Enhanced T cell receptor specificity through framework engineering

Abstract

Development of T cell receptors (TCRs) as immunotherapeutics is hindered by inherent TCR cross-reactivity. Engineering more specific TCRs has proven challenging, as unlike antibodies, improving TCR affinity does not usually improve specificity. Although various protein design approaches have been explored to surmount this, mutations in TCR binding interfaces risk broadening specificity or introducing new reactivities. Here we explored if TCR specificity could alternatively be tuned through framework mutations distant from the interface. Studying the 868 TCR specific for the HIV SL9 epitope presented by HLA-A2, we used deep mutational scanning to identify a framework mutation above the mobile CDR3β loop. This glycine to proline mutation had no discernable impact on binding affinity or functional avidity towards the SL9 epitope but weakened recognition of SL9 escape variants and led to fewer responses in a SL9-derived positional scanning library. In contrast, an interfacial mutation near the tip of CDR3α that also did not impact affinity or functional avidity towards SL9 weakened specificity. Simulations indicated that the specificity-enhancing mutation functions by reducing the range of loop motions, limiting the ability of the TCR to adjust to different ligands. Although our results are likely to be TCR dependent, using framework engineering to control TCR loop motions may be a viable strategy for improving the specificity of TCR-based immunotherapies.

Keywords: T cell receptor; framework regions; molecular dynamics; protein engineering; specificity.

Figure 1

Yeast titration confirms the high affinity of the 868-Z11 TCR variant for SL9/HLA-A2. The data shows the MFI of yeast expressing the 868-Z11 scTCR stained at 4°C with PE-conjugated SL9/HLA-A2 pMHC tetramer at concentrations from 256 pM to 32.8 nM, with the titration performed in duplicate. The resulting curve was fit to a 1:1 binding isotherm, yielding a K _D,app of 1.7 ± 0.6 nM.

Figure 2

Heatmap of 868-Z11 deep mutational scanning data. Blue cells indicate a positive fitness mutation, orange cells indicate a negative fitness mutation, white cells indicate fitness mutations that are approximately equivalent to WT 868-Z11, and black cells indicate mutations that were not significantly represented in the sequencing data (indicated as n/s). Red outlines indicate WT amino acids. CDR loops are indicated by the shaded regions below the WT 868-Z11 sequence. Framework sites are those outside of the shaded regions. Amino acid numbering is per that found in the PDB file as well as according to IMGT standards, as indicated.

Figure 3

Locations of the selected mutations in the 868-Z11 TCR in the TCR-SL9/HLA-A2 complex. The α chain is magenta; the β chain is teal. Darker colors indicate residues that were included in the deep mutational scanning.

Figure 4

Binding data for the 868-Z11 TCR and mutants. (A) Single cycle kinetic titrations of SL9/HLA-A2 binding 868-Z11 WT (black), A94αH (teal), and G104βP (purple). Fits are shown as red lines. On rates (k _on) were determined as 1.96×10⁵ M^-1 s ^-1 (WT), 7.92×10⁵ M^-1 s ^-1 (A94αH), and 9.04×10⁵ M^-1 s ^-1 (G104βP). Off rates (k _off) were also determined as 2.16 x 10^-4 s ^-1 (WT), 4.15 x 10^-3 s ^-1 (A94αH), and 2.30 x 10^-4 s ^-1 (G104βP). From the on and off rates, the K _D values were determined as 1.1 nM (WT), 5.2 nM (A94αH), and 0.3 nM (G104βP). Large spikes associated with injections and pump refills were edited out for clarity. Data are reflective of two separate titration series. (B) Steady state titrations of SL9/HLA-A2 binding 868-Z11 T96βK (black), A94αH/T96βK (red), and G104βP/T96βK (blue). The K _D values of 868-Z11 T96βK, A94αH/T96βK, and G104βP/T96βK to SL9/HLA-A2 were determined to be 2.5 ± 0.8 µM, 33 ± 9 µM, and 2.2 ± 0.9 µM respectively. Data are reflective of six separate titrations; values are the averages and standard deviations from the six experiments. Note that the fitted curve for the T96βK variant is obscured by that of the G104β/T96βK variant due to its nearly identical affinity.

Figure 5

Impacts of mutations in the 868-Z11 TCR on the binding of SL9 escape variants presented by HLA-A2. ΔΔG° values relative to the WT SL9 peptide are shown, determined from steady state K _D measurements in triplicate. In general, compared to the single T96βK the mutation, addition of the G104βP mutation enhances specificity by shifting ΔΔG° values in a positive (unfavorable) direction, whereas addition of the A94αH mutation weakens specificity by shifting ΔΔG° values in a negative (favorable) direction. ΔΔG° values were determined from three independent K _D measurements ( Table S2 ). Each K _D was converted to a ΔG° before determining the ΔΔG° relative to WT SL9. The three ΔΔG° values were then averaged, and the standard deviations determined. Statistical differences between mutations were determined using unpaired Student’s t-tests (* = significant differences with p< 0.05; ns, differences not significant).

Figure 6

Positional scanning library analysis confirms the G104βP mutation confers higher specificity. (A) The T96βK and G104βP/T96βK variants of 868-Z11 have EC₅₀ values identical within error for the SL9 peptide in functional assays measuring cytokine release. Data points are averages and standard deviations of five separate titrations; values reported are the averages and standard deviations from the five experiments. Negative control data are for co-cultures with the irrelevant Tax_11-19 peptide (sequence LLFGYPVYV). (B) Positional scanning library data for the T96βK and G104βP/T96βK variants of 868-Z11. For each peptide in the library, IL-2 production at 10 µM peptide in a co-culture experiment is normalized to that of the WT SL9 peptide as indicated by the scale on the right. Addition of the G104βP mutation results in fewer stimulatory peptides, particularly in the C-terminal half of the peptide. Data in each cell are the average of three separate co-culture experiments. (C) Fingerprint analysis from the data in panel B, showing the distribution of scores for all 1.28 billion peptides of the form XLXXXXXXL, where X is any of the 20 standard amino acids. The greater specificity conferred by the G104βP mutation is indicated by the left-shifted blue curve, further highlighted by the much smaller number of peptides with scores ≥ 0.8 as indicated by the inset.

Figure 7

The G104βP mutation reduces the magnitude of fluctuations in the 868-Z11 TCR CDR3β loop. (A) The fluctuations for all six loops of the α and β chains for the T96βK, A94αH/T96βK, and G104βP/T96βK 868-Z11 TCR variants determined from 1 μs of molecular dynamics simulations of the unbound TCR. The largest reduction with the G104βP mutation is seen in the CDR3β loop. (B) Structural view of the 868-Z11 CDR3β loop and the framework mutations above it, highlighting the backbone hydrogen bond between Gly104 and Cys91 that is lost with the glycine-to-proline mutation. (C) Cα motional cross-correlation matrices for the CDR3β loop and the framework amino acids above it (residues 87-107) from the molecular dynamics simulations described in panel (A). Amino acids of the loop are indicated by the black boxes. The green box indicates position 104β. As indicated by the scale, blue is positively correlated motion, red is negatively (anti) correlated motion. Addition of the G104βP mutation, but not the A94αH mutation, substantially reduces the magnitude of correlated motion between the center of the loop and the residues above it.