Improved antibody validation

By Dr. Liji Thomas, MDOct 16 2019

Antibodies are used in a variety of clinical and research situations to examine the structure and function, and distribution of proteins. However, many researchers have found a lack of specificity in the functioning of these antibodies. This means that any work conducted using these antibodies will not yield reliable results.

A new paper published in the journal eLife on October 15, 2019, is based on a human protein called C9ORF72 that was tested using a variety of specific antibodies. The results provide a striking illustration of the antibody specificity problem, and the researchers also demonstrate a feasible solution to ensure the antibodies used by any laboratory can first be validated.

A new study points to the need for better antibody validation, and outlines a process that other labs can use to make sure the antibodies they work with function properly. Image Credit: The Neuro

Protein detection and antibodies

Proteins are the fundamental underpinnings of life, being large biomolecules that actually execute the information in an organism’s genes in the form of actual physical structure, function and controllers of physiological processes from cellular to whole-organism level. Mutations, or unplanned errors, in the genetic code, often cause abnormal proteins to be synthesized or the failure to synthesize the protein altogether, both of which can result in disease conditions.

Antibodies gain their importance because they are directed against specific proteins, helping scientists to detect them and localize them, as well as to quantify them. As a result, there are a number of antibody manufacturers that source over a million commercial antibodies to clinical and laboratory facilities. There are also other publicly funded recombinant or monoclonal antibodies. One of the largest public projects has come up with mostly polyclonal antibodies to a total of over 24 000 antibodies directed to about 17,000 human genes.

The problem is that many studies have found these antibodies to display nonspecific binding. This stems from the lack of strenuous standardization antibody characterization procedures, in contrast to the zeal to develop new antibodies. The underlying reason may well be the high expense of characterization, beyond the limit of cost-effectiveness for these companies.

As a result, all protein researchers are united in their demand for a way to determine antibody quality in a precise and reproducible way before using them in their experimental studies.

The problem

Earlier workers in this field have been aware of this problem and have suggested several different ways to ensure that properly validated antibodies are used. However, it is not clear which of these is best and should be used preferentially, to provide the most useful information on validation. Probably, a genetic test showing antibody-protein detection in a parental cell and loss of signal in a knockout cell is suitable for validating antibodies in all criteria, but no standardized template has yet been generated to enable its application in all settings.

The study and its outcomes

The C9ORF72 protein is one which is subject to mutations that cause the well-known paralyzing condition called amyotrophic lateral sclerosis (ALS), and the variant of dementia called frontotemporal dementia (FTD). One specific mutation has been found to be linked to over 11% and 23% of ALS and FTD respectively, in North America. The localization and function of this protein is still a matter of controversy, however, and this is probably due, at least partly, to the use of non-validated antibodies in various studies.

In the current study, the researchers took advantage of the newly available CRISPR gene editing technology to create knockout cells, that lack the protein against which the antibody to be tested is directed. After identifying cells in which the level of expression of the C9ORF72 gene was high, they created knockout cells using CRISPR.

They then tested a battery of 16 antibodies which were claimed by their manufacturers as being specific detectors of this protein. However, only 1/16 was able to do so using immunofluorescent staining, when the protein-antibody binding causes a fluorescent protein to light up so that the bound complexes can be easily and accurately visualized under a microscope.

When other techniques were used, the results were just as poor: only 2/16 antibodies successfully detected the presence of the protein. And of these, none of the antibodies that actually detected the C9ORF72 molecule have been used in research.

In contrast, all the ‘failed’ antibodies have been the basis of many protein detection studies – leading to questions about the reliability of the study results. These antibodies either do not recognize the C9ORF72 protein or also recognize other proteins that do not share structural similarity with it in the least. They are either not sensitive for the detection of C9ORF72 or else they are non-specific, detecting a wider range of proteins unrelated to C9ORF72.

This leads to unreliable reporting. Moreover, it explains why many studies have not been reproducible by other scientists. This shows the urgent need to validate antibodies used in research, as most of the earlier results may now be deemed useless in the face of the current study.  Researcher Peter McPherson says, “A large part of the reproducibility crisis is because of poor antibody validation. We owe it to funders and patients to do better.”

The new protocol and its implications

The present experiment puts forward a new approach which is inexpensive, highly standardized and convenient. It corrects several misconceptions about gene expression, emphasizes the need to use immunoblot procedures to identify cell lines that have the highest level of expression of the gene in question, and quality control standards for the various procedures used during the analysis.

The researchers also stress the need to test all antibodies under all possible conditions, before validating its use, since an antibody’s specificity and sensitivity will change with the extent of target binding, which in turn depends on the test conditions and type. For instance, one antibody used in this study is best used for immunoblot and immunohistochemistry, but not in immunoprecipitation or immunofluorescence.

Though initially focused on developing methods to validate antibodies used to detect the C9ORF72 protein, the scientists found themselves drawn to developing a universal antibody validation model for any antibody used in any laboratory, instead. They are confident that the new pipeline can be used for any commercial or home-made antibody, as a whole or a few steps in isolation.

In McPherson’s words, “The procedures we use are not revolutionary, and in fact this makes our approach widely applicable to any laboratory skilled in the art, yet to my knowledge this is one of the first papers to describe a streamlined process for antibody validation.”