A biparatopic agonistic antibody that mimics fibroblast growth factor 21 ligand activity

Abstract

Bispecific antibodies have become important formats for therapeutic discovery. They allow for potential synergy by simultaneously engaging two separate targets and enable new functions that are not possible to achieve by using a combination of two monospecific antibodies. Antagonistic antibodies dominate drug discovery today, but only a limited number of agonistic antibodies (i.e. those that activate receptor signaling) have been described. For receptors formed by two components, engaging both of these components simultaneously may be required for agonistic signaling. As such, bispecific antibodies may be particularly useful in activating multicomponent receptor complexes. Here, we describe a biparatopic (i.e. targeting two different epitopes on the same target) format that can activate the endocrine fibroblast growth factor (FGF) 21 receptor (FGFR) complex containing β-Klotho and FGFR1c. This format was constructed by grafting two different antigen-specific VH domains onto the VH and VL positions of an IgG, yielding a tetravalent binder with two potential geometries, a close and a distant, between the two paratopes. Our results revealed that the biparatopic molecule provides activities that are not observed with each paratope alone. Our approach could help address the challenges with heterogeneity inherent in other bispecific formats and could provide the means to adjust intramolecular distances of the antibody domains to drive optimal activity in a bispecific format. In conclusion, this format is versatile, is easy to construct and produce, and opens a new avenue for agonistic antibody discovery and development.

Keywords: antibody engineering; biparatopic antibody; bispecific antibody; drug development; fibroblast growth factor (FGF); fibroblast growth factor receptor (FGFR); single-domain antibody (sdAb, nanobody); β-Klotho.

Figure 1.

Identification of heavy chain-only domains that interact with human β-Klotho. A, schematic representation of the structures of VHO, HCAb, and a traditional IgG antibody. B, binding of VHO1 and 2 to human β-Klotho measured by bio-layer interferometry. K_D values were estimated using a globally fitted 1:1 binding model. Data are mean ± S.E. (n = 3). C, binding of VHO1 and 2 (30 nm) to human FGFR1c measured by bio-layer interferometry. D, competition between VHO2 and VHO1 for β-Klotho binding. Human β-Klotho ECD His₆ protein was immobilized onto biosensors in an Octet RED instrument and exposed to VHO1 and VHO2 (30 nm) in two consecutive association steps. E, competition between VHOs and FGF21 for β-Klotho binding. Human β-Klotho ECD His₆ protein was immobilized onto biosensors in an Octet RED instrument and exposed to FGF21 (30 nm) and VHO1 or 2 (30 nm) in two consecutive association steps. Data are representative of three independent experiments.

Figure 2.

Bivalency of heavy chain-only domains is required to activate receptor signaling. A and B, stimulation of Elk1-luciferase activity by FGF21 (A), VHO1 and VHO2 and (B) HCAb-VH1 and HCAb-VH2 in CHO reporter cells stably expressing human β-Klotho and FGFR1c. C, binding of HCAb-VH1 and 2 to human β-Klotho measured by bio-layer interferometry. K_D values were estimated using a globally fitted 1:1 binding model. D, schematic representation of IgG-VH1+VH1, IgG-VH2+VH2, Fab-VH1+VH1, and Fab-VH2+VH2. E, stimulation of Elk1-luciferase activity in CHO reporter cells. F, binding of antibody molecules to human β-Klotho measured by bio-layer interferometry. K_D values were estimated using a globally fitted 1:1 binding model. All data are representative of three independent experiments and expressed as mean ± S.E.

Figure 3.

Generation of a biparatopic monoclonal antibody. A, schematic representation of the biparatopic antibody, IgG-VH1+VH2. The close and distant geometries of the two β-Klotho binding domains are noted. B, stimulation of Elk1-luciferase activity in CHO reporter cells. C, binding of IgG-VH1+VH2 to human β-Klotho measured by bio-layer interferometry. K_D values were estimated using a globally fitted 1:1 binding model. D, inhibition of FGF21 signaling by IgG-VH1+VH2. CHO reporter cells stably expressing human β-Klotho and FGFR1c were treated with a combination of FGF21 and IgG-VH1+VH2 for 4 h, after which luciferase activity was measured. E, rat L6 cells were co-transfected with expression vectors encoding human FGFRs and β-Klotho. Cells were stimulated with treatment molecules (100 nm) followed by analysis of cell lysates for pERK. Results are expressed as percentage of phosphorylated ERK protein over total ERK protein. Data are representative of three independent experiments and expressed as mean ± S.E.

Figure 4.

The correct orientation of VH1 and VH2 domains is required for optimal agonistic activity. A, schematic representation of IgG-VH1+sLC (sLC represents a light chain from an unrelated antibody) and IgG-VH2+VH1. B, binding of IgG-VH1+sLC and IgG-VH2+VH1 to human β-Klotho measured by bio-layer interferometry. K_D values were estimated using a globally fitted 1:1 binding model. C, stimulation of Elk1-luciferase activity in CHO reporter cells. D and E, effect of (D) IgG-VH1+sLC and (E) IgG-VH2+VH1 on FGF21 signaling. CHO reporter cells stably expressing human β-Klotho and FGFR1c were treated with a combination of FGF21 and antibody molecules for 4 h, after which luciferase activity was measured. Data are representative of three independent experiments and expressed as mean ± S.E.

Figure 5.

The Fab fragment of IgG-VH1+VH2 can activate β-Klotho/FGFR1c signaling. A, schematic representation of the Fab fragment of IgG-VH1+VH2. B, stimulation of Elk1-luciferase activity by Fab-VH1+VH2 in CHO reporter cells. C, binding of Fab-VH1+VH2 to human β-Klotho measured by bio-layer interferometry. K_D values were estimated using a globally fitted 1:1 binding model. D, effect of Fab-VH1+VH2 on FGF21 signaling. CHO reporter cells stably expressing human β-Klotho and FGFR1c were treated with a combination of FGF21 and Fab-VH1+VH2 for 4 h, after which luciferase activity was measured. Data are representative of three independent experiments and expressed as mean ± S.E.