High-affinity, Peptide-specific T Cell Receptors can be Generated by Mutations in CDR1, CDR2 or CDR3

The third complementarity-determining regions (CDR3s) of antibodies and T cell receptors (TCRs) have been shown to play a major role in antigen binding and specificity. Consistent with this notion, we demonstrated previously that high-affinity, peptide-specific TCRs could be generated in vitro by mutations in the CDR3α region of the 2C TCR. In contrast, it has been argued that CDR1 and CDR2 are involved to a greater extent than CDR3s in the process of MHC restriction, due to their engagement of MHC helices. Based on this premise, we initiated the present study to explore whether higher affinity TCRs generated through mutations in these CDRs or other regions would lead to significant reductions in peptide specificity (i.e. the result of greater binding energy gained through interactions with major histocompatibility complex (MHC) helices). Yeast-display technology and flow sorting were used to select high-affinity TCRs from libraries of CDR mutants or random mutants. High-affinity TCRs with mutations in the first residue of the Vα, CDR1, CDR2, or CDR3 were isolated. Unexpectedly, every TCR mutant, including those in CDR1 and CDR2, retained remarkable peptide specificity. Molecular modeling of various mutants suggested that such exquisite specificity may be due to: (1) enhanced electrostatic interactions with key peptide or MHC residues; or (2) stabilization of CDRs in specific conformations. The results indicate that the TCR is positioned so that virtually every CDR can contribute to the antigen-specificity of a T cell. The conserved diagonal docking of TCRs could thus orient each CDR loop to sense the peptide directly or indirectly through peptide-induced effects on the MHC.

Introduction

T cells recognize foreign and self-antigenic peptides that are bound to proteins of the host's major histocompatibility complex (MHC) through a cell-surface T cell receptor (TCR). TCRs are αβ heterodimeric membrane proteins that are generated by somatic rearrangements of multiple genes, analogous to antibodies.¹ TCR:pepMHC interactions shape the TCR repertoire early in thymic development, yielding T cells that have TCRs with low affinity for self-pepMHC ligands (positive selection).² T cells that have TCRs with higher affinity for self-pepMHC (K_D values of about 10 μM or lower) are deleted in a process known as negative selection.3, 4 TCRs also possess the ability to bind to MHC proteins that are encoded by different MHC alleles, resulting in the biological consequence of alloreactivity (i.e. T cells that are involved in transplant rejection).

Crystal structures of ten different TCR:pepMHC complexes have been reported, indicating that there are some general rules based on shared structural features.4, 5, 6, 7 In addition to binding in a diagonal (or orthogonal) orientation over the pepMHC, most contacts between TCR and peptide occur through the CDR3α and/or CDR3β. The finding that CDR3s are ideally positioned to interact with the central region of the exposed peptide correlates with the observation that CDR3s exhibit the greatest variability among the CDRs, as they are encoded by the junction of the V–J (α) or V–D–J (β) genes. In the 2C TCR structures, CDR3s of the α and β chains are positioned to interact directly with the central residues of the peptide bound to K^b.8, 9, 10, 11, 12 The liganded and unliganded structures of the 2C TCR showed a movement in the CDR3α loop of 6 Å, first indicating that TCRs are capable of extensive plasticity in their recognition of pepMHC ligands.⁹ This movement was even greater for the KB5-C20 TCR, in which the CDR3β loop was shifted 15 Å in the liganded state.¹³ CDR3 regions have also been shown to undergo considerable mobility in the NMR structure of the D10 TCR.¹⁴ Nevertheless, CDR3s can also contact the MHC helices as shown with the 2C TCR and more recently with the K^b–alloantigen complex with BM3.3 TCR, in which the CDR3α folds away from the peptide groove and interacts with Gln65 of K^b.¹⁵

