Neuropilin-1 drives tumor-specific uptake of chlorotoxin

Abstract

Background: Chlorotoxin (Cltx) isolated from scorpion venom is an established tumor targeting and antiangiogenic peptide. Radiolabeled Cltx therapeutic (¹³¹I-TM601) yielded promising results in human glioma clinical studies, and the imaging agent tozuleristide, is under investigation in CNS cancer studies. Several binding targets have previously been proposed for Cltx but none effectively explain its pleiotropic effects; its true target remains ambiguous and is the focus of this study.

Methods: A peptide-drug conjugate (ER-472) composed of Cltx linked to cryptophycin as warhead was developed as a tool to probe the molecular target and mechanism of action of Cltx, using multiple xenograft models.

Results: Neuropilin-1 (NRP1), an endocytic receptor on tumor and endothelial cells, was identified as a novel Cltx target, and NRP1 binding by Cltx increased drug uptake into tumor. Metabolism of Cltx to peptide bearing free C-terminal arginine, a prerequisite for NRP1 binding, took place in the tumor microenvironment, while native scorpion Cltx with amidated C-terminal arginine did not bind NRP1, and instead acts as a cryptic peptide. Antitumor activity of ER-472 in xenografts correlated to tumor NRP1 expression. Potency was significantly reduced by treatment with NRP1 blocking antibodies or knockout in tumor cells, confirming a role for NRP1-binding in ER-472 activity. Higher cryptophycin metabolite levels were measured in NRP1-expressing tumors, evidence of NRP1-mediated enhanced drug uptake and presumably responsible for the superior antitumor efficacy.

Conclusions: NRP1 was identified as a novel Cltx target which enhances tumor drug uptake. This finding should facilitate tumor selection for chlorotoxin-based therapeutics and diagnostics.

Keywords: Chlorotoxin; Neuropilin-1; Peptide-drug-conjugate; Tumor; Uptake.

Fig. 1

Structure of Cltx, ER-472 and related compounds. a, amino acid sequence of native Cltx which features 4 disulfide bonds, 3 lysine (K) and 3 arginine (R, in bold) residues, including amidated R at the C-terminus. b, structure of ER-472 PDC comprised of Cltx peptide linked via K27 to a novel cryptophycin analog (cytotoxic warhead), via a cleavable dimethyl disulfide linker. c, S-methyl cryptophycin, the active metabolite of ER-472 generated in tumor. d, ER-271 mimics ER-472 that lacks Cltx; composed of cryptophycin analog plus the dimethyl disulfide linker which terminates in an aryl group

Fig. 2

ER-472 antitumor activity and levels of active metabolite in 3 xenograft models; MIA PaCa-2, BxPC-3 and PC-3. Subcutaneous tumors were established in nude mice; treatment groups were n = 6 for MIA PaCa-2 and BxPC-3 and n = 5 for PC-3 efficacy studies and n = 3 for all metabolite studies. a, ER-472 was administered i.v. Q4Dx3 (dosing days denoted with long ticks on x-axis). Tumor volume data represents the mean ± SEM. b, mean active metabolite (S-methyl cryptophycin) tumor concentration versus time following a single dose of ER-472 at 2.5 mg/kg (MTD). Data represents the mean ± StdDev. c, antitumor activity of ER-472 versus ER-271 at molar equivalent (ME) dose in PC-3 model. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, one-way ANOVA followed by Tukey’s multiple comparison test; n.s. = not significant

Fig. 3

Cltx binds to NRP1 only when its C-terminal arginine is de-amidated; Cltx de-amidation occurs in tumor. Cltx-CONH₂ (native Cltx with amidated C-terminal arginine) does not bind to NRP1 at concentrations from 0 to 1750 nM (a) while Cltx-COOH (Cltx with a carboxylated C-terminal arginine) demonstrated dose responsive NRP1 binding with an affinity (K_D) of ~ 240 nM (c). Graphs are output from BLI assay with K_D determined by octet data analysis software version 8.2; graphs are representative of multiple assays. b, peptides identified in PC-3 tumor lysate by MS analysis 1 h post dose of Cltx, in order of abundance. Peptides represented in black or red have amidated versus de-amidated C-terminal arginine residue respectively. Peptides 1 and 3 are full length Cltx with amidated (native) versus de-amidated C-terminal arginine respectively. d, VEGF binding to NRP1 (fixed concentration at 390 nM) was dose dependently inhibited in the presence of increasing concentration of Cltx-COOH (0 to 800 μM), R² = 0.98; suggests that Cltx binding to NRP1 occurs at VEGF binding site

Fig. 4

High NRP1 expression in PC-3 tumors correlates with ER-472 antitumor activity; anti-NRP1 blocking antibody treatment reduced ER-472 efficacy. a, expression of NRP1 mRNA in tumor lysates from 3 xenograft models. For each model n = 4 tumors, data represents mean ± SD. b, Western blot of NRP1 protein levels in tumor lysates, for each model at least 3 lysates were pooled for analysis (50 μg loaded onto gel) and data is representative of several individual experiments. Anti-NRP1 antibody detects both full length (~ 120 kDa) and soluble forms of protein (70–80 kDa); recombinant human NRP1 protein (15 ng) represents soluble/extracellular domain of NRP1. c, blocking anti-NRP1 antibody administered i.p. 30 min prior to ER-472 (0.9 mg/kg) diminished the PDC’s antitumor effect versus control IgG pre-treatment, *P ≤ 0.05 one-way ANOVA followed by Dunnett’s multiple comparison test

Fig. 5

Knockout of NRP1 in PC-3 tumors blunts the antitumor effect of ER-472 through reducing active metabolite uptake into tumors. Subcutaneous tumors were established in NOD SCID mice; treatment groups were n = 5 or 6 and treatment schedule was Q4Dx3 (dosing days denoted with long ticks on x-axis). Tumor volume data represents the mean ± SEM. Differential sensitivity of PC-3 NRP1 WT versus KO xenografts to treatment with ER-472 (a) compared to ER-271 (b). **P ≤ 0.01, ***P ≤ 0.001, one-way ANOVA followed by Tukey’s multiple comparison test; n.s. = not significant. c, mean active metabolite (S-methyl cryptophycin) tumor concentration versus time profiles following a single dose of ER-472 at 0.6 mg/kg (¼ MTD). Data represents the mean ± StdDev. ****P ≤ 0.0001, unpaired t-test