Current approaches to fine mapping of antigen–antibody interactions

Abstract

A number of different methods are commonly used to map the fine details of the interaction between an antigen and an antibody. Undoubtedly the method that is now most commonly used to give details at the level of individual amino acids and atoms is X-ray crystallography. The feasibility of undertaking crystallographic studies has increased over recent years through the introduction of automation, miniaturization and high throughput processes. However, this still requires a high level of sophistication and expense and cannot be used when the antigen is not amenable to crystallization. Nuclear magnetic resonance spectroscopy offers a similar level of detail to crystallography but the technical hurdles are even higher such that it is rarely used in this context. Mutagenesis of either antigen or antibody offers the potential to give information at the amino acid level but suffers from the uncertainty of not knowing whether an effect is direct or indirect due to an effect on the folding of a protein. Other methods such as hydrogen deuterium exchange coupled to mass spectrometry and the use of short peptides coupled with ELISA-based approaches tend to give mapping information over a peptide region rather than at the level of individual amino acids. It is quite common to use more than one method because of the limitations and even with a crystal structure it can be useful to use mutagenesis to tease apart the contribution of individual amino acids to binding affinity.

Introduction

Antibodies are a key biological molecule, with numerous applications in the fields of therapeutics, diagnostics and biological research. Within the field of therapeutics alone, antibodies represent one of the largest growing classes of molecule, with five of the top ten selling prescription drugs in 2012 being antibody-derived.¹ Antibodies have evolved to bind with exquisite specificity to their target antigen to be able to potently neutralize ‘non-self’ intruders. Since their initial discovery, much study has focused on the precise nature of the binding between the antibody and the antigen. The molecular structures within any given target antigen that make specific contacts with the antibody are referred to as antibody or B-cell epitopes and conversely, the molecular determinants within the antibody structure that make specific interactions with the antigen epitope are often termed paratopes. Antibody paratopes are often composed primarily of the so-called complementarity-determining regions (CDRs), which can be classed in slightly different ways, based upon sequence variability or structural analyses,^2–4 but often also contain other so-called ‘framework’ residues.^5,6 Mapping the interactions between antibodies and antigens has a number of important uses, such as advancing our knowledge of the immune response and autoimmunity,⁷ predicting suitable antigens for use as vaccine components,^8,9 gaining a greater understanding of a therapeutic antibody's mechanism of action,^10,11 and securing and protecting intellectual property. Concurrent developments in the use and application of antibodies, the technologies available for antibody discovery engineering and the technologies for mapping antigen–antibody interactions have led to a rapid proliferation in publically available antibody epitope data, for example in the immune epitope database (http://www.iedb.org).¹² Experimental methodologies for determining the molecular interactions of antibodies with their cognate antigens have evolved from simple ELISA-based approaches, such as the use of linear peptides representing small stretches of the antigen protein sequence, to more complex mutant screening approaches to determine structural, discontinuous epitopes; mass spectrometric methods, such as hydrogen deuterium exchange or protease protection, or, where possible, the use of nuclear magnetic resonance (NMR) or X-ray crystallography to precisely determine the interface between antibody and antigen.

There has been much recent activity in developing in silico methods for predicting and determining antibody epitopes (reviewed in ref. 13), but B-cell epitope predictions have been shown to be suboptimal, due to a preponderance of both false-positive and false-negative results.^14,15 This review will therefore focus upon experimental methods for mapping antibody paratope and epitopes. The word ‘epitope’ can mean many things, e.g. the surface protein recognized on a virus, the individual protein bound in a protein complex, a domain of a particular protein, a small part of a protein and ultimately the contribution of individual amino acids and individual atoms. A further complication arises in that epitopes can be defined as structural epitopes or functional epitopes. A structural epitope consists of amino acids or other molecules in a region that is in close contact with the antibody usually revealed by a structure. A functional epitope is defined as those parts of a molecule that make an energetic contribution to binding such that when they are changed there is a decrease in binding affinity.¹⁶ When considering what an epitope really is it is important to know not only what residues are making contact with the paratope but also what residues are contributing to the affinity irrespective of whether they are proximal or not. There are multiple methods that have and can be used to map epitopes in the broadest sense that have been described in some detail in an earlier review.¹⁷ In this review we will focus on those methods that are most often currently used to enable a determination of the fine details of interaction, i.e. small peptide regions and individual amino acids.

