|Home | About | Journals | Submit | Contact Us | Français|
The NS1 protein of the influenza A virus is a potent virulence factor that inhibits type I interferon (IFN) synthesis, allowing the virus to overcome host defenses and replicate efficiently. However, limited studies have been conducted on NS1 function using human virus strains and primary human cells. We used NS1 truncated mutant influenza viruses derived from the human isolate influenza A/TX/91 (TX WT, where WT is wild type) to study the functions of NS1 in infected primary cells. Infection of primary differentiated human tracheo-bronchial epithelial cells with an NS1 truncated mutant demonstrated limited viral replication and enhanced type I IFN induction. Additionally, human dendritic cells (DCs) infected with human NS1 mutant viruses showed higher levels of activation and stimulated naïve T-cells better than TX WT virus-infected DCs. We also compared infections of DCs with TX WT and our previously characterized laboratory strain A/PR/8/34 (PR8) and its NS1 knockout strain, deltaNS1. TX WT-infected DCs displayed higher viral replication than PR8 but had decreased antiviral gene expression at late time points and reduced naïve T-cell stimulation compared to PR8 infections, suggesting an augmented inhibition of IFN production and human DC activation. Our findings show that human-derived influenza A viruses have a high capacity to inhibit the antiviral state in a human system, and here we have evaluated the possible mechanism of this inhibition. Lastly, C-terminal truncations in the NS1 protein of human influenza virus are sufficient to make the virus attenuated and more immunogenic, supporting its use as a live attenuated influenza vaccine in humans.
Models of influenza A infection have shown that upon initial exposure to virus, innate immune functions are activated primarily by the release of type I interferon(s) (IFN) from infected cells. These critical cytokines can be released from any infected cell and trigger the synthesis of antiviral proteins in adjacent cells, which wards off attacking virus (52). Most pathogenic viruses including influenza virus possess antagonistic proteins that thwart the protective effect of IFN and allow establishment of infection (52). Once the innate barrier has been breached, an adaptive immune response is initiated that is characterized by the production of cytotoxic T lymphocytes, T helper 1 (Th1) CD4 T cells, and a neutralizing antibody response. Recovery from primary infection is mediated by cytotoxic T lymphocytes, which are expanded and activated in draining lymph nodes, that cycle back to the infection site and kill virus-infected cells (8, 20).
Dendritic cells (DCs) are important in detecting influenza A infections and function as an essential link between innate immunity and adaptive immunity leading to viral clearance (41). Steady-state DCs residing in tissue actively take up antigen and upon recognition of unique viral structures initiate signaling cascades that activate (mature) the DCs, leading to their production of chemokines, expression of chemokine receptors (56), and production of cytokines required for leukocyte recruitment and stimulation. In addition, the maturational process leads to an increased expression of major histocompatibility complex class II (MHC-II), CD86, CD80, and CD40 needed for cognate antigen presentation to stimulate adaptive immunity (41, 45). While these events are relatively similar when initiated by other types of microbes, viruses trigger the release of large amounts of type I IFN that functions both as an antiviral alarm and as a signal that enhances the DC maturation. The coincident timing and regulation of DC maturation and type I IFN production suggests that the entire process is regulated by common pathways (15, 37).
In influenza A virus infections, the RNA helicase retinoic acid-inducible gene I (RIG-I) is the major trigger of host antiviral responses (29, 30, 34, 36, 37), binding virus-derived double-stranded RNA and single-stranded RNA bearing uncapped 5′ phosphates, which results in a conformational change that triggers the initiation of multiple antiviral signaling pathways (10, 27, 53, 64). RIG-I signals through the mitochondrial protein IPS-1 (IFN promoter-stimulating factor 1) to activate the noncanonical TANK-binding kinase and IKK (IκB kinase) kinases and the transcription factors c-Jun, ATF, NF-κB, and interferon regulatory factor 3 (IRF3) and IRF7, which upregulate IFN and inflammatory cytokines (31, 72). Secreted type I IFN from virus-infected cells amplifies the antiviral response by binding to membrane IFN receptors and activating the intracellular Jak/STAT pathway that leads to the upregulation of various cellular host products such as MxA, IRF7, cytokines, and chemokines that further stimulates innate antiviral responses and alert the immune system against viral infection (18).
