On the other hand, RIG-I interacted with V and Vcys but not with P/V, Vu, and Vu cys (Fig. 2B), suggesting that the interaction requires the entire V protein and that cysteine mutations did not affect

the interaction. Similar results were obtained in binding of the V protein with IKKɛ and IRF3 (data not shown). These results show that only MDA5 interacts with the V unique region and that RIG-I, IKKɛ and IRF3 interact with the V protein in a mode different from MDA5. We next investigated whether the V and MDA5 interaction was related to inhibition of IRF3 activation. 293T cells were transfected with an IRF3-dependent reporter plasmid, p-55C1B-EGFP, together with FL-MDA5 and one of the viral proteins. Cells were further GSK1120212 cell line transfected with poly(I:C), and IRF3 transcription activation was investigated. EGFP expression showed that the V protein significantly suppressed IRF3 activation induced by overexpression of FL-MDA5 (Fig. 3A). Expression of EGFP as well as FL-MDA5 JAK inhibitor and a viral protein was confirmed by western blotting (Fig. 3B), and light intensity of EGFP protein bands was quantitated and plotted in a graph (Fig. 3C).

The N-terminal part of the V protein lacking the V unique region (P/V) and the C protein (C) did not suppress IRF3 activation. The V protein with a single point mutation of a cysteine to alanine at the V unique region (Vcys2A: C341A and Vcys7A: C365A) also did not suppress IRF3 activation. Amino acid substitutions of the SeV V protein at those positions ameliorated viral load and pathogenicity in a mouse model (11). Influenza virus NS1 protein (NS1) did not suppress IRF3

activation in this condition. SeV C protein and influenza virus NS1 protein are known to inhibit IRF3 activation and IFN-β production (27, NADPH-cytochrome-c2 reductase 28, 29). The inhibition is thought to be due to reduction of RNA species that belongs to pathogen-associated molecular patterns. These two proteins did not inhibit IRF3 activation induced by overexpression of MDA5 and poly(I:C) treatment. A similar experiment using an IRF3-dependent GFP and luciferase reporter plasmids showed that V and Vu suppressed IRF3 activation and that the Vcys and Vu cys, which have two point mutations at the cysteine residues in the V unique region, and P/V did not suppress the IRF3 activation (Fig. 4). These results indicate that the V protein suppressed MDA5-induced IRF3 activation in a Vu-dependent and cysteine-dependent manner, corresponding to the mode of interaction of V proteins with MDA5. We previously reported SeV V mutants with attenuated pathogenicity (11, 12). SeV V-H318N, R319W, R320G, and W336G were highly attenuated in virulence by more than 25-fold in 50% mouse lethal dose, and SeV V-E321K and P339T were mildly attenuated (12). We infected FL-MDA5-transfected cells with V mutant viruses, and interaction of the V mutant protein with FL-MDA5 was then investigated by immunoprecipitation and western blotting.