NRF2 activators inhibit influenza A virus replication by interfering with nucleo-cytoplasmic export of viral RNPs in an NRF2-independent manner

Abstract

In addition to antioxidative and anti-inflammatory properties, activators of the cytoprotective nuclear factor erythroid-2-like-2 (NRF2) signaling pathway have antiviral effects, but the underlying antiviral mechanisms are incompletely understood. We evaluated the ability of the NRF2 activators 4-octyl itaconate (4OI), bardoxolone methyl (BARD), sulforaphane (SFN), and the inhibitor of exportin-1 (XPO1)-mediated nuclear export selinexor (SEL) to interfere with influenza virus A/Puerto Rico/8/1934 (H1N1) infection of human cells. All compounds reduced viral titers in supernatants from A549 cells and vascular endothelial cells in the order of efficacy SEL>4OI>BARD = SFN, which correlated with their ability to prevent nucleo-cytoplasmic export of viral nucleoprotein and the host cell protein p53. In contrast, intracellular levels of viral HA mRNA and nucleocapsid protein (NP) were unaffected. Knocking down mRNA encoding KEAP1 (the main inhibitor of NRF2) or inactivating the NFE2L2 gene (which encodes NRF2) revealed that physiologic NRF2 signaling restricts IAV replication. However, the antiviral effect of all compounds was NRF2-independent. Instead, XPO1 knock-down greatly reduced viral titers, and incubation of Calu3 cells with an alkynated 4OI probe demonstrated formation of a covalent complex with XPO1. Ligand-target modelling predicted covalent binding of all three NRF2 activators and SEL to the active site of XPO1 involving the critical Cys528. SEL and 4OI manifested the highest binding energies, whereby the 4-octyl tail of 4OI interacted extensively with the hydrophobic groove of XPO1, which binds nuclear export sequences on cargo proteins. Conversely, SEL as well as the three NRF2 activators were predicted to covalently bind the functionally critical Cys151 in KEAP1. Blocking XPO1-mediated nuclear export may, thus, constitute a "noncanonical" mechanism of anti-influenza activity of electrophilic NRF2 activators that can interact with similar cysteine environments at the active sites of XPO1 and KEAP1. Considering the importance of XPO1 function to a variety of pathogenic viruses, compounds that are optimized to inhibit both targets may constitute an important class of broadly active host-directed treatments that embody anti-inflammatory, cytoprotective, and antiviral properties.

Fig 1. Chemical structures of the four compounds used.

The reactive electrophilic carbon atoms that can potentially undergo Michael addition from nucleophilic targets are highlighted in red or blue. A. Bardoxolone methyl (BARD) is unique in that it has two reactive carbon atoms at positions 1 (red) and 9 (blue). B. Sulforaphane (SFN). C. 4-Octyl itaconate (4OI). D. Selinexor (SEL). This bona fide XPO1 inhibitor is not known to be an NRF2 agonist, but also possesses one electrophilic double bond.

Fig 2. NRF2 activators reduce release of IAV virions and inhibit nuclear export of vRNP.

A. Schematic of experimental layout. A549 cells were pretreated with the compounds (SEL, 1 μM; 4OI, 100 μM; BARD, 0.1 μM; SFN, 10 μM) for 12 h, were then infected with IAV PR8M (MOI = 0.05 in B, MOI = 1 in C-F) for 1 h and subsequently incubated in fresh buffer containing the compounds. Measurements were performed at the indicated times post infection (p.i.). B. NRF2 activators reduce release of progeny virions. Viral titers (FFU/mL) in cell culture supernatants were determined 12 and 24 h p.i. n = 3. C-F. NRF2 activators interfere with nuclear export of vRNP. Subcellular localization of viral NP was determined by immunofluorescence 4, 6, and 8 h p.i. Viral NP was visualized by indirect immunofluorescence using Cy3-labeled secondary antibody (561 nm, red) and nuclei by staining DNA with DAPI (405 nm, blue). Cells with NP staining in nucleus, cytoplasm or both nucleus and cytoplasm were quantified by visual inspection. N = 2 replicates, n = 7 digital images per replicate. C. Representative microscopic images. D. Proportion of cells with nuclear NP staining only. E. Proportion of cells with cytoplasmic NP staining only. F. Proportion of cells with both nuclear and cytoplasmic NP staining. Data are shown as means ±SEM. One-way ANOVA with Tukey’s post-hoc test. p = * ≤0.05, ** ≤0.01, *** ≤0.001.