The viral nonstructural protein NS1 is a potent virulence factor for the influenza A virus as it is implicated in “masking” viral presence and in inhibition of immunity via multiple mechanisms (14, 16, 24). NS1 binds to RIG-I/IPS-1 complexes (22, 42, 48) and blocks downstream signaling (39, 65, 71), resulting in attenuation of type I IFN and inflammatory cytokine expression. However, NS1 can also disrupt cellular mRNA processing and nuclear-cytoplasmic export by binding to the cellular mRNA processing factor cleavage and polyadenylation specificity factor (CPSF), downregulating the induction of cellular genes at a posttranscriptional level (9, 32, 46, 47). These two activities are possible due to a complex regulation of NS1 cellular localization mediated by two nuclear localization signals (at residues 34 to 38 and residues 216 to 221) and one nuclear export signal (residues 138 to 147), which allow the shuffling of this protein between the cytoplasm and nucleus of the infected cell (33). The CPSF inhibitory properties of NS1 appear to be dispensable for some influenza virus strains (32, 47, 68). Other functions of NS1 that could contribute directly or indirectly to regulation of the host antiviral response include sequestration of viral double-stranded RNA from cellular sensors (7, 25, 38), disruption of nuclear pore function (58), activation of the phosphatidylinositol 3-kinase/Akt pathway (13, 23, 62), and interactions with cellular proteins involved in translation (1, 5). Consistent with a prominent role of NS1 in regulating host responses to viral infection, truncation of the C terminus of NS1 results in influenza A virus attenuation in mice, pigs, horses, and macaques (5, 12, 54, 55, 63, 66, 71). Animals infected with NS1 mutant viruses show reduced pathogenesis and increased IFN expression with respect to animals infected with wild-type influenza virus, and animals immunized with NS1 mutant influenza viruses are protected from wild-type influenza virus infection (55, 66, 69). This suggests that NS1 C-terminally truncated mutants could be further developed as live attenuated vaccines against influenza (28, 49, 50). Interestingly, the multiple activities of the NS1 protein result in diminishing activation of DCs, blocking cytokine expression as well as T-cell activation in our model of human DCs, thus counteracting both innate and adaptive immune responses (15).
Despite several studies on the effects of the NS1 of influenza virus in viral replication and inhibition of antiviral host responses during viral infection, only limited studies have been performed using human influenza virus strains and/or human primary cells. And, though it is known that the NS1 C-terminally truncated mutants are immunogenic and serve as efficient vaccines in mice, swine, equine, and nonhuman primate models, it is yet to be demonstrated in a human system. Using a human influenza virus strain, we examined the effects of NS1 C-terminal truncations on viral replication in human primary respiratory epithelial cells and in induction of immune responses using human DCs as a surrogate model for immune activation. In an attempt to map the regions of the NS1 protein responsible for the inhibition of DC activation observed in our previous studies, we decided to generate and use recombinant influenza viruses with truncations in the NS1 protein using a human influenza virus as the backbone strain. Via reverse genetics, a human isolate of the H1N1 influenza A virus circulating in 1991 (influenza A/Texas/91 [TX WT]) and two different recombinant viruses containing NS1 C-terminal deletions (TX 1-99 and TX 1-126 expressing NS1 truncated proteins of 99 and 126 amino acids, respectively) were generated. In the present study, we show that the TX WT NS1 C-terminal deletion mutants are attenuated, with decreased viral replication in primary cells, but are efficient stimulants of type I IFN production and innate and adaptive immunity. Having generated a human ex vivo model of human influenza infection and immune response, we then compared the activation of human DCs by human influenza TX WT viruses to that previously characterized with influenza A mouse-adapted laboratory strains, influenza A/PR/8/34 (PR8) and the NS1 deletion mutant of PR8, deltaNS1 (15). TX WT replicates at a greater level than PR8 in human DCs but can display stronger inhibition on the expression of IFN-responsive genes. These results are consistent with reduced naïve T-cell stimulation with TX WT-infected DCs compared to PR8-infected DCs.
Taken together, our results support our system of human primary cell lines in being an efficient surrogate model for investigating human influenza A pathogenesis. These studies also suggest that the NS1 of human influenza A has additional adaptations for heightened inhibition of establishing the antiviral state, but truncating the C terminus is sufficient to attenuate this more potent virulence, making these mutants suitable candidates for vaccine design.
TX WT, TX 1-99, and TX 1-126 influenza viruses were generated in Peter Palese's laboratory by reverse genetics similarly to other influenza virus mutants (5, 54, 63, 66). Recombinant TX WT virus as well as TX 1-99 and TX 1-126 viruses, encoding truncated NS1 proteins of 99 and 126 amino acids, respectively, were grown in 7-day-old embryonated chicken eggs (SPAFAS; Charles River Laboratories). Influenza virus deltaNS1 was generated by reverse genetics from influenza virus PR8 as previously described (10). Influenza virus A/NY/55/04 (NY), influenza virus A/WYM/3/03 (WYM), and PR8 virus were grown in 9-day-old eggs. Influenza virus deltaNS1 was grown in 6-day-old embryonated eggs. All influenza viruses were titrated by influenza virus nucleoprotein (Flu NP) immunostaining using MDCK or Vero cells. For influenza virus titrations, 2.5 μg/ml trypsin was included in the infection medium.
