Electron Microscopy

The limit of resolution of the light microscope is roughly 2000 Å, which is insufficient to visualize cell organelles, viruses, and macromolecules of current interest. This is possible, however, with the electron microscope for which the limit is less than the diameter of а uranium atom (approximately 5 Å ) under special conditions.
There are few instruments that present such а bewildering array of knobs and meters as does the electron microscope; furthermore, few techniques require greater skill and attentiveness to detail. Nonetheless, for most applications the instrument is simple to use and sample preparation is not complicated. There is no doubt that every laboratory should have access to an electron microscope and that every biochemist should be proficient in its use. This chapter will not supply specific procedures or technical details but will indicate how electron microscopy (ЕМ) is done and what its potential is.

SIMPLE THEORY ОF OPERATION

An ordinary light microscope consists ofа light source а condenser for focusing the light on or near the object, an object holder (i.e., the slide and the stage), an objective lens for focusing the image, and an eyepiece for projecting the image formed by the objective onto the eye or photographic film. This is also true of an electron microscope, except that the light is replaced by an electron beam, the sample holder is a wire screen called a grid, and the lenses are electromagnets rather than glass.
A schematic of an electromagnetic lens, which basically consists of an axially symmetric electromagnet through which the electron beam passes, is shown in (The focusing action of a magnetic field is complex and will not explained here. For further information see the Selected References near the end of the chapter. ) A magnetic field is generated by windings of copper wire. For high magnification, lenses of short focal length are needed and it can be shown that, focal length decreases as the magnetic field increases. The magnetic field could be made larger by increasing the current in the windings, but this would generate a c considerable amount of heat. Instead the field is concentrated by enclosing the wires in a soft iron casing and by inserting two conical pieces of soft iron called pole pieces, each of which contains a small orifice through which the beam passes.
As mentioned earlier, this brief description is not intended to explain how an electromagnetic lens works, but mainly to describe some of the, parts and to introduce the terminology commonly encountered in electron microscopy. Suffice it to say that a divergent beam of electrons can be brought to focus within certain limits (i.e., there is substantial spherical aberration) at a point on the axis of the lens.
The illumination source consists of a white hot tungsten filament, which emits electrons. The potential of the anode, to which electrons are drawn, is normally from 40 to 100 kilovolts greater than that of the filament. The filament and the anode together constitute the electron gun. The anode contains a small orifice through which some of the fastest electrons pass. This hole plus a small aperture just below it collimate the electrons to form a beam. The beam is slightly divergent because the electrons m deflected toward the edge of the orifice of the anode owing to its positive potential. The divergent beam is then made to converge onto the specimen by an electromagnetic condenser lens. The beam is rarely focused sharply on the sample because an intense beam could destroy it.
The image is formed by what is often called the subtractive action of the sample. That is, some of the electrons are scattered from the atoms or the object. The pattern of this loss or electrons generates the image pattern (in much the same way that the light intensity is reduced by an absorbing object in the light microscope). The objective lens, which is adjusted so that the sample is precisely at its focal point, then refocuses the beam to produce an image. This image is then magnified in several stages by three electromagnetic lenses called the diffraction; intermediate, and projector lenses. The final projector lens forms the image on either a fluorescent screen or a photographic plate.
The electron microscope differs in three respects from the light microscope: First, because electrons do not travel very far in air, the entire microscope column must be in a high vacuum; hence, the object must always be dry and therefore dead. Second, because the magnification of an electromagnetic lens is proportional to the magnetic field, which in turn is proportional to the current in the windings, the magnification can be varied continuously by varying the current through the windings of the lens. In light optics, magnification is fixed by the set shape of the glass lens; hence, many objectives are needed to cover a range of magnification. Third, none of the primary aberrations can be corrected in standard electrometric lenses because the magnetic lenses are always convergent. (The Crewe microscope described in a later section partly corrects spherical aberration.) To reduce spherical aberration and thereby improve image quality , the lenses are operated at very small numerical apertures. This has the effect of severely limiting the resolution allowable by the Compton wavelength of the electron because the limit of resolution is 0.61 λ/NA. For an electron, the Compton wavelength, λ , is hc/ E, in which h is Planck's constant, c is the velocity of light, and E is the energy of the electron. For a 60-Kv electron this 0.03 Å. However, because the numerical aperture of a magnetic lens is normally about 0.0005, the practical limit is more like 4 Å. (As will be discussed later, the nature of the sample actually allows this limit to be reached only rarely. ) Nonetheless, the resolution is still approximately 500 times as great as that obtained with a light microscope.

