Technical Help: Successful Immunoelectron Microscopy Labeling
This page is based on a recent discussion on the MSA microscopy listserver. It includes a diverse range of equally valid opinions. The requirements for successful immunoelectron microscopy labeling vary widely with different systems, and no one approach will work for all experiments. However, by including the experiences of different researchers, we aim to present a range of different approaches to identifying and solving problems in electron microscopy immunolabeling, and to give you the best chance to find one that is relevant to your own project.
We are grateful to the Microscopy Society of America and all the participants for allowing us to excerpt their discussion here.
Questions and Issues
Many antibodies don't work for light microscopy, and even more don't work for TEM immunocytochemistry. What experiences and success stories have people had? How does antibody performance compare:
- Between different methods - ELISA, western blot or immunoprecipitation, compared with light microscopy and electron microscopy?
- Between monoclonals and polyclonal antibodies?
- How can I get an antibody which works for light microscopy to work for EM? What proportion of antibodies that work for light microscopy can I expect to work for immunoelectron microscopy?
- How do the requirements for successful immunoelectron microscopy vary between different tissues, antigens, and epitopes?
- What is required to determine a successful EM protocol?
- How is good light microscopy procedure (e.g. acetone whole-mount sections) translated to electron microscopy?
Introduction and general approaches
It is sometimes mistakenly assumed that immunoreagents are inert objects, and that their size is the only property to be taken into account. Immunoreagents are biological molecules rather than chemicals; they are charged molecules and have many complex, variable interactive and binding properties.
As a first step in immunoelectron microscopy method development, try the whole preparation and immunolabeling protocol at the LM level first. Use the same reagents and preparation protocols as you plan to use for EM, and expose to silver enhancement for a longer time so that you can see the signal by LM. If you see a signal, then you will be confident that you will see a signal when you go to examine the EM experiment.
Generally, we have found that if theres no signal by LM, the antibodies will never work when used with the same preparation protocols for EM. However, often minor modifications to these protocols can make the system work. General tips for successful EM labeling:
- Get to know the people who make the antibodies and find out as much as possible about the target that the antibodies label, and the biology of the system in which they will be used.
- Screen the antibodies by light microscopy on the system to be used for electron microscopy labeling. Thin sections of Lowicryl-embedded material make good light microscope preparations, and silver enhancement is often particularly good for visualizing protein A-gold.
As a general phenomenon, some antibodies work only for one method. Possible reasons include:
- Antibody binds antigen only in its natural conformation, not in western blots where protein is denatured, or vice versa.
- Strong fixing retains proteins around antigen, making it inaccessible.
- For antibodies to peptides: very few sequences on the target molecule will be recognized by the antibody, and antibody binding will depend upon the position and orientation of these sequences in the protein structure. Interior sequences may be inaccessible, or may be rendered so by fixation; exterior sequences that are part of a soluble component of a complex may be dislodged by washing. Because they recognize only a very small proportion of the target, sensitivity may be a problem as well as accessibility.
- Specificity is often greater for anti-peptide antibodies than for monoclonals, which in turn are more specific than polyclonals.
- Components of blocking agents can bind to the antibody. For example, protein A binds to IgG present in serum, resulting in inhibition of specific signal. Therefore, dilution of protein A in serum is not recommended!
In one very unusual case, autofluorescent organelles with a similar distribution to the target were mistaken for specific labeling in fluorescence labeling experiments; then, when the protocol was repeated for electron microscopy, no labeling was observed. The protein under study was a serum protein, and the blocking agent initially used to dilute the antibody was 10% serum; the antibody reacted with this molecule was removed from labeling solution.
Specimen and antibody treatment
- Cultured cells must be grown for long enough for sufficient antigen to be expressed. The time required will vary with the cells under study, so it is important to understand the system thoroughly. For example, some types of cells may need to be grown for 10 or 15 days, while others may express sufficient numbers of specific antigens in only a day or two.
- Storage issues: the freezing of antibodies, and especially their treatment after thawing, can affect performance. Freeze-thawing antibodies can cause aggregation, especially for IgM primaries which are large molecules with many branches; if the solution is centrifuged after thawing and there are many aggregates, especially big ones, the antibody can be effectively removed from solution. Therefore, it is worthwhile to check how the antibodies have been stored.
