Gene and protein sequence features augment HLA class I ligand predictions

Kaspar Bresser; Benoit P Nicolet; Anita Jeko; Wei Wu; Fabricio Loayza-Puch; Reuven Agami; Albert J R Heck; Monika C Wolkers; Ton N Schumacher

doi:10.1016/j.celrep.2024.114325

Gene and protein sequence features augment HLA class I ligand predictions

Kaspar Bresser, Benoit P Nicolet, Anita Jeko, Wei Wu, Fabricio Loayza-Puch, Reuven Agami, Albert J R Heck, Monika C Wolkers, Ton N Schumacher^*

^*Corresponding author for this work

Research output: Contribution to journal › Article › Academic › peer-review

Abstract

The sensitivity of malignant tissues to T cell-based immunotherapies depends on the presence of targetable human leukocyte antigen (HLA) class I ligands. Peptide-intrinsic factors, such as HLA class I affinity and proteasomal processing, have been established as determinants of HLA ligand presentation. However, the role of gene and protein sequence features as determinants of epitope presentation has not been systematically evaluated. We perform HLA ligandome mass spectrometry to evaluate the contribution of 7,135 gene and protein sequence features to HLA sampling. This analysis reveals that a number of predicted modifiers of mRNA and protein abundance and turnover, including predicted mRNA methylation and protein ubiquitination sites, inform on the presence of HLA ligands. Importantly, integration of such "hard-coded" sequence features into a machine learning approach augments HLA ligand predictions to a comparable degree as experimental measures of gene expression. Our study highlights the value of gene and protein features for HLA ligand predictions.

Original language	English
Article number	114325
Number of pages	23
Journal	Cell Reports
Volume	43
Issue number	6
Early online date	11 Jun 2024
DOIs	https://doi.org/10.1016/j.celrep.2024.114325
Publication status	Published - 11 Jun 2024

Bibliographical note

Publisher Copyright:
© 2024 The Authors

Funding

We would like to thank Mireille Toebes for helpful input during generation of the HLA class I ligandome dataset. We would like to thank Prof. Dr. Daniel Peeper for kindly providing the SK-MEL-95 and M026.X1 cell lines. This work was supported by Institute for Chemical Immunology (ICI) grant 003 (to T.N.S.). K.B. conceptualization, methodology, validation, formal analysis, investigation, data curation, writing \u2013 original draft, writing \u2013 review & editing, visualization; B.P.N. conceptualization, methodology, formal analysis, writing \u2013 review & editing; A.J. formal analysis, methodology, and investigation; W.D. data curation; F.L.-P. investigation, methodology, and resources; R.A. methodology, resources, and supervision; A.J.R.H. conceptualization, methodology, resources, supervision, and funding acquisition; M.C.W. methodology, writing \u2013 review & editing, and supervision; T.N.S. conceptualization, methodology, writing \u2013 review & editing, supervision, and funding acquisition. T.N.S. is an advisor for Allogene Therapeutics, Asher Bio, Merus, Neogene Therapeutics, and Scenic Biotech; a stockholder in Allogene Therapeutics, Asher Bio, Cell Control, Celsius, Merus, and Scenic Biotech; and a venture partner at Third Rock Ventures; all outside of the current work. We would like to thank Mireille Toebes for helpful input during generation of the HLA class I ligandome dataset. We would like to thank Prof. Dr. Daniel Peeper for kindly providing the SK-MEL-95 and M026.X1 cell lines. This work was supported by Institute for Chemical Immunology (ICI) grant 003 (to T.N.S.).

Funders	Funder number
Scenic Biotech
Institute for Chemical Immunology	003

Keywords

CP: Immunology
HLA class I
HLA ligand predictions
HLA ligandome
XGBoost
antigen presentation
epitope prediction
epitopes
machine learning

