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

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 denovo 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.

Figure 1:. Diffusion of pMHC binders.

a, pMHC structure and design challenge. The goal is to distinguish a target peptide (in this example, MAGE) from a closely related off-target (Titin). The positions that differ between the two peptides are circled in the pMHC structure model at the bottom, b, Representative diffusion trajectories and design models for three different pMHC targets. Left two columns, target pMHc identities and structures. Columns 3 and 4, intermediate steps in diffusion denoising trajectories starting from completely random residue distributions. Right column, fully denoised design model backbones, c, Partial diffusion of design scaffolds with desirable properties (interacting over the full length of the peptide, but making few interactions with MHC) to more efficiently generate binders to related targets.

Figure 2:. Design models and binding specificity.

a, Design models. Left, overall structure; right, zoom in on peptide binding region. pMHC in orange, peptide in sticks, and binder in blue. HLA allele and peptide sequence are specified above the zoom-in view, b, Flow cytometry of cells displaying the design incubated with on-target pMHC tetramer (x-axis) and 1 or 2 off-target tetramers (y-axis) at 10nM concentration. Staining in the lower right quadrant indicates specific on-target binding. Rows 1 and 2, individual designs displayed on yeast; rows 3–6, CARs incorporating designs on Jurkat cells, c, Partial diffusion of design at top to targets below (left panels) and corresponding Jurkat straining (right).

Figure 3:. Selective activation of T-cells expressing designed CARs targeting MAGE-A3.

Activation of Jurkat cells expressing the MAGE-513 CAR by 293T cells expressing HLA-A*01:01 pulsed with 5uM of different peptides measured through CD69 expression level, a, Histograms of CD69 expression levels following treatment with 5uM MAGE, the closely related Titin peptide or DMSO; the MAGE peptide leads to considerable activation while the Titin peptide is similar to the DMSO control. Horizontal black bars represent CD69 positive population; the fraction of cells within this range is indicated at top left of each panel, b, Design models of the MAGE-513/MAGE peptide/HLA-A*01:01 complex (lower panel) and zoom-in view of the key residues mediating the interaction (upper panels), c, Histograms of CD69 expression levels following pulsing with MAGE-A3 single alanine mutants (D3A indicates mutation of the Asp at position 3 in the peptide to alanine) or DMSO. d,Histograms of CD69 expression levels following pulsing MAGE-513 variant CARs with MAGE-A3 peptide or DMSO. e and f, CD69 levels of Jurkat cells expressing MAGE-513 CAR upon pulsing with top ranked peptides from yeast binding screen (e) or from sequence similarity search (f). Identity to the MAGE peptide is indicated in red.

Figure 4:. Specific activation of designed CARs by cognate pMHC complexes.

Activation of Jurkat cells expressing the CARs by 293T cells with 5uM of different peptides measured through CD69 expression level by staining (indicated by histogram or mean fluorescence intensity (MFI)). a, CD69 MFI of gp100_T3 CAR with gp100 peptide, alanine mutant peptides or DMSO. b, CD69 MFI of WT1_5 CAR with WT1 peptide, alanine mutant peptides, off-target peptides, or DMSO. c, zoom-in view of design model of WT1_5 with pMHC target, d and e, (upper) CD69 MFI of d, Mart-1_3 or e, Mart-1_43 CAR with Mart-1 peptide, alanine mutant peptides, or DMSO. (lower) design models of binders with Mart-1 pMHC antigen, f, Histograms of CD69 expression level of PRAME CARs with PRAME peptides or DMSO. g, SPR traces of PRAME binder A9 on PRAME pMHC monomer, h, CD69 MFI of Jurkat cells expressing PRAME_A9 CAR with PRAME peptide, alanine mutant peptides, four peptides with similar sequences, or DMSO.