METHODS FOR PREPARING SAMPLES AND PRODUCINC CONTRAST

Preparation of Specimen Supports

The great capability of an matter for scattering electrons requires that a sample be very thin-otherwise no beam will get through to form all image. In practice, the maximum thickness is approximately 0.1 micron (1000 Ǻ) for 100 Ǻ resolution and from approximately 50 to 100 Ǻ for 10 Ǻ resolution. This poses no real problem in observing viruses, fibrils,
or macromolecules, but for most cells, which range from 1 to 50 microns in thickness, it is necessary to make thin sections (see Embedding, Sectioning, and Staining). This requirement clearly means that the sample support (i.e., the equivalent of a microscope slide) must also be very thin, uniform in thickness, and without obvious structure at high magnification.
The specimen support, or grid for all samples consists of a disc cut from a rigid copper (in some cases, platinum) mesh with openings approximately 75 μ. per side, overlaid with a thin "electron transparent" film called the support film. Electron transparency indicates that its electron scattering power is both low and uniform. Commonly used films consist of layers from 100 to 200 Ǻ of either carbon or various plastics (Parlodion, Formvar). Unfortunately, no support film is truly structure less and for high-resolution work with macromolecules whose dimensions are comparable to film thickness, variations in the intensity of the background produce a "granularity," the grains ranging from 5 to 10 Ǻ. This seriously limits the attainable resolution. Films are prepared in one or three ways Parlodion films are prepared by placing a drop of a solution , of Parlodion in amyl acetate on a water surface. (A liquid surface is used because it is very smooth. ) The droplet spreads, the solvent evaporates, and a thin film of the plastic forms. Formvar films are prepared by dipping a smooth glass microscope slide into a solution of the plastic and then removing it. When dry, the thin film on the glass will slide onto a water surface if the slide is slowly lowered into the water. (This dipping method is preferred by some microscopists for Parlodion. ) Carbon films are prepared by evaporating onto freshly cleaved mica (which is a molecularly smooth surface, being a single plane. of a crystal), and floating the film . onto a water surface as is done in preparing Formvar films. In all cases, the film is mounted on grids in either of two ways: it can be lowered onto the grids-which had been placed on the bottom of the container before- hand-by draining the water off, or the grids can be placed on the film from above and the entire support picked up by touching the surface with a sheet of plastic or absorbent paper.
The intrinsic contrast of biological material is poor because scattering of the carbon atoms in the support film is of roughly the same magnitude as that of all the principal atoms (C, N, O, P, S) of the material. The usual method for correcting this situation is to deposit heavy metals of very high scattering power on the structure in such a way that the pattern of metal somehow indicates the features of the sample. Useful metals are osmium, platinum, lead, and uranium, although chromium, palladium, tungsten, and gold are sometimes used. Several standard methods for sample preparation and contrast enhancement follow.