MDCK, Vero, and A549 cells were grown in tissue culture medium, Dulbecco's modified Eagle's medium (DMEM; Invitrogen), supplemented with 10% fetal calf serum (HyClone), 1 mM sodium pyruvate (Invitrogen), 2 mM l-glutamine (Invitrogen), and 50 μg/ml gentamicin (Invitrogen). All cells were grown at 37°C in 7% CO2.
Peripheral blood mononuclear cells were isolated by Ficoll density gradient centrifugation (Histopaque, Sigma Aldrich) from buffy coats of healthy human donors (Mount Sinai Blood Donor Center and New York Blood Center) as previously described (15). Briefly, CD14+ cells were immunomagnetically purified using anti-human CD14 antibody-labeled magnetic beads and iron-based MiniMACS LS columns (Miltenyi Biotech). After elution from the columns, cells were plated (0.7 × 106 cells/ml) in DC medium (RPMI medium [Invitrogen], 10% fetal calf serum [HyClone] or 4% human serum [Cambrex/Lonza], 100 U/ml penicillin, and 100 μg/ml streptomycin [Pen/Strep; (Invitrogen]) supplemented with 500 U/ml human granulocyte-macrophage colony-stimulating (Peprotech), and 1,000 U/ml human interleukin-4 (hIL-4; Peprotech) and incubated for 5 to 6 days at 37°C.
After 5 to 6 days in culture, DCs were infected with influenza viruses at a multiplicity of infection (MOI) of 0.5, 1, or 2 for 60 min in serum-free RPMI medium. Cells were then plated in DC medium (described above) at 1 × 106 cells/ml maintained in culture for different time periods, depending on the experiment.
Cryopreserved normal human tracheo-bronchial epithelial (HTBE) cells were purchased from Cambrex and cultivated according to the manufacturer's instructions and a previously described protocol (21). Briefly, cells were passaged twice in bronchial epithelial growth medium (BEGM) containing necessary supplements and growth factors (Cambrex) and seeded onto Transwell-Clear Permeable filters (24-mm diameter and 0.4-mm pore size; Millipore) at a density of 1.4 × 105 cells/filter. Before the seeding step, filters were coated with collagen I derived from human placenta (BD Biosciences). Cells were grown submerged in a 1:1 mixture of DMEM and BEGM supplemented with growth factors for 1 week. Then, medium from the apical surface was removed, and cells were maintained in an air-liquid interface for 3 weeks. When cells are maintained in an air-liquid interface for long periods of time, they differentiate into different cell types present in the mucosal surface of the respiratory tract. The differentiation status of cells was routinely monitored by immunofluorescence of cells with anti-β-tubulin antibodies (Sigma) specific for ciliated cells.
HTBE cells were infected with either TX WT or TX 1-126 viruses at an MOI of 2 or mock infected for 1 h at 37°C. Prior to infection, cells were washed 10 times with growth medium. After infection cells were washed once, and medium (BEGM-DMEM) was left to cover apical surfaces of cells as well as the basal compartment of wells. At 9.5 and 25 h postinfection, medium from apical and basal compartments was collected, and cell layers were either resuspended in sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis buffer for analysis by Western blotting or fixed for immunostaining. Harvested supernatants from apical surfaces were used to determine viral titers. In order to detect the presence of IFN-α/β, a bioassay was utilized as described previously (61). Briefly, medium from the basal compartment of infected HTBE cells was dialyzed against low-pH buffer (50 mM glycine, pH 2) to inactivate virus, and neutral pH was restored by subsequent dialysis against phosphate-buffered saline (PBS). Vero cells were then incubated with different dilutions of medium from infected HTBE cells for 24 h and infected with Newcastle disease virus expressing green fluorescent protein (NDV-GFP), which is known to be sensitive to the action of IFN. Twenty-four hours later, the intensity of green fluorescence was measured. As controls, Vero cells were treated with a known quantity of IFN-β (Calbiochem). To plot the data, we considered the fluorescence level of cells incubated with medium from mock-infected HTBE cells to be 100%. In other experiments, HTBE cells were infected at an MOI of 0.001 for 1 h at 37°C, virus was removed, and cells were washed once with medium and continued to be incubated in the air-liquid interface. At various time points, medium was added to apical surfaces for 30 min and harvested for virus titration.
Capture enzyme-linked immunosorbent assays (ELISAs) for IL-6, tumor necrosis factor alpha (TNF-α) (R & D systems), IFN-α, and IFN-β (Biosource) were used according to the manufacturer's instructions to quantify the cytokines and chemokines in the DC supernatants. Plates were read in an ELISA reader from Biotek instruments.
