Background Neuropilin-1 (NRP1) is a co-receptor that complexes with diverse ligands and their cognate receptors. As such, it plays a role in multiple different biological processes, including axon guidance and angiogenesis. NRP1 contains two CUB domains (a1 and a2) involved in binding the ligand Semaphorin3A (SEMA3A), two Factor V/VIII domains (b1 and b2) involved in VEGF ligand binding and one MAM domain (c domain). While functional antibodies with anti-tumor activity have been generated against the SEMA3A and VEGF binding domains, little attention has been paid to the c domain of NRP1, which has been implicated in the dimerization of NRP1, a prerequisite for functionality. We therefore hypothesized that c domain-binding antibodies would offer an opportunity to generate functional inhibitors of both SEMA3A and VEGF signaling and therefore improved anti-tumor activity.

Methods Recombinant human NRP1 comprising all subdomains was used to identify fully human anti-NRP1 antibodies. Specific antibodies were tested for their ability to block NRP1 interactions with recombinant SEMA3A and VEGF protein in vitro. Blocking antibodies were subsequently assessed for their functional effects, such as inhibition of SEMA3A-mediated growth cone collapse. Antibodies with diverse binding characteristics were then tested for in vivo anti-tumor activity in multiple cancer models of interest.

Results Recombinant NRP1 containing the a1, a2, b1, b2 and c subdomains was used to successfully identify a series of specific monoclonal antibodies that cross-reacted with Cynomolgus monkey and mouse NRP1, but not human NRP2. Except for the a2 domain, epitope mapping showed an even distribution of mAbs for binding to each of the NRP1 subdomains, including the c domain that has been proposed to play a role in dimerization. Using biolayer interferometry, we identified antibody classes with direct SEMA3A and/or VEGF blocking properties. Further optimization of these antibodies yielded mAbs with subnanomolar affinities that showed significant tumor growth inhibition in multiple mouse models, including anti-PD1 non-responsive models.

Conclusions Here we demonstrate the identification of fully human monoclonal antibodies that specifically bind to the c domain of human NRP1. A subset of these c domain binders do not block either SEMA3A or VEGF binding to NRP1 but do show in vivo efficacy, suggesting a role for the c domain of NRP1 in the formation of functional (dimeric) complexes. Thus, c domain binding antibodies show remarkable inhibition of tumor growth in mouse cancer models and offer a novel means of therapeutic intervention in patients who are refractory to immune checkpoint inhibition.