Synergistic Effect of the PDZ and p85β-Binding Domains of the NS1 Protein on Virulence of an Avian H5N1 Influenza A Virus

Abstract

The influenza A virus NS1 protein affects virulence through several mechanisms, including the host's innate immune response and various signaling pathways. Highly pathogenic avian influenza (HPAI) viruses of the H5N1 subtype continue to evolve through reassortment and mutations. Our recent phylogenetic analysis identified a group of HPAI H5N1 viruses with two characteristic mutations in NS1: the avian virus-type PDZ domain-binding motif ESEV (which affects virulence) was replaced with ESKV, and NS1-138F (which is highly conserved among all influenza A viruses and may affect the activation of the phosphatidylinositol 3-kinase [PI3K]/Akt signaling pathway) was replaced with NS1-138Y. Here, we show that an HPAI H5N1 virus (A/duck/Hunan/69/2004) encoding NS1-ESKV and NS1-138Y was confined to the respiratory tract of infected mice, whereas a mutant encoding NS1-ESEV and NS1-138F caused systemic infection and killed mice more efficiently. Mutation of either one of these sites had small effects on virulence. In addition, we found that the amino acid at NS1-138 affected not only the induction of the PI3K/Akt pathway but also the interaction of NS1 with cellular PDZ domain proteins. Similarly, the mutation in the PDZ domain-binding motif of NS1 altered its binding to cellular PDZ domain proteins and affected Akt phosphorylation. These findings suggest a functional interplay between the mutations at NS1-138 and NS1-229 that results in a synergistic effect on influenza virulence.

INTRODUCTION

Highly pathogenic avian influenza (HPAI) viruses of the H5N1 subtype first emerged in 1997 in Hong Kong. Since 2003, these viruses have become enzootic in poultry populations in several parts of the world and have evolved rapidly through reassortment and point mutations (1–9). Recently, we carried out a comprehensive phylogenetic analysis of the genes that encode the internal proteins (i.e., the PB2, PB1, PA, NP, M, and NS genes) of HPAI H5N1 viruses isolated from 2000 to 2008 and found that viruses of the so-called genotype Z (6, 10), which was first reported in 2002, have since diverged into three subgenotypes, Z.1 to Z.3 (8). Subgenotype Z.1 includes the Qinghai Lake viruses (which caused an outbreak among wild aquatic birds at Qinghai Lake, China, in 2005 [11–13]) and their descendants, which have spread into Europe, Africa, and the Middle East. Subgenotype Z.2 comprises viruses isolated in Thailand and Vietnam, whereas most subgenotype Z.3 viruses have been isolated in Indonesia. The three subgenotypes are characterized by phylogenetically distinguishable polymerase (PB2, PB1, and PA) genes (8). In addition, the NS1 protein of subgenotype Z.1 viruses differs by several amino acids from that of subgenotype Z.2 and Z.3 viruses (8). Two of these amino acid changes (an F-to-Y change at position 138 and an E-to-K change at position 229) are highly characteristic of subgenotype Z.1 viruses (i.e., they are not commonly found among other H5N1 genotypes or among other avian influenza virus subtypes) and emerged at around the same time.

The NS gene encodes two proteins, the NS1 interferon antagonist protein (14, 15) and NS2/NEP, which functions in the nuclear export of viral ribonucleoprotein complexes (16, 17) and in virus budding (18). NS1 is a multifunctional protein that interferes with the activation of cellular innate immune responses upon virus infection and has been recognized as a virulence factor of HPAI H5N1 viruses in chickens and mice (19–21). In particular, NS1 represses RIG-I (22–24) and IRF-3 (25) activation and blocks IPS-1 (26). It also affects viral replication through its PDZ domain-binding motif (27, 28) and through activation of the phosphatidylinositol 3-kinase (PI3K)/Akt pathway (29–34).

Large-scale sequencing of influenza A viral genomes identified a PDZ domain-binding motif (PDM) at the C terminus (amino acids 227 to 230) of most NS1 proteins (35). This motif is recognized by PDZ domains, which are found in many cellular proteins that have roles in trafficking, signaling, apoptosis, and the establishment and maintenance of tight junctions and cell polarity (36; reviewed in references 37 and 38). Most influenza virus NS1 proteins possess a PDM with the sequence ESEV (NS1-ESEV, typically found in avian virus NS1 proteins) or RSKV (NS1-RSKV, typically found in human virus NS1 proteins). NS1-ESEV has high affinity for the cellular scaffolding proteins Dlg-1 and Scribble (39–41), whereas the affinity of human virus-type NS1-RSKV for these cellular proteins is low (39, 41, 42). Our phylogenetic analysis revealed a characteristic amino acid change at NS1-229 of subgenotype Z.1 viruses, which creates a PDM (NS1-ESKV) that is not commonly found among other influenza A virus NS1 proteins (8).

