Antibodies are one of the most commonly used reagents in life science research, diagnostics, and clinical tests. Despite their widespread use and the significant investment of time and money that their uses represent, there are no standard guidelines that define how they should be validated prior to use. The consequence is that many commercial antibodies are not adequately validated and/or experimental data are not accessible. This poor antibody validation has led to irreproducible results and even projects abandoned [1], causing tremendous loss of time, money and samples [2]. Thus, there is an urgent need for validating antibodies and creating universal standards for it.

Antibody validation is defined in the following points: proving specificity (the ability of an antibody to differentiate between different antigens), proving specificity in the application in which it is going to be used, proving affinity (the strength with which an antibody binds an epitope) and finally proving reproducibility. However, while this definition of antibody validation is practical, its widespread application or implementation of a standardized process is still a significant area of concern.

Several antibody-selling companies validate their products, and others do not validate them at all [3]. These differences occur because, unlike drugs, there is no agency governing what can be sold. But with a rapidly growing market, a reputation for quality is becoming part of some suppliers’ business plans and antibody validation is now viewed as a competitive advantage.

The responsibility for validating an antibody is with the vendor itself whether or not they manufacture that antibody or simply distribute it. But while vendors are responsible for the quality of the reagents they sell, antibody performance can vary for reasons that are beyond their validation efforts. For example, antibodies can get altered during shipping, or although validation can be done in biologically relevant systems with appropriate controls, the end user is most probably going to use it in their own system. Thus, final users also share responsibility in validating their antibodies. Various manufacturers have proposed validation processes [4]. Guidelines for authors and reviewers on antibody use are emerging from journal publishers [5], which may have helped antibody identification rather than actual experimental validation [6].

By taking several small steps, researchers can reduce the risks when running an experiment without critical thinking on the antibodies they acquire. As a start, researchers should be aware of the information provided in the product’s data sheet, which must clearly define the recommended usage of the antibody, including what applications it has been validated for, appropriate protocols and recommended dilutions.

Antibody validation is a crucial issue. Researchers and organizers have tried to address the problem. Antibody Society produced a 10-part webinar series for early-career antibody users. Laflamme C et al after examining the specificity of C9ORF72 antibodies, proposed "a validation workflow as follows: 1) identify all commercial antibodies against a protein of interest; 2) use PaxDB to identify cell lines that express the protein of interest at relatively high levels and are readily modifiable by CRISPR/Cas9; 3) generate a KO cell line and screen the antibodies by quantitative immunoblot; 4) use specific antibodies determined from step 3 to screen panels of cell lines; 5) based on the information in step 4 choose a cell line that has high expression of the protein of interest, is amenable to editing by CRISPR/Cas9, and has appropriate characteristics for the target of interest; 6) use the selected edited line to screen antibodies for specificity by immunoprecipitation and immunofluorescence; 7) select specific antibodies for use in more intensive processes such as immunohistochemistry" [7]. Szőnyi A et al, in their study about neuronal basis for negative experience in mouse, clearly stated how the specificity for each of two dozens of primary antibodies had been established [8].

Antibodies should be validated for specific applications, including western blot (WB), immunohistochemistry (IHC), immunocytochemistry (ICC), immunofluorescence (IF), ELISA, immunoprecipitation (IP), chromatin immunoprecipitation (ChIP), peptide array, and protein array [9], among others. Each assay has its advantages and disadvantages when it comes to validation (Table I). It is crucial for scientists to be aware of these so that they may have confidence in their results.

The criteria for what method should be used and how many of them, rely on the sensitivity, specificity, and reproducibility, but also on the context in which a specific antibody is going to be used. That means that if a given antibody is intended to be used only for WB, then they should only have to test it in that context. But if it is going to be used in another assay, it should certainly be tested for that application in the first place. This helps saving time and costs related to antibody validation processes, whether for antibody providers with portfolios of thousands of antibodies or for research laboratories.

