There are two important aspects to consider when looking for potential B cell epitopes. The first is the primary amino acid sequence of the protein and the second is the conformational structure of the protein. The surface of a protein is not a simple string of amino acids, but a complex of loops and folds determined by the interaction between side chains of the residues.
B cell epitopes consist of groups of amino acids that lie close together on the protein surface and that determine antigenicity (www.roitt.com/elspdf/Epitope_mapping.pdf).
There are two main classifications of B cell epitopes:
Conformational epitopes are thought to form the majority of strong antibody binding epitopes on most proteins. However numerous conformational epitopes may also be recognizable as linear epitopes. Linear B cell epitopes typically vary from 5 to 20 amino acids in length.
(The term 'Functional Epitope' may sometimes be used to refer to elements of a protein which are essential for antibody binding but which do not actually contact the antibody. On these pages we will define an epitope as only the region contacted by an antibody.)
This makes B cell epitopes more difficult to identify than T cell epitopes, which are wholly determined by their amino acid sequence.
Epitope Mapping Methods
When the 3-D structure of a protein is known, or can be modelled based on the established structure of a closely related protein, the structure can be used in an algorithm based approach to predict surface epitopes. Despite advances in crystallographic methods, a relatively small proportion of protein structures have been established, and structures obtained are usually those of small soluble proteins which are easily crystallized. For very small proteins, or domains of larger proteins, NMR structure may be available for mapping, and for very large assemblies (such as viruses), electron microscopy can be used to identify approximate contact regions.
The majority of B cell epitope mapping experiments make use of linear peptide fragments from
antigenic proteins. These peptides can be homologous enough to parts of
the whole antigen in order to allow antibody binding.
Enzyme-Linked Immunosorbent Assay (ELISA) is still the basis of much physical epitope mapping, and remains a very useful technology. The binding of whole proteins, protein fragments or peptides to an antibody can be quantitated, so that epitopes can be identified.
For a higher throughput approach, and even more accurate measurement of relative binding, peptide microarrays are increasingly being used. The basic technology is similar, but using microarray hundreds or even thousands of slide-immobilized peptides or proteins can be screened at once, in a very short space of time.
An overlapping peptide library can be generated to isolate the candidate antibody-binding region of a protein, or even an entire protein. B cell epitopes do not have a defined length and can vary from 5 to 20 amino acids in length. One approach would be to set a defined peptide length, e.g. 15-mer, and to generate a library of peptides with a uniform overlap e.g 3 amino acids. In this way 'hot-spot' areas of the protein can be identified and the optimal epitope can be found and defined using further experiments with more closely overlapping peptides, or an alanine scanning mutant peptide library.
The great advantage of peptide libraries is that the whole antigen does not need to be available. In cases such as high category biohazard viruses, this is even more of a benefit, as researchers are able to understand their proteins without putting themselves at risk.
Peptide libraries are increasingly easy to generate, particularly using technologies such as ProImmune's PEPscreen® platform. Peptide libraries can be manufactured with a tag to allow for plate or array immobilisation. Arrays in particular are very amenable to use with peptide libraries, as both facilitate a high-throughput approach.
Phage display can be used to express sequences derived from DNAse digestion, to generate a random peptide library. Clones expressing antigens which bind test antibody can then be sequenced, and the protein sequence of the antigen elucidated. This method is labour- intensive, but requires minimal knowledge of the starting proteome, so, for example, whole viruses can be investigated without knowledge of which of their coded proteins are expressed.
For low resolution mapping, competition can be very useful. In a typical experiment an antigen would be incubated with two antibodies sequentially, to see if they recognized the same (in which case the second would be unable to bind) or different (in which case both would bind) regions of the target.
For higher resolution mapping, peptide inhibition can be used to validate epitopes. Competing peptide or whole-protein is titrated in to a binding assay experiment, to compete with binding of antibody to plate-immobilized antigen. After washing, antibody bound to competing peptide is lost, and the signal from plate-coupled antigen bound to antibody will be reduced, confirming the specificity of binding. An irrelevant blocking peptide will have no competitive effect on the antigen binding. This is quite a time-consuming process if it is repeated for every potential epitope, so it is best used as a confirmatory strategy.
Protection of the antibody-bound antigen from acetylation, or partial protelytic digestion ("protein footprinting"), followed by sequencing of the products by mass spectrometry or N-terminal microsequencing, can be used to identify antibody-bound regions of a protein.
Partial proteolytic digestion of a protein antigen , followed by denaturation and Western blotting with the test antibody, can be used to pull out linear epitopes, and again these can be identified by sequencing or spectrometry. Denaturing versus native gel electrophoresis, followed by western blotting with the test antibody, can be a reliable way of testing whether an antibody recognizes a linear or conformational epitope.
The data generated from ELISA protocols depends upon the quality of the antibody that is being tested. The clearest results are obtained from monoclonal antibodies which, by their very nature, respond to only one epitope. With polyclonal antibodies, there is the possibility that antibodies that bind to specific peptides may be recognizing denatured antigen or cross-reacting with an unrelated antigen. Serum samples, which can contain multiple antibodies, often give very weak signals and may give responses to several peptides. Whilst the data generated from experiments with polyclonal antibody or serum samples may initially be confusing, this information can provide a good indication of the best areas of the antigen on which to carry out further investigations.