maintaining solute concentration, both of which are likely to be caused by water evaporation between the blotting of the sample. This reduces the specimen to a thin film before freezing.
An advantage of using hydrated protein crystals is that they retain a high degree of periodicity. In negatively stained protein crystals, the periodicity is rarely seen to resolutions greater than 8Å.
In addition, the images obtained by cryo-electron microscopy show details of the entire specimen. In contrast to negative staining or metal shadowing which show the surface of the sample, the image is a superimposition of all of the density variations in the sample. Although interpreting such a projected image is complicated, it is a major advantage when studying many different macromolecules.
The main limitation of cryo-electron microscopy is that it is not possible to look at the sample for prolonged periods of time because of beam damage. As such, this is a limiting factor for resolution. It is not possible, for instance, to pick out areas of interest in a sample and study each carefully for a prolonged period of time.
Surface tension in the liquid film can displace structures and even crush delicate structures if the film is too thin. In addition, during the process of crystallisation, particles may be pushed above the surface of the sample. This would cause the particles to lose their hydrated environment, and may cause these areas to be freeze-dried. Contrast can be affected by the amount of specimen that is embedded in the ice. A partially dried specimen will give greater contrast than a fully hydrated, frozen specimen. Thus, it is necessary to determine whether the sample is fully hydrated or not. Comparison of various specimens, in differing thicknesses of ice can help when deciding which image properly represents the hydrated specimen. However, a more reliable method is to warm the grid after photographing the specimen, causing the sample to become freeze-dried. This will allow comparisons to be made. The freeze-dried specimen will have less distinct fine detail when compared to the frozen-hydrated specimen.
In cryo-electron microscopy, there is an extremely low signal-to-noise ratio. Biological macromolecules are normally made up of carbon, hydrogen, oxygen and nitrogen. The electron absorption of these molecules is very low. As a result, image contrast is also very low and it is hard to detect features when dealing with just a few images.
The requirement for very rapid cooling is a limitation of the technique. The method works best with specimens that can be made very thin so that cooling is rapid. Examples of these specimens include suspensions of viruses or protein complexes. Thicker objects, such as cells, must be handled differently and represent a challenge to the field. Suspension samples also present problems if they contain solutes that interfere with vitrification. Samples containing salts or sucrose show a phase partitioning after vitrification.
Photographic detail is an additional problem. The specimen can only be observed with low magnification, subjecting the sample to a low dose of radiation in comparison to the photograph. This is necessary to enable any suitable photograph to be taken. The dose of radiation received by the sample depends on the length of time that the radiation beam remains on the specimen. Thus, by reducing the time that the sample is subjected to the radiation, a greater radiation dose can be applied. This means that a higher magnification of the specimen can be achieved. As a result, photographs are often taken without looking at the sample properly, as it is known that many will be discarded.
Movement of the specimen can also cause photographic problems. It is a result of both mechanical and radiation-induced effects. Movement is a large limiting factor in resolution. Mechanical movement can be reduced by the production of better cold stages. However, radiation-induced specimen movement may represent a severe obstacle to high-resolution imaging.
Looking at consecutive photographs, electron beam damage becomes obvious. A typical micrograph is recorded with an optical density of 0.8 at a magnification level of 36,000X. It receives an electron radiation dose of 5 electrons per Å2 50. However, after the sample has received a dose of 50-100 electrons per Å2, 'bubbling' begins to occur. Bubbling is thought to be a result of the accumulation of trapped volatile fragments. A large number of bubbles begin to form on the surface of the sample, which increase as the radiation dose increases. The smaller bubbles then fuse into a large bubble that breaks. Eventually, a hole forms throughout the depth of the specimen. Bubbling easily destroys the delicate biological structure within the specimen.
Currently, it is not known what the limit of resolution will be for cryo-electron microscopy. However, as with all techniques, it will depend on many different factors, including the signal-to-noise ratio, contrast, defocus and the specimen itself.
The Biological Applications of Cryo-Electron Microscopy
Once Taylor and Glaeser had published their results on both catalase and Spirillum serpens, many other research groups began looking at cryo-electron microscopy as a method to study other biological macromolecules. As a result, the technique has been applied to study a large, diverse range of specimens, including viruses, ribosomes, actin filaments, membrane proteins and lipid vesicles. A few of these applications will be explored below.
Recently, it has been possible to use the technique to provide detailed three-dimensional structures of individual macromolecules greater than 100kDa in size. This is an easier technique to perform than X-ray crystallography, as no crystal preparation is required. In addition, the random orientation of the particles removes the need to tilt the specimen. This makes data collection extremely simple.