The PI3K/Akt pathway regulates many different cellular events, including apoptosis, cell metabolism, and proliferation (reviewed in reference 43). PI3K has been described as an antiviral factor (29, 44, 45), but it is also critical for efficient influenza virus replication (29–34). The PI3K/Akt pathway is activated when NS1 binds to PI3K (30–32, 34, 46–48), and a recent study showed that vRNA-mediated upregulation of RIG-I also leads to PI3K activation (44). Several regions in NS1 (30–32, 34, 46, 47, 49), including a prominent loop encompassing amino acids 137 to 142 (48), may interact with the p85β regulatory subunit of PI3K. The amino acid at position 138 of NS1, a phenylalanine (NS1-138F), is highly conserved among all influenza A viruses; however, in subgenotype Z.1 NS1, the amino acid at this position is tyrosine (NS1-138Y). Here, we tested whether the subgenotype Z.1-specific mutations NS1-138Y and NS1-229K affect virus virulence.

MATERIALS AND METHODS

Cells and viruses.

Human embryonic kidney (293T), chicken fibroblast (DF-1), and human brain astrocytoma (1321N1) cells were maintained in Dulbecco's modified essential medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Madin-Darby canine kidney (MDCK) cells were grown in MEM containing 5% newborn calf serum (NCS). Human airway epithelial (Calu-3) cells were supplied with MEM/F12 containing 10% FBS. All cells were incubated at 37°C with 5% CO₂. A/chicken/South Kalimantan/UT8210/2006 (H5N1, SK/06), A/duck/Hunan/69/2004 (H5N1, HN/04), and A/Bar-headed goose/Qinghai/3/2005 (H5N1, BHG/05) viruses were described previously (11, 50, 51).

Antibodies and human IFN-β.

Rabbit polyclonal phospho-Akt (Ser-473) and Akt antibodies, as well as horseradish peroxidase (HRP)-conjugated anti-rabbit immunoglobulin G (IgG), were purchased from Cell Signaling Technology (Danvers, MA). An influenza A virus NS1 mouse monoclonal antibody, rabbit polyclonal antibodies to human Dlg-1 and Scribble, a rabbit polyclonal antibody to calnexin, and HRP-conjugated anti-mouse IgG were all purchased from Santa Cruz Biotechnology (Santa Cruz, CA). An antibody to mouse p85β was obtained from Serotec (London, United Kingdom). Human beta interferon (IFN-β) was purchased from PBL Interferon Source (Piscataway, NJ). A mouse anti-FLAG M2 antibody was purchased from Sigma (St. Louis, MO).

Construction of plasmids and reverse genetics.

All viruses used in this study were generated by reverse genetics, using eight vRNA transcription constructs and four protein-expressing plasmids as described by Neumann et al. (52). At 48 h posttransfection, supernatants derived from plasmid-transfected 293T cells were used to generate virus stocks by infecting MDCK cells. The titers of all virus stocks were assessed by means of plaque assays in MDCK cells. Viruses possessing mutations at NS1-138 and/or NS1-229 were generated by using site-directed mutagenesis and reverse genetics. In particular, NS1-138Y (encoded by the codon TAT) was converted to NS1-138F (encoded by the codon TTT) or vice versa; likewise, NS1-229K (encoded by the codon AAA) was converted to NS1-229E (encoded by the codon GAA) or vice versa.

Wild-type and mutant NS1 protein expression plasmids were generated by amplifying the NS1 coding region and inserting it into the protein expression vector pCAGGS/MCS (53). To generate FLAG-tagged NS1 protein expression constructs, the NS1 coding region was amplified and cloned into the protein expression vector N-Flag-pCAGGS/MCS (which encodes a FLAG epitope and a downstream cloning site for the generation of FLAG-tagged proteins). All plasmids and viruses were sequenced to ensure that there were no unwanted mutations.

Replication of wild-type and mutant viruses in cell culture.

For growth curves in chicken DF-1 cells, cells were infected with viruses at a multiplicity of infection (MOI) of 0.00001 and incubated at 41°C. Supernatants were collected at 12, 24, 36, 48, and 60 h postinfection and titrated in triplicate in MDCK cells. Human Calu-3 cells were inoculated with viruses at an MOI of 0.0001 and incubated at 37°C. Supernatants were collected at 12, 24, 36, 48, 60, and 72 h postinfection and titrated in triplicate in MDCK cells.

Mouse studies.

