Controls for Immunocytochemistry

Abstract

Immunocytochemistry is a highly productive method in biomedical research used to identify proteins and other macromolecules in tissues and cells. Control samples are required to show label localization is correct, but the understanding and use of immunocytochemistry controls have been inconsistent. A new classification of immunocytochemical controls is proposed that will help in understanding this most important component of the experiment. The three types of controls required for immunocytochemistry are primary antibody controls that show the specificity of the primary antibody binding to the antigen, secondary antibody controls that show the label is specific to the primary antibody, and label controls that show the labeling is the result of the label added and not the result of endogenous labeling. Publications containing immunocytochemical results must give details of how these controls were performed.

The use of antibodies to localize structures in cell cultures and tissue sections is an extremely powerful method and is responsible for many important discoveries. However, sometimes the results of immunocytochemical experiments are confusing or are inconsistent with results obtained with other methods. In these cases, it can be difficult for investigators to identify the problems and understand them. There are examples of experiments that, upon further investigation, have shown the original immunocytochemical results were incorrect (e.g., Pradidarcheep et al. 2008; Wolf et al. 2001).

In immunocytochemistry, colorful micrographs of multiple labels are so compelling that it is hard to even imagine that the information they contain could be wrong. However, scientists experienced with immunocytochemistry know that to trust a micrograph, you need controls that show the label is correct. The use of antibodies in the complex and heterogeneous environment of the cell sometimes gives unexpected binding not dependent on specific binding of the primary antibody to the correct antigen. A set of controls should show that all aspects of the incubation were done correctly and not leave questions about the reliability of the labeling.

Immunocytochemistry began in 1942 when Albert Coons and coworkers (Coons et al. 1942) reported use of a fluorescent-labeled antibody to localize pneumococcal antigen in liver sections. The use of controls evolved from the use of antibodies in other methods such as the enzyme-linked immunosorbent assay (ELISA) and, before it, radioimmunoassay (RIA). In their seminal study, Yalow and Berson (1960) described two RIA antibody controls. The first is the use of non-immune serum from the same species of animal as the primary antibody that eliminates the specific response, and the second is reacting the primary antibody with saturating concentrations of the antigen, insulin, as an absorption control. In fact, the groundbreaking experiment by Coons et al. (1942) uses these two controls. In the following years, the controls for immunocytochemistry were taken from the method of Coons et al. and other non-microscopic techniques that used antibodies to identify specific proteins.

It is important to understand interactions of the antibody with its antigen. The immunoglobulin G (IgG) is the most common antibody type used in immunocytochemistry. The antibody structure (Fig. 1) consists of the variable region (Fab portion) of the antibody that binds the epitope part of the antigen and the constant region (the Fc portion) that is specific to the animal where the antibody was raised. Thus, a rabbit anti-tubulin antibody was made in a rabbit and binds the protein tubulin, and it can, in turn, be bound on its constant region by an antibody to rabbit IgG. In this example, the antibody that binds the antigen of interest is referred to as the primary antibody, and the antibody that binds the Fc portion of the primary antibody is known as a secondary antibody.

Figure 1.

An immunoglobulin G (IgG) antibody is a single protein made from four peptides joined by disulfide bonds. There is a single constant region (white) containing the Fc portion and species-specific antigens. The variable region (gray) contains the Fab portion that binds the epitope portion of the antigen. The short protein found only in the variable region is known as the light chain; the large protein that is part of the constant and variable region is the heavy chain. The IgG can be digested by the protease enzyme, papain, at the hinge region (flexible region of the heavy chain), into an Fc end (constant end) and a Fab end (variable end). The antigen is the molecule used to immunize the animal, and the epitope is one of many portions of the antigen that can generate antibodies. Figure reproduced with permission from Burry (2010).

The literature contains a variety of different names for controls, and some controls have multiple names (Stirling 1993). One control in immunocytochemistry is the “negative control” or “procedural control,” where the secondary antibody binding is examined.