EMBEDDING, SECTIONING AND STAINING

If the material under observation is too thick for the passage of electrons, a thin slice or section must be made. To prepare a thin section, the sample must be made rigid so that it can be cleanly cut. This process, called embedding, consists of the gradual replacement of the aqueous material of the sample with an organic monomer (e.g., methyl methacrylate) that can be hardened by polymerization. After it has become solid, the plastic containing the supposedly undisrupted sample is sliced with an ultra-microtome (a kind of knife) into layers from 500 to 1000 Ǻ thick. The sections are then stained ( although staining is sometimes done before embedding) by exposure to solutions of salts of molybdenum, tungsten, lead, or uranium, or to the vapor of osmium tetroxide. (The word staining refers to the deposition of a metal by a chemical reaction or the formation of a Complex with certain components of the sample, to increase the electron density. ) These stains react with proteins and other macromolecules and aggregates and thereby put electron-dense material in the sample. Stained preparations are beautiful and appear to contain considerable detail, yet it must be realized that what is being observed is the distribution of metal atoms and therefore of the chemical groups that can react with a particular stain. An example of a type of artifact that can arise is the deposition of osmium on opposite sides of a thick membrane, producing two black lines separated by an unstained space, which can be mistaken for a double membrane. The embedding and sectioning procedures themselves can induce distortion because of uneven permeation of the sample by the organic monomer and because of the cutting itself.
Only a single layer through a sample is observed when looking at a thin section and this may not always be adequate. To get a picture of the entire sample, a large number of sections are normally examined. An elegant though tedious method is serial sectioning in which successive sections are collected in sequence and examined.

REPLICA FORMATION

The method used for observing the surface of an electron-opaque or easily destroyed specimen is called replica formation. The specimen is coated with a thin layer of platiltum and then a supporting layer of carbon (for strength), both deposited by vacuum evaporation or shadow-casting. This bilayer is then floated off onto water and picked up on a grid. The replica is thus a facsimile of the surface of the object-that is, the contours are the same as those of the sample. This method has been used to study the surfaces of viruses, membranes, and certain protein crystals that are immediately destroyed by the electron beam.

FREEZE-ETCHING AND THE CRITICAL - POINT TECHNIQUE

In replica formation, the water in the sample must be removed before preparing the replica because the production of a film by shadow-casting must be in vacuum. This presents a problem in that structures usually collapse during air drying as a result of surface tension effects accompanying the phase changes that occur during evaporation of the solvent. Freeze etching and the critical-point method avoid the production of artifacts due to drying. In freeze-etching, the sample is rapidly frozen, sectioned or fractured, and placed in a vacuum with conditions of pressure and temperature such that the water sublimes from the surface of the sample. A replica of this surface is then prepared by evaporating platinum or carbon while it is still in the vacuum.
The critical-point method makes use of the fact that no liquid phase can exist above a "critical temperature" characteristic of each substance. The procedure follows: First, a wet sample is soaked in ethanol. The ethanol is then exchanged with liquid CO2 under pressure at 15°C. The temperature of the specimen is then raised above 31°C (the critical temperature) and the liquid CO2 becomes a gas. Presumably, all three- dimensional relations are preserved. A replica can then be prepared or the sample can be observed directly in the microscope, if it had been
stained beforehand. This method is especially useful in preserving macro-molecular structures if molecules are deposited on a film rrom a solution.

SHADOW-CASTING

A great deal or electron microscopy is concerned with the structures ofparticles-such as viruses, phages, and ribosomes-and or macromolecules. The sizes of such objects as well as limited information about their structures can be obtained by shadow-casting. The particles (in solution or suspension) are applied to a grid overlaid by a support film by spraying. The liquid quickly evaporates, the sample is placed in vacuum, and a heavy metal is applied by evaporation. This requires boiling a metal using white hot tungsten and is typically done either by wrapping a metal wire around a tungsten wire or by placing small lumps of metal in a tungsten container. The metal atoms are projected in all directions and, if the vacuum is good, in straight lines. If evaporation is from an acute angle, metal will pile up on only one side of the sample and win cover the grid except in the shadow of the particle. If the vertical (H) and horizontal (L) distances from the evaporation source to the specimen are known, the height (h) of the particle above the surface of the grid can be calculated from the length (d) of the shadow cast by the specimen because h/d = H / L. Hence, the dimensions of the particle can be determined.
Shadow-casting has not been too successfu1 for smaller macromolecules, because or the small size of the shadow and the granularity of the support film. Many of the problems can be avoided by using the negative-contrast procedure described next.
A special use or shadow-casting, the Kleinschmidt technique for observing nucleic acid molecules, will be discussed later.

