TCR Binding Prediction

Recognition of pathogenic peptides contained in major histocompatibility complex (MHC) molecules is an early event in the T cell mediated immune response. Successful recognition of antigenic peptides bound to MHC (pMHCs) requires specific binding of the T Cell Receptor (TCR) to these complexes. Each chain of TCR consists of two immunoglobulin-like domains, the variable domain (Vα and Vβ) and the constant domain (Cα and Cβ). In the vertebrate immune system, tremendous T cell variants, different from each other in the TCR, are encoded by gene segments joined in a process known as v(d)j recombination, which occurs during the T cell maturation in the thymus. Gene segments are combined while nucleotides are randomly introduced within the variable domains. The binding interface of the TCR to the peptide-MHC molecule complex (pMHC) is formed by loops named as complementary determining regions (CDR), and each chain of TCR contains three CDRs. Within the TCRβ chain, the CDR1 and CDR2 loops of the TCR contact the MHC alpha-helices while the hypervariable CDR3 interact mainly with the peptide. Notably, CDR3 loops have the highest sequence diversity and are the principal determinants of receptor binding specificity.

Figure 1 Representation of the TCRpMHC complex (PDB-ID 2bnq). (Thomas Hoffmann, et al. 2018)

Considering the TCR diversity and the high polymorphism of the MHC molecules, it is of crucial importance to complement time-consuming experimental structural techniques by developing reliable structural methods. Advanced modeling approaches can help in the field of rational TCR design/optimization (e.g., adoptive T cell cancer therapy) and vaccine design.

Molecular Modeling of TCRpMHC Complexes (Computational Method)

Precise description on an atomistic level.
Compare broad sets of TCR structures with different crystal structures.
Analysis of the inter-domain angle between the Vα and Vβ TCR domains.
Rigid body optimization (a rigid body energy minimization approach).

Sequence-based Prediction Methods and Related Technologies

High-throughput DNA sequencing of TCR sequences.
Probabilistic approach for most highly cross-reactive TCRs.
Gaussian Processes, Random Forest, Convolutional Neural Networks, Recurrent Neural Networks.
LSTM based model architecture (Figure 2), Autoencoder based model architecture.

Figure2 LSTM based model architecture

Datasets of Published TCR Binding Specific Peptides

McPAS-TCR
VDJdb

Standard Method to Estimate Predictions

Single Peptide Binding—SPB.
Multi-Peptide Selection—MPS.
TCR-Peptide Pairing I—TPP-I.
TCR-Peptide Pairing II—TPP-II.
TCR-Peptide Pairing III—TPP-III.

Currents tools are based on conserved motifs and are applied to peptides with many known binding TCRs. Our expert team has employed Long short-term memory (LSTM) networks based methods for the prediction of TCR binding using different parallel encoders. It is a highly specific and generic TCR-peptide binding predictor.

References

Tonegawa S. Somatic generation of antibody diversity. Nature. 1983;302(5909):575–81.
Michielin O, Luescher I, Karplus M. Modeling of the TCR-MHC-peptide complex. J Mol Biol. 2000;300(5):1205–35.
Davis MM, Bjorkman PJ. T-cell antigen receptor genes and T-cell recognition. Nature. (1988) 334:395–402.
Krogsgaard M, Davis MM. How T cells ‘see’ antigen. Nat Immunol. (2005) 6:239–45.
Hoffmann, T., Marion, A. & Antes, I. DynaDom: structure-based prediction of T cell receptor inter-domain and T cell receptor-peptide-MHC (class I) association angles. BMC Struct Biol. 17, 2 (2018).
Springer I, Besser H, Tickotsky-Moskovitz N, Dvorkin S and Louzoun Y (2020) Prediction of Specific TCR-Peptide Binding From Large Dictionaries of TCR-Peptide Pairs. Front. Immunol. 11:1803.