Contents
Principles of electron microscopy
Before we can understand how the electron microscope (EM) works, it is important that we distinguish between resolution and magnification. Magnification is how much bigger we can make something appear. Resolution is how much better we can distinguish two closely spaced points.
A lower limit on resolution is imposed by the fact that when we view specimens in a microscope, we are viewing them using waves. We use light waves to view objects under the light microscope. It is vitally important to understand that we cannot distinguish details finer than the wavelength of these waves. This is because light diffracts (bends into what 'should' be crisp shadows) and interferes with itself when moving around very small objects (or through very small holes).
This means edges become fuzzy, and images lose their crispness.
A computer model of diffraction through a pinhole. Note that there is
not a clear light/dark division, but instead that interference fringes
make the edges fuzzy.
Limits to light microscopy
The wavelength of light (λ) imposes a limit on light microscopy that makes it unsuitable for exploring all but the largest features of most cells. The wavelength of green light is c. 500 nm. Bacteria and mitochondria are about 500 nm long. Hence, only the grossest details (nucleus and endosymbionts) of a typical eukaryotic cell are resolvable using visible light. There are several possible solutions…
By using blue/violet light (400 nm) (decrease the wavelength), we can approximately double the resolution that can be obtained using red (600 nm) or white (mean 500 nm) light, but a doubling of the resolution is hardly a great improvement. What we would really like to do is decrease the wavelength, a lot. Unfortunately, as a ribosome is about 20 nm wide, so to see reasonable detail, we need 'light' (electromagnetic radiation) of about 1 nm, which are X-rays.
However, to use X-rays in a microscope, we need to be able to focus them. Unfortunately, refracting X-rays to any significant extent, is very difficult and even if we could, the lenses would need to be perfect to within 1 nm, which is pretty infeasible. We can use X-rays for diffraction crystallography, but not easily for direct imaging. X-ray microscopes do exist, but they are not widely used in biology.
Some of the blur of a light preparation is not due to diffraction. Some is due to the thickness of the specimen, because not all the light reaching your eye comes from the focal plane on which you're focussed.
Computer processing can de-blur a stack of images at different focal planes (a blurry 3D image). This 'image deconvolution' gets us right down to 200 nm, but no further.
The confocal scanning microscope (CFSM) is very similar: it's image deconvolution by optics rather than number-crunching. The CFSM performs a Raster scan of the specimen using a laser, which is focussed on a single point in the specimen with a pinhole. Fluorescent light emitted by the specimen is focussed onto a pinhole detector. Only light from the illuminated point contributes significantly to the detector signal. This process is faster than deconvolution, but tends to cook the specimen.
However, neither of these fixes the fundamental problem that visible light (even given the best optics and chance in the world) is unable to resolve structures much more closely separated than 100 nm.
Luckily, quantum physics rides to he rescue of use poor biologists. Particles have an associated wavelength, due to the wave-particle duality expressed in the De Broglie relationship:
λ = h / m v
- me = 9.11 × 10−31 kg
- h = 6.63 ×10−34 J s
So for 10−9 m (1 nm), we need the electrons to have a velocity close to 100 000 m s−1), or about 0.2% of the speed of light. Accelerating an electron to this speed needs a very high voltage, about 100 000 V. This sounds a lot, but in fact, the energy required is rarely limiting to the resolving power of the electron microscope. We should be able to resolve to 0.002 nm with easily achievable voltages, but - as ever - there are some obstacles. Electrons are difficult to focus, and require magnetic coils for this, not glass. The observation must be performed in a vacuum to prevent scattering of the electrons by stray air molecules; and the specimen must absorb or reflect electrons, which most biomolecules don't. Despite these problems, the practical resolution limit of a (transmission) electron microscope is 1 nm (0.1 nm is very good), with a magnification of 100 000 ×, which is clearly a great improvement on light microscopy.
The solution these problems is to simply ensure that the specimen is dead.
Really, really dead.
We put the specimen through five processes to ensure this before subjecting it to transmission electron microscopy…
- Fixation. The tissue must be sectioned later, but it will rumple and become disordered if left in its native state. Therefore the tissue must be hardened and crosslinked, or 'fixed'. Osmium tetroxide (OsO4) is useful for crosslinking membranes and glutaraldehyde crosslinks proteinaceous organelles into a resistant 3D matrix (although this needs careful pH buffering, because it generates acid as it works. Sodium cacodylate is the most widely used buffer). Fixation also prevents oxidative degradation of the dying specimen.
- Staining. Organelles of interest must be electron dense to contrast with the background. Electron density increases with atomic number, so only heavy atoms work well, and there are few of these in the cell naturally. Sections are therefore often stained with osmium tetroxide, uranyl acetate and/or lead citrate.
- Dehydration. The specimen will be viewed in a vacuum, so it can't be wet, else water will evaporate into the vacuum and scatter electrons. Water is usually removed with increasing concentrations of ethanol, from 70% to 100%.
- Embedding. The dehydrated specimen is then soaked in propylene oxide containing epoxy or acrylic resin monomers. The resin is crosslinked (cured) using heat, UV light or chemical hardeners, depending on its type. The specimen is now embedded in a solid block of plastic.
- Sectioning. The embedded specimen is finally sectioned using a microtome, which is a glass or diamond knife that can cut extremely thin sections (100 nm). This is necessary because thicker specimens will not allow electrons to penetrate them at all.
Finally, the dehydrated, fixed, embedded, stained, diced and sliced specimen can be observed.
Electron microscopy techniques
There are two main electron microscopy techniques:
- Transmission electron microscopy, which essentially looks through a thin slice of a specimen.
- Scanning electron microscopy, which looks at the surface of a solid object.
