Change country

How to validate an antibody?

Every researcher in the lab will have different wishes and demands on how an antibody should perform.  Ideally, an antibody meets all demands one can think of.  Unfortunately this does not always happen.  Each antibody has its own unique characteristics. It may work very well in one or two types of assays (for example in WB and IHC), but not in others. If one needs an antibody for quantification in micro-well format, one should not test in WB or IHC.  Many WB antibodies do not work in IHC and many IHC antibodies do not work in WB. Yet, all combinations of the above are feasible, and to certain target proteins all antibodies work in all applications tested.

Sadly, there are also many proteins out there that refuse to generate antibodies fit for any application. Hence, the choice of target protein is a big factor determining the versatility of antibodies or whether it is fit for purpose at all.  The host species can also be a factor. It is commonly known that when mammals fail to produce a decent antibody one has to generate bird antibodies (chicken).  

Validation can also be restricted by a regulatory environment. Laboratories regulated by GLP will have to follow fixed procedures for antibody validation.  In such case, the quality control data on the product sheet are hardly relevant. The operator in the GLP lab will have to follow the obligatory procedures from the start anyway.  Here one has to decide on purchasing the antibody with the highest chances of success. Here the decision can go wrong when one picks an antibody with superior QC data from an irrelevant application. One tends to think that an antibody without QC data is less likely successful for the required assay in the GLP lab than an antibody with nice WB and / or IHC data. It is a difficult choice to make as each purchase has to undergo this obligatory and therefore expensive validation procedure.  The best choice is an antibody that already has proven itself in the relevant assay type. It is therefore recommended to first try a panel of different antibodies (from different manufacturers) in a much cheaper feasibility study before making that choice. It is worth asking the manufacturers (like us) for antibodies that are not on their catalogue. Many antibodies are waiting to be tested in assays not accessible to the manufacturers, and so they are not for sale (not working in WB or IHC) and a free sample is made available if feedback on the results is promised.

General validation rules

An antibody is meant to bind to the protein intended.  This confirmation is a minimal requirement for the product’s datasheet. One should never purchase a product without this confirmation on the product sheet.  Usually this confirmation is established by direct ELISA with a titre. Alternatively, a recombinant protein or purified protein (or fragment thereof) may be stained in WB. This is not yet a validation of any kind! It merely confirms that the antibody has a certain affinity to the intended target protein.

Validation starts with comparing its affinity to the intended protein with its affinity to all other proteins occurring in the natural environment of the intended target protein.  In other words, the antibody needs to be able to specifically bind to its intended target while it does not bind to the vast majority of all other molecules that naturally surrounds the target.  For this reason it is good practise to compare the binding of the antibody on two identical mixtures of proteins, one with the target and one without the target. This may translate into comparing matrix, lysate or tissue containing endogenous target levels with matrix, lysate or tissue containing knocked-down or knocked-out target levels. Or as an alternative, matrix, lysate or tissue with low target levels compared to matrix, lysate or tissue with artificially increased target levels.

More often than one would wish, the results of the above mentioned tests are not going to be black and white. When successful, a clear preference is observed to the intended target, but with a certain level of background.  At this stage one has to optimize conditions so to increase the signal/noise to acceptable levels. One has to take into account that when the antibody is binding to common epitopes, this antibody is going to be cross-reactive with related proteins sharing these epitopes when also present in the mix.  This cross-reactivity then invalidates the antibody and a better antibody needs to be identified. When the antibody is mono-specific to one defined and unique epitope, one has to take into account that such antibody will still bind to similar epitopes albeit at lower affinity. The background ocurring from proteins with such epitopes can be reduced by further diluting the antibody and by reducing the primary incubation time. Background can also derive from added reagents that are required for signal reporting. A bad secondary antibody can be appreciated by comparing the complete assay with the same assay but without the primary antibody.  Finally random noise is most likely produced by the reporting chemicals when the blocking conditions and or buffer constituents have not been optimal.

In Western blots, background can be seen as extra bands, while noise can be seen as random spots. In IHC, background will still stain certain structures in the cell (albeit different from the expected structures), while noise will show stains overlapping different cellular structures, thus showing lack of specificity of the stain itself. In fluorescence, noise will become apparent as a constant when different dilutions of the primary are compared.

