Basic Neuroscience
Antibody testing for brain immunohistochemistry: Brain immunolabeling for the cannabinoid CB2 receptor

https://doi.org/10.1016/j.jneumeth.2013.03.021Get rights and content

Highlights

  • Three antibodies for the CB2 receptor, which have been previously used to describe immunolabeling in the uninjured brain, were tested.

  • None of the antibodies were confirmed using the knockout control test.

  • Confirmation of these antibodies using other tests indicated that those tests cannot be used to validate the antibodies.

  • Claims of widespread CB2 expression in neurons in the uninjured brain are thereby cast into doubt.

Abstract

The question of whether cannabinoid CB2 receptors are expressed on neurons in the brain and under what circumstances they are expressed is controversial in cannabinoid neuropharmacology. While some studies have reported that CB2 receptors are not detectable on neurons under normal circumstances, other studies have reported abundant neuronal expression. One reason for these apparent discrepancies is the reliance on incompletely validated CB2 receptor antibodies and immunohistochemical procedures. In this study, we demonstrate some of the methodological problems encountered using three different commercial CB2 receptor antibodies. We show that (1) the commonly used antibodies that were confirmed by many of the tests used for antibody validation still failed when examined using the knockout control test; (2) the coherence between the labeling patterns provided by two antibodies for the same protein at different epitopes may be misleading and must be validated using both low- and high-magnification microscopy; and (3) although CB2 receptor antibodies may label neurons in the brain, the protein that the antibodies are labeling is not necessarily CB2. These results showed that great caution needs to be exercised when interpreting the results of brain immunohistochemistry using CB2 receptor antibodies and that, in general, none of the tests for antibody validity that have been proposed, apart from the knockout control test, are reliable.

Introduction

At its best, immunohistochemistry provides some of the most detailed and informative images of protein expression currently available: high-affinity and highly specific binding by antibody probes can map the spatial distribution of receptors, peptides, and enzymes to a micrometer resolution. However, the power of immunohistochemistry has at times been such that striking imagery has deflected attention away from the issue of antibody validity. Thus, the problems associated with poor antibody quality control have led Rhodes and Trimmer (Rhodes and Trimmer, 2006) to describe antibodies as being potentially “reagents of mass distraction.”

A number of criteria have been proposed to test for antibody specificity (Bussolati and Leonardo, 2008, Lorincz and Nusser, 2008, Moser et al., 2007, Rhodes and Trimmer, 2006). Unfortunately, these proposals are often ignored. First, the concentrations of primary and secondary antibodies, incubation times and other variables should be titrated and optimized to reduce labeling that is not specific to the primary antibody. This includes no-primary controls where the primary antibody concentration is zero. Ideally, the antibody should generate consistent results between Western blotting and immunohistochemistry analyses, with immunoblotting identifying a protein of the correct molecular weight. However, this is not always the case because the differences in the antigenicity for primary protein structures (for Western blot) and tertiary structures (immunohistochemistry) indicate that an antibody may be validated for one use but not the other. Third, labeling should be biologically plausible. For instance, antibodies for membrane-integral proteins should show membrane-restricted labeling. In addition, the labeling of restricted populations of distinct cellular phenotypes more reliably indicates specific labeling compared to ubiquitous cellular labeling (note that this is only a rule of thumb, as some proteins are expressed by all cells). Fourth, confidence is considerably increased if two antibodies raised against different epitopes of the same protein provide the same immunolabeling pattern; it is argued that a coincidental match for the labeling of two such antibodies is extremely unlikely unless they are labeling the same protein. Fifth, immunolabeling should not contradict the results generated by other independent techniques, such as autoradiography, mRNA in situ hybridization, and in situ functional assays. Sixth, high-affinity binding by the primary antibody should be discriminated from low-affinity binding via the use of preincubation with an immunizing peptide. Seventh, immunolabeling should be abolished in tissue generated from animals lacking the gene for the protein of interest, i.e., the so-called knockout (KO) control.

Although the knockout control is potentially the most powerful test, given that (a) nearly all transgenic animals at present are mice and (b) many proteins have evolved considerably within mammals, such controls are only appropriate if the epitope has been conserved and the same immunolabeling pattern has been generated in both mice and the species of interest. Although it is possible that developmental compensation could dramatically reduce the expression of a cross reactive protein following the deletion of the gene for the target protein and although this possibility is increased when the cross-reactive protein is physiologically related to the protein of interest, it seems likely that this is a rare enough occurrence where mice are an appropriate model system for the species of interest. The knockout control should thus be considered the gold standard for antibody validation.