Bamboo vascular bundles. (Left to right): light micrograph with
safranin/alcian blue stain; SEM at similar scale; TEM of fibre cells (a
few of the pink cells from the light micrograph), note the higher
magnification and resolution: lamellation of the cell wall is
clear.
Transmission electron microscopy (TEM)
In TEM the section (prepared as described above) is placed on a copper grid. The grid is then placed into the EM through an airlock and bombarded with a focussed electron beam. A 'silhouette' is projected onto a phosphor screen below the specimen. The electrons are usually produced from a heated tungsten wire ('electron gun').
Immunogold staining is used to identify proteins in TEM. A thin section is incubated with primary antibodies raised against an antigen of interest. We then use secondary antibodies tagged with colloidal gold particles to stain these primary antibodies. Immunogold staining requires careful interpretation - it only stains the very surface-most structures if applied after fixation, and antigens can be damaged by the processes described above.
Immunogold staining for α-amylase (rat pancreas vesicles)
Macromolecules can be visualised under TEM with negative staining. This is done by washing a specimen of macromolecules fixed to carbon tape with concentrated uranyl acetate. The metal stains the background and edges of the molecules, but not the molecule itself. This technique is useful for visualising viruses (bacteriophage below), DNA, RNA and filamentous proteins.
You must be extremely careful interpreting an EM-micrograph: are we really looking at what we think we're looking at?
- How much is due to the processing?
- How much is due to the slow death of the specimen and bursting of lysosomes?
- Many years of argument about whether the endoplasmic reticulum was real.
- You only look at thin sections: your single continuous intestine would look like Emmenthal cheese.
There are several solutions to these artefact worries. Firstly, negative controls must always be performed for immunogold staining (to prevent misinterpretation of stain precipitate as immunogold, for example). We can also compare TEM images to less 'processed' methods such as scanning electron micrographs (field emission SEM has a similar resolution to TEM).
Scanning electron microscopy (SEM)
SEM is used to look at the surface of a solid specimen. The resolution is usually only 10 nm (unless FESEM is used), but with 20 000 × magnification. We get very attractive 3D-looking images because of the large depth of field.
Like CFSM, the SEM raster scans the specimen (like a TV screen) with a (primary) electron beam: (secondary) electrons bounced off the specimen are detected. X-rays produced by excited atoms in the specimen can also be detected.
For SEM, critical point drying in liquid CO2 is used more often than alcohol dehydration and air-drying. This is because it prevents artefacts caused by surface tension during evaporation. Critical point drying takes CO2 directly from liquid to gas without boiling, by exploiting the fact that above a certain temperature, the distinction between a very thick vapour and a very thin liquid disappears.
The dried specimen is then now coated with 2 nm coat of gold atoms. This prevents charge accumulation in the EM and erosion of the specimen. It is also required as, just as in TEM, the specimen needs to be electron dense.
Coating is usually performed in a sputter coater. The specimen is placed on conductive carbon film on an earthed SEM stub in an inert argon atmosphere. A cathode ray tube is then used to coat the specimen. The cathode is made of gold, and as a very high voltage is applied to the tube, the argon in the sputter coater ionises. Bombardment of the gold in the cathode by ionised argon atoms sputters off gold atoms from the cathode onto the earthed specimen, until the specimen looks like Shirley Eaton in Goldfinger.
There are several techniques that are frequently used with SEM (although some may also be used with TEM). Freeze fracture is used to observe internal surface features. The specimen is rapidly frozen in liquid nitrogen, then cleaved with a sharp knife. The specimen often shears along the inside of lipid bilayers (where the membranes are weakest), revealing transmembrane proteins. This technique isn't so good for internal structures because of ice crystal formation.
The frozen fracture is not electron dense, so it needs to be coated with heavy metal (often platinum). This metal is evaporated onto the specimen from an oblique angle, to create a 'shadow' effect. The platinum replica is then thickened by coating with carbon perpendicularly.
The specimen is then dissolved away with acid, and we observe the metal replica under TEM.
Alternatively, the surface of a frozen specimen can be etched (removal of surrounding ice by sublimation), then directly shadowed with platinum and observed with FESEM or TEM.
Other techniques:
Cryo-EM is an alternative way to look at macromolecules. Specimens of macromolecules are frozen and fixed simultaneously by slamming them into a liquid-helium cooled copper block. The water freezes so fast that it cannot form crystals, and instead becomes vitreous ice (an amorphous, glass-like substance with no long-range atomic order, and hence no irritating crystals). We then create a replica as described above.
In EM-tomography, pictures from different angles in EM can be combined electronically. This is termed EM tomography, and it's useful for investigating structures that are too large or un-crystallisable for X-ray crystallography.
As mentioned earlier, electrons in SEM (and TEM) excite atoms in the specimen. X-rays characteristic of particular atoms and molecules can be detected. This allows us to identify where e.g. in a piece of wood a preservative like PCP has been deposited.
Summary
- Resolution is inversely related to wavelength. To resolve distances ≈ 100 nm, we can use confocal optics for light microscopy. To resolve distances <100 nm, we must use electron microscopy.
- Transmission EM casts a 'silhouette' of a thin slice of aldehyde-fixed, dehydrated, heavy-metal stained, resin-embedded tissue.
- Scanning EM scans the surface of a fixed, sputter-coated specimen.
- Artefacts are common. Techniques that avoid some of these problems include cryo-EM and shadowed, etched freeze-fractures (or metal replicas thereof).
Test yourself
- Why and how must EM samples be dehydrated?
- What are the pros and cons of TEM compared to confocal light microscopy?
Bibliography
- Alberts, B., et al. (2002). Molecular biology of the cell. 4th edition. Garland Science, New York. 548-570. "Looking at the structure of cells in the microscope"
This page has been peer reviewed by 2 people. Thanks to Michael
Schmeisser for his correction.