Specificity measures the degree to which an antibody differentiates between different antigens. It is possible for an antibody to be sensitive for a protein of interest but still cross-react with other proteins and therefore lack specificity. In this sense, the confirmation that an antibody binds specifically to the intended protein is its minimal requirement and the first thing to look up to in the product’s data sheet.

There are a couple of questions that need to be addressed before analyzing the specificity of a given antibody. These are, what type of immunogens were used to produce that antibody? And how is the target protein structured in the samples that will be analyzed? The first question is not always easy to answer, especially if manufacturers do not disclose this information. The importance of knowing this information rely upon the fact that determining the specificity of an antibody is in part dependent on the type of the immunogens: synthetic peptides or purified proteins. Synthetic peptides do not necessarily recapitulate the 3-D structure or post-translational modifications of the native protein [10], so antibodies generated against them can fail to bind to native proteins. If immunogens are purified proteins instead, antibodies might not bind to it when denatured. This information can help discriminate which applications antibodies would be suitable for, thus if immunogens are synthetic peptides, antibodies might be useful for WB, but not IP or IHC.

It is essential then, to purchase antibodies for which manufacturers disclose the immunogen.

Although many researchers consider WB as the first step to evaluate a new antibody, it cannot be an absolute standardization for antibody binding in assays where the antigen is in its native conformation, such as IHC. For example, a rabbit monoclonal anti-MrgprD antibody appears to be specific for Western blot; however, it recognizes non-specific epitopes with knockout control [11, 12]. This issue can be even more complicated by methods used to fix the tissue because antibodies can recognize one epitope in a fresh tissue, but a different epitope in a fixed tissue. This can occur because epitopes that were not previously exposed in the native protein are now accessible in fixed tissues and vice-versa [13, 14].

If the information available for an antibody specifies that it recognize the denatured antigen or that it is suitable for WB, then this assay functions as an appropriate first validation step. The first indication that the antibody is specific for the selected target would be observing a single band at the known molecular weight for the target. Presence of multiple bands or bands not at the proper molecular weight could represent the same target at different post-translational modification status, breakdown products, or splice variants rather than cross-reactivity. However, such observations should raise concerns about using this antibody for further experiments.

Another drawback of WB is its low throughput as it is typically restricted to testing one antibody at a time. However, recently a high-throughput version of this method, known as PAGE-MAP, was developed for antibody validation purposes by converting WB into a capture format [15]. In this method, biotinylated protein samples are subjected to preparative PAGE and size-separated proteins are incubated with microsphere-based barcoded antibody arrays for multiplexed IP, and captured proteins are labeled with streptavidin for on-bead detection by flow cytometry (i.e., microsphere affinity proteomics (MAP)). Size-separated proteins can also be used for parallel readouts shotgun mass spectrometry (MS) - which can be used to obtain a rough estimate of specificity and for use in selecting fractions with levels of antibody targets that are sufficient to allow identification by immunoprecipitation followed by MS (IP-MS). This new method is powerful for screening thousand of antibodies and also for identifying those that bind the same protein and provides the means for large-scale implementation of antibody validation.

Antibody specificity has also been evaluated by using blocking peptides especially for IHC [16]. These peptides are the same ones used to generate the antibody, and they are used to immunoneutralize the antibodies when incubating with them in significant excess. Antibodies are then used to stain the sample tissue, with non-neutralized antibodies as controls. If the antibody is specific, immunoneutralization will result in loss of staining on the tissue. An example of this technique is the validation of phospho-specific antibodies for ERα [16]. A drawback of this assay though, is that it does not demonstrate selectivity, i.e., that the antibody is specific only for that antigen since the non-specific binding activity of the antibody will also be inhibited by neutralization with the blocking peptide. Thus, blocking peptides can prove that an antibody is bad, but they cannot prove that an antibody is good [17]. Cardenas A et al, for example, found an antibody against alpha 6 nicotinic acetylcholine receptor subunit non-specific after using knockout mouse lysates, despite that the control antigen blocked all binding [18].