In LM, even low affinity antibodies can be detected, while at EM they may not. If only about 10% of an IgM primary antibody remains after centrifugation, antibody dilution has to be increased for EM detection; and if there has been any change in the antibody (such as aggregation), although it may still bind to antigens on your section, the secondary antibody may not recognize it. In LM, where there are more antigens in the field of view and amplification methods may be used, the signal may still be detected; however, by EM the signal will be absent.
- Streptavidin-gold is a tricky reagent: when used with silver enhancement, it frequently seems to detect biotin at the LM level, but gives no signal at the EM level. Reasons for this can include:
- Blocking agent: if serum is used, the gold probe will bind to biotin-like molecules in the serum.
- The probe is relatively unstable: colloidal gold separates from the protein after only short storage times. Free streptavidin in the solution may not be sufficient to inhibit labeling for LM, but removes almost all the EM label.
- Protein A-gold can have advantages over secondary antibodies. Since protein A-gold is used by everyone, turnover is faster and quality control and troubleshooting easier (However, it should be noted that protein A does not recognize all classes of IgG, so a bridging antibody may be required).
- Freunds adjuvant effects: Freund's adjuvant is a homogenate of ground-up bacterial proteins. Therefore, if the study involves looking at bacteria, some background labeling on bacteria is inevitable. If we know how the antibody was prepared, we can explain this result. If we do not know, we can try many different fixation and blocking solutions without success, only to attibute this to a poor antibody.
Different criteria for acceptable morphological preservation for LM and EM can result in differences in labeling. At the LM level, preservation of morphology may be less critical, so less stringent fixation conditions may be used; however, when these protocols are transferred to EM, more stringent fixation may be required for acceptable morphological preservation, which results in lower labeling. For example, freezing tissue without aldehyde fixation or cryoprotection, then fixing the sections with acetone or methanol and air-drying them before or after this treatment, is used for LM, but it can remove or relocate antigens on the EM scale (see: Hannah, Weiss & Huttner, "Differential extraction of proteins from paraformaldehyde-fixed cells: Lessons from synaptophysin and other membrane proteins," Methods (a companion to Meth. Enzymol.), 16, 170-181 (1988)).
The solution is to begin the development of an EM method by using the same fixation protocol used for LM. This way, labeling will be assured for EM. Then, the protocol may be adjusted until an acceptable compromise is reached between morphological preservation and labeling. The level of acceptance of poor morphology will depend on the result expected. Each system will have its own protocol for getting that particular result.
- If it is possible to label purified (or semi-purified) sub-cellular particles, this may be worthwhile. For example, mitochondria are highly recognizable, so this approach is acceptable for labeling antigens on the outside of mitochondria. If the antigen is inside the mitochondria, then sectioning may be best. If the profile that is labeled is so obviously a mitochondrion, then it doesn't matter if the cristae are not perfectly preserved (unless there is extraction or redistribution of antigen). If specific membrane markers are available, structural morphology is even less important: co-localizing the marker with the test antibody on a membrane or organelle fragment is often acceptable.
- Antigen accessibility can be increased simply by breaking open cells to let out the antigens we are not interested in (unless we are looking for cytoplasmic antigens). For example, one investigator labeled the lumen of the RER with a very specific antibody. He couldn't do this on well fixed, heathly cells because the antibody could not gain access to the luminal antigens. He noticed that some of the cells in the pellet had obviously died before being fixed, and had poor morphology. However, the RER was still recognizable, and densely labeled.
Antibody and specimen characteristics
- Many (monoclonal) antibodies provided today are very well suited for Westerns and sometimes LM because they were developed for the degenerative conditions which are applied in these techniques (up to 90% of the epitopes can be removed by embedding and sectioning).
- In small samples (a few millimeters in square with a section thickness of 50-70 nm) fixation and post-processing can lead to a loss of antigenicity.
- Very few antibodies are suitable for immuno-EM; antibodies specifically generated for immunoelectron microscopy, which recognize their epitope after aldehyde degeneration/fixation and acrylic plastic embedding, are not readily available commercially.
Tips for developing TEM immunolabeling protocols:
- Count the signals (specific signals may not be visible at X 20,000 or higher magnification).
- Test different antibodies and manufacturers.
