Some Notes for Biological Electron Microscopy

The importance of the electron microscope lies in its ability to break through the resolution barrier of the light microscope (c.a.100nm) into the world of biological ultrastructure. The pioneering work of Robley Williams in 1946 gave the first images of individual viruses. Since then, the technique has been constantly improved to the point where resolutions for biological structures in the 2-3nm range are routine. For a small number of specimens, using a combination of EM and computer image processing (electron crystallography) it has been possible to reach the level of resolution (0.3nm) previously possible only by X-ray crystallography and NMR.

The two main types of electron microscopes used in biology are the scanning electron microscope and the transmission electron microscope. As the latter is more common, we will go into more detail here into its use.

The scanning electron microscope (SEM).

In the SEM the specimen is scanned with a focussed electron spot which produces "secondary" electrons as it hits the specimen surface. These secondary electrons are detected and converted into a video image of the surface. The specimen is usually coated with a thin layer of evaporated metal to prevent electrostatic charging. This coating and the size of the scanning spot limits the normal resolution of the SEM to about 5-10nm.

The Transmission Electron Microscope (TEM)

The TEM has much in common with the light microscope in that images are produced by transmission of the illuminating radiation through a thin slice of material. This can be an ultra thin section (less than 0.2mm in thickness), dispersed particles or (in cryo-em) frozen suspensions. The specimens are normally supported by a thin carbon film which is analogous to the glass specimen slide in the light microscope. Contrast is produced either by staining the specimen with heavy metal salts (e.g. uranyl acetate, lead citrate), by evaporating a layer of metal on to the surface or by the use (as in light microscopy) of phase contrast for unstained specimens.

Resolving Power

The human eye can distinguish two objects if they are not closer than 0.1 mm at a normal viewing distance of 25 cm. This ability to separate two objects optically is called resolving power. Any finer detail than this can be resolved by the eye only if the object is enlarged. This enlargement can be achieved by the use of optical instruments such as hand lenses, compound light microscopes and electron microscopes.

Resolution in the light microscope

The resolving power of the light microscope is mainly limited by the wavelength of the light (l) used for illumination.
The Abbe "diffraction limit" for image resolution is given by:
d= l/2*N.A.

Where N.A. (numerical aperture) is a measure of the resolving power of the lens. The best oil immersion lenses have an N.A. of about 1.4, thus for visible light with a wavelength of 400nm, the resolution would be about 140nm.

It can be seen that resolving power improves (d decreases) as the wavelength decreases.

Due to this limitation of resolving power in light microscopy, other sources of illumination with shorter wavelengths than visible light, have been investigated. Early experiments using X-rays of extremely short wavelength were not persued further because of the inability to focus these rays although there have been recent attempts to revive X-ray microscopy using synchrotron radiation. So-called "lensless" imaging systems such as the scanning tunneling microscope (STM) and the atomic force microscope (AFM) have also been developed to overcome the limitation of wavelength.

Resolution in the electron microscope

The development of electron imaging became a possibility when Louis de Broglie proposed in 1924 that just as light has a dual nature behaving both as particles and waves, there could be a wave-like quality to the propagation of atomic particles.

l = h/mv where h is Planck's constant, m the mass and v the velocity

In the case of electrons which are electrically charged and can be accelerated by an electric field:

l = h/(2meV)-2 where V is the accelerating voltage, e the electron charge


In addition to this wave-like nature of moving electrons, they can be easily deflected (focussed) by magnetic or electrostatic lenses, opening up the possibility of constructing a magnifying, imaging system with very high resolution - the transmission electron microscope, first developed by Ernst Ruska in 1935. Using the formula above we can calculate that for electrons accelerated by 100 kilovolts, l = 0.0037nm. In practice, the resolution of the best electron microscopes is limited, mainly by lens defects (small N.A., spherical aberration etc.), to about 0.2nm.

Construction of the TEM

In its normal form the TEM is like an inverted light microscope. The illumination source , usually a tungsten filament (cathode) is at the top of the evacuated column. The filament is heated, causing it to emit electrons (thermionic emission). A high voltage (the accelerating voltage) of up to 400,000 volts is passed between it and the anode, so that the negatively charged electrons are accelerated towards the anode (positively charged) which is placed just below the filament. Some electrons pass through a small hole in the anode, forming an electron beam which passes down the column. The speed of the electrons and hence their wavelength depends upon the accelerating voltage.

Electro-magnetic coils, placed at intervals down the column, act as lenses to focus the electrons. Condenser lenses focus the electron beam on to the specimen which is mounted on a movable stage above the objective lens. As the electron beam passes through the specimen some electrons are scattered whilst the remainder are focussed first by the objective lens and then by two further stages of magnification (intermediate lens and projector lens) to give an image on a fluorescent screen or on photographic film. Electron lenses (unlike optical lenses) can be varied in focal length by varying the lens current through the coils which simplifies focussing (by the objective lens) and zooming of magnification (by the intermediate and projector lenses).

An aperture is placed below the objective lens to block electrons scattered to high angles (particularly by the heavy metal atoms used for staining). The removal of these electrons from the image produces amplitude contrast. In the case of unstained specimens, where there is very little amplitude contrast, it is necessary to use phase contrast, produced by defocussing (under-focussing) the image.

