Structure-based drug discovery for combating influenza virus by targeting the PA-PB1 interaction

Abstract

Influenza virus infections are serious public health concerns throughout the world. The development of compounds with novel mechanisms of action is urgently required due to the emergence of viruses with resistance to the currently-approved anti-influenza viral drugs. We performed in silico screening using a structure-based drug discovery algorithm called Nagasaki University Docking Engine (NUDE), which is optimised for a GPU-based supercomputer (DEstination for Gpu Intensive MAchine; DEGIMA), by targeting influenza viral PA protein. The compounds selected by NUDE were tested for anti-influenza virus activity using a cell-based assay. The most potent compound, designated as PA-49, is a medium-sized quinolinone derivative bearing a tetrazole moiety, and it inhibited the replication of influenza virus A/WSN/33 at a half maximal inhibitory concentration of 0.47 μM. PA-49 has the ability to bind PA and its anti-influenza activity was promising against various influenza strains, including a clinical isolate of A(H1N1)pdm09 and type B viruses. The docking simulation suggested that PA-49 interrupts the PA-PB1 interface where important amino acids are mostly conserved in the virus strains tested, suggesting the strain independent utility. Because our NUDE/DEGIMA system is rapid and efficient, it may help effective drug discovery against the influenza virus and other emerging viruses.

Figure 1

Hit compounds with anti-influenza virus activity. Hit compounds selected by in silico screening were tested to determine their anti-influenza virus activities via cell-based screening. Potent compounds with anti-influenza virus activities (MIC value < 20 μM) are shown. (a) Compounds bearing quinolinone and tetrazole moieties. (b) Compounds containing different scaffolds.

Figure 2

Direct binding between PA protein and hit compounds. Compounds selected by cell-based screening were subjected to surface plasmon resonance analysis. Anti-PA antibody was immobilised on the sensor chip and captured the recombinant PA protein before various concentrations of compounds were loaded as analytes. (a) Sensorgram and (b) fitting curves from the sensorgram were shown.

Figure 3

Effect of PA-49 on subcellular localisation of PA protein. In 24-well plates, HeLa cells were seeded on coverslips and grown to 70–80% confluence. Cells were then transfected with 0.5 μg of plasmid as indicated then 3.5 h after transfection, PA-49 was added then incubated further 25.5 h. Cells were fixed with 4% paraformaldehyde for 10 min, and were permeabilised by 0.1% NP40 in PBS for 20 min. The cells were incubated with 1% skim milk in PBS for 40 min. The cells were then reacted with anti-PA or -PB1 antibody. After washing, the cells were reacted with Alexa Flour 488-conjugated anti-rabbit IgG (Thermo Fisher Scientific K.K., Kanagawa Japan; A11008). (a) Subcellular localisation of PA and PB1 in the absence or the presence of 40 μM PA-49 were shown. Scale bar; 25 μm. (b) The percentage of cells showing greater nuclear than cytoplasmic localisation (N>C), nuclear equal to cytoplasmic localisation (N=C) or cytoplasmic localisation (N<C) was determined by direct counting. The mean and standard deviation obtained from three different areas were shown.

Figure 4

Anti-influenza virus activity of PA-49. The anti-influenza virus effects of PA-49 were evaluated as described in the Methods section. MDCK cells grown in 24-well plates were infected and treated with serial dilutions of PA-49. At 48 h after infection, the cell morphology (a) was observed (scale bar; 150 μm) and (b) the antiviral effects on cells were measured by the WST-1 assay (closed circles) or CV staining (open circles). Optical density (OD) values (%) are expressed relative to the percentage of cells without virus infection. Viral titres in the supernatant (open triangles) were measured by TCID₅₀ assays.

Figure 5

Determination of IC₅₀ values by plaque inhibition assay. (a) Inhibition of plaque formation in the presence of PA-49. Confluent MDCK cells seeded in six-well plates (2 × 10⁶ cells/well) were then infected with approximately 200 pfu of A/WSN/33 virus at 37 °C for 1 h. After removing the medium, the cells were overlaid with 4 mL of PA-49-containing agarose solution (0.8% agarose, 0.1% BSA and MEM vitamin) and incubated for 2 days. The cells were fixed and stained with 0.5% Amido black. After washing with water and air drying, the number of plaques was counted visually. Representative results from three independent experiments are shown. (b) Results of plaque formation experiments in (a) and two other experiments are shown as relative plaque numbers.

