How many antibodies do you have to trial to find just one that is specific?

Investigation of the activity of the central nervous system involves both the excitatory and inhibitory signals that are coordinated to maintain homeostasis within the brain and spinal cord. GABA is the major inhibitory neurotransmitter in the brain. It is released from inhibitory interneurons to control the rhythms of the cortical networks, which are believed to be of critical importance for information processing and can dysfunction in some neurological diseases such as schizophrenia, depression and bipolar disorder.

A recent paper (link is external) from the lab of Dr Simon Glerup at Aarhus University examines the profile of the inhibitory interneurons, or more specifically the long and tedious journey that his lab took to correctly localise these neurons by immunohistochemistry in sections of mouse hippocampus and spinal cord and in cultures of hippocampal neurons.
The inhibitory interneurons are divided into 21 different subtypes based on their expression of different markers and their electrophysiological properties. The whole population of inhibitory interneurons can be visualised by immunostaining using antibodies directed against GABA, but to narrow down the specific interneuronal subtypes the markers calreticulin, calbindin, parvalbumin and somatostatin are commonly used by researchers in the field.
During a study investigating the different interneuronal subtypes Glerup’s team came across a stumbling block of locating antibodies up to the task of specifically localising these neurons. Searching amongst the apparent several hundred different commercially available antibodies for the four interneuronal subtype markers they initially opted for the antibodies with the greatest number of citations.
Surprisingly in their hands many of these antibodies did not perform as cited in the literature. Issues of lot-to-lot variation for previously specific antibodies, antibodies released that were never specific to their target and user variability may all have contributed to this disheartening irreproducibility.
This left the researchers with the arduous and expensive task of searching through panels of antibodies that could correctly label the interneurons they were interested in. They bought in a full panel of two antibodies directed against somatostatin, three against calbindin and four against both calretinin and parvalbumin from a range of suppliers and with a mix of polyclonal and monoclonals.
The disappointing news is of all the antibodies they tested only one of the three calbindin antibodies and only one of the four calretinin antibodies showed specific staining with little background, which was in agreement with the previous literature.
The results for parvalbumin were worse- of the four antibodies tested only one showed good enough staining, but even this showed some background staining.
And for the antibodies to somatostatin though one of the two showed good staining in cultures of hippocampal neurons it was not compatible with immunohistochemistry techniques showing poor staining in spinal cord sections. This difference may be due to tissue accessibility problems of the antibody in histological sections, but does not help the researchers wanting to localise their protein of interest within these sections and ultimately means yet more costly and time-consuming experiments.
Dr Glerup recently spoke of another antibody horror story that his lab have experienced. In studies analysing the receptor tyrosine kinase TrkB, which is activated by BDNF and important in maintaining synaptic strength and plasticity in the CNS, he explained that in an effort to find an antibody specific to this target his lab spent three months and large sums of money trialling different antibodies where nine out of the ten they tested failed to work in the way promised by the suppliers.
Having developed Affimer technology to overcome many of the problems with antibodies, such as specificity, we are all too familiar with tales of antibody specificity problems. Affimer binders are targeted against the native protein state and so will recognise the functional form of the protein found in tissue samples. They are also significantly smaller than antibodies and thus perfect for applications such as immunohistochemistry, where tissue accessibility may present problems.