Cells were stained with phycoerythrin-linked CD86 according to the manufacturer's instructions (Beckman Coulter and BD-Pharmingen), and expression was determined by flow cytometry. For Flu NP staining, cells were fixed in 3% formaldehyde for 15 min on ice and then permeabilized in 0.1% saponin in PBS for 30 min on ice. Cells were then incubated with rabbit polyclonal anti-Flu NP for 1 h, and then goat anti-rabbit secondary antibody conjugated to Alexa 488 (Invitrogen) for 1 h. Cells were washed twice with 0.05% fetal bovine serum-PBS twice between each step. Data were analyzed using Flowjo software.
Samples of 1 × 105to 5 × 105 DCs were either infected with virus or mock infected, and RNA was isolated and DNase treated using an Absolutely RNA reverse transcription-PCR (RT-PCR) miniprep kit (Stratagene). RNA was quantified using a Nanodrop spectrophotometer (Nanodrop Technologies). The yields of RNA were approximately 50 to 100 μg/ml.
Quantitative RT-PCR (qRT-PCR) was performed according to a previously published protocol using SYBR Green in an ABI 7900 HT protocol(15, 73). Each transcript in each sample was assayed in duplicate, and the mean threshold cycle of three housekeeping genes (RSP11, β-actin, and α-tubulin) was used to calculate the relative copy number for each gene.
DCs either left untreated or infected with different recombinant influenza viruses were mixed together with allogeneic peripheral blood mononuclear cells from buffy coats and cultured for 4 days in 96-well plates at a ratio of 1:5 (106 DCs/ml to naive T cells). Supernatants from the cocultures were tested by ELISA for IFN-γ release at different times of culture (eBioscience).
Samples of 106 cells were lysed in 100 μl of lysis buffer (150 mM NaCl, 25 mM Tris-HCl [pH 7.4], 2 mM EDTA, 1 mM sodium orthovanadate, 10 mM sodium fluoride, 1% Triton X-100, 0.5 mM phenylmethylsulfonyl fluoride, 10 μg/ml aprotinin, and 10 μg/ml leupeptin) or 100 μl of Phosphosafe buffer (Calbiochem). Cell lysates were centrifuged (14,000 × g for 10 min) to remove cellular debris and boiled for 5 min in SDS sample buffer containing 1 mM dithiothreitol. Protein lysates were resolved on 4 to 15% SDS-polyacrylamide gel electrophoresis gradient gels (Bio-Rad), transferred to nitrocellulose membranes, and probed with rabbit polyclonal anti-NP and anti-NS1 as previously described (63) and with mouse monoclonal anti-M1 and TX WT-specific anti-HA (Mount Sinai Hybridoma center), anti phospho-IRF3 (S396) (Cell Signaling), anti-phospho-NF-κBp65/RelA (S536) (Cell Signaling), anti phospho-c-Jun (S73) (Cell Signaling), anti-NF-κBp65/RelA (Cell Signaling), anti-c-Jun (Cell Signaling), anti-IRF3 (Cell Signaling), and anti-actin (Santa Cruz). Immunoreactive bands were detected using horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit immunoglobulin G (Bio-Rad) and enhanced chemiluminescence (Bio-Rad).
We aimed to determine whether NS1 proteins in several influenza virus strains were also associated with decreased inhibition of IFN-related gene expression and decreased levels of DC activation in human influenza virus-infected DCs. Two NS1 mutant viruses in the TX WT background were generated, TX 1-126 and TX 1-99, that encode C-terminally truncated NS1 proteins of 126 and 99 amino acids, respectively, in contrast to the 230-amino-acid-long NS1 protein of the corresponding TX WT virus (as previously described in Materials and Methods). Rescue of TX 1-126 virus from plasmids was previously reported (5). TX 1-99 virus was rescued by using a truncated NS gene similar to that of the TX 1-126 virus except that the NS1 stop codon was inserted after amino acid residue 99.
In order to investigate the impact of NS1 function during infection of human respiratory epithelial cells, the primary target for influenza virus replication in humans, we first determined the levels of viral protein synthesis in the human A549 lung alveolar cell line. TX 1-99 and TX 1-126 virus-infected A549 cells showed similar levels of NP protein and reduced M1 protein levels compared to levels seen in TX WT virus-infected A549 cells (Fig. (Fig.1A).1A). However, HA protein and the respective truncated NS1 polypeptides of the mutant viruses were poorly expressed and only barely detectable in overexposed blots compared to the clearly visible levels observed in TX WT-infected cells (Fig. (Fig.1A1A).