The CB2 receptor was discovered in 1993 by Munro et al. (Munro et al., 1993) in a study that showed that CB2 mRNA was present in cells of the spleen but that it was not detectable in the brain. Later studies confirmed the presence of CB2 in cells of the immune system, such as macrophages, natural killer cells, monocytes, neutrophils, and B- and T-cells (Fernandez-Ruiz et al., 2007). Autoradiography with the high affinity CB1/CB2 agonist CP55,940 showed a distinct pattern of labeling in wildtype mouse brains and a complete absence of detectable CP55,940 binding in CB1 KO mouse brains (Ibrahim et al., 2003, Zimmer et al., 1999), although labeling was clearly present in the CB1 KO spleen (Zimmer et al., 1999). In a complementary fashion, another study used this method and found strong binding in wildtype (WT), but not in CB2 KO mouse spleen (Buckley et al., 2000).

However, in contrast to these findings, immunohistochemistry has been used to describe widespread labeling for CB2 in the uninjured brain (Gong et al., 2006, Suarez et al., 2008, Suarez et al., 2009). There is some evidence for potential functional effects for CB2 receptors in several brain subregions (Morgan et al., 2009, Van Sickle et al., 2005, Xi et al., 2011), but these hypotheses for the highly circumscribed functional effects of CB2 receptors in the brain are distinct from the hypothesis supported by the immunohistochemistry, e.g., that CB2 receptors were widely expressed in neurons in the brain, which is a controversial finding (Atwood and Mackie, 2010). In this article, we propose that this controversy is largely due to a lack of specificity of the available CB2 receptor antibodies.

To the best of our knowledge, unlike the CB1 receptor antibodies (Grimsey et al., 2008), no published study has specifically investigated the specificity of commercially available CB2 receptor antibodies. When Gong et al. (2006) reported immunolabeling for CB2 in the brain using immunohistochemistry, they used CB2 receptor primary antibodies from three different commercial companies: Cayman, Sigma, and Santa Cruz Biotechnology. Although the Cayman antibody was used to map the immunolabeling through the brain, the Sigma antibody was only used to show staining in the hippocampus, and the Santa Cruz antibody was used to show a restricted amount of staining in the brainstem in WT, but not CB2 KO, mice. The Cayman antibody, which is of central importance to the study, was not tested in the CB2 KO mice. In this study, we re-examined the CB2 immunolabeling in the uninjured brain by applying the criteria for antibody specificity discussed above.

Section snippets

Materials and methods

All of the previously described experiments were approved by the University of Otago Animal Ethics Committee or the Institutional Animal Care and Use Committee of Indiana University, Bloomington in USA.

Results

Because the C-terminus-directed Santa Cruz CB2 antibody was the only antibody in the study used by Gong et al. (2006) and was tested against sections obtained from CB2 KO mice, we initially reinvestigated the immunolabeling experiments performed in the rat brain using this primary antibody. We found strong antibody-specific labeling in the hippocampus. We took care to optimize the primary and secondary antibody concentrations as well as the incubation times (Table 1) such that the staining was

Discussion

Our results clearly demonstrate that many of the methods that have been put forward as controls for antibody specificity are inadequate. In this study, CB2 antibodies were confirmed in validation tests, including the optimization of the primary antibody's concentrations and incubation times, no-primary and blocking peptide controls, band-labeling in Western blot and discrete labeling of distinct cell types and subcellular regions in immunochemistry. Moreover, at low magnification, very similar

Conclusion

No currently commercially available CB2 receptor antibodies have been validated against convincing negative controls. Furthermore, caution needs to be exercised in using the currently available polyclonal CB2 receptor antibodies because in addition to the problems discussed above, the specificity of these antibodies may vary between batches depending on the bleed and individual animals being immunized during antibody production. Furthermore, the precise conditions for the immunohistochemical

Acknowledgements

J.H.B. was supported by a University of Otago Ph.D. Scholarship. Y.Z. was supported by a Sir Charles Hercus HRC Senior Research Fellowship. This research was supported by grants obtained from the New Zealand Neurological Foundation and Marsden Research Fund. The CB2 KO and WT brains used in this study were kindly gifted by Professor Ken Mackie of the University of Indiana.

References (22)

  • N. Hajos et al.

    Cannabinoids inhibit hippocampal GABAergic transmission and network oscillations

    Eur J Neurosci

    (2000)
  • Cited by (0)

    View full text