Approaches to fine epitope mapping and their limitations

X-ray crystallography

Undoubtedly the method of choice to determine the precise sites of interaction between an antibody and its antigen is X-ray crystallography. Crystal structures enable very accurate determination of key interactions between individual amino acids from both side chains and main chain atoms in both the epitope and paratope. Amino acids that are within 4 Å of each other are generally considered to be contacting residues. There are now hundreds of three-dimensional structures (as of 17 December 2013 there were 742 antibody epitope three-dimensional structures in the immune epitope database). Some examples are referenced here^10,18–27 and a few are discussed in much greater detail below. The process requires at least a reasonable degree of purity for both the antibody and antigen. This is generally straightforward to achieve for the antibody but can be much more difficult for the antigen. The antibody is usually made as a Fab fragment either recombinantly by co-expressing the light chain and a construct of the VH and CH1 domains or via proteolytic cleavage of the whole IgG with papain or pepsin to generate Fab or Fab’ fragments, respectively. The latter process has the disadvantage that it can result in inadvertent proteolytic cleavage of the antibody fragment itself. Other antibody fragments such as scFv^28,29 can be used.

The methodology involves first, purification of antibody and antigen, formation of the complex and then frequently purification of this complex by size exclusion chromatography. It is then generally necessary to go through successive rounds of crystallization screens and optimization to obtain diffraction-quality crystals. Structural solution is obtained following X-ray crystallography frequently at a synchrotron source. The main disadvantage, apart from the high level of expertise required and high capital cost, is that diffraction-quality crystals must be obtained and this is often not a simple task even with some simple soluble antigens. When the antigen is a membrane protein the task can be even more difficult. As an aside, antibody fragments are themselves often straightforward to crystallize and there are now numerous reported examples where antibody fragments have been used to derive a crystal structure of a protein that was otherwise proving intractable to crystallization (e.g. G protein coupled receptors (reviewed in ref. 30) exemplified with the B2-adrenergic receptor,³¹ and flexible proteins such as interleukin-17A (IL-17A).²² Nonetheless, obtaining diffraction quality crystals is often the rate limiting step and sometimes makes the task with a particular antibody–antigen pair impossible.

Nuclear magnetic resonance

Nuclear magnetic resonance can also be used to obtain very fine epitope mapping information – although it is not commonly applied in this respect, it has been well exemplified.³² The method requires a previous structure determination of the free antigen and consequent full resonance assignment of a ¹⁵N-, ¹³C- and potentially ²H-labelled protein that will need to be expressed in Escherichia coli. Once this is available two-dimensional ¹H/¹⁵N transverse relaxation optimized spectroscopy (TROSY) NMR correlation spectra of free and Fab-bound ¹⁵N-labelled antigen are obtained. Those peaks that exhibit major chemical shifts in the presence of Fab are then likely to be the amino acids in the epitope. Although feasible in some circumstances, the technical hurdles and level of expertise required are very high, which probably explains why the approach is not commonly used.

Hydrogen–deuterium exchange coupled to mass spectrometry

Backbone amide hydrogen atoms will exchange for hydrogen atoms in a solvent³³ and if the solvent contains deuterium rather than hydrogen then the same exchange process occurs such that the protein will become deuterated. The rate at which this occurs for any particular hydrogen atom is governed by solvent accessibility and the flexibility of that part of the protein. Hydrogens that are buried deep within the hydrophobic core of a rigid protein will barely exchange at all whereas those on a flexible loop will exchange almost instantaneously. The process can be used to map interaction sites such as epitopes because the rate of exchange will slow down when a particular region is in contact with a partner such as an antibody.^34,35 Following interaction and deuteration the antigen is digested with pepsin and the mass of peptides is determined to assess the level of deuteration.³⁶ The approach has the advantage that the interaction is performed with the antigen in its fully folded structure and is applicable to most proteins if they can be obtained with a moderate level of purity. Furthermore, this approach can be used with impure antibodies such as polyclonals.³⁷ Perhaps the biggest disadvantage is that it is fairly coarse epitope mapping such that, for example, it is generally not possible to say in any more detail than that a sequence of approximately 10–20 residues long contains an epitope.^38–41

Peptide-based approaches

In these methods, overlapping peptides are synthesized that cover the whole sequence of the antigen and are then immobilized onto a solid surface as an array and the binding to the antibody of interest is determined in an ELISA format.^42–45 It is ideally suited to situations where the epitope is a linear peptide sequence, although it is also possible to constrain peptides via one or more disulphide bonds to mimic discontinuous and conformation-dependent epitopes.⁴⁶ It is simple and quick to perform. A variation on the use of synthetic peptide libraries is the use of random phage libraries that can be linear or constrained by disulphide bonds.^47–51 In this approach, very large phage libraries can be ‘panned’ with the antibodies and the recombinant peptide genes from binding phage particles can be sequenced to determine the sequences and therefore putative epitopes or ‘mimotopes’ as they are sometimes called.

