Design and selection of anti-PD-L1 single-domain antibody and tumor necrosis factor superfamily ligands for an optimal vectorization in an oncolytic virus

Abstract

Arming oncolytic viruses with transgenes encoding immunomodulators improves their therapeutic efficacy by enhancing and/or sustaining the innate and adaptive anti-tumoral immune responses. We report here the isolation, selection, and vectorization of a blocking anti-human PDL1 single-domain antibody (sdAb) isolated from PDL1-immunized alpacas. Several formats of this sdAb were vectorized into the vaccinia virus (VV) and evaluated for their programmed cell death protein 1 (PD1)/PD1 ligand (PDL1) blocking activity in the culture medium of tumor cells infected in vitro. In those conditions, VV-encoded homodimeric sdAb generated superior PDL1 blocking activity compared to a benchmark virus encoding full-length avelumab. The sdAb was further used to design simple, secreted, and small tumor necrosis factor superfamily (TNFSF) fusions with the ability to engage their cognate receptors (TNFRSF) only in the presence of PDL1-positive cells. Finally, PDL1-independent alternatives of TNFRSF agonists were also constructed by fusing different variants of surfactant protein-D (SP-D) oligomerization domains with TNFSF ectodomains. An optimal SP-D–CD40L fusion with an SP-D collagen domain reduced by 80% was identified by screening with a transfection/infection method where poxvirus transfer plasmids and vaccinia virus were successively introduced into the same cell. However, once vectorized in VV, this construct had a much lower CD40 agonist activity compared to the SP-D–CD40L construct, which is completely devoid of the collagen domain that was finally selected. This latest result highlights the importance of working with recombinant viruses early in the payload selection process. Altogether, these results bring several complementary solutions to arm oncolytic vectors with powerful immunomodulators to improve their immune-based anti-tumoral activity.