Development of a peptide-based tumor-activated checkpoint inhibitor for cancer immunotherapy

Abstract

Antibody-based checkpoint inhibitors have achieved great success in cancer immunotherapy, but their uncontrollable immune-related adverse events remain a major challenge. In this study, we developed a tumor-activated nanoparticle that is specifically active in tumors but not in normal tissues. We discovered a short anti-PD-L1 peptide that blocks the PD-1/PD-L1 interaction. The peptide was modified with a PEG chain through a novel matrix metalloproteinase-2 (MMP-2)-specific cleavage linker. The modified TR3 peptide self-assembles into a micelle-like nanoparticle (TR3-M-NP), which remains inactive and unable to block the PD-1/PD-L1 interaction in its native form. However, upon cleavage by MMP-2 in tumors, it releases the active peptide. The TR3-M-NP_5k nanoparticle was specifically activated in tumors through enzyme-mediated cleavage, leading to the inhibition of tumor growth and extended survival compared to control groups. In summary, TR3-M-NP shows great potential as a tumor-responsive immunotherapy agent with reduced toxicities. STATEMENT OF SIGNIFICANCE: In this study, we developed a bioactive peptide-based checkpoint inhibitor that is active only in tumors and not in normal tissues, thereby potentially avoiding immune-related adverse effects. We discovered a short anti-PD-L1 peptide, TR3, that blocks the PD-1/PD-L1 interaction. We chemically modified the TR3 peptide to self-assemble into a micelle-like nanoparticle (TR3-M-NP), which itself cannot block the PD-1/PD-L1 interaction but releases the active TR3 peptide in tumors upon cleavage by MMP-2. In contrast, the nanoparticle is randomly degraded in normal tissues into peptides fragments that cannot block the PD-1/PD-L1 interaction. Upon intraperitoneal injection, TR3-M-NP_5k was activated specifically in tumors through enzyme cleavage, leading to the inhibition of tumor growth and extended survival compared to the control groups. In summary, TR3-M-NP holds significant promise as a tumor-responsive immunotherapy agent with reduced toxicities. The bioactive platform has the potential to be used for other types of checkpoint inhibitor.

Keywords: Anti-PD-L1; Peptide; Self-assembly nanoparticle; Tumor-activated immunotherapy.

Figure 1.. Discovery and characterization of low-molecular weight anti-PD-L1 peptides.

(A) Sequences of the truncated anti-PD-L1 peptides. (B) Heatmap of binding sites from molecular docking of truncated anti-PD-L1 peptides with the human PD-L1 protein (PDB ID: 5C3T). (C) Blocking efficacy of the peptides (10 μM) against the human PD-L1 protein. Results are presented as the mean±SD (n=3, *P<0.05, **P<0.01). (D) Representative SPR sensorgrams of the TR3 peptide to immobilized human PD-L1 ECD and albumin. The experiments were performed in three replicates. (E) K_D value of the truncated anti-PD-L1 peptides (TR1, TR2, TR3, and TR4) against human PD-L1 ECD and albumin. (F) Binding affinities of the TR3 peptide towards human MDA-MB-231 WT and MDA-MB-231 PD-L1 KO cells. Results are presented as the mean±SD. (n=3, **P<0.01). (G) Representative fluorescence images of MDA-MB-231 and MDA-MB-231 PD-L1 KO cells after 1 hour of incubation with the Cy5-labeled TR3 peptide. The scale bar is 100 μm. (H) CT26 tumor-bearing Balb/c mice (n=8, half male and half female) were intraperitoneally injected with the TR3 peptide (2 mg/kg) daily for a total of 10 injections. (I) Tumor growth curve. Tumor volumes were presented as the mean±SEM. (n=8, **P < 0.01). (J) Body weight over time. (K) Survival curve. The Gehan-Breslow-Wilcoxon test was used for analyzing survival curves using GraphPad Prism 7 software (*P < 0.05). (L) Tumor growth curve for individual mice in each group.

Figure 2.. Characterization and in vitro cytotoxicity of TR3-NPs.

(A) Molecular docking of modified TR3 peptides (TR3, LVG-TR3 and TR3-LAG) with the human PD-L1 protein (PDB ID: 5C3T). (B) K_D value of the modified TR3 peptides against human PD-L1 ECD and albumin. (C) Binding curves of the modified TR3 peptides on MDA-MB-231 cells after 1 hour incubation. (D) Flow cytometry analysis of labelling MDA-MB-231 cells after incubation with TR3 peptides (500 nM) for 1 hour. (E) Blocking efficiency of TR3, LVG-TR3 and TR3-LAG peptides (10 μM) against the human PD-1/PD-L1 interaction. Results are presented as the mean±SD (n=3, **P<0.01). (F) Sequences of each TR3-related nanoparticle and peptides. Zeta potential (G), particle size (H), TEM images (I), and CMC (J) of TR3-M-NP_-2k and TR3-M-NP_-5k. (n=3). All results are presented as the mean±SD.

