In the current age of biomedical research, antibodies are essential tools for research and diagnostic medicine. However, science is all about reproducibility, and some alarming studies have shown that there are real problems in the commercial antibody market. This issue, dubbed the ‘reproducibility crisis’, is widespread, with most researchers preferring fast and cheap products discovered based more on optimized Google searches than quality, with disastrous and expensive consequences. There are good solutions, but they require diligence and selectivity in purchasing and validating antibodies.
A Little Background
Thirty years ago, there was a choice between polyclonal and monoclonal antibodies. Polyclonal antibodies are produced by immunizing a host animal with an antigen, then collecting and purifying serum from the host after several antigen boosts. The serum contains all the antibodies in the blood of the host, including non-specific antibodies in the serum before the immunization. Since polyclonal antibodies cannot be standardized, their use in today’s research and medical applications is limited.
Monoclonal antibodies, as the name implies, are clones of a single cell. They are much more specific, in that they can be designed to target a single epitope. Monoclonal antibodies are produced by fusing a B-lymphocyte cell with a myeloma cell, producing a “hybridoma” cell line that secret antibodies into the growth medium. Unfortunately, these hybridoma cells can sometimes slough off their antibody genes, or they can stop growing or die. Thus, traditionally produced monoclonal antibodies can vary from batch to batch as well.
A 2015 feature in the prestigious journal, “Nature”, summarized what has been called “The reproducibility crisis” in antibody production. This crisis is estimated to cost the biomedical research community $350 million per year in the U.S. alone (Bradbury and Plückthun, 2015). For example, in 2006, David Rimm, a Pathologist at Yale University developed an antibody-based diagnostic test for Melanoma that stained tumor cells in biopsies. Supported by over $2M in funding, the test looked promising. Three years later, when Rimm ordered fresh antibodies from the same companies, his test didn’t work. Given this enormous loss in time, money, and clinical promise, Rimm began speaking out about the problems with antibody validation. He was not alone. A 2008 study, involved with the ambitious Human Protein Atlas found that fewer than half of the commercially available antibodies uniquely recognized their specified targets, with much of the variability correlated with the manufacturer.
The problems can have multiple causes. Rimm’s problem was batch to batch variation. No diagnostic test is reliable if different batches of supposedly the same antibody give unpredictable results. Another problem is when antibodies recognize more than the proteins they are sold to detect (non-specific). In 2009, Michel and colleagues published an investigation into G-protein-coupled receptors (GPCRs) that revealed that in a test of 49 commercially available antibodies, most bound to more than one protein. This non-specificity is especially troublesome, because GPCRs are cell-signaling proteins that are routinely targeted by drugs to treat various disorders.
Worse, sometimes antibodies marketed to detect one protein actually recognize another. Prassas and Diamandis, in a 2014 editorial, recounted their experience with an ELISA test that recognized an entirely different protein than that which it was sold to detect, costing them 2 years of work, and half a million dollars. They caution against a growing problem of “knock off” reagents produced in China, India, and Brazil are jeopardizing the integrity of science, as reported by David Cryanoski in 2017.
Glenn Begley and Lee Ellis published an alarming critique, also in “Nature“, of the way scientists are rewarded with citations for work that drug companies find impossible to reproduce. The hottest new cancer treatment target cancer cells with antibodies, so if a “landmark” study, published in a high-impact journal works, you can bet that drug companies would quickly develop it, for the public good, as well as their own profit margin. But they don’t, because they cannot reproduce the results for a wide variety of reasons, including antibody non-reproducibility.
As Rimm and others set out to raise awareness of the reproducibility crisis, they found that it was an uphill battle. The problem has two root causes. Vendors producing antibodies will continue selling products that sell. That is what businesses do if they wish to stay in business, and by 2011, the antibody production market had over 360 companies, worth $1.6 billion. And medical researchers are busy people, who trust that when they purchase an antibody labeled to detect a particular protein, it will work. Rimm states a natural expectation that “As a pathologist, I wasn’t trained that you had to validate antibodies; I was just trained that you ordered them.” However, it is clear that some researchers are taking this issue seriously, and if the scientific community pushes for reliable antibodies, vendors will respond accordingly.
A comment, published in “Nature” in 2015 by Andrew Bradbury and Andreas Plückthun, with 110 cosignatories, called for drastic changes in the production of antibodies, including defining protein-binding reagents by their sequences, and producing them through recombinant technology. Rimm too has called for six labor intensive practices that researchers should follow in order to validate their antibodies.