Specimen preparation for TEM

Most of the improvement in resolution of biological specimens has resulted from better specimen preparation techniques. There are four main problems:

a) The need to maintain a high vacuum in the microscope column
The biological specimen must be stabilized in some way so that its ultrastructure, when exposed to the vacuum, is as close as possible to that in the wet living material. This can done by chemical fixation of bulk specimens followed by dehydration and plastic embedding or by embedding a particulate specimen in a layer of stain (negative staining). Unstained specimens for electron crystallography (see below) are embedded in a layer of glucose or tannic acid which replaces the water. Alternatively, the specimen is frozen very rapidly to prevent ice-crystal damage and then transferred, still frozen, at liquid nitrogen or liquid helium temperature into the EM for observation (cryo-EM).

b) The low intrinsic contrast of biological specimens.
As mentioned above, amplitude contrast depends on the type of atoms in the specimen; electron scattering increases in proportion to the atomic number of the scattering atom. Biological specimens are composed of atoms of low atomic number (carbon, hydrogen, nitrogen, phosphorus and sulphur) and their contrast can be increased by staining with heavy metal salts or shadowing with evaporated heavy metal (e.g. platinum). This can have an added advantage that the heavy metal is usually less sensitive to irradiation by electrons (see below). The simplest method is ?negative staining? in which a sample in suspension is dried down in a layer of uranyl acetate or sodium phosphotungstate. This limits the resolution to about 2nm which is, however, sufficient to visualise the subunit structure of proteins and is useful as a routine technique which does not require specialised equipment.

c) The specimen must be thin.
The limited penetrating power of electrons means that the specimens must be very thin or must be sliced into thin sections (50 - 100 nm) with an ultramicrotome to allow electrons to pass through.

d) The damaging effect of electron irradiation of the specimen.
The radiation dose experienced by the specimen during observation in the TEM can be sufficient to cause covalent bond breakage, leading to mass loss, shrinkage and other structural changes. This can be reduced by so-called ?low-dose? techniques where the minimum electron dose is used to photograph a ?fresh? part of the specimen - all focussing and other adjustments to the microscope are made on an adjacent area. Image averaging techniques, either combining data from many single particles or from unit cells in a crystal, are also important in order to improve the signal-to-noise ratio for images taken with very low electron doses. There is also a reduction in damage (by a factor of 5-10 times) by cooling the specimen to low temperatures.

3-Dimensional Image reconstruction.
Klug in 1968 showed that because of the high depth of focus of electron lenses, images are 2-D projections of the whole object density (or the stain density) on to the image plane. The 3-D information can be recovered if a number of views of the object are recorded at different angles of observation. This can be done using a ?tilt stage? or goniometer, normally up to a maximum tilt angle of about 60 degrees. For special cases, e.g. helices or structures of high rotational symmetry it may be possible to use just a single view or a much smaller number of tilted views to reconstruct a 3-D model. In all cases, it is necessary to digitise the micrographs and then use either ?fourier? or tomographic computer algorithms to calculate the 3-D object density. The fourier methods (often referred to as ?electron crystallography?, see below) are analogous to those used in X-ray crystallography with the important difference that there is no ?phase problem? in electron crystallography since it is possible to calculate phases directly from the images.

Electron Crystallography
Parsons in 1971 showed that unstained, thin catalase crystals will give electron diffraction patterns to atomic resolution if they are maintained in a wet state in an electron microscope equipped with a hydration chamber. Although this demonstrated that very high resolution information is present in unstained specimens, it was not possible to obtain high resolution images by this method. In 1975 Unwin & Henderson showed that similar results could be obtained for bacterial rhodopsin (purple membrane) by drying the specimen in glucose. Electron diffraction gave diffraction intensities extending to about 0.3nm and the first images gave phase information extending to about 0.7nm. Combining these data for a series of images of specimens at different tilt angles they obtained the first 3-D structure of a membrane protein showing trans-membrane helices. The method has now been improved to the point where phases are also measurable to about 0.3nm allowing amino acid side-chains to be resolved. For this technique, very large single layer crystals are necessary - it is thus particularly suitable for crystals of membrane proteins confined within a lipid bilayer. Beam damage can be reduced by a factor of about 5 times by cooling the specimen to liquid nitrogen temperature (see below).

Cryo-EM.
Cryo methods were developed partly to reduce the electron beam damage which decreases at low temperatures and partly to reduce the damage (drying artefacts) caused by removal of water from biological specimens prepared for EM. In 1978 Glaeser & Taylor constructed the first cryo-transfer holder which enabled a frozen hydrated biological specimen to be observed in the EM. In 1982 Dubochet and co-workers at EMBL improved the technique by introducing fast freezing in liquid ethane to give vitreous (as opposed to crystalline) ice films. Cryo-electron microscopy has now become a routine technique for obtaining well preserved hydrated specimens with resolutions almost as good as those obtained by electron crystallography of glucose embedded material. In both techniques, low dose imaging is necessary to avoid disruption of the specimen and contrast is enhanced by defocussing the objective lens. Phase artefacts introduced by defocus - the contrast transfer function (CTF) - can be corrected during the computer image reconstruction procedure.
A disadvantage of the electron microscope compared with the light microscope is that it is not possible to observe dynamic processes. Rapid freezing and Cryo-EM, however, provide a method for "trapping" dynamic processes at millisecond time intervals, which can be useful for many biological processes.


--------------------------------------------------------------------------------------------------
Literature:

Good introductions to the methodology are provided by the two multivolume series (which are in the EMBL library)

Hayat, M.A (editor) . Principles and techniques of electron microscopy:

Glauert, A. Practical methods in electron microscopy

A review on Cryo-EM:

Dubochet et al, 1988: Cryo-electron microscopy of vitrified
specimens. Q Rev Biophys, 21:2, 129-228

A review on 3-D image processing of crystals and electron crystallography:
Amos et al (1982) , Prog. Biophys. Molec. Biol. 39, 183-231