The making of multivalent gamma delta TCR anti-CD3 bispecific T cell engagers

We have recently developed a novel T cell engager concept by utilizing γ9δ2TCR as tumor targeting domain, named gamma delta TCR anti-CD3 bispecific molecule (GAB), targeting the phosphoantigen-dependent orchestration of BTN2A1 and BTN3A1 at the surface of cancer cells. GABs are made by the fusion of the ectodomains of a γδTCR to an anti-CD3 single chain variable fragment (scFv) (γδECTO-αCD3), here we explore alternative designs with the aim to enhance GAB effectivity.

The first alternative design was made by linking the variable domains of the γ and δ chain to an anti-CD3 scFv (γδVAR-αCD3). The second alternative design was multimerizing γδVAR-αCD3 proteins to increase the tumor binding valency. Both designs were expressed and purified and the potency to target tumor cells by T cells of the alternative designs was compared to γδECTO-αCD3, in T cell activation and cytotoxicity assays.

The γδVAR-αCD3 proteins were poorly expressed, and while the addition of stabilizing mutations based on finding for αβ single chain formats increased expression, generation of meaningful amounts of γδVAR-αCD3 protein was not possible. As an alternative strategy, we explored the natural properties of the original GAB design (γδECTO-αCD3), and observed the spontaneous formation of γδECTO-αCD3-monomers and -dimers during expression. We successfully enhanced the fraction of γδECTO-αCD3-dimers by shortening the linker length between the heavy and light chain in the anti-CD3 scFv, though this also decreased protein yield by 50%. Finally, we formally demonstrated with purified γδECTO-αCD3-dimers and -monomers, that γδECTO-αCD3-dimers are superior in function when compared to similar concentrations of monomers, and do not induce T cell activation without simultaneous tumor engagement. In conclusion, a γδECTO-αCD3-dimer based GAB design has great potential, though protein production needs to be further optimized before preclinical and clinical testing.