Computational epitope binning reveals functional equivalence of sequence-divergent paratopes

Abstract

The therapeutic efficacy of a protein binder largely depends on two factors: its binding site and its binding affinity. Advances in in vitro library display screening and next-generation sequencing have enabled accelerated development of strong binders, yet identifying their binding sites still remains a major challenge. The differentiation, or "binning", of binders into different groups that recognize distinct binding sites on their target is a promising approach that facilitates high-throughput screening of binders that may show different biological activity. Here we study the extent to which the information contained in the amino acid sequences comprising a set of target-specific binders can be leveraged to bin them, inferring functional equivalence of their binding regions, or paratopes, based directly on comparison of the sequences, their modeled structures, or their modeled interactions. Using a leucine-rich repeat binding scaffold known as a "repebody" as the source of diversity in recognition against interleukin-6 (IL-6), we show that the "Epibin" approach introduced here effectively utilized structural modelling and docking to extract specificity information encoded in the repebody amino acid sequences and thereby successfully recapitulate IL-6 binding competition observed in immunoassays. Furthermore, our computational binning provided a basis for designing in vitro mutagenesis experiments to pinpoint specificity-determining residues. Finally, we demonstrate that the Epibin approach can extend to antibodies, retrospectively comparing its predictions to results from antigen-specific antibody competition studies. The study thus demonstrates the utility of modeling structure and binding from the amino acid sequences of different binders against the same target, and paves the way for larger-scale binning and analysis of entire repertoires.

Keywords: AU-PRC, Area under the precision-recall curve; Docking; Epitope; Epitope binning; IL-6, Interleukin - 6; LRR, leucine-rich repeat; PCC, Pearson correlation coefficient; Paratope equivalence; Pro, Proline; Protein binder; RMSD, Root-mean squared deviation; Repebody; SARS-CoV-2, severe acute respiratory syndrome coronavirus – 2.

Fig. 1

Computational sequence, structure, binding, and competition analysis of IL6-binding repebody clones. (a, b) Seven repebodies were selected for analysis based on experimental properties and sequence diversity in the hypervariable sites in Modules 3 and 4. (c) The amino acid differences in the repebodies manifested as structural differences in homology models. (d) The structural differences in the repebody models led to predicted complex structure differences according to docking models (for clarity, only a few models out of ∼30 per repebody are shown). (e) The docking models were analyzed by Epibin in order to predict competition for binding between pairs of repebodies. First, Epibin computes the “epitope overlap score (EOS)”, the extent of common epitope residues (0: none to 1: all). In the example, B10_1 and B3_1 represent docking models of B10-IL-6 and B3-IL-6, respectively. Comparing the epitopes of these two docking models indicates that their epitopes overlap to some extent, quantified by an EOS of 0.6. Similarly, B10_2 (third row) and B3_4 (fifth column) docking models share common epitopes, yielding an EOS of 0.7. On the other hand, docking models B10_2 and B3_1 do not have any common epitope residues and hence the EOS between them is 0. Similarly, B10_4 does not share any common epitope residues with any of the B3 docking models, yielding EOS = 0.0 for the whole fourth row. Based on the set of EOSs, the overall “Epibin score” capturing competition between two repebodies is computed as the average fraction of each repebody’s docking models that have a sufficiently strong competitor for the same epitope among the other repebody’s docking models (i.e., EOS exceeds a threshold). In the example, the cells containing the ‘Δ’ symbol indicate those pairs of models for which the EOS is above the threshold of 0.5 and hence included in the fraction. Among the illustrated pairs of docking models, we see that three of the four B10 models (rows) have a strong competitor among the B3 models (columns), and all four of the B3 models have a strong competitor among the B10 models. For consistency with experimental data, the average fraction of sufficiently overlapped models is subtracted from 1 to yield the Epibin competition score, here giving a score of 1 – 0.5*(3/4 for B10 + 4/4 for B3) = 0.125.

Fig. 2

Computational and experimental evaluation of repebody competition for IL-6 binding. (a) Relative competition and (b) clustering of repebodies based on pairwise sequence similarity (in terms of BLOSUM62 substitution scores across the six hypervariable positions), pairwise structural similarity as measured by C_β RMSD, pairwise complex similarity as evaluated by Epibin with an EOS threshold of 0.65, and competitive immunoassays. For both Epibin and the immunoassays, lower values indicate increased competition. (c) Correlation between Epibin-predicted and experimentally-assayed competition of the seven repebodies.

Fig. 3

Competition between A4 and B10, but not A10, can be attributed to the presence of similar amino acids in “paratope-equivalent” positions. (a) Full view of the A4, B10, C6, and A10 structural models superposed onto one another, with the general location of IL-6 (from the D3 crystal structure, PDB ID: 4J4L). (b) Zoomed-in view illustrating the positions of similar amino acids in paratope-equivalent positions of Modules 3 and 4 for (i) A4 and B10, (ii) B10 and A10. Hypervariable sites 1, 2, and 3 are represented as sticks. (c) Amino acids present in key structural positions highlight the similarity of position pairs ‘β’ and ‘δ’ (denoted by asterisks) in A4 and B10 and the presence of dissimilar residues in these positions of the non-competing repebody A10. The (o) and (x) signs indicate competition of the respective repebody with B10.

Fig. 4

Competitive immunoassays andisothermal titration calorimetry (ITC)results confirm the significance of paratope-equivalent positions ‘β’ and ‘δ’. (a) Correlation between Epibin-predicted competition (x-axis) and corresponding observed competition (y-axis) of all designed variants competing against the wild-type forms A4, B10, and C6 (separate panels by wild-type). Pearson’s correlation coefficient is denoted as r. (b) Binding affinity measurement of the wild-type and variant forms of A4, B10, and C6 against IL-6 as measured by ITC. Bars show the means and SDs over three replicates.

Fig. 5

Mutational studies confirm that epitopes of A4 and B10 partially overlap.Triple-mutant variants of IL-6 were designed by Episcope to localize repebody epitopes by disrupting hypothesized binding interfaces while maintaining IL-6 stability. (a) Residues selected by Episcope for mutation mapped onto the structure of IL-6 (PDB ID: 1ALU) and colored by general regions. (b) Bar plots representing the normalized binding of IL-6 variants, relative to wild type IL-6, to each of the six repebodies. Colors of mutated residues refer to colors of regions in panel (a). Values are the means of three replicates.