The key to proving antibody specificity is the correct use of controls. As shown in the validation of the cannabinoid CB2 receptor, the common practice of validating antibodies with positive controls only, is insufficient to ensure antibody reliability [19]. In this study, despite gathering many results in favor of an anti-CB2 antibody validity using several techniques such as WB, mass spectrometry and blocking peptides, as well as strong positive controls, when the antibody was tested with knockout negative controls, it became clear that this antibody was not specific for CB2 receptor protein. This can occur because of the similarity between proteins with respect to the epitope region, or because multiple epitopes exist for the antibody [20]. Closely related proteins can serve as good negative controls as well. For example, Simons IM et al used close-related proteins GABARAPL1, GABARAPL1, LC3A, LC3B, and LC3C to test the selectivity of a rat monoclonal antibody against GABARAP in ELISA, dot blots, and Western blots [21]. In this regard, polyclonal antibodies, due to its inherent nature of multiple epitopes, are likely to be non-specific, when examined with knockout negative controls (it is estimated that only 10-20% polyclonals are specific using knockout negative controls). Also, polyclonals are liable to shifts in epitope binding between batches/lots. The specificity of a polyclonal antibody preparation can be improved through purification with, for example, a negative bacterial lysate.

The best negative control is knockout cells/animals, which do not express the protein of interest and the best positive controls are the non-expressing cells transfected with the protein of interest. For example, VG Magupalli et al validated anti-NLRP3 and anti-ASC antibodies with respective knockout cells [22]. If knockout cells are too difficult to acquire, alternatives are siRNA or shRNA knockdown controls. Although negative controls (e.g., no primary antibody) are necessary, they are not sufficient and positive controls must be present as well. Another negative control is the mutated antigen, as used by L Cantuti-Castelvetri et al [23].

Several methods have emerged that give antibodies a quantitative score on their performance, in contrast to typical qualitative or semi-quantitative methods described above. In this way, a researcher’s opinion on whether a stain looks strong enough is geared towards more sensitive methods that allow validation and comparison between antibodies. An automated method has been devised to validate antibodies against synaptic proteins for array tomography [27]. Deming et al demonstrated a multiplexing IHC platform, tissue-based cyclic immunofluorescence (t-CyCIF), to validate antibodies for IHC experiments [28]. Simons IM et al used several approaches to validate rat monoclonal antibody 8H5 against GABARAP (gamma-aminobutyric acid type A receptor-associated protein), a crucial protein involved in autophagy, for immunocytochemical applications involving knockout cell lines and co-localization of YFP-GABARAP with antibody staining [21]. Simons IM et al also examined the specificities of several commercial GABARAP antibodies, and found them to be specific for western blot, but not for immunocytochemistry, for example, GABARAP (E1J4E) Rabbit mAb #13733, from CST [21]. Below, we discuss the validation strategies for immunoprecipitation and Chromatin Immunoprecipitation in some detail.

A highly sensitive technique to identify and quantify proteins is mass spectrometry (MS), especially for amino acid modification-based antibodies, such as pTyr antibodies [29], and to date, two MS-based methods have been developed to evaluate antibodies for immunoprecipitation. The first one consists of a method for scoring IP antibody quality. The authors used 1,124 recombinant antibodies raised against 152 chromatin-related human proteins, immuno-precipitated target antigens from a cell lysate and then performed mass spectrometry on the precipitates to quantify the pulled-down proteins [30]. With this method, the authors determined that only 31% of the analyzed antibodies pulled down their targets.