Six-week-old female BALB/c mice (Jackson Laboratory, Bar Harbor, ME) were used for these experiments. To determine the dose lethal to 50% of infected mice (MLD₅₀), we anesthetized mice (3 per group) with isoflurane and inoculated them intranasally with 10-fold dilutions of virus (from 10¹ to 10⁶ PFU) in a volume of 50 μl. Mice were monitored for 14 days for mortality and changes in body weight. Any mouse that lost more than 25% of its starting body weight was euthanized. For virus titration in organs, groups of mice (6 per group) were infected intranasally with 10⁵ PFU of wild-type or mutant viruses. Three mice per group were euthanized on days 3 and 6 postinfection, respectively, and organs (lung, brain, nasal turbinate, kidney, and spleen) were collected for virus titration by means of plaque assays in MDCK cells. Data shown are the mean virus titers ± standard deviations.

Reporter gene assays.

To assess the interferon-antagonist activity of wild-type and mutant NS1 proteins, 293T cells were transfected with the respective pCAAGS-NS1 protein expression plasmid and the reporter plasmid pGL-IFN-β, which encodes the firefly luciferase protein under the control of the IFN-β promoter (54). Twenty-four hours later, cells were infected with Sendai virus at an MOI of 5 for 1 h. Cells were incubated for 24 h and then lysed with Glo lysis buffer (Promega, Madison, WI). Firefly luciferase expression was determined using Steady-Glo assay buffer (Promega). Data were obtained from three independent experiments carried out in triplicate. The relative fold change in firefly luciferase expression levels was normalized to cells expressing wild-type NS1. In another set of experiments, 293T cells were transfected with the respective pCAGGS-NS1 protein expression plasmid and the reporter plasmid pISRE-Luc (Promega, Madison, WI), which encodes the firefly luciferase protein under the control of an interferon-regulated promoter. Twenty-four hours later, cells were treated with 1 × 10⁴ U/ml of human IFN-β for another 24 h. At 48 h posttransfection, cells were harvested and tested for luciferase protein expression as described above.

Phospho-Akt activation and virus replication in 1321N1 cells.

Human brain astrocytoma 1321N1 cells were grown in serum-free medium overnight. Confluent serum-starved monolayers of 1321N1 cells were then infected at an MOI of 1 (calculated based on virus titration in MDCK cells) with wild-type or mutant viruses. Supernatants were harvested and total cell lysates prepared at 3, 7, 12, and 24 h postinfection. Virus titers were assessed in triplicate by means of plaque assays in MDCK cells. Cell lysates were used in duplicate to assess the levels of phosphorylated Akt, total Akt, calnexin (as an internal control), and viral NS1 protein by Western blotting.

Western blotting.

For Western blotting, infected cells were washed once with cold phosphate-buffered saline (PBS). The cells were incubated with lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.5% NP-40, 1 mM phenylmethylsulfonyl fluoride, protease and phosphatase inhibitor cocktail tablets [Roche, Indianapolis, IN]) on ice for 20 min. Cell lysates were cleared by centrifugation, and the protein concentration was determined by using the bicinchoninic acid (BCA) kit as directed by the manufacturer. Equal amounts of total protein (20 μg/each) for each sample were directly subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and subsequent blotting. Cellular and viral proteins were detected with the corresponding antibodies as described above.

Coimmunoprecipitation analysis.

293T cells were transfected with N-terminally Flag-tagged NS1 protein expression constructs. Cell extracts were prepared 36 h later as described above. Cell lysates were incubated with mouse anti-FLAG M2 (4 μg; Sigma, St. Louis, MO), antibody-conjugated Dynabeads (Invitrogen, Carlsbad, CA), and an equivalent amount of normal rabbit IgG (Santa Cruz Biotechnology, Santa Cruz, CA) for 30 min at room temperature. The Dynabeads-antibody-antigen complex was washed three times in PBS and released from the Dynabeads by incubation in SDS-PAGE sample buffer (160 mM Tris-HCl [pH 6.8], 4% SDS, 20% glycerol, 10% 2-mercaptoethanol, and 0.04% bromophenol blue) for 5 min at 95°C. The immunoprecipitated complex was analyzed by means of Western blotting with antibodies against p85β, Dlg-1, and Scribble.

RESULTS

Mutations at NS1-138 and in the NS1 PDM synergistically affect the virulence of avian H5N1 influenza viruses.

