Abstract:Class I MHC molecules present peptides derived from intracellular antigens on the cell surface for immune surveillance, and specific targeting of these peptide-MHC (pMHC) complexes could have considerable utility for treating diseases. Such targeting is challenging as it requires readout of the few outward facing peptide antigen residues and the avoidance of extensive contacts with the MHC carrier which is present on almost all cells. Here we describe the use of deep learning-based protein design tools to de novo design small proteins that arc above the peptide binding groove of pMHC complexes and make extensive contacts with the peptide. We identify specific binders for ten target pMHCs which when displayed on yeast bind the on-target pMHC tetramer but not closely related peptides. For five targets, incorporation of designs into chimeric antigen receptors leads to T-cell activation by the cognate pMHC complexes well above the background from complexes with peptides derived from proteome. Our approach can generate high specificity binders starting from either experimental or predicted structures of the target pMHC complexes, and should be widely useful for both protein and cell based pMHC targeting.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to design small proteins with high specificity to bind to peptide - major histocompatibility complex class I molecules (peptide - MHC - I complexes, pMHCs). Specifically, the research aims to overcome the following challenges: 1. **Difficulties in targeting pMHC**: pMHC displays peptides from intracellular antigens on the cell surface for immune surveillance. However, specific targeting of these pMHC complexes is very challenging because it is necessary to recognize the exposed peptide antigen residues while avoiding extensive contact with the MHC carriers present on almost all cells. 2. **Limitations of existing methods**: - **T - cell receptor (TCR) screening**: The method of screening TCRs from patient samples is limited by the availability of donor HLA, and central tolerance may eliminate the most reactive TCRs against endogenous targets. In addition, the discovery of high - affinity TCRs is relatively rare. - **Single - chain variable fragment (scFv) library screening**: This method usually produces non - specific binders, and even the most successful specific binders tend to adopt a docking geometry similar to that of TCR, indicating the importance of the docking interface for sufficient peptide contact and binding specificity. 3. **The need to cover a wide range of patient groups**: Due to the diversity of MHC alleles and antigen peptides in different patient groups, hundreds or thousands of effective and specific binders need to be developed to achieve wide patient coverage. Existing methods are costly, labor - intensive and time - consuming, and it is difficult to economically cover all pMHC targets, especially small patient subgroups. To solve these problems, the authors use deep - learning - driven protein design tools to de novo design small proteins that can span the central peptide - binding groove of pMHC complexes and make extensive contact with their peptides. Through this method, high - specificity and high - affinity binders can be generated for therapeutic applications, such as chimeric antigen receptor (CAR) T - cell therapy and bispecific T - cell engagers (BiTEs). ### Main objectives of the study - **Design high - specificity binders**: Generate small - protein binders for specific pMHC complexes through computational design, ensuring that they can specifically recognize the target peptides without cross - reacting with other peptides presented on the same MHC. - **Improve stability and manufacturability**: The designed small proteins should have high stability, be easy to express in Escherichia coli, and their binding patterns can be predicted computationally, thus eliminating the need for time - consuming and costly experimental structure determination. - **Expand to multiple pMHC targets**: This method is applicable not only to experimentally determined pMHC structures but also to predicted pMHC structures, thereby greatly expanding the range of potential therapeutic targets. Through the above methods, researchers hope to develop an efficient, economical and widely applicable pMHC - binder design platform, providing new tools for immunotherapy of cancer and other diseases.

Design of high specificity binders for peptide-MHC-I complexes

De novo design of high-affinity binders of bioactive helical peptides

A general platform for targeting MHC-II antigens via a single loop

De novo designed pMHC binders facilitate T cell induced killing of cancer cells

In silico design of high-affinity antigenic peptides for HLA-B44

Machine learning application to predict binding affinity between peptide containing non-canonical amino acids and HLA0201

Design of High Affinity Binders to Convex Protein Target Sites

Ranking-based Convolutional Neural Network Models for Peptide-MHC Binding Prediction

BindCraft: one-shot design of functional protein binders

Ranking of binding and nonbinding peptides to MHC class I molecules using inverse folding approach: Implications for vaccine design

Design of Peptide Binders to Conformationally Diverse Targets with Contrastive Language Modeling

De Novo Design of Peptide Binders to Conformationally Diverse Targets with Contrastive Language Modeling

Development of a Stable Peptide-Major Histocompatibility Complex (MHC) via Sortase and Click Chemistry

Peptide-binding specificity prediction using fine-tuned protein structure prediction networks

DeepMHCI: an anchor position-aware deep interaction model for accurate MHC-I peptide binding affinity prediction

Preparation empty peptide-receptive MHC class I complex for large-scale detection through photolabile peptide ligands

Peptide binder design with inverse folding and protein structure prediction

RLpMIEC: High-Affinity Peptide Generation Targeting Major Histocompatibility Complex-I Guided and Interpreted by Interaction Spectrum-Navigated Reinforcement Learning

A comprehensive review and performance evaluation of bioinformatics tools for HLA class I peptide-binding prediction

PepMLM: Target Sequence-Conditioned Generation of Therapeutic Peptide Binders via Span Masked Language Modeling

DeepMHC: Deep Convolutional Neural Networks for High-performance peptide-MHC Binding Affinity Prediction