The second method was developed to improve microsphere multiplexed detection with commercially available antibodies. The reason is that multiplex analysis requires highly specific antibodies, especially for complex samples, and the frequency of commercial antibodies useful for it may be as low as 5% [31]. The creators of this method showed that this frequency could be increased by 20% when cellular proteins are fractionated by size exclusion chromatography (SEC) first [32]. After this step, the lysate’s proteins are biotinylated and immunoprecipitated (in this work the authors used a pool of 1725 antibodies to cellular proteins), each type attached to a different color of bead. Running the beads in a flow cytometer, each antibody can then be identified by the bead’s color and the amount of protein bound by the streptavidin (which binds to biotin) signal.

[enlarge]

Figure 1. Schematic view of the validation of antibodies for immunoprecipitation using mass-spectrometry (MS).

The use of MS can be of great advantage in particular for novel proteins with limited literature or proteins with many paralogues or orthologues, which are quite tricky for antibody development. Moreover, it allows characterizing antibodies in the absence of overexpressing cell lines or of any previous knowledge of the antigen. But it is worth noting that although both methods evaluate antibodies for their use in IP, it does not indicate success in other applications. A general schematic view of the use of MS for the validation of antibodies for use in IP is shown in Figure 1.

ChIP and ChIPseq are techniques widely used in the study of epigenetics and transcriptional regulation of gene expression, and it is one of the most dependent on antibodies. The ENCODE and modENCODE consortia have published guidelines for the standards required of ChIP-seq experiments [33], that suggests a two-step validation procedure: an initial immunoblot or immunofluorescence assay test, followed by at least one secondary validation assay.

The most often used primary validation is WB, which can be performed on cell or nucleate lysates, chromatin preparations or immunoprecipitated material. ENCODE accepts that the primary reactive band should contain at least 50% of the signal observed in the blot. It is expected that this band corresponds to the size of the expected protein, but shifts may appear as discussed previously in this article, and further validation must be done in such cases. It is important to know that, when using chromatin-immunoprecipitated material, it is highly probable that the target protein will immunoprecipitate with other proteins of the same complex, thus it would not be the assay of choice for analyzing antibody specificity. In cases where ChIP antibodies do not work on WB, IF may be used as an alternative. A drawback of this technique is that it does not show cross-reactivity of antibodies, so further validations must be carried out.

Further validation can be carried out by performing the aforementioned assays on KO or KD systems. ENCODE accepts data if the immunoreactive signal is reduced by at least 70% in WB or IF, or if the ChIP-seq signal is reduced by at least 50% in any of the two systems. Another validation is to use multiple antibodies directed towards different regions of the protein of interest or members of the same complex. What is expected from this assay is significant overlap of enriched loci. Historically, ENCODE has accepted an overlap of 75% of common targets. The last validation assay is motif enrichment, which consists of searching for known DNA sequences among bound peaks, known to be targets of DNA-binding proteins, such as transcription factors. ENCODE accepts data if the motif is over fourfold enriched compared to all other accessible regions and present in more than 10% of peaks. In the case of histone antibody validation, there are further recommendations. For example, in WB specific histone band should be at least 50% of the blot signal, with at least a 10% enrichment over any other individual band and the recombinant unmodified histone band. Secondary tests should include at least one of the following tests: peptide competition assays, in which histone tail peptides with particular modifications should have a binding signal 10 times higher than with other modifications; MS, where the target histone should account for at least 80% of the immunoprecipitated sample; mutants defective in histones modifying proteins, in which antibody signal should be reduced to below 10% of the wild-type signal; Annotation enrichment, which applies to ChIP-seq, consists in observing overlaps between the localization of histone modifications with previous ChIP-seq datasets.

Antibody binding affinity is another parameter that can be used for antibody validation. It refers to the strength with which an antibody molecule binds an epitope. It is typically reported by the equilibrium dissociation constant (KD), which is the ratio of the antibody dissociation rate or k_off (how quickly it dissociates from its antigen), to the antibody association rate or k_on of the antibody (how fast it binds to its antigen). In a diluted solution of antibody and antigen, the binding events can be described as a stochastic process. The antibody-antigen complex form by an association reaction when the epitopes of the antigen collide with a binding site of the antibodies. Briefly afterward, dissociation of this complex will occur due to the thermally induced motion. Both reactions will eventually reach an equilibrium state with equal amounts of binding and dissociation events per second. The concentrations of the components in this steady state are strongly related to the affinity of the binding reaction. Ranawakage DC et al developed HiBiT-qIP assay for determining antibody affinity under immunoprecipitation conditions and evaluated the affinity of several commonly used tags and their cognate antibodies [34].

