Enhancing NK cell-mediated tumor killing of B7-H6+ cells with bispecific antibodies targeting allosteric sites of NKp30

Abstract

In this work, we report the discovery and engineering of allosteric variable domains of the heavy chain (VHHs) derived from camelid immunization targeting NKp30, an activating receptor on natural killer (NK) cells. The aim was to enhance NK cell-mediated killing capacities by identifying VHHs that do not compete with the natural ligand of NKp30:B7-H6, thereby maximizing the recognition of B7-H6⁺ tumor cells. By relying on the DuoBody technology, bispecific therapeutic antibodies were engineered, creating a panel of bispecific antibodies against NKp30xEGFR (cetuximab moiety) or NKp30xHER2 (trastuzumab moiety), called natural killer cell engagers (NKCEs). These NKCEs were assessed for their killing capacities on B7-H6-expressing tumor cells. The results demonstrated an enhancement in NK killing capacities for both EGFR-expressing (HeLa) and HER2-expressing (SK-BR-3) cells, indicating the significance of the natural NKp30/B7-H6 axis in tumor recognition by the immune system. Notably, engineering NKCEs to allow natural recognition of B7-H6 was found to be more effective in promoting NKCE-mediated killing of B7-H6⁺ tumor cells via enhancement of cytokine release. This study highlights the potential of an enhanced-targeting approach, wherein tumor cell surface antigens are targeted while still enabling the natural recognition of the activating ligand (B7-H6) by the immune cells.

Keywords: DuoBody; EGFR; MT: Regular Issue; NKCE; NKp30; allosteric site; bispecific antibodies; bsAbs; cancer therapy; natural-killer cell engager.

Graphical abstract

Figure 1

Discovery of non-B7-H6 competitive anti-NKp30 VHHs (A) Overview of the workflow. Camelids were immunized with NKp30, and a yeast surface display library was constructed. The library was subsequently incubated with NKp30 and B7-H6, as well as their detection antibodies. Selected hits were reformatted as DuoBody with cetuximab arm (anti-EGFR) or trastuzumab arm (anti-HER2). (Created with BioRender.com.) (B) FACS-based discovery of non-B7-H6 competitive VHHs. A three-color-based sorting approach was applied: a two-dimensional gate to identify functional display in combination to NKp30 binding, and then, a second two-dimensional gate was applied to select NKp30 binders that were not competing with B7-H6. From the first round of sorting, 0.19% of the library was binding NKp30, while 0.1% was binding NKp30 in the presence of B7-H6. The 0.1% was selected for sorting. For the second and final round, 41.7% of the library was NKp30 binders. We applied a more stringent sorting gate to select VHHs that were better positively modulating NKp30 affinity for B7-H6 and selected 3.5% of the library for sorting. Applied sorting gates and corresponding cell populations (as percentage of total cells) are shown. Plots were generated using FlowJo. (C) Sequence homology of NKp30 binders. Sequence alignment of NKp30 binders showed a heterogeneous diversity. The alignment was generated with Geneious Prime.

Figure 2

Functional binding assay to NK-92 cells evaluated by FACS NKp30-expressing NK-92 cells were incubated with decreasing concentrations of anti-NKp30 antibodies. Antibody binding was evaluated via a secondary detection antibody (anti-Fc Alexa Fluor 488 conjugate) on an iQue3.

Figure 3

Dual binding of DuoBody determined by BLI and by FACS G10xcetuximab (A and C) or G10xtrastuzumab (B and D) are shown as examples. (A and B) Dual binding determined by BLI. DuoBody molecules were loaded on AHC2 biosensors. After a baseline step in kinetic buffer, binding to EGFR (A) or HER2 (B) was tested and simultaneous binding capacity to NKp30 was subsequently highlighted. (C and D) Dual binding determined by FACS on tumor cells. HeLa (C) or SK-BR-3 (D) cells were incubated with the corresponding DuoBody and subsequently incubated with NKp30 His-tagged protein. NKp30 His was detected for both constructs, showing simultaneous binding.

Figure 4

Characterization of NKCE targeting NKp30 and EGFR in terms of killing capacity and cytokine releases on HeLa cells (A) Fluorescence-based killing assay. HeLa cells were incubated with NK cells at a 1:5 ratio in the presence of the respective NKp30xEGFR targeting NKCE. Cetuximab with or without (LALA) effector-attenuated mutations were used as controls. All data have been normalized to the maximum killing of cetuximab. The graph shows normalized means ± SEMs of n = 6 healthy donors. Two-way ANOVA was performed; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001. (B) NK cell-mediated target cell-dependent cytotoxic properties determined after data fitting. (C) IFN-γ release. NK cell-mediated IFN-γ release was determined by HRTF analysis of the supernatant of the killing assay. The graph shows normalized means ± SEMs of n = 6 healthy donors. One-way ANOVA was performed; ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001. (D) TNF-α release. NK-cell mediated TNF-α release was determined by HRTF analysis of the supernatant of the killing assay. The graph shows normalized means ± SEMs of n = 6 healthy donors. One-way ANOVA was performed; ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001.

Figure 5

Killing assay of NKCE targeting NKp30 and HER2 on SK-BR-3 cells (A) SK-BR-3 cells were incubated with NK cells at a 1:5 ratio in the presence of the respective NKp30xHER2 targeting NKCE. Trastuzumab with or without (LALA) effector-attenuated mutations were used as controls. All data have been normalized to the maximum killing of trastuzumab. The graph shows normalized means ± SEMs of n = 6 healthy donors. two-way ANOVA was performed; ∗p ≤ 0.05; ∗∗∗p ≤ 0.001. (B) NK cell-mediated target cell-dependent cytotoxic properties determined after data fitting.