HPAI H5N1 viruses of the Qinghai Lake lineage (which form subgenotype Z.1) have killed wild waterfowl (11–13), sporadically infect humans, and have become enzootic in poultry populations in the Middle East; hence, viruses of this lineage present a potential threat to humans and wild life. Our phylogenetic analysis of HPAI H5N1 viruses identified two amino acid changes in NS1 that are highly characteristic for these viruses: an F-to-Y mutation at position 138 and an E-to-K mutation at position 229 (Fig. 1); the respective nucleotide changes in the NS RNA segment do not cause amino acid changes in the NS2 protein. To assess the biological significance of these subgenotype Z.1-characteristic amino acid changes in NS1, we selected the following three viruses: (i) A/duck/Hunan/69/2004 (H5N1; HN/04), which is an ancestor of subgenotype Z.1 viruses that possesses some, but not all, of the subgenotype Z.1-characteristic mutations; in particular, HN/04 encodes NS1-138Y and NS1-ESKV but lacks the subgenotype Z.1-characteristic PB2-627K mutation that facilitates efficient replication of HPAI H5N1 viruses in mammals (55) (Fig. 1); (ii) A/bar-headed goose/Qinghai/3/2005 (H5N1; BHG/05), a subgenotype Z.1 virus that possesses the NS1-138Y and NS1-ESKV sequences that are characteristic of this virus group; and (iii) A/chicken/South Kalimantan/UT8210/2006 (H5N1; SK/06), a subgenotype Z.3 virus that encodes NS1-138F (highly conserved among all influenza A viruses) and NS1-ESEV (highly conserved among avian influenza viruses) (Fig. 1).

Fig 1.

Overview of the characteristic amino acid changes in NS1 and the viruses used in this study. Genotype Z viruses (possessing NS1-138F and NS1-ESEV) have diverged into three subgenotypes, Z.1 to Z.3. The NS1 proteins of subgenotype Z.1 viruses encode two characteristic amino acid changes that result in NS1-138Y and NS1-ESKV. Here, we analyzed the following three viruses: A/bar-headed goose/Qinghai/3/2005 (H5N1; BHG/05), a subgenotype Z.1 virus encoding NS1-138Y and NS1-ESKV; A/chicken/South Kalimantan/UT8210/2006 (H5N1; SK/06), a subgenotype Z.3 virus that encodes the conserved NS1-138F and NS1-ESEV sequences; and A/duck/Hunan/69/2004 (H5N1; HN/04), an ancestor of subgenotype Z.1 viruses, which bears NS1-138Y and NS1-ESKV but lacks the subgenotype Z.1-characteristic PB2-627K mutation that facilitates efficient replication of HPAI H5N1 viruses in mammals (55).

For these three viruses, we generated variants possessing amino acid changes at position NS1-138, in the NS1 PDM, or at both positions (Table 1). First, we tested the virulence of wild-type and mutant viruses in mice inoculated with different doses of viruses. For wild-type HN/04 virus, only the highest dose tested (10⁶ PFU) killed a subset of infected animals (Fig. 2A), resulting in an MLD₅₀ of 10^6.3 PFU (Table 1). HN/04 viruses encoding NS1-138F or NS1-ESEV were slightly more pathogenic (MLD₅₀, 10^5.0 and 10^5.3 PFU, respectively) (Table 1 and Fig. 2A). In contrast, an HN/04 virus encoding both NS1-138F and NS1-ESEV had an MLD₅₀ of only 10^2.5 PFU (Table 1 and Fig. 2A).

Table 1.

Overview of viruses tested and their lethality in mice

Virus	NS1-138a	PDZ domain-binding motifb	MLD50c [log10 PFU]
HN/04d	Y	ESKV	6.25
HN/04-NS1-138F	Fg	ESKV	5.0
HN/04-NS1-ESEV	Y	ESEV	5.3
HN/04-NS1-138F-ESEV	F	ESEV	2.5
BHG/05e	Y	ESKV	0.9
BHG/05-NS1-138F	F	ESKV	0.5
BHG/05-NS1-ESEV	Y	ESEV	0.5
BHG/05-NS1-138F-ESEV	F	ESEV	0.5
SK/06f	F	ESEV	3.3
SK/06-NS1-138Y	Y	ESEV	3.5
SK/06-NS1-ESKV	F	ESKV	3.5
SK/06-NS1-138Y-ESKV	Y	ESKV	5.5

^aShown is the amino acid at position 138 of NS1; F is highly conserved among all influenza A viruses.

^bShown is the sequence of the PDZ domain-binding motif; most avian virus NS1 proteins possess the ESEV motif.

^cMLD₅₀ is the amount of virus required to kill 50% of infected mice.

^dHN/04, A/duck/Hunan/69/2004 (H5N1).

^eBHG/05, A/bar-headed goose/Qinghai/3/2005 (H5N1).

^fSK/06, A/chicken/South Kalimantan/UT8210/2006 (H5N1).

^gMutated amino acids are shown in boldface.

Fig 2.

Survival rates of mice infected with wild-type or mutant viruses. BALB/c mice (three/group) were inoculated intranasally with the indicated doses of HN/04 (A), BHG/05 (B), or SK/06 (C) virus. Infected animals were monitored daily for body weight changes and survival.

