Conventional and Chemically Programmed Asymmetric Bispecific Antibodies Targeting Folate Receptor 1

Abstract

T-cell engaging bispecific antibodies (biAbs) can mediate potent and specific tumor cell eradication in liquid cancers. Substantial effort has been invested in expanding this concept to solid cancers. To explore their utility in the treatment of ovarian cancer, we built a set of asymmetric biAbs in IgG1-like format that bind CD3 on T cells with a conventional scFv arm and folate receptor 1 (FOLR1) on ovarian cancer cells with a conventional or a chemically programmed Fab arm. For avidity engineering, we also built an asymmetric biAb format with a tandem Fab arm. We show that both conventional and chemically programmed CD3 × FOLR1 biAbs exert specific in vitro and in vivo cytotoxicity toward FOLR1-expressing ovarian cancer cells by recruiting and activating T cells. While the conventional T-cell engaging biAb was curative in an aggressive mouse model of human ovarian cancer, the potency of the chemically programmed biAb was significantly boosted by avidity engineering. Both conventional and chemically programmed CD3 × FOLR1 biAbs warrant further investigation for ovarian cancer immunotherapy.

Keywords: CD3; FOLR1; bispecific antibodies; catalytic antibodies; folate; ovarian cancer.

Figure 1

Design of asymmetric biAbs for conventional and chemically programmed cancer cell targeting. Schematic diagram of v9 × Farl (A), v9 × h38C2 (B), and v9 × (h38C2)₂ (C). All three asymmetric biAbs share a scFv module of v9, a humanized mouse anti-human CD3 mAb, and a Fc module of human IgG1 combining one mutation for aglycosylation in the C_H2 domain (N297A) and two or four mutations in the C_H3 knob (S354C and T366W) and C_H3 hole (Y349C, T366S, L368A, and Y407V) domains, respectively, for heterodimerization. The Fab arm of the conventional biAb (A) is derived from humanized anti-human FOLR1 mAb farletuzumab (Farl). The single (B) or double (C) Fab arm of the chemically programmable biAbs is derived from humanized catalytic antibody 38C2 (h38C2).

Figure 2

Cell surface binding and catalytic activity of chemically programmed FOLR1-targeting biAb with a single Fab arm. (A) Flow cytometry analysis of unprogrammed and chemically programmed biAbs binding to CD3+ FOLR1– human T-cell line Jurkat and FOLR1+ CD3– human ovarian cancer cell line IGROV-1 using 20 nM biAbs followed by Alexa Fluor 647-conjugated goat anti-human IgG-Fc pAbs. Compounds 1a and 2 are monovalent and bivalent folate derivatives, respectively. (B) Catalytic retro-aldol activity of unprogrammed and chemically programmed h38C2-containing biAbs. The signal is reported in relative fluorescent units (RFU; mean ± SD of triplicates). PBS was used as negative control. (C) Titration curve of chemically programmed biAbs binding to IGROV-1 cells detected with Alexa Fluor 647-conjugated goat anti-human IgG pAbs (left panel). Saturation analysis of chemically programmed biAbs binding to IGROV-1 cells (right panel). Shown are mean ± SD values from independent triplicates.

Figure 3

Avidity engineering. In chemically programmed biAbs, the avidity effect of two Fab arms of conventional mAbs can be mimicked by either doubling the chemical or biological component. (A) Bivalent FOLR1 targeting with v9 × h38C2_(folate)₂. The two folate groups are spaced by a flexible PEG linker. (B) Bivalent FOLR1 targeting with v9 × (h38C2_folate)₂. The structural formula shows the amide bond formed by the reaction of β-lactam hapten derivatives of (folate)₂ and folate, respectively, with the ε-amino group of the uniquely reactive lysine residue (K) in the hapten binding site of h38C2.

