In the present study we have carried out a detailed functional analysis of influenza A virus NS1 protein NES element, which is extremely well-conserved among all presently known influenza A virus strains and represents a typical consensus NES-type sequence. We show that mutations in the NES element, with the exception of mutation L146A, lead to nuclear retention of NS1 at late stages of infection, indicating an inactivation of the NES. NES-inactivating NS1 mutations (L144A, L146A) and (A149V) were associated with severe inhibition of viral growth properties and inability of the mutant virus to inhibit host cell interferon production.
Regulation of the mechanisms of NS1 NES function during infection is unknown, although the amino acid region immediately after the consensus-like NES element has been suggested to play a regulatory role in the nuclear export of NS1 . This may take place by masking of the export signal at early times of infection and by allowing the exposure of the NES element at later time points by some activating event. The fact that some activation event is required for triggering the nuclear export of NS1 protein is well supported by the structure of NS1 (Figure 1C). Leucine residues in the NES element are of hydrophobic nature and for them to become in contact with export molecules relatively drastic conformation changes are required. Inactivation of the NES by the A149V mutation, which renders the protein strongly nuclear at late time points (Figure 2), also supports the view that some masking sequence is regulating the accessibility of the consensus leucine residues to the export machinery. Interestingly, the A149V NS1 mutation has been linked to pathogenesis in chickens when it was discovered in the context of a natural H5N1 virus strain that was circulating in poultry .
The observed reduction in viral replication of the NS1 mutant viruses (A149V) and (L144A, L146A) (Figure 4) was likely due to impaired IFN antagonist functions of mutant NS1 proteins. These viruses readily induced IFN production into cell culture supernatant, as evidenced by the ELISA assays (Figure 6A and B), and also showed induction of IFN induced MxA protein (Figure 6C). Upon further investigation the elevated IFN levels were found to correlate with very high virus activated p-IRF3 levels as compared to wild-type virus, suggesting the inability of these mutant NS1 proteins to inhibit the RIG-I pathway. Strong expression of p-IRF3 was also observed with the other nuclearly localized NS1 mutant viruses (L141A) and (F138A, L141A), which might be explained by the nuclear NS1 being unable to inhibit cytoplasmic RIG-I. However, this correlation remains unclear since strong phosphorylation of IRF-3 was also associated with the L146A NS1 mutant virus, which shows a similar NS1 localization pattern as the wild type virus. Lack of potent IFN production despite strong p-IRF3 activation by the (L141A) and (F138A, L141A) NS1 mutant viruses indicates, that in the case of at least some nuclearly retented mutant NS1 proteins, the nuclear functions of NS1 in the absence of cytoplasmic NS1 are sufficient to inhibit IFN production. Detectable levels of MxA were observed with all the mutant viruses despite low IFN production by some of them, which is probably due to even low IFN production being enough to cause some MxA production. It was also of interest that Western blot analysis revealed diminished p-Akt levels for the (L144A, L146A) and (A149V) mutants adding further evidence that multiple NS1 functions may be affected by the mutations. This suggests a more extensive structural or functional importance of the NES region. In a study investigating the structural conservation of the whole NS1 protein, Leu144 and Ala149 were found to be very well-conserved and were suggested to play a role in stabilizing the protein structure . The observed reduction of almost 50% in the levels of NS1 (L144A, L146A) compared to wild type NS1 (Additional file 3: Figure S3) supports the idea of a stabilizing role for the NES region, but this is unlikely to have an impact on viral growth properties or IFN production since very low levels of NS1 have been shown to be sufficient for normal replication and IFN antagonism . The NES region has also been suggested to be involved in CPSF binding , but an X-ray crystal structure published on the NS1-CPSF complex somewhat later does not support this view . Nevertheless, we analyzed whether any of our inserted mutations affected CPSF-binding of NS1 on GST pull-down assay and found that binding to CPSF30 was similar to that of wild-type NS1 (Figure 7).