BHG/05, a prototype subgenotype Z.1 virus, had an MLD₅₀ of 10^0.9 PFU (Table 1 and Fig. 2B). Introduction of NS1-138F and/or NS1-ESEV (which increased virulence in the background of HN/04 virus) had no significant effect, likely because the parental BHG/05 virus was already highly virulent. In contrast, the MLD₅₀ of SK/06 (a subgenotype Z.3 virus) was 10^3.3 PFU (Table 1 and Fig. 2C), and the introduction of the subgenotype Z.1-characteristic sequences of NS1-138Y or NS1-ESKV had no detectable effect on virulence individually (Fig. 2C); however, the combined introduction of these two changes markedly attenuated the virus, resulting in an MLD₅₀ of 10^5.5 PFU (Table 1 and Fig. 2C).

We next examined the virus titers of wild-type and mutant HN/04 and SK/06 viruses in mice on days 3 and 6 postinfection. Wild-type and mutant BHG/05 viruses were excluded from this and further experiments, since only minor differences in virulence were detected (as described above). Animals were intranasally infected with 10⁵ PFU of virus, and titers were assessed in different organs on days 3 and 6 postinfection (Fig. 3A and B). Wild-type HN/04 virus was confined to the lungs on days 3 and 6. Introduction of NS1-138F or NS1-ESEV resulted in virus replication in the nasal turbinates or spleen, respectively, on day 3 postinfection; by day 6, virus was cleared from these organs. In contrast, HN/04 virus encoding NS1-138F and NS1-ESEV caused systemic replication on days 3 and 6 postinfection (Fig. 3A), with appreciable titers in the brain on day 6 (Fig. 3A). Hence, the amino acid changes tested here converted a virus that was confined to the respiratory tract to one that causes systemic infection. Wild-type SK/06 virus and both single mutants replicated in several organs, including brain, on day 6 postinfection (Fig. 3B). The NS1-138Y and NS1-ESKV sequences in combination, however, restricted SK/06 to the respiratory tract (lungs and nasal turbinates) (Fig. 3B). Collectively, these data demonstrate that the subgenotype Z.1-characteristic sequences NS1-138Y and NS1-ESKV have a synergistic effect on the virulence of HPAI H5N1 viruses.

Fig 3.

Replication of wild-type and mutant viruses in mice. Six mice per group were intranasally inoculated with 10⁵ PFU of HN/04 (A) or SK/06 (B) virus. Three mice were euthanized on days 3 and 6 postinfection. Lungs, nasal turbinates, brains, spleens, and kidneys were collected and virus titers were determined by means of plaque assay in MDCK cells. Data shown are the mean virus titers with standard deviations. The dotted line indicates the lower limit of detection of infectious virus. The P values were calculated by using Student's t test, comparing the virus titers between the wild-type and mutant viruses detected in the indicated organs. **, P < 0.001; *, P < 0.05.

Mutations at NS1-138 and in the NS1 PDM synergistically affect virus replication in cell culture.

We tested wild-type and mutant HN/04 and SK/06 viruses in human respiratory epithelial (Calu-3) and chicken fibroblast (DF-1) cells. Calu-3 and DF-1 cells were infected with virus at an MOI (as determined in MDCK cells) of 10⁻⁴ or 10⁻⁵ and incubated at 37°C (Calu-3 cells) or 41°C (DF-1 cells; to match the body temperature of chickens). At the indicated times after infection, viruses were titrated in MDCK cells (Fig. 4). In both cell lines tested, the introduction of NS1-138F or NS1-ESEV increased the replicative ability of HN/04 virus slightly to moderately, whereas a more pronounced effect was detected when both mutations were tested together (Fig. 4A and B). Thus, these data are consistent with our findings in mice (Fig. 2 and 3). On the other hand, the introduction of NS1-138Y and NS1-ESKV into SK/06 virus did not attenuate virus replication in cell culture (Fig. 4C and D), in contrast to the attenuating effect of these mutations in mice (Fig. 2 and 3). Thus, the effect of these mutations on virus replication in cell culture was viral background dependent.

Fig 4.

Replication of wild-type and mutant HN/04 and SK/06 viruses in chicken DF-1 and human Calu-3 cell cultures. DF-1 cells were infected with the indicated viruses at a multiplicity of infection (MOI) of 10⁻⁵ PFU and incubated at 41°C. Calu-3 cells were inoculated at an MOI of 10⁻⁴ PFU and incubated at 37°C. Aliquots of supernatants were collected at 12, 24, 36, 48, and 60 h for DF-1 cells or 12, 24, 36, 48, 60, and 72 h for Calu-3 cells and were titrated by means of plaque assays in MDCK cells. The titers shown are the means from three samples infected in parallel.

Mutations at NS1-138 and in the NS1 PDM do not affect the interferon antagonist activity of NS1.