It is important to know that affinity determination of monoclonal antibodies can be made with high accuracy since they are homogeneous and selective for one epitope, but in the case of polyclonal antibodies, only an average affinity can be obtained, since they are heterogeneous and contain a mixture of antibodies with different affinities for various epitopes. For determining antibody affinity, several methods have been elaborated. Among them are methods based on ELISA, and other biophysical methods such as microscale thermophoresis (MST) and surface plasmon resonance (SPR) [35].

They are the most popular methods for studying antibody affinities. They do not require considerable quantities of antibodies and antigens nor their purification. In this method, a fix antibody concentration is incubated with antigen in solution until steady state is reached. Then, the concentration of unbound antibody is measured by indirect ELISA [36]. This method need previous setup experiments though, to ascertain that the antibody concentration used is in the linear range of the ELISA response (so that the absorbance is proportional to the Ab concentration) and that only a small fraction of the total free antibody in solution is retained on the plate (so that the measurement does not significantly affect the equilibrium in solution). Variations of these methods include live cell ELISA [37], ELISA for bivalent antibodies [38] and ELISA with non-linear regression [39]. Some examples of antibodies that have been analyzed by ELISA are human Pan-IgG [40] and anti-human PCSK9 [41].

[enlarge]

Figure 2. Determination of antibody affinity by Microscale Thermophoresis (MST). In MST, a microscopic temperature gradient is induced by an infrared laser in small capillaries that hold the sample in solution. The directed movement of particles across the temperature gradient induced by an IR-laser is then detected by fluorescence. Binding of antibodies to antigens results in an altered movement. This technique retrieves information of binding affinities between the molecules in solution.

MST is a biophysical method that can access molecular affinities in a wide concentration range. It measures the motion of molecules along microscopic temperature gradients (thermophoresis) induced by an infrared laser (Figure 2). This movement depends on a number of factors including the hydration shell, charge and size of the molecules. These molecules are initially distributed evenly and diffuse freely in solution. When the IR laser is switched on, unbound molecules typically move out of the heating spot. Binding of one molecule to another, such as antibodies and antigens, results in an altered movement across the temperature gradient. To follow the movement of molecules, MST uses fluorescence, which can be intrinsic of the proteins, fluorescent fusion proteins or an attached dye [42]. An example of antibodies analyzed by this technique is the recombinant scFv against human transferrin [43], a mouse monoclonal antibody against cocaine used for cocaine immunotherapy studies [44], and an anti-myc tag antibodies derived from 9E10 clone [45].

[enlarge]

Figure 3. Determination of antibody affinity by Surface Plasmon Resonance (SPR). SPR occurs when light is reflected from a metal-coated interface between two media of different refractive index (a glass prism and a solution). The angle of reflection changes when an interaction between an immobilized ligand (e.g., an antibody, Y) and an antigen in solution (filled circles) occurs. This technique retrieves information of binding affinities between antibodies and antigens.

SPR is an optical technique utilized for detecting molecular interactions, where one of the molecules is mobile, and the other is immobilized in a metal film [46]. Binding of these molecules changes the refractive index of the film. Thus, when polarized light impact upon the film, the angle of extinction of light is altered which can be monitored by an optical detector (Figure 3). Some of the antibodies, whose affinities have been determined by SPR are human epidermal growth factor (hEGF) monoclonal and polyclonal antibodies, and therapeutic monoclonal antibodies against proprotein convertase subtilisin/kexin type 9 (PCSK9), progranulin (PGRN), and fatty acid binding protein (FABP4) in human serum at its endogenous concentrations [47].