Both of these approaches are more accessible without specialist expensive equipment and expertise, can be used without the necessity of purified antigen or antibody, and are ideal for linear and some relatively simple conformational epitopes. However, complex conformational epitopes that involve tertiary and/or quaternary structure of the antigen are unlikely to be identified.

Mutagenesis

A further approach to putatively determine epitopes, which does not require a high level of expertise and expensive equipment, is to analyse the binding of antibody to mutated forms of the antigen. Mutagenesis also has the advantage that it is possible to screen many hundreds or thousands of proteins (using Ala scanning or ‘shotgun mutagenesis’) quickly and is more straightforward with difficult proteins such as membrane proteins.⁵² Loss of binding to a particular mutant is usually interpreted to mean that those mutated amino acid(s) constitute(s) an epitope; however, the disadvantage of this approach is that it can be difficult to know whether the mutation has disrupted the folding of the protein and therefore the interaction with the antibody, or is genuinely a key interacting residue. Regardless, mutagenesis is often used very powerfully alongside other approaches such as crystallography to dissect out the contribution of individual amino acids to the affinity of the interaction.^{11,19,23,26,27,38}

Confidence in alanine scanning mutagenesis can clearly be increased in the presence of a structure and then only changing those residues at the surface of the protein. These changes are unlikely to induce major changes in the overall folding of the protein. Changing internal hydrophobic residues would clearly involve a much greater risk of adversely affecting the folded structure of the protein; although as these are not likely to be involved in an epitope there is little reason to do this. Even in the presence of a structure it can sometimes be difficult to interpret data, particularly the various effects on the free and bound states of the antigen and the effects of interacting residues in the antigen on the overall affinity to the antibody. These various issues are discussed in much more detail in ref. 16.

Confidence in this approach can also be further increased by making targeted mutants of structurally related proteins such as orthologues from different species or homologous proteins from the same species. Essential to the success of this method is that the fine specificity of the antibody is known and that homologous proteins are not recognized. In ‘knock-out’ mutagenesis, segments of the target antigen are replaced by the homologue that is not recognized and, as described above, loss of binding is interpreted to mean that those regions are involved in binding. However, as the two proteins are structurally similar it is less likely that the mutants will have misfolded.^11,22 An even more compelling approach is ‘knock-in’ mutagenesis where changes confer binding of an antibody to a structurally similar protein that is not typically recognized. Here, the same non-binding protein sequence used in knock-out mutagenesis is encoded as the acceptor antigen with the key residues of the target antigen encoded in place of the corresponding acceptor sequence.^11,37

Examples

There are many more examples than could be covered in this review and readers are referred to the immune epitope database for other specific examples (http://www.iedb.org/). Those antibody–antigen pairs discussed below have been chosen to exemplify some of the approaches that have been taken with particular proteins and the value and limits of the information that has been so derived.

Interleukin-17A

CAT-2200 is an antibody that was identified by panning scFv libraries with fully folded IL-17A (a disulphide-bonded homodimer).²² The fine details of both epitope and paratope were revealed primarily by X-ray crystallography and are shown in Fig. 1. The crystal structure showed that all CDRs were involved in recognition and that the epitope involved the quaternary structure of IL-17A. It also enabled some speculation on a molecular understanding of the process of affinity optimization of the antibody. The epitope data were also confirmed using hydrogen deuterium exchange coupled to mass spectrometry (HDX-MS) and mutagenesis. In the HDX-MS approach two major peptides were identified, 45–53 and 71–87. The former contains an Arg that makes several interactions with the heavy chain and the latter contains adjacent Tyr and His residues that make multiple interactions with the light chain. The mutagenesis was based on the HDX-MS data showing an epitope between residues 71 and 87, and took advantage of the observation that the antibody bound human IL-17A but not murine IL-17A. The mutant with all the residues exchanged in this region to their murine equivalents was active but failed to bind CAT-2200. Because of the complex conformational nature of the epitope it is clear that peptide-based approaches would be unlikely to be successful.

Figure 1.

