Engineering a high-affinity peptide binding site into the anti-CEA mAb M5A

Abstract

We have previously identified a cyclic peptide called meditope which binds to the central cavity of the Fab portion of cetuximab and shown that this peptide binding site can be grafted, or 'meditope-enabled', onto trastuzumab. This peptide has been shown to act as a hitch for the non-covalent attachment of imaging agents to meditope-enabled antibodies. Herein, we explore the process of grafting this peptide binding site onto M5A, an anti-CEA antibody in clinical trials for cancer diagnostics. In order to explore the contributions of the amino acids, we sequentially introduced pairs of amino acid substitutions into the Fab and then we reverse-substituted key residues in the presence of the other substitutions. We demonstrate that Pro40Thr, Gly41Asn, Phe83Ile and Thr85Asp in the light chain are sufficient to recreate the meditope binding site in M5A with single-digit micromolar affinity. We show that Pro40 abrogates peptide binding in the presence of the other 12 residue substitutions, and that the presence of all 13 substitutions does not interfere with antibody:antigen recognition. Collectively, these studies provide detailed insight for defining and fine-tuning the binding affinity of the meditope binding site within an antibody.

Keywords: CEA; antibody engineering; antibody-drug conjugate; grafting; meditope.

Fig. 1

Meditope-enabling M5A. (A) Sequence comparison of cetuximab, parental trastuzumab, meditope-enabled trastuzumab and parental M5A. Residues selected for substitution are highlighted in magenta. (B) Model of meditope (green surface) binding to an IgG1. (C) Stereoview of meditope-enabled trastuzumab with the substituted residues depicted as magenta spheres. (D) In stereo, the superposition of the parental trastuzumab Fab (yellow carbons) and meditope-enabled trastuzumab Fab (light blue carbons) bound to the meditope (green carbons) highlights multiple steric clashes between the residues from the parental Fab and the meditope.

Fig. 2

Characterization of parental and 13M M5A variants. (A) Size Exclusion Chromatography elution profiles of parental M5A IgG (solid line) and 13M M5A (dashed line) (Superdex 200 10/300 GL) with a corresponding SDS-PAGE (P – parental M5A IgG; 13M – M5A 13M IgG; HC – heavy chain; LC – light chain). (B) Melt curves of parental M5A (black line), parental M5A with meditope (dashed black line), 13M M5A (red line), and 13M M5A with meditope (dashed red line). Tm was calculated to be 67.6 ± 0.4°C and 62.3 ± 0.8°C for parental and 13M M5A, respectively. In the presence of meditope, the Tm was found to be 66.6 ± 0.1°C and 66.2 ± 0.4°C for parental and 13M M5A with meditope, respectively.

Fig. 3

SPR sensograms and graphical comparison of meditope binding affinities with M5A variants. (A) Parental M5A with meditope. (B) M5A 4M with meditope. (C) M5A 13M with meditope. (D) M5A LC40rev with meditope. (E) Comparison of association constants with increasing number of substitutions in M5A in the vicinity of meditope including 4, 6, 8, 10, 12 and all 13 substitutions. (F) Comparison of the association constants of M5A 13M with selected reverse variants back to parental residues.

Fig. 4

Overlay of the structures of apo parental (yellow) and meditope-bound memAb trastuzumab (gray) (PDB ID 1N8Z and 4IOI, respectively). Meditope is shown in green. (A) LC Lys42Gly and Ala43Ser. (B) LC Ser9Ile and Ser10Leu. (C) HC Ala40Ser. (D) HC Val89Ile. (E) LC Lys45Arg.

Fig. 5

Grafting of meditope site does not affect binding to CEA. Representative SPR sensograms of (A) parental M5A and (B) M5A 13M binding to CEA in the absence and presence of meditope. (C) A representative xy scatter plot is depicted for Fluorescence-activated cell sorting (FACS) analysis conducted on LS174T cells. M5A antibodies (Alexa Fluor 647) show binding to CEA antigen and Meditope-Protein L (MPL6) conjugated to GFP added in presence of antibodies shows interaction with M5A 13M (~7× over parental M5A). (D) Confocal microscopy of parental and M5A 13M binding to CEA antigen on cells and simultaneously binding to bivalent meditope-Fc (MFc). (E) Graphs show FACS data in (C) from duplicate experiments for average percent of total cell number with antibody binding to antigen (Q1 + Q2) and meditope binding to bound antibody (Q2).