Figure 4

Cell surface binding and catalytic activity of chemically programmed FOLR1-targeting biAb with a tandem Fab arm. (A) Catalytic retro-aldol activity of unprogrammed and chemically programmed h38C2-containing biAbs. The signal is reported in relative fluorescent units (RFU; mean ± SD of triplicates). PBS was used as negative control. (B) Flow cytometry analysis of unprogrammed and chemically programmed biAbs binding to CD3+ FOLR1– human T-cell line Jurkat and FOLR1+ CD3– human ovarian cancer cell line IGROV-1 using 20 nM biAbs followed by Alexa Fluor 647-conjugated goat anti-human IgG-Fc pAbs. Compound 1b is a monovalent folate derivative. (C) Titration curve of chemically programmed biAbs binding to IGROV-1 cells detected with Alexa Fluor 647-conjugated goat anti-human IgG pAbs (left panel). Saturation analysis of chemically programmed biAbs binding to IGROV-1 cells (right panel). Shown are mean ± SD values from independent triplicates.

Figure 5

Cross-linking and cytotoxicity mediated by FOLR1-targeting biAbs in vitro. (A) Cross-linking of 1 × 10⁶ Jurkat cells (stained with CellTrace Far Red) and 2 × 10⁵ IGROV-1 cells (stained with CellTrace CFSE) in the presence of 200 nM FOLR1-targeting biAbs and negative control. Double-stained events were detected by flow cytometry. (B) Quantification of double-stained events from three independent triplicates (mean ± SD). An unpaired two-tailed t-test was used to analyze significant differences (^*p < 0.05; ^**p < 0.01; ^***p < 0.001). (C) Cytotoxicity of biAbs tested with ex vivo expanded primary human T cells (effector cells) and IGROV-1 cells (target cells) at an effector-to-target cell ratio of 10:1 and measured after 16-h incubation. Shown are mean ± SD values from independent triplicates. Chemically programmed biAb v9 × h38C2_1b (red circles) was significantly (p < 0.001; extra sum-of-squares F-test) less potent than chemically programmed biAb v9 × (h38C2_1b)₂ (blue squares) and conventional biAb v9 × Farl (yellow triangles). (D) Corresponding cytotoxicity toward SKOV-3 cells. Shown are mean ± SD values from independent triplicates. Again, chemically programmed biAb v9 × h38C2_1b (red circles) was significantly (p < 0.001; extra sum-of-squares F-test) less potent than chemically programmed biAb v9 × (h38C2_1b)₂ (blue squares) and conventional biAb v9 × Farl (yellow triangles). The negative control, 0 × (h38C2_1b)₂, revealed no cytotoxicity in either experiment.

Figure 6

T-cell activation mediated by FOLR1-targeting biAbs in vitro. Ex vivo expanded primary human T cells (effector cells) were incubated with 50 nM of the indicated biAbs in the presence of IGROV-1 or SKOV-3 cells (target cells) at an effector-to-target cell ratio of 10:1 for 16 h. Shown in the upper panel is the percentage of activated T cells based on CD69 expression after incubation with IGROV-1 (A) or SKOV-3 cells (B) measured by flow cytometry. Shown in the lower panel is the cytokine release measured by ELISA for IFN-γ (C), IL-2 (D) and TNF-α (E). All experiments are shown as mean ± SD values from independent triplicates. An unpaired two-tailed t-test was used to analyze significant differences (^*p < 0.05; ^**p < 0.01; ^***p < 0.001).

Figure 7

In vivo activity of FOLR1-targeting biAbs. Five cohorts of mice (n = 5) were inoculated with 1 × 10⁶ IGROV-1/ffluc cells via i.p. injection. After 6 days, 1 × 10⁷ ex vivo expanded primary human T cells and the indicated biAbs or vehicle alone (PBS) were administered by the same route. The mice received a total of three doses of T cells every 8 days and a total of six doses of biAbs or vehicle alone every 4 days. (A) Bioluminescence images of all 25 mice from day 6 (before treatment) and days 28 and 32 (after treatment) are shown. (B) The weight of all 25 mice was recorded on the indicated days (mean ± SD). (C) Starting on day 6, all 25 mice were imaged every 3–5 days and their radiance was recorded (mean ± SD). Significant differences between cohorts treated with biAbs (colored graphs) and vehicle alone (black graph) were calculated using an unpaired two-tailed t-test (^***p < 0.001). Red arrows indicate the three T-cell doses and black arrows the six biAb or vehicle alone doses. (D) Corresponding Kaplan-Meier survival curves with p-values using log-rank (Mantel-Cox) test (^**p < 0.01).