How to put an end to the reproducibility crisis
Poorly characterized antibodies waste researchers’ time and resources, and lead to low-quality science. Biophysical antibody ‘fingerprinting’ can renew confidence in these essential tools.
This article was originally published in Nature in May 2024.
Antibodies are both blockbuster biopharmaceuticals and some of the most versatile reagents in the laboratory toolkit. But many antibody lines are not consistent or do not work as described.
“Antibodies have incredible promise as drug-like molecules and fantastic tools, but they are also at the very root of the reproducibility crisis in biomedical science,” says Anita Bandrowski, neuroscientist at the University of California San Diego, who was a member of the ad hoc International Working Group for Antibody Validation (IWGAV).
“These are incredibly finicky research reagents, with considerable lot-to-lot variability, and their authenticity is difficult to track and validate.”
“Antibodies have incredible promise as drug-like molecules and fantastic tools, but they are also at the very root of the reproducibility crisis in biomedical science.”
Anita Bandrowski, neuroscientist, University of California San Diego
A pervasive problem
The fallibility of antibodies is not news; there are widespread reports of researchers failing to replicate or having to retract findings because of poorly characterized antibodies or differences between batches1.
These headline cases are symptomatic of a much more pervasive and pernicious problem: researchers are losing confidence in antibodies.
It’s a familiar headache for Will Howat, vice-president of validation and technical quality at Abcam, a global producer of research reagents. As a former molecular pathology team leader at a pharmaceutical company, he recalls having to modify assays every time he got a new antibody batch. “I’d have an assay set up, and have titrated the antibody in my system across different tissue types, only to find it would no longer work the same with the next batch of antibody. So, research stops while you titrate again. It’s a lot of work.”
What adds to this problem is a lackadaisical approach to antibody reporting across the scientific literature, something Bandrowski is working hard to change.
“In order to define good antibodies, or even bad antibodies, one needs to first identify them,” says Bandrowski. “The unfortunate problem is that authors do not identify the antibodies they use an estimated 20% to 50% of the time.”
Initiatives such as the Antibody Registry2 and the Research Resource Identification Initiative3 have already markedly improved antibody identifiability. However, if those antibodies are not properly characterized, results may still be questionable.
Taking an antibody's fingerprint
Using recombinant antibodies, which are produced from a specific genetic sequence that does not vary over time, removes the variability inherent with hybridoma-produced antibodies, but quality control (QC) is still essential to ensure no natural product variation creeps in.
Antibodies produced for therapeutic purposes are subjected to rigorous biophysical QC methods, including liquid chromatography-mass spectrometry, dynamic light scattering and high-performance liquid chromatography, to ensure they meet regulators’ good manufacturing practices for biopharmaceuticals. But there are no agreed QC standards for research-use only (RUO) antibodies.
“It’s critical to know that you’ve got the confidence of a pure antibody product and a reproducible supply”
Will Howat, Vice-President of Validation and Technical Quality, Abcam
The IWGAV recommended application testing and knock-out validation as key pillars for the validation of RUO antibodies4, which Abcam has been doing for many years. The company has now raised the bar even higher by introducing stringent biophysical QC standards across its catalogue of recombinant rabbit monoclonal RUO antibodies, with 75% tested to date.
“Biophysical characterization is not about how the antibody works in an assay, it’s about its identity,” says Howat. “By confirming its mass, its aggregation, its purity — all of which are unique to a specific antibody — we can pull those features together into an antibody fingerprint, so you know the reagent is exactly what it should be.”
Abcam is taking on these additional tests in order to help overcome the reproducibility crisis. “Whether you’re a pharmaceutical scientist who needs to seamlessly move from discovery to development, or a facility manager concerned by consistency of product quality and performance, it’s critical to know that you’ve got the confidence of a pure antibody product and a reproducible supply,” says Howat.
For all antibody research, starting with identified, quality antibodies is paramount, according to Bandrowski, and should always be followed by confirmation that the antibody performs as intended. “If you’re not validating your reagents, you can’t be confident in what they’re telling you,” she warns. “And if you publish those data, you can’t be confident in what you’re contributing to the scientific record."
References
1. Baker, M. Nature 521, 274‒276 (2015).
2. Bandrowski, A. et al. Nucleic Acids Res. 51, D358‒D367 (2023).
3. Bandrowski, A. et al. J. Comp. Neurol. 524, 8‒22 (2016).
4. Uhlen, M. et al. Nature Methods 13, 823‒827 (2016).