Figure 3.. Enzyme cleavage of TR3-NPs.

UV and Mass spectrometry of TR3-M-NPs were monitored before and after specific cleavage by the MMP-2 enzyme. UV and Mass spectrometry of TR3-M-NP-_2k (A) and TR3-M-NP-_5k (C) before cleavage. UV and Mass spectrometry of TR3-M-NP-_2k (B) and TR3-M-NP-_5k (D) after MMP-2 enzyme cleavage. (E) Cleavage curves of TR3-M-NPs. (F) Release curves of TR3-LAG peptide from TR3-M-NPs. (G) Blocking efficiency of TR3-M-NP-_2k after cleavage by MMP-2 enzymes. (H) Blocking efficiency of TR3-M-NP-_5k and MMP-2 uncleavable TR3-G-NP-_5k after a 2-hour enzyme cleavage reaction. (I) IC₅₀ of TR3-LAG peptide against the human PD-1/PD-L1 interaction. All results are presented as the mean±SD. (n=3, *P < 0.05, **P < 0.01).

Figure 4.. The anti-PD-L1 peptides reduce T cell apoptosis.

Fresh hPBMC were co-cultured with DU145 cells (1:2 ratio) and then incubated with TR3 and TR3-LAG peptides. (A) Identification of CD3⁺ T cells from hPBMC. (B) Flow cytometry analysis of apoptotic CD3⁺ T cells. (C) Quantitative analysis of CD3⁺ T cells. Anti-PD-L1 peptides restore the secretion of IFN-γ (D), GM-CSF (F), IL-6 (G), and IL-10 (H), but not TNF-α (E) from hPBMC after co-culturing with DU145 cells. All results are presented as the mean±SD. (n=3, *P<0.05, **P<0.01).

Figure 5.. In vivo biodistribution and pharmacokinetic study of TR3-M-NPs.

TR3 peptide and TR3-M-NPs were labeled with Cy5 for biodistribution and pharmacokinetic studies. (A) Fluorescence images of the liver, spleen, tumor, heart, lung, kidney, and muscle were taken 24 hours after intravenous and intraperitoneal injection. (B) Mean fluorescence of each organ was normalized to that of the heart. Fluorescence images (C) and mean intensity analysis (D) of major organs after 6, 24, and 48 hours of i.p. injection. (E) Pharmacokinetic study of Cy5-labeled TR3 peptide and TR3-M-NPs after intravenous injection at a dose of 8.15 nmole peptide/kg. All results are presented as the mean±SD. (n=3, *P < 0.05, **P < 0.01).

Figure 6.. Anti-tumor activity of TR3-M-NP-_5k.

CT26 tumor-bearing Balb/c mice (n=8, half male and half female) were treated with TR3-LAG peptide, TR3-M-NP-_5k, and TR3-G-NP-_5k at an equivalent dose of 2 mg/kg TR3 peptide via intraperitoneal injection every two days for a total of 6 injections. (A) Tumor growth curve. Tumor volumes were presented as the mean ± SEM. Images (B) and weights (C) of tumors harvested on day 16. (D) The fold change of CD8⁺ T cells in tumors. (E) The expression of IFN-γ in tumors (n=8). The levels of ALT (F), AST (G), and total proteins (H) in the plasma (n=6–7). All results are presented as the mean ± SD. (I) Survival curve of the mice treated with TR3-M-NP-_5k for 6 injections from Day 5 to Day 15. The Gehan-Breslow-Wilcoxon test was used to analyze survival curves using GraphPad Prism 7 software. (J) Tumor growth curve for individual mice in each group. (n=8, *P < 0.05, **P < 0.01)

Scheme 1.. Schematics of the tumor-activated anti-PD-L1 nanoparticles.

(A) The N-terminus of the TR3 peptide is modified with PEG and a cleavable linker specific to the MMP-2 enzyme. These modified peptides self-assemble into micelle-like nanoparticles known as TR3-M-NP. (B) Following administration, the nanoparticles accumulate in the tumor microenvironment due to the EPR effect. Subsequently, the active TR3 peptide is released from TR3-M-NP through cleavage by MMP-2. This liberated TR3 peptide binds to PD-L1 on the surface of cancer cells to block the PD-1/PD-L1 pathway. In contrast, in normal tissues where MMP-2 is absent, we hypothesize that the nanoparticles are degraded into random peptide fragments that cannot bind to PD-L1.