Fig 3. Effects of XPO1 knock-down on IAV infection, cellular responses, and antiviral activity of the compounds.

A549 cells were transfected for 24 h with specific siRNA targeting XPO1 mRNA or nonspecific siRNA. Cells were then pretreated with the compounds (SEL, 1 μM; 4OI, 100 μM; BARD, 0.1 μM; SFN, 10 μM) for 12 h, infected with IAV PR8M (MOI = 1) for 2 h, and then incubated in fresh buffer containing the compounds for 22 h. A-C. Efficiency of XPO1 knock-down. A. XPO1 mRNA (RT-qPCR). B. XPO1 protein (immunoblot). C. Densitometry of B. D. Viral HA mRNA expression with reference to HPRT1 mRNA as internal control (RT-qPCR). E. Viral NP (immunoblot). F. IAV titers in cell culture supernatants (foci-forming assay, foci-forming units [FFU]/ml). G, H. IFIT1 and CXCL10 mRNA (RT-qPCR). I. Mitochondrial ROS (flow cytometry). J. Expression of NFE2L2, HMOX1, SLC7A11, AKR1B10, GCLM, and KEAP1 mRNAs (RT-qPCR, internal control HPRT1 mRNA). The heat map is based on log₂ fold change (scale as indicated in the color legend) with respect to expression in wild-type uninfected cells. Bar graphs for each target gene are shown in S2 Fig for additional clarity. n = 3, means ±SEM. One-way ANOVA with Tukey’s post-hoc test, using infected untreated wild-type or knock-down cells as reference. * ≤0.05, ** ≤0.01, *** ≤0.001, **** ≤0.0001.

Fig 4. Antiviral effects of the four compounds are NRF2-independent.

hiPSC-derived wild-type or NRF2^-/- vascular ECs were pretreated with the compounds (SEL, 1 μM; 4OI, 100 μM; BARD, 0.1 μM; SFN, 10 μM) for 12 h, infected with IAV PR8M (MOI = 1) for 2 h, and then incubated in fresh buffer containing the compounds for 22 h. A. NFE2L2 mRNA (RT-qPCR). B. Viral titers in cell culture supernatants (foci-forming assay, FFU/mL). C, D. IFIT1 and CXCL10 mRNA (RT-qPCR). E. Expression of HMOX1, SLC7A11, AKR1B10, GCLM, and KEAP1 mRNAs (RT-qPCR, internal control HPRT1 mRNA), heat map based on log₂ fold change (as indicated in the color legend) with respect to expression in wild-type uninfected cells. Column graphs of these data are shown in S3 Fig for additional clarity. F-H, Knocking down KEAP1 expression reduces viral titers, but does not affect the antiviral effect of the compounds. F, Expression of KEAP1 mRNA (RT-qPCR). G, Expression of viral HA mRNA (RT-qPCR). H, Viral titers (foci-forming assay, FFU/ml). n = 3, means ±SEM. One-way ANOVA with Tukey’s post-hoc test. * ≤0.05, ** ≤0.01, *** ≤0.001, **** ≤0.0001.

Fig 5. Biochemical and predicted ligand-target interactions with XPO1.

A,B. “Click-chemistry” pull-down assay demonstrating covalent binding of an alkynated 4OI probe (4-OI-alk) to XPO1 (A) and KEAP1 (B) in Calu-3 cells. At the indicated time points after addition of the probe to the cells, proteins complexed with the probe were detected by immunoblot for XPO1 or KEAP1. C-J. Ligand-target modeling studies of the compounds with the active site of XPO1 containing the functionally critical Cys528 (marked with a white asterisk *). Predicted binding energies are shown in Table 1. 3D models and the corresponding 2D interaction diagrams are shown in A,B (SEL), C,D (4OI), E,F (SFN), and G,H (BARD C1). A more detailed binding pose of 4OI to this site, as well as superimposed binding poses of 4OI and leptomycin B, are shown in S9 Fig. * = Cys528.

Fig 6. Predicted ligand-target interactions of the four compounds with the BTB domain of KEAP1.

Ligand-target modeling studies of the compounds with the active site of the BTB domain of KEAP1 containing the functionally critical Cys151 (marked with a white asterisk *). Predicted binding energies are shown in Table 2. 3D models and the corresponding 2D interaction diagrams are shown in A,B (SEL); C,D (4OI); E,F (SFN); G,H (BARD C1); and I,J (BARD C9). * = Cys151.