NS1 is a known interferon antagonist (14, 15; reviewed in references 16 and 56). To determine whether NS1-138Y and/or NS1-ESKV affect the interferon antagonist function of NS1, we transfected human 293T cells with plasmids expressing wild-type or mutant HN/04 or SK/06 NS1 proteins and with a plasmid expressing the luciferase reporter protein under the control of an IFN-β promoter (to assess the ability of NS1 to interfere with IFN-β synthesis) or an IFN-β-stimulated interferon-stimulated response element (ISRE) promoter element (to assess the ability of NS1 to interfere with the synthesis of IFN-β-stimulated genes). Twenty-four hours later, cells were stimulated through Sendai virus infection or treatment with human IFN-β. After a further 24 h, luciferase assays were carried out. Stimulation with Sendai virus or IFN-β resulted in high levels of luciferase expression, which were efficiently blocked by wild-type HN/04 and SK/06 NS1 proteins (data not shown). The mutant HN/04 and SK/06 NS1 proteins were as efficient as wild-type NS1 proteins in blocking IFN-β-stimulated luciferase expression (data not shown). Hence, the observed differences in virulence and pathogenicity between wild-type and mutant HN/04 and SK/06 viruses were not caused by the interferon antagonist activity of NS1.

Mutations at NS1-138 and in the NS1 PDM affect the PI3K/Akt pathway.

Based on computational modeling, a loop in NS1 encompassing amino acids 137 to 142 is thought to interact with p85β, the regulatory subunit of PI3K (48). To assess the effect of NS1-138F and NS1-138Y on PI3K activity, we tested the phosphorylation levels of Akt, a major target of PI3K. These experiments were carried out in human brain astrocytoma (1321N1) cells, which express low intrinsic levels of phosphorylated Akt after serum starvation and are widely used for Akt phosphorylation studies. First, we verified that influenza viruses replicate in 1321N1 cells. Wild-type HN/04 virus replicated to appreciable titers in these cells (Fig. 5A); again, HN/04 virus encoding both NS1-138F and NS1-ESEV replicated more efficiently than wild-type virus or the single mutants, consistent with our findings in mice (Fig. 2 and 3) and in Calu-3 and DF-1 cells (Fig. 4).

Fig 5.

Induction of Akt phosphorylation. (A) Human 1321N1 cells were infected (MOI, 1) with wild-type or mutant HN/04 viruses. At the indicated times postinfection, virus titers in the supernatants were titrated by means of plaque assays in MDCK cells. The titers shown are the means from three samples infected in parallel. (B) Serum-starved 1321N1 cells were treated with 5% fetal bovine serum for 1 h; control cells were mock treated. Cell lysates were analyzed by Western blot analysis with antibodies to phospho-Akt (Ser-473), total Akt, calnexin, or NS1. (C) Serum-starved 1321N1 cells were infected (MOI, 1) with wild-type and mutant HN/04 viruses. Supernatants and total cell lysates were harvested at 3, 7, 12, and 24 h postinfection. Proteins were detected as described for panel B. (D) Coimmunoprecipitation of NS1 with p85β. 293T cells were transfected with plasmids for the expression of N-terminally Flag-tagged wild-type or mutant HN/04 NS1 protein. Cell extracts were prepared 36 h later; cell lysates and immunoprecipitates were examined by means of immunoblots with antibodies to NS1 or p85β.

To assess the levels of Akt phosphorylation, we serum starved 1321N1 cells overnight, which resulted in low levels of phosphorylated Akt (Fig. 5B). Induction with fetal bovine serum for 1 h induced high levels of phosphorylated Akt, as expected (Fig. 5B); total Akt and calnexin expression levels served as controls. Infection with wild-type HN/04 virus (MOI, 1) stimulated Akt phosphorylation, although appreciable levels were not observed until 12 h postinfection (Fig. 5C). Compared to wild-type virus, HN/04 encoding NS1-138F induced slightly higher levels of Akt phosphorylation. Interestingly, replacement of NS1-ESKV with NS1-ESEV resulted in even greater upregulation of Akt phosphorylation (Fig. 5C), suggesting that the NS1 PDM affects the regulation of the PI3K/Akt pathway. The double mutant encoding NS1-138F and NS1-ESEV did not cause a further increase in Akt phosphorylation levels, even though it appeared to be expressed at higher levels than the wild-type or single-mutant NS1 proteins. These differences in NS1 protein levels did not correlate with the Akt phosphorylation levels; for example, NS1-ESEV induced higher levels of Akt phosphorylation than did NS1-138F, even though their expressions levels were comparable. In addition, Akt phosphorylation studies in 1321N1 cells transfected with plasmids expressing wild-type or mutant HN/04 NS1 proteins yielded similar data (not shown). Infection of 1321N1 cells with wild-type or mutant SK/06 viruses did not reveal differences in Akt phosphorylation levels (data not shown), consistent with the lack of differences in their replication kinetics in Calu-3 and DF-1 cells (Fig. 4).

