Inhibiting the Keap1/Nrf2 Protein-Protein Interaction with Protein-Like Polymers

Abstract

Successful and selective inhibition of the cytosolic protein-protein interaction (PPI) between nuclear factor erythroid 2-related factor 2 (Nrf2) and Kelch-like ECH-associating protein 1 (Keap1) can enhance the antioxidant response, with the potential for a therapeutic effect in a range of settings including in neurodegenerative disease (ND). Small molecule inhibitors have been developed, yet many have off-target effects, or are otherwise limited by poor cellular permeability. Peptide-based strategies have also been attempted to enhance specificity, yet face challenges due to susceptibility to degradation and lack of cellular penetration. Herein, these barriers are overcome utilizing a polymer-based proteomimetics. The protein-like polymer (PLP) consists of a synthetic, lipophilic polymer backbone displaying water soluble Keap1-binding peptides on each monomer unit forming a brush polymer architecture. The PLPs are capable of engaging Keap1 and displacing the cellular protective transcription factor Nrf2, which then translocates to the nucleus, activating the antioxidant response element (ARE). PLPs exhibit increased Keap1 binding affinity by several orders of magnitude compared to free peptides, maintain serum stability, are cell-penetrant, and selectively activate the ARE pathway in cells, including in primary cortical neuronal cultures. Keap1/Nrf2-inhibitory PLPs have the potential to impact the treatment of disease states associated with dysregulation of oxidative stress, such as NDs.

Keywords: antioxidant; biomaterial; drug delivery; peptides; polymers.

Figure 1.

Keap1/Nrf2 Protein-Protein Interaction, Antioxidant Pathway and the PLP inhibitor. A/B) Keap1/Nrf2 PPI wherein Neh2 domain of Nrf2 (B) interacts with Keap1 homodimer (A) Kelch domains (red) at the low-affinity DLG region (teal) and the high-affinity ETGE region (blue). Keap1 and Nrf2 structures are derived from Alphafold Q14145 and Alphafold Q60795, respectively. C) Keap1/Nrf2 pathway. Under oxidative stress Keap1 conformation change prevents Nrf2 binding leading to Nrf2 activation of the ARE DNA promoter region. Similarly, Keap1/Nrf2-inhibitors, lead to Nrf2 accumulation and have been proposed as antioxidant therapeutics. D) All-atom molecular dynamics simulation showing the globular structure of the PLP, with the hydrophobic polynorbornene backbone (purple) linked to and surrounded by hydrophilic Keap1 binding peptides derived from Nrf2 (gray ribbons with ETGE amino acid motif highlighted in blue).

Figure 2.

Keap1/Nrf2-Inhibiting PLPs are easily synthesized, bind Keap1, and activate the ARE pathway A) Predicted structure by in silico MARTINI simulation of the Kelch domain with the 8^th peptide side chain (blue) of PLP-10 (DP=15, purple backbone and grey unbound sidechains) docked. B) Synthesis of PLP homopolymers via ROMP with varying peptide groups (P-1, P-10, P-FS, P-SR). C) TR-FRET binding assay for Keap1/Nrf2-inhibiting PLPs and controls (n=3), showing no inhibition of Keap1 by scrambles and enhanced inhibition of Keap1 by PLPs as compared to free peptides. D) ARE-Luc HepG2 Reporter Assay wherein PLP-10_Low [EC₅₀ = 3.76 μM (95% CI: 2.96–4.56)] and PLP-10_High [EC₅₀ = 3.56 μM (95% CI: 2.19–4.94)] show dose-dependent activation 24 hours post-treatment and are compared to the small molecule inhibitor positive control, tBHQ (100 μM) as well as scramble PLPs (n=5). Dotted line represents baseline activation of the untreated controls (relative luminescence of 100%). The activation seen for each group is modeled using a nonlinear regression over the tested concentrations as shown, along with the SEM for each tested concentration group.

Figure 3.