Additional problems with controls for immunocytochemistry are the wide range of methods in which antibodies and labels are used to localize proteins in cells. Indirect immunocytochemistry with an unlabeled primary antibody and a species-specific labeled secondary antibody is most popular. Immunocytochemistry for multiple primary antibodies needs controls that show each secondary antibody binds to the correct primary antibody. Avidin-biotin complex (ABC) uses a third incubation after the secondary antibody to add the enzyme horseradish peroxidase (HRP) and is also very popular because of its high sensitivity and relative ease to perform. Today, with many new methods of using antibodies in immunocytochemistry, a new approach is needed to classify and use controls.

The controls described below are applicable to immunocytochemistry regardless of the labeling method used. Labeling methods used in immunocytochemistry can be grouped in three types: fluorescence, enzymes, and particulate. Fluorescence relies on light emitted from a fluorophore with a different wavelength than the light used to excite the fluorophore. Histological enzymes are proteins that convert an uncolored water-soluble substrate to a colored water-insoluble product, with HRP a commonly used enzyme. Particulate labels are heavy metals such as colloidal gold or small gold particles that are used mainly for electron microscopic immunocytochemistry.

To simplify the use and discussion of immunocytochemistry controls, three types of controls were proposed (Burry 2010). This organization of the controls is based on the function of the individual control and its use in the immunocytochemical procedure. The new terms for the controls are designed to better describe their function in the immunocytochemical procedure. With a better understanding of the function of the controls and the new terms, it is hoped that the use of controls will become more consistent.

Immunocytochemistry Controls

• Primary antibody controls: Show the specificity of primary antibody binding to the antigen

• Secondary antibody controls: Show the label is specific to the primary antibody

• Label controls: Show the labeling is the result of label added during the procedure and not endogenous labeling or reaction products

Primary Antibody Controls

The primary antibody control is a specificity control and confirms that the primary antibody binds to the correct epitope on the expected antigen. Ideally, this control demonstrates binding specificity under the same conditions of the fixed cells or tissue section. Thus, the primary antibody control is not just showing the antibody is specific for the antigen, but it also shows the effects of fixation and detergent treatment on the tissue or cells. A broad range of methods have been used to show specificity of the primary antibody for the antigen. The lack of a single widely accepted method for determining the specificity of the primary antibody confirms the difficulty of demonstrating this point. Four methods are discussed as primary antibody controls.

The best primary antibody control is to use a genetic approach to manipulate the expression of the antigen protein. Within the genetic approaches, the first method is to use tissue from a knockout animal (Saper and Sawchenko 2003). This approach removes the protein and allows the tissue to be prepared in the same manner as the experimental sample. With a nonfunctional gene, the protein of interest is not expressed, and the primary antibody should not bind to the tissue. Tissue from these animals can be fixed and prepared using the same procedure as the experimental animals to minimize experimental conditions affecting the result of the control.

However, there are problems with knockout as a primary antibody control. Not only are knockout animals for specific proteins not common, but even if one is found, it may be difficult to get access. Some knockout animals are functional knockouts that still express the protein, but the protein is mutated or non-functional (Lorincz and Nusser 2008). In these cases, the primary antibody to the protein could still bind to the defective protein in the knockout animal.

The second genetic method used for a primary antibody control is the transfected cell line expressing the antigen for the primary antibody (Saper 2009). This control includes the conditions of fixation and shows the primary antibody works with the method used. The untransfected cells serve as a negative control because they do not express the protein. Also, using siRNA to knock down the expression of the antigen protein can be used (M. C. Willingham, personal communication, 2010).

A third genetic method used as a primary antibody control is the combination of immunocytochemistry and fluorescent in situ hybridization (Rhodes and Trimmer 2006). The concept is that synthesis of a protein from mRNA should be close to the site where the antibody detects the protein. One drawback to this approach is that the methods for preserving tissue for immunocytochemistry and in situ hybridization are not fully compatible, leading to less than optimal signal. This problem is combined with the fact that mRNA for some proteins in developing cells is silenced with RNA binding proteins, and the mRNA is transported within the cell before translatiion.

