Immuo-  Electron Microscopy
George Posthuma, Cell Biology UMC Utrecht

The transmission electron microscope is a versatile instrument that can provide information at a (sub) nanometer scale about the composition and form of many different substances ranging from atoms to molecules in geological, chemical and biological specimens.   The best attainable resolution is at the moment 0.05 nm.


Figure 8.1: Immunolabeling with ProteinA-Gold

As soon as biological research in the 1940’s became aware of the fact that an antibody (AB) can be used to mark the molecules to which they were directed (hence called antigens) they became an indispensable tool in science.

In theory it is a simple procedure: First an antigen is purified, next it is injected into an animal of a different species and nature does its job. The animal will recognize the antigen as a foreign molecule and will produce antibodies which will recognize their antigen out of millions of other molecules.

Another widely used approach is genetic labeling. In this case the protein of interest is tagged genetically with a relative small protein like hemaglutinin, green fluorescent protein or with the smallest tag available at the moment: tetracystein(Gaietta, Giepmans et al. 2006). The latter would be in theory the most suitable tag, but there are as yet no antibodies available and Reash reagents in combination with DAB cytochemistry must be used for detection in the electron microscope.  When AB’s are available, they can be used to visualize antigens on gels, blots, tissues, cells and (ultra-thin) sections with unprecedented accuracy provided they are coupled to a suitable marker.

To be able to see the antibodies in the electron microscope an electron dense marker is necessary. At first antibodies were coupled to enzymes that could produce an electron dense precipitate (after osmium tetroxide fixation). Although extremely valuable, this technique has its limitations: the reaction products are diffusible, sometimes difficult to discern from the surrounding tissue elements and difficult to quantify. Visualization of more than one antigen in a specimen is not possible.

Figure 8.2: Immunolabeling with IgG Gold

For immuno-electron microscopy a very small discrete electron dense marker would be ideal. Faulk and Taylor (Faulk and Taylor 1971) introduced the gold particle as a marker for the electron microscope. Gold particles can be made in different sizes  (Slot and Geuze 1981) and can be coupled to a wide variety of biological active molecules :  enzymes, antibodies, toxins, lectins, protein A and many more. Since that time immunogold labeling is the most widely used technique to visualize molecules in the (transmission or scanning) electron microscope. For obvious practical reasons the primary antibody is not coupled to an electron dense marker (availability of antibodies,  shelf life). The primary antibody is visualized by means of Protein A coupled to gold (Figure 8.1) or an IgG (Figure 8.2).

How many gold particles can be expected?
Before an immuno-labeling study is performed, one should always realize that there should be enough molecules present to be visualized. In a thin section only a very small part of the cell is visible. A 50 nm thick section of a “standard” 10x10x10 µm cell only represents 0.5% of the total volume or 0.3% of the plasma membrane! Furthermore molecules will change during their lifetime in a specimen: they are trimmed, glycosylated, phosphorylated, polymerization takes place, etc.  Moreover, the antigenic determinants can be destroyed by fixation and may not be accessible to antibodies in the specimen thus further reducing the number of available molecules. The highest reported labeling efficiency (LE) is approximately 10%(Griffiths and Hoppeler 1986), but often it is less than 1 %. When this standard cell has 30,000 molecules on its plasma membrane, in the electron microscope between 0.9 (LE=0.01) and 9 (LE=0.1) gold particles can be observed along the membrane.

Some background information

Figure 8.3: Bjerrum plot of antibody-antigen binding. Small Kd :strong binding, larger Kd weaker binding

Molecules bind to each other, either very weak (no binding or even repulsion) or very tight, the quality of this binding is described by a dissociation constant or Kd. A Kd < 10-8 describes a strong bond between AB and antigen whereas a Kd > 10-3 means that the two are not tightly bound to each other. In the Bjerrum Plot in Figure 8.3 different theoretical plots are displayed.

The curve with the solid circles represents a genuine AB-antigen binding where, at a concentration of  5×10-8 mg/ml, 95% of all the available antigen molecules are bound to an AB. The other curves represent the binding of that same AB to other molecules.  At a concentration of 10-4 mg/ml every available antigen molecule is bound to an AB, but at the same time the AB binds to other molecules as well due to electrostatic or hydrophobic interactions: background.

The solution to the background problem is to dilute the AB to the appropriate concentration and block possible nonspecific binding sites with molecules that do not bind to the specific AB but do bind to molecules in the specimen.  A blocking agent should be selected depending on the nature of the nonspecific binding. When antigens are very abundant it is often quit obvious whether the correct AB concentration is used. However when that is not the case, statistics will give a clear answer whether the labeling is specific or not as recently described in a brief yet comprehensive tutorial by Mayhew (Mayhew 2005).

Figure 8.4: Whole mount immunolabeling.EM Picture of an Enterococcus adhered to a Formvar-Carbon coated grid. After fixation a surface antigen has been labeled with primary antibody and protein A-Gold. Final staining with MC/UA. Bar=200 nm

Immuno-labeling strategies
In general, 3 different strategies can be discerned: whole mount immuno-EM, Pre-embedding labeling and post-embedding labeling.
Whole mount immuno-EM is in fact the fastest method. Specimens are adhered to grids or cultured on grids, (mildly fixed) and labeled. When the labeling is performed without any detergents present in the incubation media it is most likely that only the outside of a specimen will be labeled. In this procedure the entire cell is present on the grid and later in the electron microscope. Therefore only rather thin, non-electron dense, specimens can be used in 60-120 KV electron microscopes.  In Figure 8.4 an Enterococcus has been immuno-labeled for a surface antigen. The same technique can be used to visualize antigens on viruses, blood platelets and the rims of flat cells.

