Broad-spectrum activity against mosquito-borne flaviviruses achieved by a targeted protein degradation mechanism

Abstract

Viral genetic diversity presents significant challenges in developing antivirals with broad-spectrum activity and high barriers to resistance. Here we report development of proteolysis targeting chimeras (PROTACs) targeting the dengue virus envelope (E) protein through coupling of known E fusion inhibitors to ligands of the CRL4^CRBN E3 ubiquitin ligase. The resulting small molecules block viral entry through inhibition of E-mediated membrane fusion and interfere with viral particle production by depleting intracellular E in infected Huh 7.5 cells. This activity is retained in the presence of point mutations previously shown to confer partial resistance to the parental inhibitors due to decreased inhibitor-binding. The E PROTACs also exhibit broadened spectrum of activity compared to the parental E inhibitors against a panel of mosquito-borne flaviviruses. These findings encourage further exploration of targeted protein degradation as a differentiated and potentially advantageous modality for development of broad-spectrum direct-acting antivirals.

Fig. 1. Design of bivalent degraders targeting the DENV E protein.

A Structure of the soluble DENV2 prefusion E protein with βOG bound in the pocket located between domains I and II (PDB:1OKE). Note that only one monomer of the prefusion E is illustrated here for visual simplicity. B Structure of the DENV2 immature prM-E proteins (PDB:3C6E). Only one monomer of the immature trimer is shown. Domains I (red), II (yellow), III (blue), βOG (green), and prM (pink) are shown. C Chemical structures of GNF-2 and 2-12-2. D DENV E PROTACs GNF-2-deg, 2-12-2-deg, and their corresponding negative control compounds, GNF-2-deg-BUMP and 2-12-2-deg-BUMP, in which truncation of the carbonyl group at the glutarimide disrupts the CRBN-binding moiety.

Fig. 2. GNF-2-deg and 2-12-2-deg induce CRBN- and proteasome-dependent depletion of DENV E in infected cells.

A Schematic of antiviral assay utilized in the study. B Western blots show concentration-dependent depletion of E in the presence of GNF-2-deg and 2-12-2-deg. Depletion of E is not observed for negative control compounds GNF-2-deg-BUMP or 2-12-2-deg-BUMP or when the experiment is conducted in CRBN-deficient cells. The representative results are shown from n = 3 independent experiments for GNF-2 derivative compounds in WT and n = 2 for all other conditions. C Schematic of experiments probing the mechanism of E degradation. Cells were infected with DENV2 at an MOI of 1 and treated with the E degrader (GNF-2-deg or 2-12-2-deg) along with the CRBN ligand lenalidomide or the neddylation inhibitor MLN4924 from 1 to 24 h post-infection. Due to cytotoxicity in the presence of proteasome inhibitor MG-132, cotreatment was limited to 21 to 24 h post-infection. Western blots show that depletion of E in the presence of GNF-2-deg and 2-12-2-deg is dependent on CRBN-binding (D), neddylation activity to activate CRL4^CRBN (E), and proteasome activity (F). Representative results of lenalidomide treatment are shown from n = 4 and n = 2 independent experiments for GNF-2 and 2-12-2 derivatives, respectively. Representative results of MLN4924 treatment are shown from n = 3 and n = 2 independent experiments for GNF-2 and 2-12-2 derivatives, respectively. Representative results of MG-132 treatment are shown from n = 2 independent experiments. Figures 2A and 2C were created with Biorender.com.

Fig. 3. GNF-2-deg and 2-12-2-deg have CRBN- and proteasome-dependent antiviral activity against DENV.

A Culture supernatants from the experiments depicted in Fig. 2 were harvested at 24 h post-infection to allow quantification of viral yield by plaque-formation assay. Antiviral EC₉₀ values were determined by nonlinear regression analysis of the data for both E degraders (GNF-2-deg and 2-12-2-deg) and the parental inhibitors. GNF-2-deg and 2-12-2-deg exhibit an increase in antiviral potency compared to GNF-2 and 2-12-2 that is CRBN-dependent. The representative results are shown from n = 3 independent experiments for GNF-2 derivative compounds in WT and n = 2 for all other conditions. Data are presented as means normalized to DMSO ± standard deviation. The antiviral activity of GNF-2-deg and 2-12-2-deg is blocked by the addition of excess lenalidomide (p value < 0.0001 for GNF-2-deg and 0.0004 for 2-12-2-deg) (B), the addition of the neddylation inhibitor MLN4924 to block CRL4^CRBN activity (p value of 0.0287 for GNF-2-deg and 0.0274 for 2-12-2-deg) (C), or the addition of the proteasome inhibitor MG-132 (p value of 0.0505 for GNF-2-deg and 0.0985 for 2-12-2-deg) (D). Representative results of lenalidomide treatment are shown from n = 4 and n = 2 independent experiments for GNF-2 and 2-12-2 derivatives, respectively. Representative results of MLN4924 treatment are shown from n = 3 and n = 2 independent experiments for GNF-2 and 2-12-2 derivatives, respectively. Representative results of MG-132 treatment are shown from n = 2 independent experiments. Data are presented as means normalized to DMSO ± standard deviation. Asterisks indicate that the differences between samples are statistically significant, using the unpaired two-tailed t test. Data are presented as means normalized to DMSO ± standard deviation.

Fig. 4. GNF-2-deg and 2-12-2-deg inhibit viral particle production by a TPD mechanism.

Wild-type or CRBN^−/− Huh7.5 cells were transfected with the VLP expression plasmid, then treated with compounds at 4 h post-transfection. At 24 h post-transfection, culture supernatants and cell lysates were harvested to allow analysis of secreted VLPs and intracellular E. Representative results from n ≥ 2 independent experiments are shown. A Both the degraders and parental inhibitors cause depletion of intracellular E associated with reduced VLPs in the culture supernatant; however, the activity of GNF-2-deg and 2-12-2-deg is lost in the CRBN^−/− cells. B, C Structures of the βOG pocket on soluble, mature, prefusion dimeric E protein (PDB:1OKE) and prM-E protein on virions (PDB:3C6E), respectively, with domains I, II, III, and prM colored red, yellow, blue, and pink, respectively. Substitutions were introduced in the pocket at the hinge region at residues F193, M196, and F279 (green). D GNF-2-deg’s and 2-12-2-deg’s TPD-based inhibition of VLP formation is unaffected by single point mutations in the βOG pocket but is reduced when both F193L and M196V are introduced to the pocket. Cell lysates and purified VLPs were analyzed by Western blot to determine the effect of inhibitors and degraders on intracellular E abundance and VLP production. Representative results are shown from n = 2 independent experiments.

Fig. 5. E degraders exhibit greater antiviral potency and broader spectrum antiviral activity compared to parental E inhibitors.

Cells were infected with the tested virus at an MOI of 1, and infected cells were treated with the indicated compounds from 1 to 24 h post-infection, after which the yield of infectious virus in culture supernatants was quantified by plaque formation assay. Viral yields were normalized to those of the DMSO-treated controls. The antiviral EC₉₀, corresponding to the compound concentration resulting in a 90% reduction in viral titer, was determined by nonlinear regression. The data are presented as means of n = 2 independent experiments.