Precise prediction of major histocompatibility complex class II-peptide interaction based on peptide side chain scanning.

Abstract

SummaryWe describe here a new method for predicting class II major histocompatibility complex-binding peptides, based on the preferences observed in a systematic series of peptide binding experiments where each position in a "minimal" peptide was replaced individually by every amino acid. The DRBI*0401 peptide binding preferences were determined and incorporated into a computer program that looks through sequences for potential epitopes and assigns each a score. These scores correlate well with previously determined T cell epitopes of foreign antigens and endogenous peptides from self proteins. Our findings hold implications for the design of subunit vaccines and in the identification of autoantigenic peptide regions within self proteins.M HC class II molecules are highly polymorphic proteins that bind peptides derived from processing of antigens and present them to T cells (1). Methods to predict regions in protein sequences capable of binding to human MHC class II molecules would be very valuable for many immunological applications. A quantitative prediction of MHC binding sites, however, requires the understanding of the rules governing peptide--MHC class II interaction. The use of M13 peptide display libraries to identify aUele-specific motifs (2-4), together with structural analyses of HLA-DRl-peptide complexes (5, 6) provided the basis to predict MHC class II-peptide interaction: (a) the identification of precisely spaced anchor positions in large HLA-DR-selected peptide pools and the x-ray crystallographic studies suggested an overall similar, sequence-independent peptide conformation (3, 5); (b) the fact that the anchor at position 1 (pl) 1 was found obligatory for high affinity binding allowed us to define frames in HLA-DR bound peptides (4); (c) the position of a particular side chain, with respect to the pl-frame,'determines whether a given amino acid behaves as anchor residue, inhibitory residue, or residue neutral for binding (4). This information led us to develop a new method for predicting class II MHC-binding peptides. The method is based on the hypothesis that peptide binding to HLA-DR correlates with the net result of all side chain effects in any given peptide, and that most side chain effects depend on their relative position to the pl anchor, rather than on the remaining peptide se- quence. Pl-anchored designer peptide libraries were thus used for scanning the relative peptide positions 2-9 for the effects of each amino acid side chain on binding to the human class II molecule DRBI*0401. When values, individually obtained by side chain scanning, were assigned to the amino acid residues of the selected protein region, the sum of these values correlated with the MHC-binding affinity of the respective peptide. The processing of the side chain scanning data into a software allowed a rapid and nearly perfect prediction of DRBI*0401 binding regions in any protein sequence. Materials and MethodsPreparation of HLA-DRBI*0401 Molecules. DRBI*0401 molecules were isolated from human lymphoblastoid cell...

Specific and general HLA-DR binding motifs: comparison of algorithms

“…As shown in Tables 2 and 3, when these algorithms were tested against several panels of peptides, it was found that they were in most cases able to predict binding of peptides to HLA-DR molecules with similar or higher sensitivity and specificity than other reported algorithms [32,40,42]. This was found, not only when the comparison was made against panels of peptides used to develop our algorithms, but most important, when independent panels of peptides were used.…”

“…We believe that a key step in the deduction of our algorithms is the use of binder and non-binder peptides as well as correcting for the relative abundance of amino acids in proteins. Since our algorithms are based on blocks of 8 amino acids, but the algorithm of Southwood et al [42], of Hammer et al [40], and of Marshall et al [32] use 9, 9, and 11 amino acid blocks, respectively, we adapted their algorithms to the prediction using 8 amino acid blocks. This was done because when their algorithms were tested using blocks of 9 or 11 amino acids like in their original algorithm, the sensitivity of the predictions against the 9-mer peptides eluted from DR1 [29], DRB1*1101, and DRB1*0401 [30], was lower than the one attained using blocks of 8 amino acids (data not shown).…”

“…g Algorithm of Honeyman et al [41]. h Algorithm adapted from [40] to predict binding using blocks of 8 amino acids. i Algorithm adapted from [42] to predict binding using blocks of 8 amino acids.…”

Dissection of the HLA-DR4 Peptide Repertoire in Endocrine Epithelial Cells: Strong Influence of Invariant Chain and HLA-DM Expression on the Nature of Ligands

“…peptides were eluted with 0.1% trifluoroacetic acid according to standard procedures (27). Pellets processed contained 1.5 ϫ 10 9 cells from DR4 and DR4Ii transfectants and 3 ϫ 10 9 cells from DR4DM and DR4IiDM transfectants.…”

Searching for borrelial T cell epitopes associated with antibiotic-refractory Lyme arthritis

“…For example, the algorithm TEPITOPE, which is based on in vitro MHC/ peptide binding of designer peptides, can predict the binding potential of T cell epitopes to 25 different DRB molecules. (Hammer et al, 1994;Hammer et al, 1997;Sturniolo et al, 1999) However, the ability of this program to identify potential disease-associated epitopes with a specific MHC binding profile has not yet been tested.…”