Development of a strategy and computational application to select candidate protein analogues with reduced HLA binding and immunogenicity

Abstract

Unwanted immune responses against protein therapeutics can reduce efficacy or lead to adverse reactions. T-cell responses are key in the development of such responses, and are directed against immunodominant regions within the protein sequence, often associated with binding to several allelic variants of HLA class II molecules (promiscuous binders). Herein, we report a novel computational strategy to predict 'de-immunized' peptides, based on previous studies of erythropoietin protein immunogenicity. This algorithm (or method) first predicts promiscuous binding regions within the target protein sequence and then identifies residue substitutions predicted to reduce HLA binding. Further, this method anticipates the effect of any given substitution on flanking peptides, thereby circumventing the creation of nascent HLA-binding regions. As a proof-of-principle, the algorithm was applied to Vatreptacog α, an engineered Factor VII molecule associated with unintended immunogenicity. The algorithm correctly predicted the two immunogenic peptides containing the engineered residues. As a further validation, we selected and evaluated the immunogenicity of seven substitutions predicted to simultaneously reduce HLA binding for both peptides, five control substitutions with no predicted reduction in HLA-binding capacity, and additional flanking region controls. In vitro immunogenicity was detected in 21·4% of the cultures of peptides predicted to have reduced HLA binding and 11·4% of the flanking regions, compared with 46% for the cultures of the peptides predicted to be immunogenic. This method has been implemented as an interactive application, freely available online at http://tools.iedb.org/deimmunization/.

Keywords: MHC/HLA; T cell; antigen/peptides/epitopes; bioinformatics; regulation/suppression.

Figure 1

Comparison of erythropoietin (EPO) immunogenicity based on in vitro and in silico studies. (a) Prediction profiles of different in silico approaches to predict immunogenicity based on previous published experimental studies.51 (b) ROC plots of the performances of median percentile rank for 7‐allele, 15‐allele and 26‐allele methods. The dotted line shows the performance with chosen threshold (20%ile) for the 26‐allele method. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 2

DRB1*01:01 antigenicity of alanine containing HA 307–319 analogues. The binding capacity for each amino acid residue after alanine mutation is shown in relation to the wild‐type (WT) peptide. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 3

Schematic presentation of de‐immunization tool. The de‐immunization process is summarized in its two‐step process. In the first step, the immunogenic regions are identified using the 26‐allele method (median percentile rank), whereas in the second step all the possible variants of immunogenic peptides are screened to identify those with decreased (predicted) binding affinities, compared with the wild‐type sequence. As a final step, all the lower binding mutants are then ranked according to their predicted effect on neighbouring peptides. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 4

Web‐tool interface for 26‐allele method immunogenicity calculations. The web de‐immunization tool interface is available in the IEDB Analysis Resource (http://tools-dev.internal.iedb.org/deimmunization/). The tool comprises an input page where sequence and parameter can be added or selected (a), an intermediate page displays the potentially immunogenic regions predicted (b) and a final page that summarizes the overall protein de‐immunization results (c). [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 5

T‐cell reactivity against peptides with increased or decreased immunogenicity. In vitro ELISPOT reactivity measured as the sum of interferon‐γ (IFN‐γ) and interleukin‐5 (IL‐5) producing cells after 2 weeks stimulation of peptides with no decreased immunogenicity (red), decreased immunogenicity (blue) or belonging to neighbouring peptides (black). (a) Frequency of responses per peptide analysed expressed as mean ± SD. (b) Magnitude of responses per peptide analysed expressed as geomean with 95% CI. Statistical analyses are performed using one‐tailed Mann–Whitney U‐test. [Colour figure can be viewed at wileyonlinelibrary.com]