Centaur antibodies: Engineered chimeric equine-human recombinant antibodies

Abstract

Hyper-immune antisera from large mammals, in particular horses, are routinely used for life-saving anti-intoxication intervention. While highly efficient, the use of these immunotherapeutics is complicated by possible recipient reactogenicity and limited availability. Accordingly, there is an urgent need for alternative improved next-generation immunotherapies to respond to this issue of high public health priority. Here, we document the development of previously unavailable tools for equine antibody engineering. A novel primer set, EquPD v2020, based on equine V-gene data, was designed for efficient and accurate amplification of rearranged horse antibody V-segments. The primer set served for generation of immune phage display libraries, representing highly diverse V-gene repertoires of horses immunized against botulinum A or B neurotoxins. Highly specific scFv clones were selected and expressed as full-length antibodies, carrying equine V-genes and human Gamma1/Lambda constant genes, to be referred as "Centaur antibodies". Preliminary assessment in a murine model of botulism established their therapeutic potential. The experimental approach detailed in the current report, represents a valuable tool for isolation and engineering of therapeutic equine antibodies.

Keywords: anti-toxins; anti-venoms; antibody engineering; botulinum; equine V-genes; phage display.

Figure 1

Study outline. (A) Schematic workflow of the development of horse-derived recombinant monoclonal antibodies (mAbs). Two horses, Agnes (♀) and Baloo (♂), immunized by administration of (C) botulinum inactivated neurotoxin (toxoid) serotype A or B (BoNT/A or BoNT/B, respectively) served as a source of PBMCs-cDNA. Two scFv-PD libraries were constructed based on Ab V-segments that were amplified using a newly designed primer set. Following enrichment by panning, specific mAbs were isolated, cloned and expressed in CHO cells as recombinant horse-human chimeric Centaur mAbs. (B) Schematic representation of BoNT/A, BoNT/B and their related recombinant antigens. BoNT protein is composed of a 50 kDa light chain (LC), which encodes the effector molecule (responsible for the BoNT toxicity) connected by a disulfide bridge to a 100 kDa heavy chain (HC). Receptor-mediated endocytosis followed by translocation of the light chain across the membrane into the neuronal cytosol is facilitated by the two HC functional domains, H_C and H_N, correspondingly. (C) Binding capacity of Agnes’s and Baloo’s sera, was evaluated by ELISA against indicated BoNT/A- and BoNT/B- recombinant antigens. Serially diluted serum samples were tested against the indicated antigens.

Figure 2

Characterization of the diversity of the equine PD library repertoires. (A, B) Length distribution of CDR-H3 (A) and CDR-L3 (B), represented by V-segment sequences of Agnes- and Baloo-PD libraries. The histograms depict the amino-acid length of library sequences exhibiting a prevalence higher than 0.1%. (C) Prevalence rate of V-segment sequences (on the ordinate axis) assigned to each cluster (on the abscissa axis), as calculated by the clonotype analysis. Clusters composed of less than 100 sequences were not included in the chart presentation. The number of unique sequences present in each cluster is reflected by the sphere size (see also Supplementary Table S1 for the detailed cluster analysis data).

Figure 3

Characterization of the anti-BoNT Centaur mAbs. (A) Amino acid sequences of the VH and VL CDR segments of the anti-BoNT selected mAbs. CDR positions are indicated according to the Kabat system (36). (B) Specificity of the nine selected mAbs determined by ELISA against the indicated BoNT subunits. Data represent average of triplicates ± SD. Skim milk (SM) was used as control protein, as indicated in the in-set legend. (C, D) Binding ability of the selected mAbs was evaluated against BoNT native toxins. Each indicated mAb was serially diluted and tested in duplicates by indirect ELISA against BoNT/A (C) and BoNT/B (D). Data represent average ± SD.

Figure 4

Biolayer interferometry (BLI) analyses of Centaur mAbs. (A, B) Binding kinetic of the selected mAbs. Each tested mAb was biotinylated, immobilized on a streptavidin sensor and incubated with increasing amounts of targeted antigen. Binding kinetics were fitted using the 1:1 binding model. Each BLI analysis was independently-repeated at least two times, resulting in highly-similar calculated values. (A) Representative assay results are shown for ALC-HN_18 mAb. Gray sensograms represent binding profiles in the presence of each indicated antigen concentration. Fitting curves are depicted in red. Binding BLI curves of the nine tested mAbs are provided in Supplementary Figure S3. (B) Table of the binding kinetic parameters of the nine selected mAbs, presented as average values. (C, D) Identification of two distinct epitopes, recognized by the six selected anti-LC-H_N/A mAbs by epitope binning. Each mAb was biotinylated (bio-mAb), immobilized on streptavidin sensor and saturated with LC-H_N/A recombinant protein. The complex was then incubated with each one of the indicated mAbs. Background signals were obtained from parallel sensors incubated with the tested mAb itself (non-biotinylated) or non-specific mAb. (C) Time 0 represent the binding to bio-ALC-HN_2-LC-H_N/A complex, demonstrating a unique epitope targeted by this Ab. (D) Time 0 represent the binding to bio-ALC-HN_4-LC-H_N/A complex. This representative data set, indicate that ALC-HN_4, 5, 6, 12 and 18 recognize a competitive epitope, (see Supplementary Figure S4 for all binning data).

Figure 5

Protection of mice against BoNT intoxication by the Centaur mAbs. Mice (n = 6) were intraperitoneally-injected with 5xLD₅₀ of BoNT/A (A) or BoNT/B (B) pre-incubated with 50 μg of each indicated Centaur mAb. Control groups included “no Ab control” and “non-specific isotype control” (anti-ricin mAb). Mice survival was monitored for 10 days post-administration.