The second primary antibody control method is immunoblot (Western blot), a very reliable and most common method of determining the specificity of the primary antibody. In this case, the antibody labels one protein at the correct molecular weight. These immunoblots are relatively easy to do and thus are the most common form of specificity control seen for commercial antibodies. This approach is relatively inexpensive and straightforward. The problem with immunoblots is that the protein is not fixed but is denatured in SDS, and so they lose their secondary and tertiary structure. Some antibodies bind only to denatured protein immunoblots, and some antibodies bind only to native proteins in immunocytochemistry (Willingham 1999). When obtaining an antibody, it is advisable to consider only those that have been tested or used for immunocytochemistry.

A third method for primary antibody controls is colocalization with the original primary antibody and an additional label to show that they both bind to the same structure. In this case, two different primary antibodies to different epitopes on the same antigen (frequently to different epitopes, such as the C- and N-termini of the same protein) give the same label site to confirm the primary antibody is specific for a protein. Another way to perform double labeling is to take advantage of a fluorescent protein as part of a specific protein to show double labeling with immunocytochemistry. Here, the transfected cells with a fluorescent protein (e.g., green fluorescent protein [GFP]) can be processed for fluorescent immunocytochemistry, and the primary antibody should show colocalization with the GFP. The general problem with colocalization is that although it shows that the labels are in the same spot, it does not show that they bind to the same protein because the resolution of the light microscope is not enough to resolve the location of a single protein.

The fourth method for primary antibody controls is absorption controls. These are derived from the early use of antibodies with RIA and ELISA where the primary antibody is mixed with the purified antigen in a tube. The absorbed antibody can no longer bind to antigens in the section (Fig. 2A; Yalow and Berson 1960; Petrusz et al. 1976; Saper and Sawchenko 2003). The primary antibody’s function is lost when it binds the antigen prior to incubation with the cells or tissue. To perform this control, correctly purified antigen or peptide antigen and not a crude homogenate is critical. The next step is to show that no unbound primary antibodies are present after the absorption. In a separate experiment, a titration curve is done to determine the concentration of the isolated antigen that saturates the antibody, leaving no unbound primary antibodies.

Figure 2.

Absorption control is the incubation of the primary antibody with the antigen used to generate the antibody. (A) The primary antibody incubated with excess antigens binds all of the Fab sites capable of binding the antigen in the tissue (arrow). (B) If the correct antigen and an incorrect antigen have the same epitope (arrows), then binding to both is inhibited by the absorption control. (C) In some cases, the antigen adsorbed by the antibody binds to proteins in the tissue, and the adsorbed antibody appears to bind to a protein (#4) independently of the antibody. Figure reproduced with permission from Burry (2010).

Two major problems with absorption controls limit its usefulness (Burry 2000). First, if a specific primary antibody binds to the same epitope on the protein of interest and a second protein, absorption by one protein in the absorption incubation inhibits binding of the antibody to all potential proteins (Fig. 2B). The absorption control does not exclude the binding of the primary antibody to proteins other than the one used for incubation. This would be a false negative, indicating the antibody is specific for one antigen. Examples of such false negatives occur when the epitopes are shared on several antigens and bind to all antigens by absorption (e.g., Swaab et al. 1977). To ensure a negative absorption control is correct, an immunoblot must show that the epitope is found on a single protein (e.g., Wolf et al. 2001). Thus, the absorption control needs an additional control to eliminate the possibility that the primary antibody is bound to multiple antigens.

A second problem with absorption controls is that the proteins, even when bound to the primary antibody, can retain the ability to bind other proteins in cells. The bound protein could then bind to the tissue independent of the antibody and give a false-positive labeling (Fig. 2C). One solution is to bind the antigen to beads and remove the adsorbed antibody from the solution (Storm-Mathisen and Ottersen 1990). However, the best absorption control is the use of small peptides for antigens because only the epitope of the antigen is used for absorption.