As a result of cryo-electron microscopy, a resolution of 7.6Å has been determined for the fold of the hepatitis B capsid protein. This was done by Crowther and Steven, using the icosahedral reconstruction of hepatitis B core particles. Thousands of images were taken to produce the visualised helices of the structure.
Recently, bacterial ribosomes have been characterised in its empty state. This involved reconstructing the image of the ribosome from 2,447 separate projections of single particles. These were then classified into groups and averaged to produce the final structure. Furthermore, in 1997, the same group published information on the structure of the ribosome in pre-translational and post-translational configurations. This was possible with the use of a technique known as EF-G chasing. However, no reconstructions of the tRNA crystal structure bound to the ribosome were presented.
This achievement was of major significance because of the asymmetry of the molecule. Work is continuing, and a resolution of 11.5Å has been obtained for the 70S Ribosome.
Joachim Frank has been a fundamental researcher in the investigation needed to determine a high-resolution structure of the ribosome. In 1997, he published a review of the steps that had been taken to obtain a high-resolution image of the ribosome between 1995 and 199799. This included the work published by many scientific groups. Furthermore, in July 2000, he published findings in Nature that showed the detection of ratcheting rotation within ribosomes. This enables the advancing of the molecule along the messenger RNA. Using cryo-electron microscopy, the rapid rotation of one of the subunits was detected. This was possible by adding a non-functional GTP analogue as well as elongation factor, which resulted in protein synthesis being halted. The specimen was then frozen and studied. Analysis was able to show that when the elongation factor and GTP bind to the 30S subunit, it rotates around 6 degrees with respect to the 50S subunit. After the GTP chemical reaction, the 30S subunit rotates back to its original orientation within the ribosome.
Many viral structures have been studied by cryo-electron microscopy, and they were one of the first particles to be studied in vitreous ice. They are well suited because of their symmetry, making it easier for computer programs to combine data, producing a clearer structure. Examples include rotavirus, SFV, SV40, adenovirus, sindbis virus, bacteriophage T4, nucleocapsids of herpesvirus, Semliki Forest virus, influenza virus, and TMV.
The morphology of the capsid protein is the most obvious feature obtained from cryo-electron microscopy. The size, shape and lattice symmetry is revealed, and it is often possible to determine how the subunits are arranged. The nuclear acid organisation can be visualised, a feature not possible with negative staining or other methods. Time-dependent studies can be used to determine how the viral structure mutates into a mature particle. Finally, it is possible to perform virus-antibody and virus-receptor complex studies, which are extremely useful in determining how the virus interacts with its environment.
The three-dimensional reconstruction of bacteriophage T4 shows that the frozen-hydrated specimen is very similar to that produced by negative stain analysis. However, it is easier to interpret and there is no uneven staining. The particles are well preserved, unlike in the negative stain. This assists the process of identifying key orientations.
DNA is denser than protein. As such, it can be visualised by the addition of solutes such as metrimazide. Bacteriophage T4 was studied and the DNA was clearly visible, and a variety of patterns were observed.
In 1992, the morphopoietic mechanism in the P2-P4 bacteriophage system was characterised. This determined the symmetry of both systems as well as structure of the capsid protein and how the domains are arranged. The paper then determined the changes that occur to the systems, and how the various protein products interact.
- Influenza Virus
Influenza has been the enveloped virus that has been examined to the greatest detail and depth. X-ray diffraction has identified the structure of glycoprotein spikes, and many different strains have been examined in different conditions by electron microscopy. Cryo-electron microscopy has visualised the lipid bilayer of the B Hong Kong strain, as well as the M-protein and the spike size and number were also estimated. A strains have been studied using cryo-electron microscopy. In addition, Nahoaki published evidence that a different membrane structure exists in an influenza A virus strain.
- Vesicular Steatites Virus
Through the use of cryo-electron microscopy, the shape and surface has been visualised. The G-protein spikes that cover the surface have been shown to be trimeric, and surface conformational changes have been seen. Defective-interfering particles have also been visualised.
Three-dimensional reconstruction has been used to produce high-resolution structures for both herpesvirus and SFV. Structures have been produced to a resolution around 10Å. This enables secondary structures to be identified.
Studies of complexes of rotavirus and monoclonal antibodies raised against the spike protein have been performed. These were the first of their kind, and revealed a large amount of information about binding sites, conformational variations of Fab fragments and the dimeric nature of the spike proteins.
Helical cytoskeletal structures have been studied by cryo-electron microscopy. This has resulted in many features being clarified.