Identifying the right fixation conditions
Immunoelectron microscopy is becoming important for resolving localized GFP transfected proteins which are insufficiently well resolved by confocal microscopy. In pre-embedding experiments with anti-green fluorescent protein (GFP) immunolabeling with Nanogold® and silver enhancement, it was helpful, before attempting immunoEM, to do a fixation series of formaldehyde and glutaraldehyde dilutions, and see where the fluorescent signal was lost. Fixation vs. penetration is always a compromise. Recommended reading: G. Griffiths, "Fine Structure Immunocytochemistry." Springer Verlag, Heidelberg & Berlin (1993).
It was also suggested that pre-embedding labeling with ultrasmall gold conjugates and silver enhancement has made it easier to translate LM success to the EM level. In some cases, the localization is only relevant when the cellular and tissue organization is intact, such as with brain tissue or in the study of cellular RNA/antigen/gene distribution, and the morphology must not be compromised beyond a certain point - which may be beyond the threshold that allows antigen survival or antibody access. However, ultrasmall gold conjugates and silver enhancement reagents can label antigens in fixed, wet tissue with excellent ultrastructure. It was pointed out that the initial fixation determines the starting point of ultrastructure quality (limited by antigen tolerance level), while the immunolabeling procedure itself is responsible for preventing deterioration of the ultrastructure. Use of Fab' conjugates allows less harsh permeablization of the tissue, and a near neutral pH silver enhancement reagent reduces the harm of pre-osmium enhancement on morphology.
Mixed results were described; one reply gave a success rate of 5 to 10 % for immunoelectron microscopy labeling (TEM thin sections) with antibodies which worked for fluorescence confocal microscopy. It was also suggested that investigators using fee-for service facilities may be unwilling to take the time necessary to develop an effective method for their project.
Questions and problems
For in situ hybridization detection (in the specific example cited, using digoxigenin-labeled RNA probes with mouse anti-digoxigenin antibody, followed by anti-mouse ultrasmall gold-conjugated secondary antibody and enhancement):
- Is a two-step detection method more sensitive than a single-step (gold-labeled anti-digoxigenin primary) method? One review attributes higher sensitivity for two-steps (J. Histochem. Cytochem., 45, 481-92 (1997)).
- For a two-step method, would gold-conjugated antibodies (against biotin probes) have advantages over streptavidin-gold?
- Will glutaraldehyde levels greater than 0.05% inhibit hybridization? For example, Punnonen et al. labeled rat glioma monolayers using 4% paraformaldehyde and 0.05% glutaraldehyde for 30 minutes (J. Histochem. Cytochem., 47, 99-112 (1999)), but protocol included 0.1% saponin permeabilization for 30 min.
ISH is another form of affinity localization labeling protocol; the probe is a nucleic acid sequence instead of an antibody, and much the same rules apply.
- Two step, sequential detection methods (primary label, followed by secondary with attached visualization probe) are usually more sensitive. This was first observed by Coons (generally considered to be the father of immunoctyochemistry) in the 1950s, and this sequential labeling method can be applied to almost any labeling system.
- If you are using antibodies to digoxigenin, then a secondary antibody-gold probe should be sufficient. If you choose to use Nanoprobes (Fab' conjugates), it might be difficult to later perform multiple labeling experiments. If you want to look for biotin labeled nucleic acid probes, it may be worthwhile to try anti-biotin antibodies, and cold water fish skin gelatin as a blocking agent rather than serum.
- Each system has its own unique qualities and sensitivity to fixatives. If it bothers you to include glutaraldehyde, then leave it out. Fix with a strong glutaraldehyde solution after hybridization and silver enhancement.
Associate Professor of Biological Sciences
Director, Molecular Cytology Core Facility
University of Missouri
E-mail: [email protected]
Web site: http://www.biotech.missouri.edu/mbp/cores/index.html.
Scientist II & Director
Ahmanson Advanced Electron Microscopy & Imaging Center
House Ear Institute, Los Angeles, CA
E-mail: [email protected]
Web site: http://www.hei.org/htm/aemi.htm.
Department of Anatomy I
University of Cologne
E-mail: [email protected], or: [email protected]
Web site: http://www.uni-koeln.de/med-fak/anatomie/einrichtungen/anatomie1
Stanford University School of Medicine, Stanford, CA
E-mail: [email protected]
Albert Einstein College of Medicine, Neuroscience Dept., Bronx, NY
E-mail: [email protected]
Any tips for other users? Things you'd like to see us make? Or, if you want to tell us your question or problem, E-mail us at [email protected].
2001 Nanoprobes, Incorporated.