Recommendations for primary antibody controls:

• Best choice is tissue from a knockout animal used with the primary antibody.

• Most commonly used controls are immunoblots with the tissue of interest to determine whether the primary antibody can bind to a single protein of the correct molecular weight.

• A good choice is immunocytochemistry with the primary antibody and cells labeled for GFP or with a second primary antibody to the same antigen.

• Absorption controls should be used with caution and combined with other controls.

• Ideally, use several methods to show the specificity of the primary antibody.

Secondary Antibody Controls

The secondary antibody control shows that the labeling observed is due only to binding of the secondary antibody to the primary antibody. This control is done by either eliminating the primary antibody or replacing it with the same amount of normal serum from the same species. With no primary antibody to bind the secondary antibody, no labeling should be seen. The secondary antibody control needs to be run in parallel with each experiment.

There are several types of problems detected with this control: nonspecific binding of the secondary antibody, unique binding of the antibody to Fc receptors, binding of secondary antibodies to same-species primary antibodies, and interactions between antibodies in samples incubated with multiple primary antibodies (Fig. 3A).

Figure 3.

(A) Incubation with antibodies should show antibody binding to only the correct antigen. Nonspecific binding can result from charged groups that bind proteins, including antibodies. The tissue can have Fc receptors that bind to the Fc region of any antibody. Some tissues can have exposed endogenous antibodies. In experiments with multiple primary antibodies, incorrect binding by other antibodies can occur. (B) There are blocking agents that block each of the sites that cause nonspecific binding. The primary antibody binds to the correct antigen and is not affected by any of the blocking agents. Charged groups can be quenched by any protein, and BSA is commonly used because it is not a source of antibody binding. The Fc receptors must be quenched by the Fc end of an IgG antibody that has no ability to bind other antigens; normal serum from the same species as the secondary antibody is commonly used. Endogenous antibodies are blocked by incubations with antispecies Fab fragments that are used only when tissue from injured animals is processed. In experiments with multiple primary antibodies, where incorrect binding of labeled secondary antibodies is suspected, the secondary antibody control shows this incorrect binding. Reproduced with permission from Burry (2010).

Binding of secondary antibodies to charged groups occurs in fixed tissue sections and cell cultures (Fig. 3A). Charged groups in cells are generally positively charged and bind proteins with negative charged groups. The charged groups are unbound aldehydes from the fixative or cell components such as histones. The solution to this nonspecific binding is to block the charged groups with a protein containing no important antigens (e.g., bovine serum albumin [BSA]).

Binding of the secondary antibody to an Fc receptor can occur (Fig. 3A). The Fc receptors bind antibodies at their Fc portion and are found on immune cells such as macrophages and natural killer cells. Nonspecific binding to Fc receptors is observed most often at injury sites or areas of inflammation where high levels of immune cells accumulate. The use of normal serum (serum from an animal not immunized) from the same species as the secondary antibody inhibits this binding (Fig. 3B).

Although rare, secondary antibodies can bind endogenous antibodies at sites of inflammation or when systemic inflammation occurs (Fig. 3A). This is the problem of using a mouse primary antibody on mouse tissue or the “mouse-on-mouse” problem. Here the secondary anti-mouse antibody binds to naturally occurring mouse antibodies in the inflamed mouse tissue. If endogenous antibodies are suspected to be present, the block is made by the Fab portion from an antibody raised to the species of the animal used as the secondary antibody (Fig. 3B).

When using multiple primary antibodies on a single sample, the secondary antibody control also detects cross-reactivity between the antibodies (Fig. 3A). In a single primary antibody experiment, the control examines only nonspecific binding of a single secondary antibody. With multiple primary antibodies, the secondary antibody control also determines if the secondary antibody is binding to the correct primary antibody and whether the secondary antibodies are binding incorrectly to each other. Binding of one secondary antibody to multiple secondary antibodies appears as colocalization but is incorrect. The control for multiple primary antibodies requires that a sample is processed with the omission of each primary antibody, but the subsequent incubation must contain all of the secondary antibodies (Fig. 3B). An example is shown in Table 1 of two primary antibodies and the expected labeling. If binding of any secondary is incorrect, these controls show labeling in the absence of the primary antibody.

