De novo design and structure of a peptide-centric TCR mimic binding module

Karsten D. Householder, Xinyu Xiang, Kevin M. Jude, Arthur Deng, Matthias Obenaus, S. Wilson, Xiaojing Chen, Nan Wang, K. Christopher GarcíabioRxiv 2024 · DOI

TL;DR

Researchers have developed a novel, computationally designed alpha-helical protein module that mimics T cell receptors (TCRs) to specifically target cancer cells. This "peptide-centric" mimic achieves high affinity and specificity for a tumor antigen, demonstrating a promising new approach for developing targeted cancer immunotherapies with potentially fewer side effects.

What Problem Does This Paper Address?

Cancer immunotherapy aims to harness the immune system to fight cancer. A key strategy involves targeting tumor-specific antigens presented on the surface of cancer cells by peptide-major histocompatibility complex (pMHC) molecules. T cell receptors (TCRs) naturally recognize these pMHCs, but their low affinity often limits their therapeutic efficacy. TCR mimics (TCRm) are engineered molecules designed to bind pMHCs with higher affinity, offering a potential solution. However, existing TCR mimics, often antibody-based, can suffer from cross-reactivity with healthy tissues due to interactions with the MHC molecule itself, leading to toxic side effects. The challenge is to design TCR mimics that are highly specific for the peptide presented by the MHC, thereby minimizing off-target interactions and improving the therapeutic window.

Methods

The study employed a multi-pronged approach combining computational protein design with experimental validation:

  • De Novo Protein Design: They utilized RFdiffusion, a deep learning model, to generate diverse alpha-helical protein scaffolds. These scaffolds were then further refined using ProteinMPNN to design specific amino acid sequences. The design process was guided by the target pMHC complex (NY-ESO-1 presented by HLA-A*02), focusing on creating a "peptide-centric" interaction.
  • Computational Docking and Scoring: AlphaFold2 was used to predict the structure of the designed TCR mimics bound to the target pMHC. An interface predicted aligned error (iPAE) metric was employed to score and rank these designs.
  • Experimental Screening: Yeast display was used to express and screen libraries of designed TCR mimics for binding to the target pMHC.
  • Biophysical Characterization: Surface plasmon resonance (SPR) was used to measure the binding kinetics and affinity of the lead TCR mimic.
  • Structural Biology: High-resolution X-ray crystallography was employed to determine the atomic structure of the TCR mimic bound to the pMHC complex, providing crucial insights into the interaction.
  • Structure-Guided Off-Target Prediction: The crystal structure was used to computationally predict potential off-target peptides from a large database.
  • Functional Assays: Flow cytometry was used to assess binding to target and off-target peptides. T cell engager (TCE) assays were performed using a Jurkat reporter cell line to evaluate the functional activation of T cells in response to the designed TCR mimic.

Datasets

  • Protein Data Bank (PDB): Existing structures of NY-ESO-1 presented by HLA-A*02 were used as input for the design process.
  • MHC Motif Atlas: A dataset of 14,363 mass spectrometry-detected 9-mer HLA-A02 peptides was used for in silico* off-target screening.

Key Insights and Ideas

  • Alpha-helical scaffolds are viable TCR mimic platforms: Contrary to traditional antibody-based approaches, alpha-helical bundles offer a compact, modular, and rigid platform that can be effectively engineered for specific pMHC recognition.
  • "Peptide-centric" design is achievable: By focusing the design on interactions with the peptide rather than the MHC helices, the researchers successfully created a mimic with high specificity for the target peptide.
  • Structure reveals a unique binding mode: The crystal structure showed the TCR mimic adopting a rigid, TCR-like docking mode that specifically engages the "up-facing" side chains of the NY-ESO-1 peptide, particularly the Met4-Trp5 motif. This focused interaction is hypothesized to reduce off-target binding.
  • Computational prediction of off-targets is effective: The structure-guided in silico screen successfully identified potential off-target peptides, which were then experimentally validated. This demonstrates the power of combining structural information with computational screening for predicting and mitigating off-target effects.
  • A wide therapeutic window is possible: The designed TCR mimic, when incorporated into a T cell engager, showed potent activation against the target peptide while exhibiting minimal or no activation against tested off-target peptides, suggesting a promising therapeutic window.

Key Results

  • High Affinity and Specificity: The designed TCR mimic (mini-TCRm 1.1) achieved a nanomolar affinity (K<sub>d</sub> = 9.5 nM) for the NY-ESO-1/HLA-A02 complex and showed no detectable affinity for MART-1/HLA-A02.
  • Structural Validation: The crystal structure at 2.05 Å resolution confirmed the designed binding mode and revealed specific interactions with the NY-ESO-1 peptide.
  • Successful Off-Target Prediction: An in silico screen of over 14,000 peptides correctly predicted two off-target peptides that showed weaker binding than the intended target.
  • Functional T Cell Engager Activity: The TCR mimic, when used in a T cell engager format, potently activated T cells against cancer cells presenting the NY-ESO-1 peptide, with minimal activation against cells presenting off-target peptides.

Limitations and Future Work

  • Limited Off-Target Validation: While the in silico prediction and experimental validation of two off-target peptides were successful, further large-scale validation is needed to fully establish the accuracy and limitations of this approach.
  • Further Optimization: The authors acknowledge that further optimization campaigns might be necessary to further reduce any remaining off-target binding.
  • Clinical Translation: While promising, the developed TCR mimic is a starting point and will require extensive preclinical and clinical development before it can be used as a therapeutic.
  • Broader Applicability: The study focused on a single pMHC target. Future work could explore the generalizability of this alpha-helical scaffold design approach to other tumor antigens and MHC alleles.
  • Computational Advancements: The authors anticipate that continued advancements in computational methods may reduce the reliance on structural validation in future discovery efforts.

Summary updated

Abstract

T cell receptor (TCR) mimics offer a promising platform for tumor-specific targeting of peptide-MHC in cancer immunotherapy. Here, we designed a <i>de novo</i> α-helical TCR mimic (TCRm) specific for the NY-ESO-1 peptide presented by HLA-A*02, achieving high on-target specificity with nanomolar affinity (K<sub>d</sub> = 9.5 nM). The structure of the TCRm/pMHC complex at 2.05 A resolution revealed a rigid TCR-like docking mode with an unusual degree of focus on the up-facing NY-ESO-1 side chains, suggesting the potential for reduced off-target reactivity. Indeed, a structure-informed <i>in silico</i> screen of 14,363 HLA-A*02 peptides correctly predicted two off-target peptides, yet our TCRm maintained a wide therapeutic window as a T cell engager. These results represent a path for precision targeting of tumor antigens with peptide-focused α-helical TCR mimics.
BibTeX
@misc{householder2024i,
  author = {Karsten D. Householder and Xinyu Xiang and Kevin M. Jude and Arthur Deng and Matthias Obenaus and S. Wilson and Xiaojing Chen and Nan Wang and K. Christopher García},
  title = {{<i>De novo</i> design and structure of a peptide-centric TCR mimic binding module}},
  year = {2024},
  doi = {10.1101/2024.12.16.628822},
  howpublished = {bioRxiv},
}
Figure