Wall-eyed stereo representation of the interleukin-17A (IL-17A) epitope recognized by CAT-2200. Residues from the IL-17A homodimer are shown in pale or dark yellow. Residues from the Fab fragments are shown with the light chain in blue and with the heavy chain in green. Labelled residues allow the interactions listed in table 2 from ref. 22 to be readily located. Reproduced with permission from ref. 22.

This fine mapping clearly illustrates why the antibody inhibits IL-17A/IL-17 receptor interactions. The recently published structure of IL-17A with IL-17RA shows that there is a high degree of overlap between the regions responsible for binding both CAT-2200 and IL-17RA fibronectin domain D2.⁵³ The IL-17A:CAT-2200 structure also exemplifies the value to using an Fab to solve the structure of a protein that had been difficult to crystallize by itself because all the crystal contacts are mediated through the Fab itself.²²

Tumour necrosis factor-α

Antibodies to tumour necrosis factor (TNF-α) are one of the longest established biological drugs and are used as treatments for inflammatory diseases. In addition to the four antibodies (adalimumab, infliximab, golimumab and certolizumab pegol) a fifth marketed anti-TNF-α reagent is a soluble TNFR2-Fc fusion (etanercept). Surprisingly, it has only been very recently that the interactions of some of these monoclonal antibodies have been mapped in fine detail.^24,54,55 Figure 2 highlights the key interaction sites of three of these biological agents together with the key interactions of the related TNF-β with TNFR1. What is striking is the high degree of overlap between adalimumab interactions with TNF-α and those of TNFR2. In contrast, the overlap between infliximab and the receptors is less marked although presumably sufficient to block the interaction between TNF-α and the receptor. It is suggested that this might account for the greater efficacy of adalimumab.⁵⁴ Interestingly a previous report on epitope mapping studies of infliximab had used deletion and point mutagenesis studies linked to yeast display and subsequent binding studies of the antibody.⁵⁶ The point mutagenesis approach successfully identified some key residues towards the C-terminus (137–141) that are also identified in the crystal structure but did not identify the key interaction site in the E-F loop and also, in the absence of the crystal structure, left concerns that the mutant molecules might be folded incorrectly.

Figure 2.

A comparison of the interface between tumour necrosis factor-α (TNF-α) and receptors and infliximab/adalimumab Fab complexes. A comparison of the interface between TNF-α and receptors and monoclonal antibodies is shown. TNF-α from the complex structures is represented as a coloured surface with TNFR2 and the monoclonal antibody Fabs interface highlighted in red at one of three interfaces on the TNF-α trimer. The E-F loop region, which is missing in the TNF-α–TNFR2 (a) complex because of a lack of interaction, is labelled. The TNF-β from the TNF-β–TNFR1 (b) complex structure is shown as a coloured surface with one of the TNFR1-binding sites highlighted in red. The TNF-α–infliximab Fab (c) and the TNF-α–adalimumab Fab (d) interfaces are shown as coloured surfaces with TNF-α-binding sites highlighted in red. Reproduced with permission from ref. 54.

Botulinum neurotoxin

Antibodies to botulinum toxin from Clostridium botulinum are of interest to treat botulism either as a consequence of natural infection or potential bioterrorism. Several monoclonal antibodies have been developed to neutralize botulinum neurotoxin serotype A (BoNT/A). A knowledge of their domain and fine-level epitope mapping have been shown to be helpful in understanding the synergy achieved when using all three antibodies together rather than separately.⁵⁷ This mapping was achieved by using yeast surface display of a series of domain and truncation mutants coupled with flow cytometry. This, together with a crystal structure of the BoNT/A, enabled fine mapping to be achieved by Ala scanning across small regions combined with accurate K_d determination of specific Ala mutants in comparison to the wild-type sequence. This then enabled a model to be created suggesting that it was possible for all three antibodies to be bound at the same time. Hence fine epitope mapping helped to explain antibody function.

