Function-based high-throughput screening for antibody antagonists and agonists against G protein-coupled receptors

Abstract

Hybridoma and phage display are two powerful technologies for isolating target-specific monoclonal antibodies based on the binding. However, for complex membrane proteins, such as G protein-coupled receptors (GPCRs), binding-based screening rarely results in functional antibodies. Here we describe a function-based high-throughput screening method for quickly identifying antibody antagonists and agonists against GPCRs by combining glycosylphosphatidylinositol-anchored antibody cell display with β-arrestin recruitment-based cell sorting and screening. This method links antibody genotype with phenotype and is applicable to all GPCR targets. We validated this method by identifying a panel of antibody antagonists and an antibody agonist to the human apelin receptor from an immune antibody repertoire. In contrast, we obtained only neutral binders and antibody antagonists from the same repertoire by phage display, suggesting that the new approach described here is more efficient than traditional methods in isolating functional antibodies. This new method may create a new paradigm in antibody drug discovery.

Fig. 1. POC study in U2OS/APJ β-arrestin reporter cell line.

a Flow cytometry of Tango/APJ β-arrestin assay cells transduced with recombinant lentivirus containing the genes encoding sdAb E3, or JN241, or JN126P3. b, c Flow cytometry of Tango/APJ β-arrestin assay cells following β-arrestin recruitment assay at antagonist (b) or agonist (c) mode. The gated window and the percentage of gated cell population are shown. In the assay at antagonist mode, apelin 13 at EC₈₀ dose was added to the plates. Antibody antagonists negatively affect apelin 13 binding to the receptor, which causes decreased β-lactamase production and subsequent reduced substrate hydrolysis, leading to more substrate fluorescence (at 530 nm) and less product fluorescence (at 460 nm). In the assay at agonist mode, no ligand was added to the plates.

Fig. 2. Function-based cell sorting and screening for sdAb antagonists.

a Function-based cell sorting for antagonist-expressing cell clones following the stimulation with apelin 13 for 12 h. The gated window and the percentage of gated cell population in the 3rd round cell sorting is shown. b Antagonist activity of each round sorted cell population in Tango/APJ β-arrestin assay. Data are expressed as mean with SD. Statistical analysis was done using unpaired two-tailed t test. The lower product/substrate ratio, the higher antagonist activity. c Single-cell clones from the 3rd round sorted library were characterized by Tango/APJ β-arrestin assay. Stimulation by 100 nM apelin 13 and 0.1% DMSO were set as 100% activation and 0% activation, respectively. d IC₅₀ values of nine soluble sdAbs cloned from positive cell clones by PathHunter β-arrestin assay. IC₅₀ values and 95% confidence intervals are shown. e Correlation between the inhibitory activity of the cell clones and the IC₅₀s of the corresponding soluble sdAbs by β-arrestin assay. Inhibitory activity of the cell clones was done by Tango assay. Pearson correlation analysis was used in the association study. R square, P value and 95% confidence bands are shown.

Fig. 3. Characterization and epitope localization of isolated sdAb antagonists.

a, b Potency of isolated sdAb antagonists by PathHunter β-arrestin assay (a) and by cAMP assay (b). c Binding of the isolated sdAb antagonists to APJ/DOR domain-swapped mutants or APJ E174A site mutant relative to wild type (WT) APJ by flow cytometry. The 1st group includes JN400, JN402, JN413, and JN414 that bind to ECL2. Only JN400 is shown as a representative. The 2nd group include JN406 and JN412 that bind to both ECL1 and ECL2. Only JN406 is shown as a representative. The 3rd group include JN411 that binds to ECL1 only. The 4th group include JN401 and JN409 that are sensitive to the changes either in the N-terminus or in any of the ECLs. JN409 is shown as a representative. The 5th group include JN404 that is sensitive to the changes either in the N-terminus or in ECL1 or ECL2. APJ/DOR N-term: APJ and DOR N-terminus-swapped mutant; APJ/DOR ECL1: APJ and DOR ECL1-swapped mutant; APJ E174A: APJ E174 alanine substitution mutant; APJ/DOR ECL3: APJ and DOR ECL3-swapped mutant. Data are expressed as mean with SD in a, b.

Fig. 4. Function-based cell sorting and screening for agonistic cell clones.

a Dot-plot of Tango/APJ sdAb cell-displayed library following substrate loading. Cells with high ratio of product/substrate (460 nm/530 nm) were sorted out. The gating window and the percent of gated cell population in the 3rd round cell sorting are shown. b β-arrestin recruitment activity of the parental library and the 1st, 2nd and 3rd round sorted cell populations by Tango/APJ β-arrestin assay. Data are expressed as mean with SD. Statistical analysis was done using unpaired two-tailed t test. The higher product/substrate ratio, the higher agonist activity. c Percentage activation of monoclonal cells by Tango/APJ β-arrestin assay. Stimulation by 100 nM apelin 13 and 0.1% DMSO were set as 100% activation and 0% activation, respectively. d Confocal microscopy of single-cell clones tested for β-arrestin recruitment activity. Negative control: Parental Tango/APJ β-arrestin assay cells. Positive control: Parental Tango/APJ β-arrestin assay cells treated with 15 nM apelin 13. Two representative single-cell clones, 9E10 and 4C6, are shown. A scale bar of 250 µm is shown in each image.

Fig. 5. Characterization of sdAb agonist JN300 for binding and function.

a Flow cytometry of 293FT/HA-APJ and 293FT/HA-SSTR2 stable cell pools stained with JN300. b Binding of JN300 to WT APJ, APJ/DOR domain-swapped mutants, and APJ E174A mutant relative to its binding to WT APJ by flow cytometry. c–e Agonist activity of soluble sdAb JN300 in PathHunter β-arrestin assay (c), LANCE cAMP assay (d) and PathHunter-based receptor internalization assay (e). An irrelevant sdAb served as isotype control in (a and c–e). Data are expressed as mean with SD.