Choice of Peptide Sequence for antibody production
If there are no constraints on antigen choice for a cytosolic or soluble polypeptide or the rationale for choosing an epitope is ambiguous, a pragmatic approach to choosing a peptide sequence for immunization is to synthesize the N-terminal and C-terminal sequences of the protein. These sequences are often found to be solvent exposed and mobile in crystallographic structures of proteins. There may be a higher likelihood that an antibody prepared against these sequences will work well in immunoblot analysis and in analyses of native proteins by immunoprecipitation. When using an N-terminal peptide, conjugation to the carrier should be achieved through the carboxyl terminus, so that the peptide will mimic its position in the protein. Similarly, a C-terminal peptide is best conjugated at its amino terminus.
Computer algorithms are frequently used to select epitopes from predicted protein sequences. The most frequently used are those that predict predominantly hydrophilic and hydrophobic portions of polypeptides, helping identify solvent-exposed regions on the surface of the protein.11 Programs such as those provided by the Wisconsin Genetics Computer Group, commercial programs, and others available on the World Wide Web12 are helpful. However, protein folding is not entirely understood. Unless a prediction of surface residues is based on modeling of a known protein three-dimensional structure, these hydrophilicity plots are not experimental data but only useful guides. If possible, more than one of these sites should be chosen to produce antibodies specific to a desired protein. More information about which peptide sequences bind to class I and class II major histocompatibility complex molecules is also being discovered.13,14This knowledge base may help future peptide antigen design.
It is possible that a desirable peptide antigen may share by chance a region of homology or identity with a sequence in a totally unrelated protein. This can lead to experimental findings that are misleading or that require much analytic work to understand.22 Searching several databases using Web-based protocols such as BLAST can help prevent the occurrence of this type of problem.23 Modifications to the standard search protocols are usually necessary when short amino acid sequences are used in the query and have been well described elsewhere.17
A major design concern about preparing sequence-specific antibodies is the choice of determinants common to a family of proteins or unique to one member of that family. This selection depends on the experiments a laboratory needs to perform. Specific and general reagents may be required. Programs located on the World Wide Web 18,19facilitate alignment of the amino acid sequences of multiple members of a protein family. Sequences or functional domains common to several related polypeptides can be highlighted. This alignment makes identification of sequences common to the same protein from multiple species straightforward, ensuring the general utility of the antibody reagent developed and enhancing its potential usefulness for discovering other related proteins. These same sequence alignments permit ready visualization of sequences that are found in only one member of a protein family or in one species of that family, aiding development of highly specific reagents.
Many common sequence motifs have been exploited during evolution to maintain specific protein functions. These can include helix-loop-helix sequences, GTP binding sites, RGD recognition motifs, SH2 domains, phosphorylation consensus sites, and others. Analyzing the predicted or known protein sequence by Web-based programs can exclude these regions as candidates for antigens.20 The programs that identify hydrophilic and hydrophobic regions within a polypeptide can help exclude the hydrophobic transmembrane regions of intrinsic membrane proteins. There are specific programs that predict these regions. Even if these sequences may elicit an antibody, the antibodies will not be as useful as antibodies to exposed portions of the membrane polypeptide.
Antibodies that discriminate between phosphorylated and unphosphorylated versions of a protein (or other modifications) can be powerful functional and positional markers in cell biology experiments.21 It is now relatively straightforward to synthesize phosphopeptides. Monoclonal antibodies directed against phosphorylated sequences can be readily prepared. Polyclonal antisera against phosphopeptides may contain antibodies to the unphosphorylated sequence, because some of the peptide becomes dephosphorylated after injection. However, use of peptide-affinity resins should permit purification of either or both antibodies to phosphorylated and unphosphorylated forms of the peptide or protein. Alternatively, monoclonal antibodies can be used to develop reagents specific for only one form of the peptide.21
If left with a set of peptide candidates, which ones should be chosen? If only one peptide will be made, the sequence that can provide the most straightforward synthesis and can be most soluble and easy to handle should be chosen. If funds for peptide synthesis, coupling, and immunization protocols are not limiting, at least two sequences, preferably from different regions of the polypeptide, should be chosen. Because epitope prediction programs are only partly based on experimental data, preparation of several antigens provides some assurance that the needs of a wide range of antibody applications can be met. Many proteins are processed proteolytically within the cell, often in a cell-specific manner. The availability of more than one sequence-specific antibody ensures that the protein can be universally detected and provides tools for studying in vivo processing of the polypeptide.
A special set of sequence constraints should be considered for peptides with hormonal activity. Endocrine peptides are generated from longer precursor polypeptides after cleavage by a specific set of proteolytic enzymes in a defined order, perhaps in a unique subcellular compartment. Many endocrine precursors are multivalent, capable of generating several different sets of biologically active peptides in different cell types or different physiologic conditions. Peptides synthesized based on the predicted cleavage products can be most useful. These peptides often have unique posttranslational modifications important for their biologic activity. One of the most common is C-terminal amidation, suggested by the presence of a C-terminal glycine just before the basic proteolytic processing site. Capping by acetylation and amidation are used to eliminate a potential charge in the peptide antigen that was not present in the native protein from which the peptide sequence was derived from. Extra electrostatic charges may affect the proper folding of the protein [16] and, therefore, potentially alter antibody specificity 1-12. Biologic activity and antibody specificity can depend on these modifications. Particularly for endocrine peptides, a Tyr residue is often incorporated into a portion of the peptide synthesis to provide a labeling site for later radioimmunoassay development.
Other concerns arise when posttranslational modifications are suspected. Unless these issues are the topic of the investigation, initial attempts at antibody preparation should avoid regions rich with probable disulfide bonds or modified residues.
As a final precaution before initiating synthesis of an antigen, the sequence selected should searched through available databases to ensure that the sequence is unique and that the antibody reagent developed will not later be found to cross-react with an unrelated protein with a chance identity in a small region of its sequence. 15