EphA2

Antibodies to the erythropoietin-producing hepatoma (Eph) family of receptors hold much interest, due to the critical role that ephrins and their receptors play in modulating cell interactions and cell migration. In particular, EphA2 has been shown to play a key role in many cancers, with over-expression being associated with poor prognosis of disease.^58,59 Eph receptors are type I transmembrane receptor tyrosine kinases, containing an extracellular region made up of an N-terminal ligand domain, a cysteine-rich region and two fibronectin type II repeats.⁶⁰ An EphA2-specific neutralizing, internalizing antibody, 1C1 has been described as a potential antibody drug conjugate, for delivery of a toxic payload to EphA2-expressing cells. This antibody was observed to induce rapid phosphorylation, internalization and subsequent degradation of EphA2 upon binding.⁶¹ To better understand these biological effects, the epitope was determined using both X-ray crystallography and a mutational approach.¹¹ Given the agonistic characteristics of the monoclonal antibody, Peng and colleagues first employed ELISA to confirm 1C1 IgG and Fab bound to EphA2 in the ligand-binding domain. Taking advantage of the fine specificity of the antibody and the lack of binding to EphA4 they then constructed a series of 30 ‘knock-out’ or ‘knock-in’ mutants using the arrangement of secondary structural elements in EphA2 to exchange short segments of the EphA2 sequence with their EphA4 counterparts or vice versa. In this way several regions of EphA2 were determined to be involved in the binding and specificity of 1C1. Combinatorial mutants encoding amino acid substitutions from these identified regions demonstrated that the epitope for 1C1 was structural and comprised several discontinuous stretches of sequence. Further engineering allowed the identification of three critical positions responsible for the EphA2 specificity of 1C1 that is lacking in the promiscuous Ephrin family of ligands. To validate this mutational approach and map the antibody paratope the co-crystal X-ray structure of the 1C1 Fab with the EphA2 extracellular domain was solved. This is shown in Fig. 3 and revealed good agreement with the epitope determined using the mutant screening approach, demonstrating the strength of this methodology. Additionally, the co-crystal structure identified the CDR3H loop of 1C1 as the paratope solely involved in direct antigen contact and provided a molecular understanding of 1C1 agonistic properties. This is but one example of how combining multiple approaches ensures success and strengthens conclusions.

Figure 3.

Interactions between Fab 1C1 and EphA2. (a) Three-dimensional view of the Fab 1C1/EphA2 complex. Fab 1C1 heavy chain and light chain are shown in magenta and beige, respectively. Human EphA2 ligand binding domain (LBD) is shown in cyan. (b) Stereographic representations of the intermolecular contacts between human EphA2 and Fab 1C1 CDRH3. Fab 1C1 and human EphA2 are shown in magenta and cyan, respectively. Nitrogen and oxygen atoms are shown in blue and red, respectively. The corresponding interface includes several hydrogen bonds shown as black dotted lines. (c) Fab 1C1 CDRH3 penetrates into a channel of the EphA2 molecule via its predominantly hydrophobic tip. Sulphur atoms are shown in yellow, whereas the rest of the colour code is identical to that in (b). (d) The maximum likelihood weighted 2mFo-DFc electron density map is shown around the area of Fab 1C1 CDRH3 penetration into EphA2. Colour code is identical to that in (b). The map is contoured at 1·5 σ. Reproduced with permission from ref. 11.

Interleukin-1β

Interleukin-1β is a key molecule in the orchestration of inflammatory and immune responses and as such has been an important target for the development of therapeutic antibodies. The functionally competent receptor is a heterotrimeric complex of IL-1β and two membrane proteins, the high affinity IL-1 receptor type 1 (IL-1R) and the IL-1 receptor associating protein (IL-1RacP). Blech et al.³² mapped the fine interactions of two different monoclonal antibodies (canakinumab and gevokizumab) using both X-ray crystallography and NMR. The methods used to determine the crystal structure of a binary complex of IL-1β and a Fab fragment of the two antibodies were standard protocols described above. Mapping of the interactions using NMR used the previously published backbone resonance assignments of Driscoll et al.⁶² Two-dimensional TROSY NMR correlation spectra were then obtained of the free and Fab-bound ¹⁵N-labelled IL-1β. By reference back to the full backbone resonance assignment it was then possible to identify those residues with major chemical shifts that were assumed to be in contact with the Fab. An example of the two-dimensional TROSY NMR spectra is shown in Fig. 4. A comparison of the contact residues from the two approaches taken is shown in Table 1. When the X-ray structure was used as reference there was 80% concordance between the two methods. Reasons for the discordance are primarily that a crystal structure is a rigid structure that allows antigenic epitopes to be defined by clear distance criteria, whereas an NMR structure is of a dynamic protein that makes interpretation more subtle. In addition, proline residues do not have an amide hydrogen and will therefore not be visible using ¹⁵N NMR. The study enabled the authors to suggest a set of guidelines to apply in the interpretation of NMR epitope mapping data although they concede that these rules will differ with different complexes and different experimental set ups. It is also worth pointing out again that the technical hurdles of requiring expression of isotopically labelled protein in E. coli and a previous backbone resonance assignment are difficult to overcome.