Figure 6

PA-49 suppresses the expression of viral proteins. MDCK cells (2 × 10⁵ cells/well) were seeded in 24-well plates and infected with A/WSN/33 at an MOI of 1 in the absence or presence of PA-49 (0.5–20 μM). (a) At 9 hpi, the cells were observed by phase contrast microscopy. Scale bar, 150 μm. (b) At 3 or 9 hpi, the cells were then lysed and subjected to 10% SDS-PAGE, and then transferred onto a polyvinylidene fluoride membrane. The membrane was incubated with a 1:30,000 dilution of anti-NP or anti-M1, 1:5,000 dilution of anti-HA or anti-PA, 1:3,000 dilution of anti-PB1, or 1:150 dilution of anti-actin for 4 h, followed by treatment with the biotinylated secondary antibody and streptavidin alkaline phosphatase, and visualisation using BCIP and NBP. (c) The band intensity was quantified by Image J software. The expression level of actin was used for normalisation. To calculate the relative intensity (%) for 3- and 9-h PA-49-treated samples, the band intensity of 9-h- infected cells without PA-49 treatment was shown as 100%. Data represent the means and standard deviation from two independent experiments.

Figure 7

Mode of action of PA-49. (a) Time-of-addition experiments. Approximately 200 pfu of virus in MEM vitamin was used for infection. The detailed procedures for each treatment are as follows. (i) Pre-treatment of cells: before the plaque inhibition assays, MDCK cells were pre-treated with the test samples at 37 °C for 3 h. After removing the medium, cells were washed with MEM and infected by adding the viral suspension in MEM vitamin. (ii) Pre-treatment of virus: approximately 10⁷ pfu/mL of virus stock was pre-incubated with the test samples on ice for 3 h. The mixture was subsequently diluted in MEM vitamin and 400-μL aliquots (200 pfu) of the diluted mixture were used for infection. (iii) Simultaneous: 200-μL aliquots of the test samples in MEM vitamin were added to MDCK cells, followed by 200 μL of the virus suspension. The cells were then incubated for 1 h. (iv) In agarose: after viral infection for 1 h, the cells were overlaid with 4 mL of agarose solution containing the samples and MEM supplemented with 0.8% agarose, 0.1% BSA and 1% 100 × vitamin solution. Two days later, the cells were fixed and stained. Relative plaque numbers based on three independent experiments are shown. Grey and white bars show the 2 and 1 μM PA-49 treatments, respectively. **Indicates p < 0.01; ***indicates p < 0.001. (b) PA-49 does not inhibit the hemagglutination activity of viral HA protein. Twofold serial dilutions of A/WSN/33 and A/Aichi/2/68 in PBS, as well as 4, 20 and 80 μM of PA-49 in PBS were prepared. In 96-well plates, 25 μL of each serially diluted virus and PA-49 solution were mixed and then incubated at 4 °C for 1 h before adding 50 μL of 5% (v/v) chicken RBCs in PBS. Hemagglutination was observed after incubating at room temperature for 1 h.

Figure 8

Binding mode of PA-49 compounds with influenza PA protein based on in silico calculations. (a,b) The crystal structure of the PA (grey)–PB1 (light blue) complex (PDB code: 2ZNL) and the binding structure of PA-49 from our docking simulations (pink) were overlaid: (a) side view and (b) top view. Side chains of PB1 protein are shown with stick representation. Hydrogen atoms are omitted for clarity. In order to obtain binding structure of PA–PA-49 complex, hydrogen atoms were added to the complex structure obtained from the docking simulation, followed by 1,000 steps of energy minimisation without any restraint using AMBER 10 software package. For minimisation, AMBER ff99SB and GAFF force fields were used for the protein and ligand, respectively. Colour code for atom: N (blue), O (red), S (yellow) and F (light green). (c) Amino acid sequence alignment of influenza virus PA protein (residues 601–716) including H1N1, H3N2 and H5N1 subtypes. Residues that are required for PA–PB1 interface are labelled with yellow. Molecular structures in (a) and (b) were drawn using UCSF Chimera package.