Structural insights into phosphoinositide 3-kinase activation by the influenza A virus NS1 protein

Abstract

Seasonal epidemics and periodic worldwide pandemics caused by influenza A viruses are of continuous concern. The viral nonstructural (NS1) protein is a multifunctional virulence factor that antagonizes several host innate immune defenses during infection. NS1 also directly stimulates class IA phosphoinositide 3-kinase (PI3K) signaling, an essential cell survival pathway commonly mutated in human cancers. Here, we present a 2.3-A resolution crystal structure of the NS1 effector domain in complex with the inter-SH2 (coiled-coil) domain of p85beta, a regulatory subunit of PI3K. Our data emphasize the remarkable isoform specificity of this interaction, and provide insights into the mechanism by which NS1 activates the PI3K (p85beta:p110) holoenzyme. A model of the NS1:PI3K heterotrimeric complex reveals that NS1 uses the coiled-coil as a structural tether to sterically prevent normal inhibitory contacts between the N-terminal SH2 domain of p85beta and the p110 catalytic subunit. Furthermore, in this model, NS1 makes extensive contacts with the C2/kinase domains of p110, and a small acidic alpha-helix of NS1 sits adjacent to the highly basic activation loop of the enzyme. During infection, a recombinant influenza A virus expressing NS1 with charge-disruption mutations in this acidic alpha-helix is unable to stimulate the production of phosphatidylinositol 3,4,5-trisphosphate or the phosphorylation of Akt. Despite this, the charge-disruption mutations in NS1 do not affect its ability to interact with the p85beta inter-SH2 domain in vitro. Overall, these data suggest that both direct binding of NS1 to p85beta (resulting in repositioning of the N-terminal SH2 domain) and possible NS1:p110 contacts contribute to PI3K activation.

Fig. 1.

Physiological constraints suggest that the NS1 ED binds β-iSH2 as a monomer. (A) Tetrameric arrangement of the ED:β-iSH2 complex that is observed in the asymmetric unit. Central to the arrangement is an NS1 ED helix–helix dimer that is the predominant multimeric form in solution (20) and is observed in previous ED crystal structures (21, 36). One ED:β-iSH2 complex is colored as in Fig. 2 and the other dark gray. (B) Model of how the tetrameric complex would orient with respect to the lipid membrane. One heterodimeric ED:β-iSH2 complex has been orientated in a manner analogous to that proposed for the p85:p110 holocomplex (see Fig. 3B in ref. 5) and colored as in Fig. 3. The other heterodimeric complex is colored gray for p110 and dark gray for the ED:β-iSH2 complex. It is apparent from this orientation that a tetrameric ED:β-iSH2 complex is incompatible which the proposed mode of membrane association of PI3K.

Fig. 3.

Model of the NS1:PI3K heterotrimer. (A) Left. Model of the interaction between NS1 ED and the β-iSH2:p110α complex. Right. Structure of a p85α nSH2-iSH2 fragment in complex with p110α as determined by Mandelker et al. (7). Individual domains are colored separately and labeled. Regions linking domains are colored white. (B) Close-up view of the NS1 ED location in the modeled heterotrimeric complex. The location of cancer-associated mutations in p85α/p110α that cause constitutive kinase activation are highlighted. (C) Close-up view of the nSH2 domain location in the structure by Mandelker et al. (7).

Fig. 2.

Structure of the ED:β-iSH2 complex. (A) Cartoon representation of the ED:β-iSH2 complex. (B) Cartoon representation of the four-helix bundle formed at the complex interface. (C) Residues at the ED:β-iSH2 interface. Potential hydrogen bonds are highlighted. Individual residues are shown in stick representation. NS1 ED is colored gold, and β-iSH2 is colored red.

Fig. 4.

Roles of the NS1 acidic α-helix in both PI3K activation and IFN-antagonism. (A) Potential interaction between the NS1 ED and the p110α kinase activation loop. Domains of p85β, p110α, and NS1 ED are shown in cartoon representation. The edges of the disordered p110α activation loop are highlighted by spheres colored magenta. Specific residues are shown in stick representation. (B) The rUd-E96/97A virus does not activate PI3K. 1321N1 cells (which lack the PIP₃ phosphatase, PTEN) (37) were infected with rUd WT, rUd-Y89F, or rUd-E96/97A viruses at an MOI of 5 PFU/cell for 10 h prior to analysis of total PIP₃ levels. Bars show mean values obtained from triplicate samples assayed in duplicate. Error bars represent standard deviation. (C) A549 cells were infected as for (B). Cells were lysed and proteins analyzed by immunoblotting. The level of PI3K activation was determined by detection of phospho-Akt (Ser473). Viral NS1 and M1 proteins were detected as markers of infection. β-actin was used as a loading control. (D) GST-pulldown experiment showing the interaction of NS1 with β-iSH2. GST-NS1 proteins were immobilized onto agarose beads and incubated with 293T cell lysates expressing myc-tagged β-iSH2. NS1:β-iSH2 complexes were separated by SDS-PAGE and detected by immunoblotting using anti-NS1 and anti-myc antibodies. (E) Mean plaque size of recombinant rUd WT, rUd-Y89F, and rUd-E96/97A viruses in MDCK cells. Plaques were immunostained 3 days postinfection, and mean plaque size was determined (± SD). (F) Mean plaque size of recombinant rUd WT, rUd-Y89F, and rUd-E96/97A viruses in MDCK-V cells (targeted degradation of STAT1) (24). Plaques were immunostained 3 days postinfection, and mean plaque size was determined (± SD).