Choosing Peptide Sequences

Many peptide sequences can be immunogenic, but not all are equally effective at eliciting antibodies that crossreact with the intact cognate protein (we term these crossreactive peptides). There is no guarantee that antibodies raised against a particular synthetic peptide will crossreact with the intact protein from which the sequence is derived. In our experience the probability of generating a successful anti-protein antibody by the methods outlined is approx 50%. Many factors can influence the success of using peptide immunogens to raise antiprotein antibodies. These include elements such as the number of peptides from one protein sequence to be used and the number of animals available for immunization (both of which may be determined by existing resources); the availability and accuracy of sequence data, the predicted secondary structure of the intact protein and finally, the ease of synthesis of specific sequences. Continual improvements to synthesis methodologies means that the latter aspect is less significant than in the past, although certain sequences can still be problematic (see Subheading 1.2.). Despite these potential reservations, there are a number of ways of improving one's chances of success (see Subheading 1.1.1.-1.1.3).

1.1.1. Predicted Structure of the Whole Protein

There is a wide range of predictive algorithms available that can provide data on antigenicity, hydrophilicity, flexibility, surface probability, and charge distribution over a given amino acid sequence. The algorithms of Chou and Fasman and of Robson and Garnier (2,3) have provided a basis for many secondary structure predictive algorithms that can give a good idea of where regions of particular secondary structure, such as a-helix, p-sheet, turns, and coils are likely to form. For example, the proteomics server of the Swiss Institute of Bioinformatics (http://www.expasy.ch/) provides access to primary and secondary structure analysis tools via the Expert Protein Analysis System (ExPASy). Other prediction scales include the Turn scales of Pellequer and Westhof (4). These are based on the occurrence of amino acids within turns. The level of correctly predicted antigenicity using this program is high (70%), but the number of predicted antigenic sites per protein is smaller than for other programs. In general, however, there is rather poor correlation between amino acid type and secondary structure with similar folds able to be made by sequences with only 20% identity. The relative merits of different predictive scales is discussed in depth elsewhere (5).

Primary amino acid sequences can also indicate consensus sequences that may be sites of posttranslational modification (e.g.,

O- and ^-linked glycosylation sites and sites of phosphorylation) and that may therefore be immunologically unavailable in the fully mature protein. Clearly, accessibility on the external surface of the intact protein is, overall, the most important requirement for a cross-reactive peptide. Very frequently, the C-terminus of a protein, although often not a region of strongly predicted secondary structure, is exposed, and this sequence makes a good first choice. However, the C-terminus occasionally forms the membrane anchoring region of some membrane-bound proteins and in these cases would generally be too hydrophobic to consider. The N-terminus of a protein can also prove to be a good candidate sequence, but in our experience is a less reliable choice than the C-terminus and may be modified or truncated. Regions with too high a charge or hydrophi-licity are sometimes not as effective as might be expected, probably because almost all known antibody combining sites make contact with their epitope via polar and Van der Waal's bonds and not-ionic interactions. Hydrophilic a-helical regions can be good pep-tide epitopes because, provided the synthetic peptide is itself long enough to form a helix, it often assumes an identical conformation to that in the intact protein.

1.1.2. Specific Requirements

By their nature, antipeptide antibodies are site-directed probes for proteins. Both the sequence and position of the antibody epitope is predefined. Indeed, the technique of "epitope tagging" exploits the existence of an antibody with specificity for a given linear pep-tide epitope that can be expressed in the context of a fusion protein (6). It is, therefore, possible to target antipeptide antibodies to specific regions of interest in the intact protein, such as areas of high conservation to identify additional members of a protein family; or areas of hypervariability in order to unambiguously identify a particular family member. The increasing reliability of synthesis of, for example, phosphopeptides means that sites of posttranslational modification can also be analyzed. Antibodies that recognize both degenerate and specific consensus phosphorylation motifs are avail able commercially and antibodies raised against a specific phosphopeptide have been used as tools to recognize novel phosphorylation targets (7,8). When selecting a peptide to produce a phosphospecific antibody, it is preferable to localize the phospho-rylated residue close to the middle of the peptide to reduce the likelihood of producing an immunodominant epitope containing nonphospho amino acid sequence. Other functional or regulatory regions of a protein, such as binding sites, transmembrane domains or signal sequences may also be targeted. However, factors , such as hydrophilicity and secondary structure, may affect the success of any given peptide immunogen.

1.1.3. Immunological Requirements

Peptides of 10-20 amino acids are optimal as antigens and our standard is approx 15 residues. Short peptides (less than approx 7 residues) are probably of insufficient size to function as epitopes. Larger peptides may adopt their own specific conformation (that is often immunodominant over any primary structural determinants), which may not be reflected in the conformation of the sequence within the intact protein. Given the previous criteria, it is possible to say that almost all peptide sequences are immunogenic if presented to the immune system in the right way (see Subheading 1.3.), but that not all will generate cross-reactive antibodies. Probably the most important factors in optimizing one's chances of making useful antibodies to a protein of interest are to use several peptides from different regions of the protein sequence and to immunize more than one animal with each peptide. Different animals within the same group frequently respond differently to the same immunogen. In addition, a given antipeptide antibody may sometimes work well in one assay, for example, Western blotting, but not in another, for example, immunoprecipitation.

1.1.4. Synthesis Requirements

The chemical difficulties of synthesizing certain amino acid sequences can be complex. In general, hydrophilic sequences are more soluble and easier to synthesize (and are more likely to be exposed on the surface of the intact molecule). There appears to be little requirement for a high degree of purity for peptide immunogens. Our experience is that peptides of 75% purity, or sometimes even less, generate effective polyclonal antisera, although criteria may need to be more stringent when making monoclonal antibodies.

0 0

Post a comment