We next tested if the amino acid changes at position 138 and in the PDZ domain binding motif directly affected the interaction with p85β, the regulatory subunit of PI3K (48). Human 293T cells were transfected with N-terminally Flag-tagged wild-type or mutant NS1 protein expression plasmids for 24 h, and immunoprecipitated complexes were assayed with an antibody to p85β (Fig. 5D). Indeed, the nature of the amino acid at position 138 (but not the sequence of the PDM) affected interaction with p85β.

These findings demonstrate that the amino acid changes tested here affected the levels of Akt phosphorylation in HN/04 virus-infected and plasmid-transfected cells. The upregulation of the PI3K/Akt pathway is important for efficient influenza virus replication (29–34); thus, the weak stimulation of Akt phosphorylation by genotype Z.1 viruses encoding NS1-138Y and NS1-ESKV may contribute to the attenuating effect of these mutations.

Mutations at NS1-138 and in the NS1 PDM affect the interaction of NS1 with Dlg-1 and Scribble.

NS1 proteins encoding the avian virus-type ESEV motif bind to several cellular PDZ domain proteins (40, 41, 57, 58), including Dlg-1 (39–41) and Scribble (39–42); in contrast, the human virus-type NS1-RSKV motif has very low affinity for these proteins (39, 41, 42). To assess the interaction of the subgenotype Z.1-specific ESKV motif with Dlg-1 and Scribble, we expressed N-terminally Flag-tagged HN/04 NS1 proteins in 293T cells and performed coimmunoprecipitation experiments with antibodies to Dlg-1 and Scribble (Fig. 6). Wild-type HN/04 NS1 protein (possessing NS1-138Y and NS1-ESKV) showed negligible interaction with Dlg-1 and Scribble, whereas the introduction of the avian virus-type ESEV motif resulted in strong interactions between NS1 and Dlg-1 or Scribble, as reported elsewhere (39–42). Thus, the ESKV motif found in subgenotype Z.1 viruses has lower binding affinity for Dlg-1 and Scribble than does the ESEV motif found in most HPAI H5N1 viruses. Replacement of NS1-138Y with the highly conserved NS1-138F residue increased the interaction of NS1 with Scribble but not that with Dlg-1 (Fig. 6). Hence, amino acids in NS1 that play a role in the activation of the PI3K/Akt pathway also affected the interaction with cellular PDZ domain proteins, a functional connection not previously known.

Fig 6.

Coimmunoprecipitation of NS1 with Dlg-1 and Scribble. 293T cells were transfected with plasmids for the expression of N-terminally Flag-tagged wild-type or mutant HN/04 NS1 protein. Cell extracts were prepared 36 h later; cell lysates and immunoprecipitates were examined by means of immunoblots with antibodies to NS1, Dlg-1, or Scribble.

DISCUSSION

Previous studies have found that the influenza A virus NS1 protein affects virulence through the activation of the PI3K/Akt pathway (29–34) and through its interaction with cellular PDZ domain proteins (27, 28), among other mechanisms (reviewed in references 16 and 56). Here, we showed that a mutation in the NS1 PDM also alters PI3K/Akt pathway activation, and that a mutation in a domain thought to affect Akt phosphorylation also alters NS1 binding to cellular PDZ domain proteins. These mutations in NS1 synergistically affected HPAI H5N1 virulence in mice.

Several classes of PDM have been identified (42, 59, 60). Influenza virus NS1 proteins possess a type I PDM, characterized by the sequence x-S/T-x-V (where x is any amino acid). Although positions −1 and −3 can accommodate different amino acids, clear preferences exist based on analyses of PDMs in human, mouse, nematode, zebrafish, and fruit fly proteins (59). Glutamate is most frequently found at positions −1 and −3 (59), forming the ESEV motif that is present in most avian virus NS1 proteins. The human virus-type RSKV motif and the subgenotype Z.1-specific ESKV motif are rare among cellular PDZ domain-binding motifs (59).

The significance of the NS1 PDM for influenza virulence is not fully understood. The NS1 PDM found in pandemic 1918 virus (KSEV) and most avian viruses (ESEV) increased virulence slightly in the background of the laboratory-adapted A/WSN/33 (H1N1) virus, which possesses an NS1-RSEV sequence (27). In addition, an H7N1 virus possessing the avian-type ESEV motif was more virulent in mice than the same virus encoding RSKV (28). In contrast, another study found that HPAI H5N1 viruses encoding NS1-ESEV or -RSKV did not differ significantly in virulence or pathogenicity in mice or chickens (61); however, this study was carried out in the background of a virus that possesses a truncated NS1 protein and may have evolved to function without the NS1 PDM. The significance of the NS1-ESKV motif found in subgenotype Z.1 viruses had not previously been addressed experimentally; here, we demonstrated that this motif attenuates HPAI H5N1 viruses and acts synergistically with the amino acid at position 138 of NS1.