Access to Document

10.1016/j.celrep.2024.114325Licence: CC BY-NC-ND

mmc2 Final published version, 7.1 MBLicence: CC BY-NC-ND

Cite this

@article{182dc40ad8054e5f9858f2d00645d71f,

title = "Gene and protein sequence features augment HLA class I ligand predictions",

abstract = "The sensitivity of malignant tissues to T cell-based immunotherapies depends on the presence of targetable human leukocyte antigen (HLA) class I ligands. Peptide-intrinsic factors, such as HLA class I affinity and proteasomal processing, have been established as determinants of HLA ligand presentation. However, the role of gene and protein sequence features as determinants of epitope presentation has not been systematically evaluated. We perform HLA ligandome mass spectrometry to evaluate the contribution of 7,135 gene and protein sequence features to HLA sampling. This analysis reveals that a number of predicted modifiers of mRNA and protein abundance and turnover, including predicted mRNA methylation and protein ubiquitination sites, inform on the presence of HLA ligands. Importantly, integration of such {"}hard-coded{"} sequence features into a machine learning approach augments HLA ligand predictions to a comparable degree as experimental measures of gene expression. Our study highlights the value of gene and protein features for HLA ligand predictions.",

keywords = "CP: Immunology, HLA class I, HLA ligand predictions, HLA ligandome, XGBoost, antigen presentation, epitope prediction, epitopes, machine learning",

author = "Kaspar Bresser and Nicolet, \{Benoit P\} and Anita Jeko and Wei Wu and Fabricio Loayza-Puch and Reuven Agami and Heck, \{Albert J R\} and Wolkers, \{Monika C\} and Schumacher, \{Ton N\}",

note = "Publisher Copyright: {\textcopyright} 2024 The Authors",

year = "2024",

month = jun,

day = "11",

doi = "10.1016/j.celrep.2024.114325",

language = "English",

volume = "43",

journal = "Cell Reports",

issn = "2211-1247",

publisher = "Elsevier Saunders",

number = "6",

}

TY - JOUR

T1 - Gene and protein sequence features augment HLA class I ligand predictions

AU - Bresser, Kaspar

AU - Nicolet, Benoit P

AU - Jeko, Anita

AU - Wu, Wei

AU - Loayza-Puch, Fabricio

AU - Agami, Reuven

AU - Heck, Albert J R

AU - Wolkers, Monika C

AU - Schumacher, Ton N

PY - 2024/6/11

Y1 - 2024/6/11

N2 - The sensitivity of malignant tissues to T cell-based immunotherapies depends on the presence of targetable human leukocyte antigen (HLA) class I ligands. Peptide-intrinsic factors, such as HLA class I affinity and proteasomal processing, have been established as determinants of HLA ligand presentation. However, the role of gene and protein sequence features as determinants of epitope presentation has not been systematically evaluated. We perform HLA ligandome mass spectrometry to evaluate the contribution of 7,135 gene and protein sequence features to HLA sampling. This analysis reveals that a number of predicted modifiers of mRNA and protein abundance and turnover, including predicted mRNA methylation and protein ubiquitination sites, inform on the presence of HLA ligands. Importantly, integration of such "hard-coded" sequence features into a machine learning approach augments HLA ligand predictions to a comparable degree as experimental measures of gene expression. Our study highlights the value of gene and protein features for HLA ligand predictions.

AB - The sensitivity of malignant tissues to T cell-based immunotherapies depends on the presence of targetable human leukocyte antigen (HLA) class I ligands. Peptide-intrinsic factors, such as HLA class I affinity and proteasomal processing, have been established as determinants of HLA ligand presentation. However, the role of gene and protein sequence features as determinants of epitope presentation has not been systematically evaluated. We perform HLA ligandome mass spectrometry to evaluate the contribution of 7,135 gene and protein sequence features to HLA sampling. This analysis reveals that a number of predicted modifiers of mRNA and protein abundance and turnover, including predicted mRNA methylation and protein ubiquitination sites, inform on the presence of HLA ligands. Importantly, integration of such "hard-coded" sequence features into a machine learning approach augments HLA ligand predictions to a comparable degree as experimental measures of gene expression. Our study highlights the value of gene and protein features for HLA ligand predictions.

KW - CP: Immunology

KW - HLA class I

KW - HLA ligand predictions

KW - HLA ligandome

KW - XGBoost

KW - antigen presentation

KW - epitope prediction

KW - epitopes

KW - machine learning

UR - https://www.scopus.com/pages/publications/85195455692

U2 - 10.1016/j.celrep.2024.114325

DO - 10.1016/j.celrep.2024.114325

M3 - Article

C2 - 38870014

SN - 2211-1247

VL - 43

JO - Cell Reports

JF - Cell Reports

IS - 6

M1 - 114325

ER -

Gene and protein sequence features augment HLA class I ligand predictions

Abstract

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this