All in all, any single approach for antibody validation is probably not sufficient. It is a combination of techniques that are necessary to make sure an antibody fulfills the specificity and reproducibility requisites for its use and they typically include positive and negative cell lines or tissues, modulating the expression or localization of the target by using specific inhibitors or activators, siRNA or knockout technologies, and other more sophisticated biophysical techniques to determine binding affinities and identifying antigens, such as MST, SPR, and MS.

An essential criterion for validation is antibody reproducibility. That is, the usage of the same antibody over time, even with different lots, would yield similar results. A common mistake is to assume that new aliquots of an antibody, whether they come from the same or different lots or from different manufacturers will yield similar results. One of the most shocking examples is the case of David Rimm from Yale University [1]. He had developed an antibody-based test to guide an effective treatment of melanoma and was ready to move the test towards the clinic. But when he ordered a new set of antibodies from the same company, he could not reproduce the original results, having to give up his work. Thus, it is crucial for researchers to always test the reproducibility of antibodies. This gains even more importance when using polyclonal antibodies, since often same catalog number from a supplier may mean different antibodies.

An example of poor correlations between antibodies from different lots can be exemplified on the reproducibility validation of five antibodies against Met, the receptor of the hepatocyte growth factor [48]. In that study, only one out of five commercial antibodies was shown to be reproducible. Moreover, two different lots of one of them (3D4), despite having similar profiles on WB, had different staining patterns-one stained the nuclear and one membranous and non-specific, as seen on an array of 640 breast cancer cases. Another example of non-reproducible antibodies is the IHC analysis of the growth factor VEGF on lung carcinoma tissues [17]. Although first staining demonstrates the specificity of the VG-1 clone, staining under the same experimental conditions on a serial section of the same tissue microarray, with the same lot of antibody, was not reproducible. But reproducible results have also been described, as in the case of the validation of antibodies against HER2 [49]. When comparing the IHC staining of 283 breast adenocarcinomas on tissue microarrays with the 4B5 rabbit monoclonal antibody to a previously established CB11 mouse monoclonal antibody, no significant differences were found in terms of sensitivity and specificity, thus retrieving reproducible results. Brandone N et al also compared a new clone QR1 against PD-L1 with three established clones 22c3, Sp263 and E1L3N for IHC on lung adenocarcinomas using a microtissue array and established the specificity of the new clone [50].

A multiparameter analysis is a crucial step for the development of immunoassays. Various characteristics, which may influence the results, include the specificity of the selected antibody, the structure of the antigen, the sample medium and the conditions of the immunoassay [51].

The specificity of an antibody depends on the potential cross-reactivity with compounds structurally similar to the specific antigen. With regard to illicit compounds, amino, aromatic and aldehyde groups may cause cross-reactivity [52]. In addition, binding to the related molecules was demonstrated by the study of piperazines [53]. Binding specificity may also be affected by the method of generation of the antibodies. The structure of the carrier compound is also important, since there is a risk that antibodies may bind to the carrier, rather than to the molecule of interest [54]. Therefore, antibody production should include the analysis for identification and removal of unspecifically binding carriers, such as bovine serum albumin (BSA).

Due to the potential binding of the applied antibodies to various serum components, including other immunoglobulins, the sample medium may also affect the efficiency of immunoassays. Besides, different carrier proteins present in serum may influence specific binding of the antibodies. However, these potential problems were shown to be reduced by dilution of the sample medium [55]. In addition, the results of an immunoassay may depend on the type of the chosen platform [56]. Western blot remains to be an effective method for evaluating the specificity of antibodies. A specific type of an immunoassay platform defines how the molecules bind to the surface. In particular, different platforms have different surface structures. Thus, various characteristics, including the specificity, sample medium, and details of the essay platform, should be evaluated during quality control for immunoassays.