A general system for targeting MHC class II-antigen complex via a single adaptable loop

Abstract

Major histocompatibility complex class II (MHCII) bound to a peptide antigen mediates interactions between CD4⁺ T cells and antigen-presenting cells. Targeting peptide-MHCII with T cell antigen receptors (TCRs) and TCR-like antibodies has shown promise for autoimmune diseases and microbiome tolerance. To develop a general targeting approach, we introduce targeted recognition of antigen-MHC complex reporter for MHCII (TRACeR-II) for the rapid development of peptide-specific MHCII binders. TRACeR-II binders have a small helical bundle scaffold and use a single loop to recognize peptide-MHCII, which offers versatility and enables structural modeling of the interactions to target MHCII antigens. We demonstrate rapid generation of TRACeR-II binders to multiple molecules with affinities in the low-nanomolar to low-micromolar range, comparable to best-in-class TCRs and antibodies. Through computational protein design, we created specific binding sequences in silico from only the sequence of a severe acute respiratory syndrome coronavirus 2 peptide. TRACeR-II provides a straightforward approach to target antigen-MHCII without relying on combinatorial selection on complementarity-determining region loops.

Fig. 1 |. Design scheme of the TRACeR-II platform.

a, Facile development of peptide-focused MHCII binders by mutating a single adaptive loop. Stars represent the five amino acids on the ARE. Star colors represent different amino acid sequences. Different peptides are represented by colored circles. Colors on MHCII represent different alleles. The highly conserved MHCII $\mathrm{\alpha }$-chain is represented with the same blue color. b, Left: ternary complex structure of the MAM–MHC–TCR complex (PDB ID: 2ICW). Right: binding structure between the truncated MAM N-terminal domain and pMHCII. c, Structural optimization by RosettaRemodel.

Fig. 2 |. Specificity determination and biophysical characterization of TRACeR molecules.

a, Sequence convergence of the ARE from the final sorting round. The most enriched clone for each target from the final sorting round was carried over for biophysical characterization and specificity determination. b, Target specificity determined with yeast surface display. Peptide-focused pMHCII binders showed high specificity for their cognate antigens and minimal cross-reactivity with other targets; staining concentration: on target, 25 nM tetramer; off target, 100 nM tetramer. c, Binding kinetics of TRACeR molecules determined by SPR. pm, picometer. Fitting data are summarized in Supplementary Table 2. d, Relative enrichment heat map of SSM data on the CLIP peptide stained by ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DR1}}^{\text{CLIP}}$. Relative enrichments were calculated after three rounds of FACS. Enrichment (red) indicates higher affinity and depletion (blue) indicates lower affinity than the wild-type peptide. Boxed cells depict wild-type identity. The structure shown on the bottom illustrates the side chain orientations of the peptide. e, Binding signal quantification relative to labeled anti-His background. Soluble ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DR1}}^{\text{CLIP}}$ was used to stain nontransfected T2 cells (HLA-DR1⁻CLIP⁻, black) and T2 cells transfected with HLA-DR1 (DR1⁺CLIP⁺, red) or HLA-DR1 and HLA-DM (DR1⁺CLIP⁻, blue). Data are presented as median ± s.d. ($n=3$ technical replicates); MFI, median fluorescence intensity.

Fig. 3 |. Mutiallelic compatibility of the TRACeR platform.

a, Histogram of ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DR1}}^{\text{CLIP}}$ binding the CLIP peptide frame (87–101) presented by different HLA-DR molecules. ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DR1}}^{\text{CLIP}}$ is displayed on the yeast surface and is stained with 50 nM pMHC tetramer. b, Titration of HLA-DR4 MHC tetramer on yeast displaying ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DR1}}^{\text{CLIP}}$ (right) or ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DR1}}^{\text{HA}}$ (left). Data are presented as median ± s.d.; $n=3$ technical replicates. c, Donor HLA class II types and binding signal quantification for TRACeR staining. Each group contains three independent donors (donor information is listed in Supplementary Fig. 15). Each data point shows the median ± s.d. from three technical replicates. d, Left: ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DR3}}^{\text{HER2}}$ displayed on the yeast surface stained with HLA-DR3–HER2 (positive control) and HLA-DR3–CLIP (negative control). Right: binding kinetics of ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DR3}}^{\text{HER2}}$ with HLA-DR3–HER2 determined by SPR. e, Left: ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DR15}}^{\mathrm{\alpha }\text{Syn}}$ displayed on the yeast surface stained with HLA-DR15–$\mathrm{\alpha }$-synuclein (HLA-DR15–$\mathrm{\alpha }\mathrm{S}\mathrm{y}\mathrm{n}$; positive control) and HLA-DR15–CLIP (negative control). Right: binding kinetics of ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DR15}}^{\mathrm{\alpha }\mathrm{S}\mathrm{y}\mathrm{n}}$ with HLA-DR15–$\mathrm{\alpha }$-synuclein determined by SPR. f, Positions on HLA-DQ2.5 $\mathrm{\alpha }$-chain, which does not share the same side chain identity with HLA-DRA1*01:01. Only residues within 5 Å from TRACeR are shown. g, Left: ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DQ2.5}}^{\text{glia-}\mathrm{\alpha }1a}$ displayed on the yeast surface and stained with 50 nM pMHC tetramer with different peptides presented on HLA-DQ. Right: binding kinetics of ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DQ2.5}}^{\text{glia-}\mathrm{\alpha }1a}$ with HLA-DQ2.5–glia-$\mathrm{\alpha }$1a determined by SPR.

Fig. 4 |. Cryo-EM structure reveals a unique peptide recognition mechanism through a single loop.

a, Cryo-EM structure of ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DR1}}^{\text{CLIP}}$–HLA-DR1–CLIP–c44H10 Fab ternary complex. b, Superposition of wild-type MAM on the TRACeR cryo-EM structure in a; WT, wild-type. c, Comparison of the designed TRACeR model to the cryo-EM structure. Left to right: global superposition, redesigned $\mathrm{\alpha }$-turn, designed disulfide and repacked helix; r.m.s.d., root mean squared deviation. d, Zoomed-in view of the ARE–peptide interface. e, Computational design scheme for ${\mathrm{T}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{e}\mathrm{R}}_{\text{MHCII,DR1}}^{\text{SARS-CoV-2}}$. Detailed descriptions of the methods and design scripts are available in Supplementary Appendix 1. f, Experimental validation of designed TRACeRs displayed on the yeast surface stained with 50 nM HLA-DR1–SARS-CoV-2 and HLA-DR1–CLIP tetramers. Data are presented as median ± s.d. ($n=3$ technical replicates). The asterisk (*) indicates that both WLHNV and WLHRV turned into WLHGV when mutating the fourth residue to G.