Immuno-labeling of the interior of a specimen is possible when the specimens are treated with detergents like saponin or Triton X100. This will generate small pores in the surrounding membranes and (part of) the cytoplasm and making the antigens accessible.(van Dam and Stoorvogel 2002).

Figure 8.5: Pre-embedding labeling. Cells were permiabilized and immunolabeled for AP1and protein A-gold. Thereafter the cells were embedded in EPON sectioned and stained with UA and lead. Arrow: clathrin coat, bar=200 nm

Closely related to the previous technique is the pre-embedding technique. The specimens are immuno labeled before they are embedded in a resin and sectioned. This is a very useful technique when the distribution of surface antigens is studied. To access the interior of a specimen (cell) the plasma membrane has to be opened prior to the labeling procedure. This can be achieved by detergents or repeated freeze-thawing, which are rather mild treatments. Cells can also be opened with an osmotic shock or by mechanical forces like scraping them from their support. Even after such a harsh treatment, many organelles are still intact but the cytoplasm is no longer present. The absence of cytoplasm during labeling increases the availability of antigenic determinants of cytoplasmic tails from membrane associated molecules.

It has been used for instance to visualize molecules on synaptic vesicles (De Camilli, Harris et al. 1983). In Figure 8.5 one of the coat proteins (AP1) present at the Trans Golgi Network of HepG2 cells is labeled with 5 nm gold particles. In this particular case the cells were fixed wit 1.0% formaldehyde for 5 min and scraped from the Petri dish and embedded in 1% agarose. Small blocks were incubated overnight consecutively with antibodies, gold probes and afterwards embedded in EPON.

The last approach is post-embedding immuno-labeling. After the specimens have been supported by a resin, methacrylate or ice, the sections are cut and the labeling is performed on the section. With a thickness of ± 70 nm most organelles will be cut open and their interior is exposed to antibodies. It is obvious that the way in which the specimens were treated will influence the labeling efficiency.

Figure 8.6 : Immuno-double labeling of a Hela cell, which was transfected with lung surfactant protein C (SPC) . Labeling: anti SPC > protein A-gold 10 nm and  anti PDI (resident ER protein, mouse antibody) >rabbit anti mouse> protein A-gold 5 nm. Bar =200 nm.
Figure 8.6 : Immuno-double labeling of a Hela cell, which was transfected with lung surfactant protein C (SPC) . Labeling: anti SPC > protein A-gold 10 nm and anti PDI (resident ER protein, mouse antibody) >rabbit anti mouse> protein A-gold 5 nm. Bar =200 nm.

Not only the fixation will influence the number of available antigenic determinants, but also the chemicals in the embedding procedure will destroy or mask epitopes. The immune reaction takes place at the very surface of the section and the availability of epitopes is quite different for each embedding method. Sections from specimens embedded in an epoxy resin have a rather smooth surface which means that there are not many epitopes available. The surface of hydrophobic methacrylates (Lowicryl HM20, LR gold) resembles that of the epoxy resins, the more hydrophilic methacrylates (Lowicryl K4M, LR white) have a rougher surface. Thawed frozen sections usually will give the best LE because the antigenic determinants have not been exposed to organic solvents and embedding media. Moreover their surface is rough and antibodies can penetrate into the section (depending on the matrix density of the organelle). Unfortunately gold particles ranging from 5 to 20 nm hardly penetrate into a section due to their size and charge. (Stierhof and Schwarz 1989) This problem can be tackled(partially) by using ultrasmall gold particles ranging from 0.8 to 2 nm. Due to their small size, these particles are hardly visible in the electron microscope and need to be increased in size by means of silver or gold enhancement (Yi, Leunissen et al. 2001).
Immuno-double labeling

Due to the particulate nature and the well-defined sizes of gold markers it is possible to label more than 1 epitope on a specimen.(Figure 8.6 and Figure 8.7)  In theory the number of different epitopes that can be labelled is only limited by available gold particles. However, with an increasing number of immunoreagents the greater the chance of background labeling will be. In practice double labeling is frequently used and occasionally also a triple labeling.


Faulk, W.P. and G.M. Taylor, An immunocolloid method for the electron microscope. Immunochemistry, 1971. 8(11): p. 1081-3.
Slot, J.W. and H.J. Geuze, Sizing of protein A-colloidal gold probes for immunoelectron microscopy. J Cell Biol, 1981. 90(2): p. 533-6.
Mayhew, T., How to Count your Gold: A Tutorial on TEM Immunogold Label Quantification. Microscopy and Analisys, 2005. 19(2): p. 9-12.
van Dam, E.M. and W. Stoorvogel, Dynamin-dependent transferrin receptor recycling by endosome-derived clathrin-coated vesicles. Mol Biol Cell, 2002. 13(1): p. 169-82.
De Camilli, P., et al., Synapsin I (Protein I), a nerve terminal-specific phosphoprotein. II. Its specific association with synaptic vesicles demonstrated by immunocytochemistry in agarose-embedded synaptosomes. J Cell Biol, 1983. 96(5): p. 1355-73.
Slot, J. W. and H. J. Geuze (2007). “Cryosectioning and immunolabeling.” Nat Protoc 2(10): 2480-91.
Stierhof, Y.D. and H. Schwarz, Labeling properties of sucrose-infiltrated cryosections. Scanning Microsc Suppl, 1989. 3: p. 35-46.
Yi, H., et al., A novel procedure for pre-embedding double immunogold-silver labeling at the ultrastructural level. J Histochem Cytochem, 2001. 49(3): p. 279-84.