Addition of PLP-10 disrupts Nrf2-Keap-1 interaction and induces Nrf2 translocation in the nucleus. A) HEK293T cells were transiently transfected with a KEAP-1-FLAG plasmid. These cells were then treated with either 10 μM PLP-10 (Mn 29.2 kDa) or PLP-SR (32.5 kDa) for 24 hrs, prior to capture of Keap1-FLAG from lysates via FLAG-tagged magnetic beads followed by detection of Keap1-FLAG via an anti-FLAG antibody. In addition, we analyzed these samples for co-eluted Nrf2. In this manner, we confirmed that, in the presence of PLP-10, Nrf2 does not co-elute with Keap1-FLAG, while PLP-SR treatment had no effect, with Nrf2 clearly co-eluting with its binding partner. Further, to confirm induction of nuclear translocation of Nrf2 by PLP-10, we performed nuclear translocation assays by isolating nuclei with cellular fractionation following treatment in three cell lines: (B) SK-N-SH, Neuroblastoma; (C) CCF-STG1, Astrocytoma; (D) HMC3, Microglial. In all three lines, we confirm significantly higher detected cytosolic and nuclear concentrations of Nrf2 upon treatment with PLP-10 compared to the scrambled control, PLP-SR. Representative blot from two independent experiments is shown. Migration of molecular weight standards (M) is depicted on the left as kDa.

Figure 4.

Keap1/Nrf2-Inhibiting PLPs activate ARE and downstream product expression A) ARE-hPAP reporter assay in murine primary cortical cultures treated with Keap1/Nrf2-inihibiting PLPs and controls, showing dose-dependent activation of PLP-10_Low [EC₅₀ = 0.68 μM (95% CI: 0.55–0.74)] and PLP-10_High [EC₅₀ = 0.17 μM (95% CI: 0.11–0.24)] with no activation produced by the scrambled controls (PLP-FS and PLP-SR) after 48 hrs (n=6). The activation seen for each group is modeled using a nonlinear regression over the tested concentrations as shown, along with the SEM for each tested concentration group. B-C) Western blot analysis of HMOX1 (Panel B) and NQO1 (Panel C) expression in primary cortical cultures treated with PLP-10_Low, PLP-10_High and PLP-SR after 48 hrs (n=3). The polymers were directly compared, correcting for molecular weight, at a concentration of 0.1 mg/mL of each PLP. This is equivalent to a concentration of 5 μM each on a per peptide basis. Groups were compared via one-way ANOVA with multiple comparisons (**** = p < 0.0001, *** = p < 0.001, ** = p < 0.01, * = p < 0.05) D) Tandem Mass Tagged (TMT) Global Proteomic analysis of primary cultures treated with tBHQ via comparison of the Log₂ fold change of cells treated with vehicle. E) TMT Global Proteomic analysis of primary cultures treated with PLP-10_Low via comparison of the Log₂ fold change of cells treated with scramble control, PLP-SR. F) TMT Global Proteomic analysis of primary cultures via comparison of the Log₂ fold change of cells treated with PLP-10_Low (0.67 μM) relative to PLP-SR (0.67 μM) as compared to cells treated with tBHQ (10 μM) relative to vehicle.

Figure 5.

Keap1/Nrf2-Inhibiting PLPs permeate Neurons and Glia and show Glial-dominant Activation A) Cellular uptake of F- PLPs at 0.1 mg/mL (5 μM on a per peptide basis). Left to right: Untreated, F-PLP-SR, F-PLP-10_Low, F-PLP-10_High; F-PLPs labeled with Cy5.5 are shown in magenta. In the top row, astrocytes are labeled with GFAP and shown in green. The bottom row shows neurons labeled with MAP2, shown in green. All cell nuclei are labeled with Hoescht 33258 and shown in blue. All tested PLPs were internalized by both neurons and astrocytes. Magnified images are given for PLP-10_Low and PLP-10_High samples at 60X as insets. The provided scale bar is 30 μm for the inset images. B) hPAP reporter levels following administration of PLPs at 0.1 mg/mL (5 μM on a per peptide basis). Left to right: Untreated, PLP-SR, PLP-10_Low, and PLP-10_High. In the top row, astrocytes are labeled with GFAP and shown in green. The bottom row shows neurons labeled with MAP2, shown in green. All cell nuclei are labeled with Hoescht 33258 and shown in blue. Activation was assessed using the hPAP reporter which is shown in red. Magnified images are given for PLP-10_Low and PLP-10_High samples at 60X as insets. The provided scale bar is 30 μm for the inset images. While the untreated and PLP-SR controls do not show any activation in the neurons or astrocytes, both PLP-10_Low and PLP-10_High show reporter activation. Colocalization of the red hPAP signal with the green GFAP signal from the astrocytes suggests astrocyte-predominate Nrf2 pathway activation. Of note, these cultures contain cells from both ARE-hPAP positive and negative E15–16 pups from a WT mouse crossed with a hemizygous ARE-hPAP positive mouse. Thus, not all of the astrocytes in culture are ARE-hPAP positive, explaining the absence of hPAP staining in some astrocytes.