Microtubles have been found to have very a similar structure to that determined by X-ray diffraction patterns. This shows that the structure is maintained in the freezing process, even though depolymerisation occurs in vitro at low temperatures. As the microtubules are not flattened as they are during negative staining, it is easier to identify their dimensions. The determination of the dimensions is also aided by the fact that stain does not concentrate at the edge of the microtubules. A supertwist has also been identified within the helix, which had a twist of several micrometers. The speed of the freezing process also aided in the exploration of the depolymerisation process in vitro. Mandelkow and Mandelkow visualised frozen-hydrated microtubules in the course of time-dependent reactions during the disassembly process. This study showed the mechanism of microtubule breakdown to occur not only at both ends of the microtubule, but also from inside the body. Short protofilaments were found to be among the early breakdown products.
Cryo-electron microscopy has enabled a better estimate of the diameter of actin filaments, and clearer images can now be visualised. For tropomyosin, the overlap between consecutive molecular ends has been studied in detail. Through the use of staining, the overlap was unclear. Cryo-electron microscopy also enabled a mercury label attached to a thiol residue to be identified. This was not possible in negative stained specimens because the protein signal was too weak to establish molecular positions.
The use of cryo-electron microscopy has enabled the characterisation of the calcium release channel from skeletal muscle. It is a major component of the triad junction, and is the site of excitation contraction coupling. The structure was determined from a set of 1,665 images of the channel, taken using cryo-electron microscopy. Three classes of four-fold symmetrical molecules were identified, and three-dimensional images were able to be calculated for two of them. These were almost identical and only differed by a 180° rotation.
The calcium release channel was found to consist of a large cytoplasmic assembly and a smaller transmembrane assembly that protrudes 7nm from one of the faces. A low density cylinder is found down the centre of the transmembrane assembly, which may correspond to the transmembrane calcium-ion conducting pathway. The cytoplasmic assembly is constructed from about 10 domains loosely packed together. This is well adapted to maintain the structural integrity of the triad junction, whilst still allowing ions to freely diffuse to and from the transmembrane assembly.
In the May 2002 Edition of the Journal of Microscopy – Vol. 206, Part II – a paper was published which characterised the structural changes in rabbit muscles fibres that occur after a rapid jump in temperature. This demonstrated that cryo-electron microscopy could be used to determine differences in structure that only exist for a matter of milliseconds. It also demonstrated that cryo-snappers, a device that can be used to clamp small specimens, if they have a series of small grooves in the jaws, do not compress the specimen to damage it.
Studies have been carried out to determine the structure of the haemoglobin found in the polychaete worm Alvinella pompejana. These worms are the most thermophilic metazoan known, and reconstructions of their haemoglobin structure have, until recently, been difficult to determine. It was found that the haemoglobin from the worm is structurally different to other annelid haemoglobin molecules. Usually, the vertices of the upper and lower hexagonal layers are at 16° to each other. However, in Alvinella, the vertices almost eclipse one another. The central linker complex is surrounded by 12 hollow globular substructures. It was seen that the linker complex consists of a central hexagonal toroid, two internal bracelets onto which the hollow globular substructures are built, and six connections between the two hexagonal layers. A local psuedo 3-fold symmetry was also recognised for the first time on the hollow globular structures. It is thought that the slight differences in the structure enable the worm to survive at such high temperatures.
Cryo-electron microscopy has been used to study the conformational changes that occur in the bacterial chaperonin GroEL. GroEL exists as two back-to-back heptamer rings in the absence of ATP and ADP, and when ATP or ADP are present, the GroES heptamer binds to one end of the GroEL cylinder. The series of conformational changes that take place as GroEL binds to GroES and hydrolyses ATP has been studied by freezing the intermediates.
Recently, researchers at Imperial College have been able to identify the structures of the RNA polymerase core and holoenzyme to resolutions of 11Å and 9.5Å126. Conformational changes between the core and the holoenzyme forms of the polymerase were identified to be mostly associated with the b’ subunit. They were able to identify the positions of a2bb' within the holoenzyme, and the location of the s70 subunit was also suggested.
This group of macromolecules have been widely studied by both negative staining and glucose embedding. However, as the whole structure of the membrane can studied in cryo-electron microscopy, many new insights that are important in understanding the function of the molecule have been identified.
Gap junctions have been extensively investigated and two calcium-sensitive states have been examined. The entire subunit was able to be identified, rather than just the external portion identified by negative staining. The continuous structure made identification of calcium-induced conformational changes more easy and straightforward. The three-dimensional reconstructions suggested that the conformational change occurs by a co-operative rearrangement with the tilt of the subunits. This may explain why calcium is able to affect gap junction permeability in vivo.
Gap junction membranes consist of a pair of membrane plaques, both of which contain hexameric connexon units arranged in a hexagonal array. Variations in the structure, orientation and packing of the connexons have been studied.