 

Western Blot (WB)

Western blot is seen by many as the easiest and most straightforward type of immune assay. This systematic underestimation is thought to holding back science at a great scale. With so many pitfalls unrecognized, many good antibodies are dismissed by one or two poor experiments. Here follows a check list to rule out trivial reasons why a WB experiment failed:

Poor choice of tissue type or cell line

Proteolysis in the lysate

Inadequate reducing conditions to resolve the proteins in SDSPAGE

Underloading/overloading of protein content

Inadequate transfer to the membrane

Inadequate dilution of the primary antibody

Too long incubation times

Bad or wrong secondary antibody (or bad conjugated enzyme)

Bad or wrong substrate for staining

 

Immunohistochemistry (IHC)

Human tissues are usually fixed by paraformaldehyde and subsequently embedded in paraffin for thin sectioning and long term storage. These conditions may mask the epitopes for proper binding to the antibodies. Under such circumstances, epitope (antigen) retrieval is required for the antibody to work. Most common procedures involve heating the sections in buffers at either pH6 or pH9. The performance of the antibodies will depend on the pH value, the heating duration, and the method of heating during the epitope retrieval procedure.

When fresh tissues have been fixed in alcohols or acetone, one usually do not store the samples, but cryosectioning and probing with the antibodies go ahead in one flow. Then epitope retrieval is not on the charts. I would however recommend epitope retrieval if the tissues have been stored in the alcohol or acetone. The dehydration of the tissues may mimic the paraffin embedding procedure, particularly in the case of storage in the hydrophobic acetone. Hence, rehydration of the tissue during the removal of the alcohol/acetone may still be not enough to expose the epitopes fully (again) after long term storage.

 

Immunofluorescense (IHC, ICC, FC)

Fluorescent labels are commonly used in Flow Cytometry (FC), ImmunoCytoChemistry (ICC) and in IHC as well. The major issue with fluorescence is the noise.  Because of the high sensitivity of this detection type, noise is much more prevalent compared to other less sensitive detection types.

As with all assays, different dilutions of the primary (including one without primary) will have to be compared in order to appreciate the level of constant noise. One has to be careful in choosing the right blocking agents, but more importantly, one has to take into account the presence of endogenous molecules (even at low abundance) that bind to either the fluorophore itself or to its conjugated carrier (for example avoid using streptavidin-fluorophore when there is endogenous biotin, and avoid using anti-IgG-fluorophore when there are endogenous Ig-receptors).

 

Immunoprecipitation (IP)

A network of antigen and polyclonal antibodies is precipitated and analyzed. This is the classic method of IP. However, when the target protein is bound to one or more other proteins, the possibilities for the antibody to bind to the target protein become limited and the network may not be stable enough for the classic approach. It is therefore recommended to have a control of denatured protein IP side-by-side with the actual experiment. 

One also may consider using epitope-specific antibodies (monoclonal or peptide-polyclonal) brought down by beads. This approach makes the scientist independent on an antibody-antigen network to form. When the right epitope-specific antibodies are used, extra information can be generated about the binding sites for the interacting other proteins.

 

ELISA (EIA)

This assay type represents any micro-well formatted immunoassay. They all have in common that one reagent is coated to a stationary phase (usually the bottom of the well), and other reagents react with the coated reagent proportionally to the content of the analysed material within a natural matrix. Matrix could be a body fluid (plasma, serum, urine, etc), a culture supernatant, or a buffered solution spiked with biological material from which one constituent needs quantifying.

The biggest hurdle in such assays is the notorious matrix effect. To put it simple: the matrix contains proteins or other molecules that interfere with the reaction between the antibodies and the antigen that needs quantifying (analyte). This interference can be visualized by making serial dilutions of the matrix.  When the matrix is diluted with a factor 2 and the measured analyte therein does not read a reduction by a factor 2 along the way, you have established matrix effect.

It is essential to establish matrix effect every time a new assay is set up.  When a new antibody is introduced in an already existing assay, the matrix effect may respond differently from the previous antibody.  A calibration curve therefore needs to be made in the same matrix and run in parallel with each assay. If the sensitivity of the assay and concentration of the analyte allows it, it is recommended to minimize the matrix effect by carrying out the assay in a diluted matrix. Some detergents may also minimize matrix effect. 

One needs to be aware that most commercial primary antibodies (including ours), have not been preadsorbed to serum proteins. This would be a compulsory step before using the antibody in matrix containing high levels of serum protein.

 

Jan Voskuil 2013