The avian virus-type NS1-ESEV motif interacts efficiently with Dlg-1, whereas the human virus-type NS1-RSKV motif does not (39, 41). NS1-ESEV also interacts with the cellular Scribble protein (39–42). We found that the subgenotype Z.1-characteristic ESKV motif did not coimmunoprecipitate Dlg-1 or Scribble (Fig. 6); as expected, the introduction of the ESEV motif into the HN/04 NS1 protein restored these interactions (Fig. 6). Interestingly, mutation of NS1-138Y (as found in HN/04) to NS1-138F (commonly found among influenza A NS1 proteins) also restored the interaction of NS1 with Scribble but not with Dlg-1 (Fig. 6).

We also found that NS1-ESEV and -ESKV differed in their ability to stimulate the PI3K/Akt pathway, which is critical for efficient influenza virus replication (29–34). A previous study showed that the adenovirus protein E4 stimulates Akt phosphorylation in a PDM-dependent manner (62, 63); interestingly, this effect was mediated through the interaction of E4 with Dlg-1 (63). Therefore, we speculate that, through a currently unknown mechanism, the interaction of NS1-ESEV (but not NS1-ESKV) with Dlg-1 stimulates the PI3K/Akt pathway, resulting in efficient virus replication. While this study was under way, Li et al. (64) described an H5N1 influenza virus that no longer activated the PI3K/Akt pathway. The NS1 protein of this virus possesses the NS1-ESKV motif, consistent with our finding that NS1-ESKV affects PI3K/Akt activation.

The amino acid at position 138 of NS1 is part of an exposed loop that is thought to interact with the p85β subunit of PI3K (48); however, X-ray crystallographic analysis of the NS1-p85β complex showed no direct interaction between the highly conserved NS1-138F and p85β (47). Moreover, replacement of NS1-137N/138F with NS1-137A/138A did not affect NS1 binding to p85β (48). In contrast, our data demonstrate that the amino acid at NS1-138, in combination with the NS1 PDM, affects HPAI H5N1 virulence. In fact, we found that the large, polar tyrosine residue found in subgenotype Z.1 NS1 proteins confers stronger interaction with p85β than NS1-138F (Fig. 5D) but weaker PI3K/Akt activation (Fig. 5C). We speculate that the large tyrosine residue causes a minor conformational change in PI3K that affects efficient activation, and/or that strong binding by NS1-138Y locks the complex in this state and interferes with the kinase activity of PI3K.

Our results demonstrate synergistic effects between the mutation at NS1-138 and that in the NS1 PDM. Inspection of the Influenza Research Database (IRD) and the Los Alamos National Laboratory influenza virus sequence database (65) identified only seven avian non-H5N1 influenza A viruses with NS1-138Y; all seven also encode NS1-ESKV. One of these is an N1 virus of unknown hemagglutinin subtype. Five are H9N2 viruses (isolated from chickens in Pakistan in 2005 to 2007) with NS genes from either H5 or H9 subtypes (66), suggesting that their NS gene originated from a subgenotype Z.1 virus. Finally, A/turkey/Israel/446/2006 (H9N2) also possesses NS1-138Y and NS1-ESKV. Thus, with the exception of subgenotype Z.1 viruses (which are now prevalent in poultry populations in the Middle East), tyrosine at NS1-138 is extremely rare and is typically found in combination with NS1-229K.

NS1-ESKV is also rarely detected among avian non-H5N1 viruses. In addition to the viruses listed above, the Mexican-lineage H5N2 viruses (which caused a major HPAI outbreak in 1993 [67–70]) also encode NS1-ESKV. Some of them also possess NS-138F, whereas others encode NS1-138S or NS-138L, which are very rare at this position and are different from the tryptophan residue found in subgenotype Z.1 viruses. In addition, only five other avian non-H5N1 viruses with NS1-ESKV were identified in the IRD and Los Alamos National Laboratory influenza virus sequence database, all of which encode NS1-138F: A/parrot/CA/6032/2004 and A/chicken/Ibaraki/1/2005, both of which belong to the Mexican lineage of H5N2 viruses (71, 72), together with A/chicken/Guatemala/270475-4/2003 (H4N2), A/yellow-headed Amazon/California/500658/2007 (H5N2), and A/lesser scaup/Wisconsin/3964/2009 (H10N3).