High-Throughput Antibody Characterization

0

September 1, 2017 (Vol. 37, No. 15)

Yasmina Abdiche Ph.D. Chief Scientific Officer Carterra

Epitope Binning for Next-Generation Biotherapeutic Discovery

Monoclonal antibodies (mAbs) are one of the most successful therapeutic drug classes, generating lucrative sales for the pharmaceutical industry. In 2016, more than 20 therapeutic antibodies gained blockbuster status, as defined by their annual sales exceeding $1 billion, with the top-tier drugs—adalimumab (AbbVie), infliximab (Johnson & Johnson and Merck), rituximab, bevacizumab, and trastuzumab (Roche/Genentech)—each generating annual sales between $6–16 billion. Significant investment in antibodies to treat cancer, autoimmune disease, infectious diseases, rheumatoid arthritis, pain, heart disease, and many more therapeutic areas, has resulted in a clinical pipeline of about 500 investigational drugs, of which more than fifty are in late-stage clinical evaluation with nine under review for first market approval in 2017. Additionally, antibody-based assays are important in the diagnostic reagents industry.

The innate ability of antibodies to bind their targets with exquisite specificity and high affinity has been leveraged in the discovery of therapeutic and diagnostic antibodies, so selecting antibodies with appropriate binding characteristics is an important step in early-stage research programs. While an antibody’s epitope largely dictates its biological function, this property can neither be predicted by in silico methods nor shifted rationally by engineering, so it must be selected empirically by screening a myriad of clones routinely produced by modern antibody libraries. Arguably, the epitope is a more relevant screening parameter than affinity, since the latter can be optimized by standard protein-engineering methods, whereas the former cannot. Indeed, it is highly desirable to discover antibodies with unique epitopes that may offer mechanistically differentiated modes of action and intellectual property opportunities. To meet the demand for evaluating these enormous antibody libraries, analytical tools are evolving to accelerate screening while minimizing sample consumption.

This article introduces how array-based surface plasmon resonance imaging (Array SPRi) methods can be used to perform high-throughput “epitope binning” assays that characterize the “epitope landscape” of an antibody library and facilitate the triaging of hits to leads, ultimately driving R&D costs down.


Biophysical Methods for Epitope Characterization

The gold standard for defining an antibody’s epitope is with structural data, available from biophysical tools such as X-ray crystallography, cryo-electron microscopy, nuclear magnetic resonance, or mass spectrometry. However, while these techniques give a precise, albeit static, description of an epitope with atomic-level resolution, they are low-throughput, resource-intensive, and often require highly specialized staff to operate them, limiting their use to confirming leads at late-stage research, rather than screening for potential leads in early-stage research. Functional (or loss of function) methods for epitope characterization include antigen mutagenesis and peptide-based epitope mapping approaches, but they require the production of specialized reagents, which may not be readily available.

Epitope binning is a competitive immunoassay that is used to assess the epitope diversity of an antibody library, in a relative sense, by clustering antibodies into epitope families or “bins” based on their ability to block one another’s binding to their specific antigen, in a pairwise and combinatorial manner. Since “bin buddies” likely share similar (or near-identical) functional characteristics, bin representatives can be chosen to distill a large panel of clones to a small subset that retains the epitope diversity of the whole panel. By merging epitope binning data with the results from orthogonal assays, such as functional cell-based data, binding affinity for the specific target, cross-reactivity to orthologs, and sequence data, a more comprehensive picture of the antibody library can emerge to inform the selection of leads worthy of further characterization. Identifying sandwich pairs by choosing antibodies from discrete, non-overlapping “bins” is fundamental to the design of diagnostic reagents that are used to detect biomarkers and support clinical programs.

Label-free biosensors, such as those employing surface plasmon resonance (SPR) or biolayer interferometry (BLI) detection, are biophysical tools well suited for performing epitope binning analyses, and the results from these studies help elucidate an antibody’s mechanism of action and identify clones with unique epitopes for therapeutic or reagent purposes. However, the limited throughput, high sample requirements, and costly consumables of traditional biosensor platforms make it impractical to perform epitope binning assays on panels of antibodies larger than about twenty or so because the size of the experiment scales geometrically with the antibody panel.


Introducing Carterra’s Approach

In contrast, Array SPRi technology by Carterra (formerly Wasatch Microfluidics) allows epitope-binning assays to be performed on much larger panels of antibodies in a facile manner. By employing a one-on-many configuration, samples are analyzed in a highly parallel style, which significantly accelerates throughput while conserving precious samples, allowing routine binning on 96-antibody arrays and expansion to 384-antibody arrays. Using specifically designed software, the results of a binning experiment are represented graphically in various ways, such as via proprietary network plots, in which blocking relationships between antibodies are indicated with chords, and bins are inscribed by envelopes (Figure A).

Furthermore, by merging data from independent assays, networks can be colored by various parameters, providing an intuitive visualization tool for organizing multi-parameter information and facilitating the discrimination of clones with unique behaviors. For example, merging epitope binning results with cell-based function-blocking data and sequence data on a panel of human antibodies produced from four healthy donors revealed two subsets of clones (Figure A and B). These clones displayed unique modes of neutralization against a Staphylococcus aureus virulence factor, exhibiting a strongly biased germline usage. This remarkable finding underscored an elegant example of convergent evolution, as the antibodies populating each bin cluster shared strikingly similar sequences that resulted in their nearly identical binding mechanism, despite their isolation from the naïve B cells of different individuals. The resolution afforded by the binning assay, therefore, enabled a deeper appreciation of the evolution of the human immune repertoire, expanding the concept of germline-restricted usage of antibodies to bacterial pathogenic proteins.1

Recently, Carterra announced the production of their LSA platform that integrates printing and imaging with automated switching between single-channel (large flow cell) and multichannel (96-printhead) modes, allowing routine binning of a panel of 384 antibodies in a single, unattended run, using less than 2 µg per antibody (1 µg for printing and 1 µg for sandwich pairing). This simplified workflow will not only enable higher-order epitope binning assays but will also accelerate kinetic screening by enabling capture-based kinetics against multiple targets on 1152 crude antibodies per unattended run. Considering the cost of development for a therapeutic antibody (from the bench to commercial release) is estimated to be $1 billion, the pharmaceutical industry is demanding higher-throughput analytical methods to facilitate the identification of product candidate leads with mechanistically differentiated modes of action as early as possible.


Figure. Examples of network plots used to visualize blocking relationships between antibodies and their deduced epitope clusters or “bins.” A) Antibodies are colored by bin; the dotted chords represent a unidirectional (or asymmetric) blocking relationship (reproduced from Y.N. Abdiche et al., 2017, PLOS ONE). B) Antibodies colored by functional activity where red, yellow, and green indicate a block, partial block, or no block in a cell-based assay, respectively (reproduced from Y.N. Abdiche et al., 2014, PLOS ONE). The arrows indicate two discrete subsets of germline-encoded function-blocking antibodies.




























Yasmina Noubia Abdiche, Ph.D., is chief scientific officer at Carterra.

Reference
1. Y.A. Yeung et al., 2016, Nat Commun.

This site uses Akismet to reduce spam. Learn how your comment data is processed.