For the next series of studies, we compared one of the NS1-truncated mutant viruses, TX 1-126, with TX WT virus in differentiated HTBE cells grown in an air-liquid interface as a surrogate system of the human upper respiratory epithelium. Upon infection at an MOI of 2, TX 1-126 virus replicated less efficiently than TX WT virus, as confirmed by different methods. Immunofluorescence of infected cells (Fig. (Fig.1B)1B) showed equal levels of infection by both viruses at 9.5 h postinfection but a much lower number of infected cells in the case of TX 1-126 virus at 25 h. Western blotting of HTBE lysates at 25 h postinfection was consistent with the immunofluorescence data (Fig. (Fig.1C).1C). A truncated 126-amino-acid NS1 protein in infected HTBE cells was not detected, which is indicative of low expression of the truncated NS1 in these cells. We also detected lower virus titers in apical supernatants from TX 1-126 virus-infected HTBE cells at 25 h postinfection (Fig. (Fig.1D)1D) than in TX WT-infected HTBE cells. Differences in viral replication between TX 1-126 and TX WT virus in HTBE cells were even more apparent when cells were infected at an MOI of 0.001, which allowed for multicycle replication of the viruses. At 24 and 48 h postinfection, virus titers in the supernatants were determined, and the TX WT virus reached titers around 106 at 48 h postinfection. However, TX 1-126 viruses were detected at very low levels at 24 h postinfection and not detected at all at 48 h postinfection (Fig. (Fig.1E).1E). Attenuation of replication of this virus in human primary respiratory epithelial cells is likely caused by an impaired ability to counteract type I IFN production in the host cells due to NS1 truncation. To confirm this hypothesis, we measured IFN-α/β in the medium of HTBE cells infected with TX WT and TX 1-126 viruses using a bioassay (as described in Materials and Methods). Indeed, increased levels of IFN-α/β in the medium of HTBE cells infected with TX 1-126 virus were observed, as demonstrated by antiviral activity against NDV-GFP replication in Vero cells (Fig. (Fig.1F).1F). This result indicates that the TX 1-126 virus is defective in blocking type I IFN production in human primary respiratory epithelial cells, which are believed to be the main targets for human influenza virus replication in vivo.
We then infected DCs, which support replication of influenza viruses but do not produce infectious particles (human data not shown, but for mouse DCs see reference35) and examined viral protein expression by Western blotting. Levels of NP in NS1 mutant virus-infected DCs were equal to that of TX WT virus-infected DCs (Fig. (Fig.2A).2A). But in DCs infected with the TX 1-99 and TX 1-126 mutant viruses, a significant reduction in the levels of HA, M1, or NS1 proteins compared to TX WT virus-infected DCs was seen (Fig. 2A and B). Thus, truncations in the NS1 protein strongly impact the expression levels of other viral proteins in infected human DCs. To examine the kinetics of viral protein synthesis in human DCs, we performed a time course experiment examining viral protein levels at 4 h, 8 h, and 16 h postinfection (Fig. (Fig.2B).2B). TX WT, TX 1-99, and TX 1-126 viruses showed similar rates of expression for NP protein at all time points tested (Fig. (Fig.2B).2B). NS1 and M1 proteins were detected as early as 4 h postinfection in TX WT-infected DCs. Again, DCs infected with TX 1-99 or TX 1-126 produced drastically reduced levels of M1 and nearly undetectable amounts of their respective NS1 truncated proteins (Fig. (Fig.2B).2B). When expression of the late viral protein HA was examined in DCs infected with TX WT, this protein was detectable as early as 8 h postinfection, and the levels were increased at 16 h postinfection. However, DC infection with TX 1-99 and TX 1-126 showed poor expression of HA even at 16 h postinfection (Fig. (Fig.2B).2B). In summary, the NS1 C-terminal deletion mutants are compromised in their ability to express late viral proteins (such as HA and M1) while expressing normal levels of early viral proteins (such as NP) in human DCs. While several viral proteins are affected by the presence of an intact NS1, the truncation mutants are able to produce early proteins in human DCs, which could still be presented to T cells by the infected DCs, thus contributing to the initiation of adaptive immunity against the virus in vivo.
Activation of RIG-I leads to induction of type I IFN and secretion of other proinflammatory cytokines by activation of NF-κB/RelA, c-Jun, and IRF3 (51, 60). It has been demonstrated that phosphorylation of IRF3 at S396 corresponds to its activation (59). Since the RIG-I pathway and DC activation can be inhibited by the influenza virus protein NS1 (15, 42, 48), we sought to determine how effective TX 1-99 and TX 1-126 mutant viruses would be at stimulating these processes. When phosphorylation and activation of IRF3 (at residue S396) were examined, an increase in phosphorylation of this transcription factor in TX 1-99- and TX 1-126-infected DCs was observed compared to TX WT-infected conventional DCs at 4 and 8 h postinfection (Fig. (Fig.3A).3A). These results are consistent with the proposed role of the carboxy-terminal region of the NS1 protein in promoting NS1 dimerization and enhancing its ability to prevent the activation of transcriptional factors (4). These results strongly suggest that the inability of the NS1 truncation mutants to inhibit IRF phosphorylation contributes to their enhanced immunogenicity while conferring on them an attenuated phenotype.