Figure 4.

Nuclear magnetic resonance (NMR) epitope mapping by backbone amide chemical shift analysis. The representative superposition of the two-dimensional ¹⁵N TROSY NMR spectra of free (shown as black NMR resonances) ¹⁵N isotope-labelled human interleukin-1β (IL-1β) and in complex with the unlabelled gevokizumab Fab (shown as red NMR resonances) with the assignment as indicated. Two small regions from the spectra demonstrate NMR resonances undergoing no chemical shift (e.g. C8, E64, and L10), minor chemical shifts (e.g. V19, K27, S34, S70, and F146), line broadening (e.g. K27), and disappearance or incapability of being reassigned (e.g. V72 and Y121) upon binding. Reproduced with permission from ref. 32.

Table 1.

Comparison of interleukin-1β (IL-1β) contact residues identified by X-ray or nuclear magnetic resonance (NMR). Reproduced with permission from Ref. 32

The conclusions of the study by Blech et al.³² confirm and very precisely define much of what had been previously published and so give a clear picture of the fine details of these interactions. Hence, previous mutagenesis studies of canakinumab^63,64 are confirmed and described in much greater detail. The competitive nature of the inhibition of canakinumab and IL-1β is clearly revealed. Finally, a molecular description for the previously suggested allosteric mechanism of gevokizumab⁶⁴ is clearly demonstrated.

Interleukin-13

Interleukin-13 is a T helper type 2 cytokine with a key role in some inflammatory diseases such as asthma and allergy. For this reason, a number of different antibodies are being evaluated for their efficacy to treat these diseases. Interleukin-13 exerts its function through a heterodimeric complex of IL-13Rα1 and IL-4Rα. The crystal structures of two of these potential therapeutic antibodies (CNTO607 and lebrikizumab) have been published recently.^10,27 These structures reveal that the two antibodies have differing mechanisms of inhibition, CNTO607 binding an epitope that overlaps with the receptor binding site to IL-13Rα1 and IL-4Rα whereas lebrikizumab inhibits binding to IL-4Rα but almost certainly not to IL-13Rα1 because binding is too distant from that site. The authors speculate that the differing epitopes on IL-13 may be one contributory factor in the different effects seen in clinical studies with different antibodies. Both studies use mutagenesis approaches to add further detail to the crystal structures. With lebrikizumab key residues of the paratope, i.e. the antibody, were mutated to dissect out the relative contribution of different residues. In the CNTO607 study,²⁷ alanine scanning mutagenesis studies were used on the IL-13 contact residues to examine the effect on the binding affinity and thereby differentiate which were part of the structural epitope and which were part of the functional epitope (Fig. 5). In this structure 13 residues on two adjacent α-helices were visualized as being in contact with the CNTO607 Fab. A total of 21 solvent-accessible residues on these two helices were mutated individually to alanine and the effect on binding affinity was analysed. Six mutants exhibited significantly weaker binding but only three of these residues were in contact so presumably the other three were part of a functional epitope but not a structural epitope. Of the other 15 mutations there are examples of all permutations, i.e. residues that are part of the structural epitope and that influence binding; those in the structural epitope that have no effect on binding; residues that are not in apparent contact but that do influence binding and are therefore considered to be a part of a functional epitope; and lastly residues outside the structural epitope that have no effect when mutated. This example therefore illustrates the caution that must be taken when interpreting and explaining data from different approaches.¹⁶

Figure 5.

Alanine-scanning mutagenesis of interleukin-13 (IL-13). (a) Relative binding of the fluorescence-labelled CNTO607 monoclonal antibody to the IL-13 muteins measured by ELISA. Wild-type IL-13 is 100%. (b) Ribbon presentation of IL-13 with mutated residues shown as sticks (relative binding of 0–30%, yellow; 30–70%, green; > 70%, cyan). Reproduced with permission from Ref. 27.

Summary

There are many ways to map the fine interactions between an antibody and its antigen. It is useful to have a number of these available to characterize different antibodies with different antigens given that no one method will suit every situation. If feasible, X-ray crystallography is certainly the method of choice, as illustrated by the examples above, although even a structure needs further experimental approaches to understand which residues either inside or outside the structural epitope are contributing to binding affinity and therefore forming the functional epitope.

Disclosures

The authors declare that they have no financial or commercial conflicts of interest.