EGATIVE-CONTRAST TECHNIQUE

The negative-contrast (incorrectly and commonly called negative staining} procedure of Brenner and Horne consists of embedding small particles or macromolecules in a Continuous stain or electron-opaque film. The stain penetrates the interstices of the particle but not the particle itself. The image is a result of the relative intensity of the beam at every point, which is proportional to the thickness of the opaque material at that point. Hence, contrast is achieved by virtue of the particle reducing, the effective thickness of the opaque film-that is, the particles are seen in
outline. In this procedure the sample is either mixed with the stain and sprayed on the grid or sprayed on the grid first and then sprayed with the stain. The interpretation of negatively stained samples is sometimes difficult because various patterns can ,be observed depending on (1) the thickness of the stain; (2) whether it has penetrated the interstices 9f the particle, (3) whether it lies above and below the particle, and (4) whether any of it has adsorbed specifically to the sample (positive staining). For this reason it is usually necessary to look at a large number of preparations and particles. Nonetheless, the negative-contrast method has been used successfully for a wide variety of phages, viruses, and proteins.
The most useful stain for negative contrast ill a phosphotungstic acid salt (although there is some indication, that this might ultimately be replaced by cadmium iodide). It should be realized of course that resolution in the negative-contrast method depends on the size of the opaque atoms c (i.e., approximately 5 Å).

POSITIVE STAINING

Positive staining has not had widespread use for most macromolecules because it is not usually possible to attach a sufficiently large n umber of heavy atoms to obtain good contrast, although it has been possible with large molecules and structures such as ribosomes, DNA, RNA polymerase, and collagen. The collagen results are especially beautiful and deserve description because they indicate the great analytical power of this technique. When tropocollagen molecules are allowed to aggregate side-by-side to produce collagen and the resulting structure is stained with phosphotungstic acid at pH 4.2 (binding to positive groups). If uranyl acetate, which binds negative groups, is used, the banding pattern is exactly the same. Therefore the sideways aggregation probably includes the conjunction of positive and negatively charged groups. Analysis of the banding pattern of several different types of collagen aggregate shows that the tropocollagen molecule can form head-to-head or head-to-tail fibers and that in the standard structure the sideways aggregation involves a displacement of one-quarter of the molecular length from one tropocollagen to the next.

KlEINSCHMIDT SPREADING WITH POSITIVE STAINING AND ROTARY SHADOWING

Probably the most spectacular method of sample preparation of the past decade has been the Kleinschmidt procedure for visualizing DNA. In a single step, artifacts due to drying are eliminated and extraordinary contrast is obtained. This technique is now used in almost all biochemical laboratories and can be learned in an afternoon. A drop or a DNA solution in 0.5- to 1.0-M NH4 acetate containing 0.1 mg/ml cytochrome c is allowed to flow down a glass slide onto the surface of 0.15- to 0.25-M NH4 acetate. As the drop touches the surface, a film of denatured cytochrome c spreads across the surface. This film contains somewhat extended DNA molecules to which a thick (100-200 Å) layer of denatured cytochrome c binds. If a grid is touched to the denatured protein film, a drop containing a part of the film is transferred to it. When the grid with the adhering drop is immersed in alcohol, the aqueous phase is removed and the film adheres tightly to the support film on the grid. As used at present, the technique requires a preliminary positive staining with uranyl acetate; the protein adsorbed to the DNA becomes stained as well as the background film but, owing to the excess protein, good contrast is achieved. Contrast is enhanced (or created, if staining is not used) by shadow-casting a metal (usually platinum) at a very small angle while the sample is rotating. Because the DNA coated with protein projects above the protein film, metal piles up against the DNA-protein complex like snow drifting against a fence-but on both sides and on all molecules, regardless of orientation, because of the rotation. This method can be used to determine the length of DNA and whether it is circular or super coiled. Under certain conditions (usually by incorporation of the denaturant formamide into all solutions), single-stranded polynucleotides become extended and are easily visualized. Single-stranded DNA and RNA are distinguishable from native DNAby their relative thinness and kinkiness.
In a variation of this. method, the diffusion method, a cytochrome , film is formed on a DNA solution and DNA molecules diffuse upward and adhere to the film. This is a slow process but allows much smaller DNA concentrations to be used.
One important application of the Kleinschmidt technique is seen in the heleroduplex method. Here single strands from two different DNA molecules are allowed to hybridize. Homologous regions (i.e., regions having complementary base pairs) show up as double-stranded DNA but no homologous regions remain as single-strand loops.
A recently developed variant of the Kleinschmidt technique replaces the cytochrome c with benzalkonium chloride, a surfactant that also forms a film to which DNA adheres. The DNA is shadowed with platinum but, because the DNA is not coated with protein, it appears much narrower than when cytochrome c is used. Positive staining is still used but the uranium atoms bind directly to the DNA. This procedure gives better resolution of macromolecules (e.g., RNA polymerase) bound to the DNA and will probably gain widespread use.