Besides being involved in regulating splicing and nuclear export of mRNA , NS1 has also been shown to influence temporal regulation of viral RNA and protein synthesis . Thus we explored whether the introduced NES mutations might have an effect on viral protein expression kinetics (Figure 5). In addition to NS1, we included an abundant early gene product (NP), a late gene product (M1) and a spliced gene product (NEP). At a glance, expression kinetics of all the analyzed proteins appear similar between the wild type and mutant viruses (Figure 5), but there are some differences in expression levels at certain time points. The growth deficient viruses (A149V) and especially (L144A, L146A) produce somewhat lower levels of NS1 and NEP, whereas the (L141A) and (L146A) viruses are even stronger inducers of NEP than the wild type virus (Figure 5). NP and M1 expression levels are mostly equal between the different viruses with the exception of (A149V), (L146A) and (F138A, L141A) viruses showing slightly reduced M1 levels. All the observed differences are modest however, and despite critical roles for these proteins during infection, unlikely to cause any defects in replication. To confirm this conclusion we used immunofluorescence analysis to visualize NP localization during infection with all our mutant viruses (Additional file 1: Figure S1 and Additional file 2: Figure S2). Since NEP and M1 are involved in the nuclear export of viral ribonucleoproteins [24, 32], any major flaw in their function would show as nuclear retention of NP. All the viruses display equal cytoplasmic localization of NP (Additional file 1: Figure S1 and Additional file 2: Figure S2), so we can conclude that the observed growth deficiencies are not due to any defects in viral mRNA splicing or protein expression.
The PI3K pathway has a dual role in influenza A virus infection by mediating processes that are both beneficial and detrimental for virus replication. Several studies have shown that the PI3K pathway supports viral replication by promoting viral entry , inhibiting apoptosis [34–36] and enhancing nuclear export of vRNP complexes . In addition, PI3K signalling is also involved in RIG-I mediated activation of IRF3 thus contributing to the induction of type I interferon production . PI3K is activated at two points during influenza A infection, first at a very early stage stimulated by viral attachment and then at a later stage of infection by NS1 binding to the p85β subunit of PI3K . NS1 has also been suggested to directly bind to Akt, a downstream effector of PI3K involved in diverse cellular processes including cell differentiation, proliferation and apoptosis . Diminished levels of cellular p-Akt observed in (L144A, L146A) and (A149V) mutant virus-infected cells point towards defects in the NS1 PI3K activating functions (Figure 6C), which correlates with a previous observation of the region around NES being involved in PI3K activation . Li and coworkers found the amino acid region 137–142 to be involved in NS1-p85β interaction and showed interference of the interaction by mutation of the amino acids 141 and 142. In a structural study of the p85β iSH2-NS1 -complex the Glu 142 of NS1 was speculated to be involved in salt bridge formation at the complex interface . The inability of NS1 (L144A, L146A) and (A149V) mutant viruses to activate PI3K signalling could result from NS1 nuclear retention, but since this phenotype is not shared by other viruses with inactivated NS1 NES signals, it must be due to an effect independent of NS1 localization. It is plausible that mutations close to the NS1-p85β binding site could cause structural changes big enough to disrupt the salt bridge formation between NS1 Glu 142 and p85β or just influence the interaction enough to disrupt PI3K/Akt activation. Such disruption by itself could cause the observed reductions in growth and IFN antagonistic properties as evidenced by other studies where NS1-p85β interaction has been investigated [12, 41, 42]. Also certain mutations in the amino acid position 138 have been shown to influence the NS1-p85β interaction [43, 44], so it is feasible that the mutations L144A, L146A and A149V might affect it as well.
Nuclear export of proteins containing the classical leucine-rich export signal is usually mediated by CRM1, but with influenza A virus NS1 this is likely not the case. It was shown previously by transfection experiments that NS1 does not interact with CRM1  and here we show that the CRM1 inhibitor LMB fails to prevent or at least partially impair NS1 export during the infection (Figure 8). Examples of proteins showing nuclear export independent of CRM1, despite harboring leucine-rich export signals, include protein kinase inhibitor (PKI) and glucocorticoid receptor, which are exported by the Ca2+-binding protein calreticulin through a mechanism similar to CRM1-mediated export [18, 19]. It is likely that a similar alternative export pathway exists for NS1 as well, but this remains to be verified by further experiments.