In contrast to CDR3s, the CDR2 loops of Vα and Vβ are positioned almost exclusively over the MHC helices and rarely have been shown to have direct contacts with peptide.⁹ On the other hand, CDR1s are typically located over the ends of the peptide (CDR1α over the N terminus and CDR1β over the C terminus) such that they can contact the peptide and/or the MHC helices. While most peptide contacts are with CDR3s, it was shown recently that CDR1α and CDR2α in the EBV-specific LC13 TCR were shifted from a position observed in their unliganded state toward the N terminus of the peptide.¹⁶ Movements in the CDR1 and CDR2 of the LC13 TCR appear to play a role in the structural rearrangements observed in other CDR loops, including the CDR3s. However, the rules that govern which CDR loops of a particular TCR undergo movement and plasticity are completely unknown and are not predictable a priori. Although it can be said with certainty that the level of diversity is greatest for CDR3s, the impact of other loops on peptide specificity is not clear. Alanine scanning mutagenesis of the 2C TCR:QL9/L^d and the 2C TCR:SIYR/K^b interactions have shown that the CDR1 and CDR2 loops contributed approximately two-thirds of the binding free energy.17, 18 As the structure of the 2C TCR:SIYR/K^b complex¹⁰ and the model of the 2C TCR:QL9/L^d complex¹⁹ show that these contacts are largely with MHC, one could reasonably predict that peptide specificity was dictated largely by CDR3 interactions. This notion has been extended in a recent proposal, based on kinetic and mutagenesis studies, suggesting that the CDR1 and CDR2 regions may bind to MHC initially, followed by reorganization and binding of the CDR3s.²⁰

The binding affinities of TCRs obtained from normal peripheral T cells are uniformly low, in the range of 1–100 μM K_D values.3, 21, 22 In previous reports, our lab showed that high-affinity TCRs could be generated in vitro using site-directed mutagenesis of the CDR3α of the 2C TCR.23, 24, 25 The 2C TCR recognizes both the peptide SIYRYYGL (SIYR) bound to K^b26 and the peptide QLSPFPFDL (QL9) bound to L^d 27, 28 as strong agonists. The structure of the 2C TCR has been solved in the unliganded state²⁹ and in complexes with a weak agonist peptide EQYKFYSV (dEV8)³⁰ bound to K^b,⁹ dEV8/K^bm3,¹² and SIYR/K^b.¹⁰ A model of the 2C TCR:QL9/L^d complex has been presented¹⁹ and its features are consistent with a range of mutagenesis studies that have been performed on the 2C TCR17, 18, 31 and the peptide.32, 33 The high-affinity TCRs generated by mutations in CDR3α maintained their antigenic peptide specificity (for SIYR or QL9) and were shown to exhibit considerable peptide fine specificity.23, 24, 25 While CDR3 loops are located in positions that can interact with peptide, it was our expectation that affinity mutations in other regions of the 2C TCR (especially CDR2s that are located directly over the MHC helices and not the peptide) would yield TCRs with reduced peptide specificity (i.e. greater cross-reactivities with other peptides bound to the same MHC). As the majority of the pepMHC surface is generated by residues of the MHC helices, an approach involving mutagenesis of the entire TCR surface should, in principle, yield some TCR mutants with hot spots of binding energy directed against the MHC.

To test this hypothesis, we generated individual libraries of the other five CDRs of 2C (CDR1α, CDR1β, CDR2α, CDR2β and CDR3β), as well as a library of random mutants across the entire single-chain TCR (scTCR, Vβ–linker–Vα). The results showed that higher-affinity TCRs could be generated by manipulating any of five CDR loops of 2C. The higher-affinity TCR mutants included those in regions predicted to make minimal or no contact with the peptide. In addition we isolated higher-affinity, peptide-specific mutants of 2C with single amino acid changes in the CDR3s and a single amino acid change in the first residue of the Vα. While we expected mutations in CDR1 and CDR2 to involve enhanced interaction with MHC, these higher-affinity TCRs maintained peptide specificity and fine specificity as measured by binding to peptides with single amino acid changes (as with altered peptide ligands, APLs³⁴). The possible molecular mechanism of enhanced affinities for several of the mutations could be rationalized by enhanced electrostatic interactions but the molecular mechanism(s) by which every mutant retained exquisite peptide fine specificity remains to be determined.

Although structural studies are required to understand the exact orientation and position of the CDR loops that contain these mutations, it is clear that almost any region can be manipulated to facilitate engineering of peptide-specific TCRs. Whereas it is very difficult to generate antibodies against specific peptide-MHC,35, 36, 37 we show here that TCR mutants generated from a scaffold of a normal, thymic-selected TCR retained exquisite peptide specificity, regardless of the region that was mutated. The conserved docking orientation of TCRs, perhaps unlike most antibodies directed at MHC products, may position the CDRs to optimally sense regions of the pepMHC surface that are influenced by the bound peptide. Understanding the molecular basis of the findings with various CDR mutants represents an opportunity to gain insight into fundamental recognition processes that determines T cell specificity.