Table 1.

Controls for Multiple Primary Antibodies

Recommendations for secondary antibody controls:

• Perform secondary antibody controls for each experiment or labeling.

• Secondary antibody controls for multiple primary antibodies must be designed for each experiment.

Labeling Controls

Labeling controls or detection controls help identify the contribution of endogenous fluorescence or enzyme to the observed label; even though it seems unlikely, the label (fluorescence or enzyme) can occur in cells endogenously. To perform a labeling control, include a sample of tissue sections or cultured cells that is incubated in all of the buffers and detergents used in the experiment but no antibodies, enzymes, or dyes. These controls should be performed with each new sample or fixation or change in incubation condition.

Autofluorescence is fluorescence observed in labeling controls (Billinton and Knight 2001). An important characteristic of autofluorescence is that it has a broad range of wavelengths for both excitation and emission, which makes it difficult to work around when using other standard fluorophores. The appearance of autofluorescence in images has one of two general patterns: (1) irregular or particulate autofluorescence (elastin, lipofuscin, NADH, flavins, chlorophyll, hemoglobin, etc.) or (2) uniformly diffuse autofluorescence caused by aldehydes in formalin-fixed tissue and glutaraldehyde-fixed tissue (not a problem with paraformaldehyde fixation). For evaluating the labeling in control slides, it is important to use the wide-field fluorescent or confocal microscope with the intensity setting and exposure times that are the same as those used for examining a correctly labeled sample. It is possible by increasing the sensitivity of the microscope to give an inappropriate fluorescence with a sample that contains no fluorescent label.

With the increasing use of transgenic animals expressing fluorescent proteins, errors in animal handling have resulted in animals that are mislabeled as untransfected or “clean” but actually contain a fluorescent protein. Transgenic animals are most commonly screened by PCR for a specific construct, but if the construct is not known, then the probe for the PCR must be to a common region for the fluorescent protein gene to rule out this possibility. This issue gets very complex when animals transfected with a single construct are bred to give animals with multiple constructs and are used for immunocytochemistry. When erroneous fluorescent labeling in a sample is suspected to be a fluorescent protein, perform immunocytochemistry with a primary antibody to the fluorescent protein and a secondary antibody with a different fluorescent emission.

Endogenous histochemical enzyme activity is detected by incubations without the antibody-bound enzyme. The appropriate control is to include a sample incubated with the substrate reagents but without enzyme-labeled antibodies to show the existence of endogenous peroxidase activity. This histochemical enzyme activity can be controlled by blocking the endogenous enzyme. For example, endogenous peroxidase activity is detected due to red blood cells or even white blood cells that remain in the tissue. Another cause of endogenous peroxidase activity is membrane-bound organelles of the lysosome family, which are found particularly in phagocytic cells at injury sites.

Recommendations for labeling controls:

• For each new tissue, cell culture, fixative, detergent, or a change in the combination of these, examine a section with the same microscope and settings used for the experimental sample.

Conclusions

When planning an experiment, all three of the control types (primary, secondary, and labeling) must be included. The primary antibody control is included for each new antibody and does not need to be repeated for each experiment. The secondary antibody control is designed for and included with each experiment. The labeling control is included if a condition of the procedure is changed, the sample is changed, or when unexpected labeling is found. A strict set of controls for immunocytochemistry is essential to point out problems with the experimental protocol and assist in their solution.

All three types of controls confirm that the labeling in the experiment can be trusted. When preparing a manuscript, it is critical that the controls used are listed in the methods section to demonstrate to the reviewers and readers that the authors were aware of the controls and that they performed them. Not only must these controls convince the scientist doing the experiments, but they also must convince the reader that the results are correct.