To further examine the level of DC activation stimulated by these NS1 mutant influenza viruses, we infected DCs and performed qRT-PCR on total mRNA isolated at different times postinfection, measuring upregulation of multiple genes involved in DC maturation, including type I IFN and selected proinflammatory cytokines and chemokines. When DCs were infected with the mutant viruses TX 1-99 and TX 1-126, there was increased expression of type I IFN genes (Fig. (Fig.3B;3B; see also Fig. S1 in the supplemental material), as well as IFN-regulated genes such as IP10 and ISG54 (Fig. (Fig.3D)3D) compared to that in TX WT virus-infected DCs. Similarly, there was an increase in expression of proinflammatory cytokine genes such as TNF-α, IL-6, IL-12p35, and MIP-1β in TX 1-99- and TX 1-126-infected DCs versus TX WT-infected DCs (Fig. (Fig.4A;4A; see also Fig. S1 in the supplemental material). Nevertheless, a cohort of genes (TANK-binding kinase, IKK, and Toll-like receptor 3) were not affected during infection of DCs with either of the viruses (data not shown). Next, DC supernatants were analyzed by ELISA for secreted proteins. Corroborating the findings from our qRT-PCR experiments, IFN-β secretion (Fig. (Fig.3C),3C), as well as TNF-α and IL-6 secretion (Fig. (Fig.4B),4B), was augmented in DC infections with TX 1-99 and TX 1-126 viruses compared to TX WT virus infection. This secretion of type I IFN and proinflammatory gene products suggests that the mutant viruses are ineffective in blocking pathogen-associated molecular patterns that trigger signaling pathways leading to inflammatory responses and an IFN-mediated antiviral state. The results from the DCs infected with the NS1 mutant influenza virus directly correlate with the low levels of expression of the NS1 proteins in DCs infected by those viruses (Fig. (Fig.2).2). These data highlight the immunogenicity of the NS1 truncated mutants as potential vaccine candidates since they induce potent activation of human DCs.
We next questioned if DCs infected with viruses carrying the C-terminal deletions in NS1 can be better antigen-presenting cells than those infected with the wild-type virus. When costimulatory surface expression was examined by flow cytometry, we observed increased surface CD86 (Fig. (Fig.5A)5A) and CD83 and MHC-II molecules (data not shown) in NS1 mutant influenza virus-infected DCs compared to TX WT-infected DCs, suggesting enhanced T-cell stimulatory function by those DCs. We next assessed the activation of infected DCs by analyzing the expression of Flu NP protein and CD86 in DCs. Figure Figure5B5B shows that mock-infected DCs had poor staining for Flu NP, but about 10% of cells expressed CD86 at high levels (CD86hi). At 16 h postinfection at an MOI of 1, TX WT infections of DCs were 58.6% Flu NP positive (Flu NP+) (Fig. (Fig.5B),5B), but only ~23% of total cells were CD86hi (Fig. (Fig.5B;5B; see also Fig. S2A in the supplemental material), with 20.2% of total cells being dual Flu NP+ CD86hi. When TX 1-99- and TX 1-126-infected DCs were analyzed, Flu NP+ cells were 64.1% and 67.5% of total cells, respectively (Fig. (Fig.5B;5B; see also Fig. S2B in the supplemental material), and CD86hi cells were 47.7% and 53%, respectively, with 39.4% Flu NP+ CD86hi cells in TX 1-99 infections and 45.5% Flu NP+ CD86hi cells in TX 1-126 infections (Fig. (Fig.5B;5B; see also Fig. S2A in the supplemental material). Also, there were higher Flu NP-negative (Flu NP−) CD86hi cells in TX 1-99 and TX 1-126 infections (8.3% and 7.6%, respectively) than in TX WT infections (2.5%). These results demonstrate that DCs infected with the truncated mutants have higher number of double NP+ CD86+ cells than DCs infected with the wild-type virus, suggesting that the mutant viruses overall will induce better immune antiviral responses than the wild-type virus, and this activation is primarily due to the actual infection.
We then tested the ability of virus-infected DCs to stimulate T cells in an allogeneic coculture assay. To assess this, DCs were infected and incubated with naïve allogeneic CD4 T cells for 3 days at a DC-to-CD4 T-cell ratio of 1:5. The cell cocultures were analyzed for T-cell proliferation and cytokines indicative of Th1 (IFN-γ) and Th2 (IL-5 and IL-10) responses. No significant differences were detected in either T-cell proliferation or in IL-5 and IL-10 secretion in T-cell cultures with DCs infected with TX WT or the NS1 deletion mutant viruses (data not shown). However, the T-cell cultures containing DCs infected with mutant NS1 influenza virus showed increased levels of IFN-γ compared to those containing DCs infected with TX WT (Fig. (Fig.5C).5C). Thus, the increased DC activation from TX 1-99 and TX 1-126 virus infections resulted in increased induction of IFN-γ production by T cells. These results suggest not only increased T-cell activation when NS1 function is impaired but also increased Th1 polarity in adaptive immune responses, which is known to be optimal for the clearance of influenza virus from the host (45). Our data show that the truncated NS1 mutants were more efficient at priming T cells toward a Th1 polarity than the wild-type virus with intact NS1.
