In the mid-1980s Townsend et al. (1) demonstrated that viral Ags recognized by class I restricted CTL could be mimicked by incubating target cells with short synthetic peptides representing viral sequences. Soon thereafter, Wiley and coworkers (2) published the structure of the HLA-A2.1 molecule. To the amazement of immunologists, the class I MHC molecule contained an Ag binding groove that was formed by two α helices composed of segments of the MHC molecule that were known to contain a high degree of allelic variation. The groove was filled with electron-dense material. Such a structure elegantly explained both MHC restriction and allele specificity. However, because the material in the groove was unresolved, the identity of the antigenic peptides contained therein remained a mystery. Bevan and coworkers (3) demonstrated that essentially any protein within the cytoplasm was a potential source for peptides that were presented by the MHC. Therefore, it was likely that the material in the groove could not be resolved because it contained a mixture of thousands of different peptides derived from proteins synthesized within the cell.
The identification of antigenic epitopes in viruses, tumors, and alloantigens was of considerable importance, and by the late 1980s many laboratories were trying to identify their favorite class I epitopes. Although the region that contained an epitope could be narrowed down somewhat by the expression of smaller portions of a protein, ultimately it required the synthesis of a large array of overlapping synthetic peptides and the testing of each for recognition by Ag-specific CTL. Several investigators observed that there may be sequence motifs shared among epitopes recognized in association with the same MHC molecule (4). If confirmed, this could lead to algorithms that could help predict where epitopes were likely to be found within a protein. However, there was no information available on the relationship between the synthetic peptides that could drive CTL activation in vitro and the sequences of the peptide endogenously presented by a MHC molecule. In fact, it was not known how many different versions or different lengths of the same epitope were processed and presented by a cell.
Rammensee and coworkers (5, 6, 7) realized that it was important to learn the identity of the epitopes that were endogenously processed and presented. His group embarked on a series of experiments that were reported in a flurry of papers in 1990 and 1991, culminating in the manuscript that is presented as the “Pillars of Immunology” paper. They developed techniques to strip peptides directly out of the grooves of class I MHC molecules and to fractionate them by reversed phase HPLC. They infected cells with influenza virus, stripped the peptide from the grooves of MHC molecules, and displayed the material by HPLC. They observed that the natural peptide that was recognized by CTL as specific for the viral nucleoprotein eluted as a single peak (6). Surprisingly, when they compared this to the chromatographic properties of the synthetic peptides widely used to stimulate these same CTL, they found that the CTL activity did not correspond with the bulk of the material in the synthetic peptide but instead resided in a minor contaminant that was smaller than the synthetic peptide. These results suggested that there was a unique epitope and that it was relatively small. Once the sequence of the active peptide sequence was identified, it proved to be potent at concentrations heretofore unimaginable. Immunologists began talking about fentamoles.
Rammensee and coworkers realized that Kd binding peptides not only contained common features but that they were also likely to be the same length. As stated in the introduction, “Assuming that all naturally processed Kd-restricted epitopes are nonapeptides, all those peptides containing Kd-restricted epitopes could be aligned with a Tyr residue at the second position and an amino-acid residue with a side-chain methyl group… at the last position, suggesting a Kd-specific peptide motif.” If indeed each MHC molecule bound peptides of a preferred length that contained a common binding motif, then, they reasoned, it should be possible to identify the sequence preferred by each MHC molecule by simply stripping the peptides out of the groove with acid and sequencing the mixtures of endogenous peptides that were released by Edman degradation, a process that results in sequential degradation from the amino terminus. This was both audacious and highly successful. The startling results were published in the “Pillars of Immunology” article that follows (8). Rammensee and coworkers were able to identify the presence of preferred amino acid residues within the peptides stripped from each MHC molecule they investigated. At certain positions there were strong dominant residues, as for example the preference for Tyr at position 2 within the Kd binding peptides, whereas in other positions the preference was less distinct. This led to the identification of preferred binding residues at key positions that served as anchor residues for each MHC molecule. Conserved residues at the C terminus and at one or two internal positions were “anchor” residues and defined the motif for each class I molecule. Each MHC molecule selected its own anchor residues.
These results also had implications regarding the mechanism of Ag processing and presentation. The authors state that, “As all epitopes appear to have hydrophobic C termini, whereas the N termini are highly variable, the hydrophobic end might be a site for an MHC-independent protease, producing larger precursor peptides, whose N termini are trimmed after binding to MHC.” Indeed, such a C termini preference is exhibited by immunoproteosomes, which are responsible for peptide cleavage and production of C termini in the cytosol. Furthermore, the amino terminus is indeed trimmed later in the endoplasmic reticulum to permit optimal binding to individual MHC molecules (9, 10).
These results made it possible to use verified binding motifs to help predict sequences within a protein that could bind individual MHC alleles and thereby narrow the universe of potential class I epitopes. This discovery has helped to identify many hundreds of epitopes (11).