SPECIAL MECHANISMS OF IMAGE FORMATION

Dark-field Electron Microscopy

Dark-field electron microscopy allows for substantially increased contrast, as is the case with the light microscope. A dark field can be obtained either by using hollow-cone illumination, as in light microscopy, or by making the beam fall on the sample at an angle such that it does not enter the imaging system. The image is formed by the scattered and diffracted electrons alone. Dark-field electron microscopy has not been used very much, but for observation of macromolecules it deserves greater attention. A simple example indicates its usefulness.

Example 3.1. The binding of proteins to DNA.
DNA has a diameter of 20 Å, yet when coated with cytochrome c, as in the Kleinschmidt procedure, its diameter is nearly 200 Å. To observe the site of binding of a protein to DNA, the protein would have to be much larger than 100 Å, or it would be buried in the cytochrome coating. However by using ultrathin electrically charged support films, DNA or DNA plus a bound protein can be drawn from an aqueous solution without the need for the supporting protein film. Both DNA and the bound protein molecule of interest can be positively stained with uranyl acetate. However, even though the support film is not stained, contrast is poor because, in the absence of cytochrome c, the total amount of bound uranium is not very great. Using ordinary illumination, the DNA is nearly invisible. However, it is easily seen by dark-field electron microscopy because the background is unilluminated.
It should be noted that dark-field electron microscopy does not improve the resolving power of the microscope but eliminates some of the factors in sample preparation that prevent making use of the existing resolving power. It is not yet clear what the relative merits of dark-field versus the benzalkonium chloride variant are.

The Crewe Microscope

In the ordinary transmission electron microscope, the incident beam covers the entire sample. As the electrons interact with the sample, they do so in several different ways: (1) no interaction at all (i.e., traveling through the interatomic spaces), the most abundant class of electrons; (2) elastically scattered (i.e., without loss of energy) by the orbital electrons of the atoms of the sample, the second most abundant; and (3) inelastically scattered (i.e., with loss of energy) from atomic nuclei, the least abundant. The ratio of the last two classes is a characteristic of each element, because the size of. the nuclear target and hence the cross section for inelastic events in- creases greatly for the larger elements.
A special new microscope has been designed to take advantage of these facts. To do this, the beam is collimated into a very small (approximately 5 Å) spot. The spot is swept across the sample as in a television set. As the beam moves, the ratio of the latter two classes of electrons is measured at each point by an electron-energy spectrometer. This ratio is converted into an image on a television screen by suitable electronic circuits; the information attained can also be processed by computer analysis. This new and important step in electron microscopy gives a new element of analysis to electron microscopy because individual atoms can be identified. In some cases, the limit of resolution is improved and pictures at 2 Å resolution have been obtained.

The Backscatter Scanning Microscope

The scanning electron microscope (SEM) is a device that has produced the many beautiful photographs of cell surfaces seen in the past few years. This microscope is limited to about 200 Å in resolution and operates on a very different principle from that of the transmission electron microscope. Like the Crewe microscope, the beam is collimated into a small (100 Å) spot, and the spot is swept across the sample surface, which has been coated with a thick (200 Å) layer of gold or other heavy metal. As the beam impinges on the metal and penetrates into it a short distance, electrons are emitted from the gold either as secondary emissions or as directly backscattered electrons from the beam, Because of an unexpected angular relationship between the number of electrons emitted and the angle of the surface to the incident beam, which is close to, but not identical with, the way that light reflects from the surface of an object, the image formed by the collected electrons gives dramatic images of the surface being examined.