We have previously shown that the laboratory strain of influenza virus PR8 does not induce significant levels of type I IFN in cultured human DCs, whereas a mutant PR8 virus with the NS1 gene deleted (deltaNS1 virus) is a good stimulant of IFN production in human DCs (15). The NS1 from PR8 influenza virus did not exhibit a global inhibitory effect on DCs as shown by microarray, qRT-PCR, and ELISA but, rather, a specific inhibitory effect on genes involved in DC activation (15). Conversely, the NS1 protein from the human isolate of influenza virus TX WT has been shown to induce a strong shutoff of host protein expression in infected cell lines (32). From these results, we inferred that human influenza A virus strains may display a greater inhibitory effect on human DC activation than the PR8 mouse-adapted strain. To test this hypothesis, human cultured DCs were infected with PR8 and TX WT as well as other human influenza A virus strains used in the 2005/2006 trivalent influenza A vaccine (WYM and NY) (see Materials and Methods) and analyzed by qRT-PCR for type I IFN as well as IFN-induced genes at 16 h postinfection. As previously described, PR8 infection of human DCs induced minimal levels of IFN-α and IFN-β but showed induction of IFN-responsive genes RIG-I, IP10, STAT1, MxA, and IFI56K (IFN-inducible protein of 56 kDa), most likely due to the residual type I IFN production observed (Fig. (Fig.6).6). DC infection with WYM, NY, and TX WT strains also resulted in minimal type I IFN, but the levels of IFN-responsive genes were reduced compared to PR8 infection of human DCs (Fig. (Fig.6).6). These observations support the idea of a stronger inhibitory property of human influenza A viruses in human DC activation and type I IFN induction by DCs than of the mouse-adapted PR8 influenza A virus laboratory strain. All the human strains of influenza virus tested in our system showed similar levels of inhibition of IFN-inducible genes and other genes involved in DC activation while the laboratory strain PR8 showed weaker inhibition of those genes, possibly as a result of multiple passages and mouse adaptation.
Using TX WT as a representative human influenza A virus to elucidate the differences in PR8 infections of human DCs, infected DCs were analyzed for viral transcription and protein synthesis. qRT-PCR for viral NP and NS1 at 1, 4, 8, and 16 h postinfection (Fig. (Fig.7A)7A) and Western blotting for viral NP and NS1 proteins at 4 and 8 h postinfection (Fig. (Fig.7B)7B) show that these viral genes were more highly expressed in TX WT infections than in PR8 virus-infected DCs. To determine if the absence of NS1 protein would affect viral replication during infection, human DCs were infected with respective NS1 mutants from either background (TX 1-99 for TX WT and deltaNS1 for PR8) in the same experiment and assessed by qRT-PCR and Western blotting. TX 1-99 viral NP gene expression levels were comparable to TX WT at the mRNA and protein levels (Fig. (Fig.7A7A and and2B).2B). TX 1-99 viral NS1 gene expression was different from TX WT, with lower levels of transcription seen at earlier time points postinfection but with higher levels at 16 h postinfection. Nevertheless, expression of the TX 1-99 NS1 protein was barely detectable (Fig. (Fig.7B7B and and2B).2B). When viral replication in deltaNS1 DC infections was analyzed, NP expression mimicked that of PR8 (Fig. 7A and B), and as expected, there was no NS1 mRNA or protein detected. As previously shown by other investigators using permissive cell lines (57), deltaNS1-infected DCs displayed aberrant viral late gene expression, i.e., reduced HA and M1 viral protein production (data not shown). Taken together, this result supports the idea that the NS1 protein is required for proper viral gene expression in human DCs.
The ability of NS1 mutants to stimulate DC activation was first assessed by examining the phosphorylation of transcription factors RelA, IRF3, and c-Jun known to be important for induction of type I IFN (51, 60), which is linked to DC activation (36). Human DCs were infected with PR8, deltaNS1, TX WT, and TX 1-99 viruses at an MOI of 1 and harvested at different times postinfection, and respective lysates were analyzed by Western blotting for RelA phosphorylation at S536, IRF3 phosphorylation at S396, and c-Jun phosphorylation at S73 (Fig. (Fig.8A).8A). As determined before (Fig. (Fig.3A;3A; also data not shown), RelA, IRF3, and c-Jun phosphorylation at 4 and 8 h postinfection was greater in human DCs infected with TX 1-99 than in DCs infected with TX WT (Fig. (Fig.8A).8A). Data shown in Fig. Fig.8A8A indicate that transcription factor activation in human DCs after PR8 and deltaNS1 infections displayed a similar trend of increased phosphorylation in the absence of NS1 (Fig. (Fig.8A).8A). Interestingly, when transcription factor activation by TX 1-99 and deltaNS1 were compared, similar amounts of RelA and IRF3 phosphorylation at 4 and 8 h after TX 1-99 and deltaNS1 infections were seen. However, levels of c-Jun phosphorylation in TX 1-99 infections were greater at 4 and 8 h than in deltaNS1 infections. Furthermore, it was observed, as in previous studies (32), that the NS1 of influenza TX WT allowed minimal activation of transcription factors, more so than PR8 infections. These results suggest faster kinetics of human DC activation in TX 1-99 infections and corroborate previous evidence that NS1 of TX WT does allow the same activation of the type I IFN induction pathway as the NS1 from PR8 (32).