The nicotinic acetylcholine receptor from the ray Torpedo has also been studied using cryo-electron microscopy. Three-dimensional reconstructions of tubes of receptors showed that the molecule is more elongated and angular than previously thought. The study showed that the receptor consists of five rod-like subunits, which arrange with pentagonal symmetry to form a cylinder. More material is found at the synaptic surface than at the cytoplasmic surface.
Other studies include the lumenal plasma membrane of the mammalian urinary bladder, and the pore-forming OmpC protein from Escherichia coli outer membrane. These gave views of the portion of the protein subunits located in the membrane in addition to the external material, which can be viewed with negative staining. Also, bacteriorhodopsin and plant light-harvesting complex have been resolved to 3Å. Currently, many other membrane proteins including visual rhodopsin, membrane receptors, ion pumps and channels, are being studied with freeze-hydration techniques.
Other MacromoleculesLipids and Lipid Vesicles
Cryo-electron microscopy has been used to study both lipids and lipid vesicles. In 1985, Lepault was able to examine the structure of the lipid bilayer. Frederik and Siegel both studied the transition from bilayer to non-bilayer. Furthermore, membrane fusion processes have also been extremely well characterised.
Nucleic Acids and Protein-Nucleic Acids Complexes
Cryo-electron microscopy has been used to study crotoxin complex, T4 DNA helix destabilising protein as well as Fc fragment. High-resolution images, around 3Å resolution or better, have been recorded of each of the above complexes.
As technology improves, advances can be made in existing techniques. Cryo-electron microscopy is no exception, and a number of improvements can be expected.
These are likely to include better electron sources with higher electron accelerating voltages (up to 400kV). In addition, mathematical corrections for lens defects such as spherical aberration, phase contrast transfer, and magnification defects are expected. Increased computer power and memory will obviously allow the processing of higher resolution information. Better scanners will reduce errors in transferring the projections onto a computer.
The development of a superfluid helium cooled stage has been published, and this is currently being refined into a commercially available product. It allows the stage to be cooled to 1.5K, and enables extremely high-resolution images to be captured.
It is likely that steps taken by Studer to improve high-pressure freezing will further provide alternatives to cryo-electron microscopy. In his studies, samples were able to be pressurised to 2000 bar whilst being simultaneously cooled. The equipment was smaller and lighter than previously possible. A wide range of samples were able to be studied, including yeast cells, liver and nerve tissue, and plant leaves.
CCD-based detectors have become very well established in electron crystallography. However, there are a few shortcomings in the CCD's performance. This has prompted research into other alternatives. As CCDs are normally used in an indirect mode, the incident radiation formed is a visible light image in a phosphor, which is imaged on to and recorded by a cooled scientific grade, low noise CCD via tapered fibre optics. Unfortunately, light scattering in the phosphor leads to an unacceptable loss in spatial resolution.
Semiconductor detectors, which eliminate the intermediate light conversion step in the phosphor, offer a number of potential advantages over CCD detectors. Hybrid pixel detectors, such as Medipix1 and Medipix2, are some of the most promising new semiconductor detectors. They were originally developed for particle physics experiments, but are now being re-designed for applications to X-ray imaging and electron crystallography.
These detectors have an absence of readout noise, which leads to true photon counting, improvement in spatial resolution due to an absence of light scattering and very fast readout. Slightly less important features include a compact size, the ability to cover large areas in different geometries and room temperature operation. These semiconductor detectors are expected to help in the task to gain better resolution pictures.
Software improvements are also eagerly anticipated. Features, including the ability of computers to automatically select a particular particle in a sample, and align projections of that particle to produce a structure in a short space of time, will benefit the scientific community.
However, in the future, it is likely that a microscope will be designed that will increase the degree of contrast in thicker specimens. This may remove the need for computer processing.
Having almost completed this dissertation, it has occurred to me that cryo-electron microscopy could be utilised in the determination of causes of certain types of cancer. Cancerous cells could be studied and compared with healthy cells, to determine whether there are any obvious metabolic differences. Extensive searching of the Internet and other methods of research resulted in gaining no information on the suitability of the technique.
1 Conformational changes are changes in the structure of a molecule.
2 When molecules interact with one another to form a new compound, there is one or more intermediate stages, known scientifically as transitionary states.
3 Hydrophilic, from the Greek, meaning water loving.
4 The heavy metal ions have many electrons surrounding the nucleus. Thus, they give a larger signal in the electron microscope.
5 This is really quite good. At a level of 3Å resolution, XXX can be identified.
6 Science-speak for ‘He looked at individual components of the human red blood cell membrane’.
7 Without the separation, it would not have been possible to maintain the saturated water environment.
8 The buffer has far fewer electrons than the protein. Thus, a contrast is created.
9 Boils, collects the evaporated acetone and then reboils it.
10 The amount of background noise.