The differences in transcription factor activation were also reflected in type I IFN production by DCs after infection with the TX WT and TX 1-99 influenza viruses (Fig. (Fig.8B).8B). TX 1-99 infection initially resulted in the fastest secretion of IFN-α at 4 h (29.8 ± 0.96 pg/ml) and 8 h (89.5 ± 1.15 pg/ml) postinfection, but these levels were overtaken by IFN-α secretion from human DCs infected with deltaNS1 after 8 h postinfection with 2,785 ± 33.8 pg/ml and with TX 1-99 at 274 ± 1.46 pg/ml at 12 h (Fig. (Fig.8B).8B). Also corroborating previous data for type I IFN transcription in infected human DCs (Fig. (Fig.6),6), TX WT infections caused slightly more type I IFN secretion than PR8 infections: TX WT infection levels were 16.8 ± 0.29 pg/ml at 4 h, 15.0 ± 0.09 pg/ml at 8 h, and 23.7 ± 3.16 pg/ml at 12 h, and PR8 infection levels were below the detection limit at 4 and 8 h but at 23.0 ± 0.37 pg/ml at 12 h.
We then examined the kinetics of STAT1 Y701 phosphorylation as a marker of type I IFN signaling. STAT1 phosphorylation in TX 1-99-infected human DCs was seen as early as 1 h postinfection and also at 4 and 8 h postinfection (Fig. (Fig.8C),8C), whereas STAT1 phosphorylation was detected only from 8 h postinfection with Delta NS1, again supporting the observation of faster kinetics of DC activation and establishment of an antiviral state in TX 1-99 infections. Surprisingly, STAT1 phosphorylation in TX WT infections of human DCs was comparable to that seen in human DCs infected with TX 1-99, and in both cases STAT1 phosphorylation was markedly increased over the minimal level in PR8 infections of human DCs (Fig. (Fig.8C).8C). Here, we show not only that type I IFN signaling in human DCs is highly sensitive, with immediate STAT1 phosphorylation from relatively low IFN-α levels (as in TX WT infections) (Fig. 6A and B and 8B and C), but also that TX WT and PR8 show different potentials of stimulating type I IFN signaling in human DCs.
As an end product of type I IFN signaling and STAT1 phosphorylation, the induction of type I IFN-responsive genes MxA and IP10 was assessed by qRT-PCR (Fig. (Fig.8D).8D). Corresponding to the transcription factor activation, IFN-α secretion and STAT1 activation, there was a rapid induction of MxA, IP10 (Fig. (Fig.8D),8D), RIG-I, and IRF7 (data not shown) in TX 1-99 infections of human DCs. MxA and IP10 induction in deltaNS1 infections lagged behind that of TX 1-99 infections (Fig. (Fig.8D).8D). Remarkably, though the STAT1 phosphorylation in TX WT-infected human DCs was similar to that in TX 1-99-infected DCs (Fig. (Fig.8C),8C), the rate of MxA and IP10 induction was reduced, and induction even became blunted toward later time points (Fig. (Fig.8D).8D). When MxA and IP10 induction levels in TX WT infections of human DCs were compared to those in PR8 infections, gene expression was greater at earlier time points. However, at later time points (8 h onward) (Fig. (Fig.8D),8D), PR8 expression levels of IFN-inducible genes exceeded those from TX WT infections as previously illustrated (Fig. 6A and B). This suggests that although TX WT is “leaky” in its inhibition of type I IFN induction and pathway activation, its ability to limit IFN-responsive gene induction and the establishment of an antiviral state is stronger than the antiviral induction capability of PR8 virus.
Having established that the human strain TX WT is a better inhibitor of the induction of the antiviral state than the mouse-adapted laboratory strain PR8 in human DCs, we sought to investigate what impact this would have on infected DCs to stimulate the adaptive immune response. Human DCs infected with PR8 or TX WT at an MOI of 1 were cocultured with naïve T cells for 2 and 4 days and assessed for IFN-γ expression. Cocultures with human DCs infected with PR8 virus had strikingly greater amounts of IFN-γ secretion than cocultures with TX WT-infected DCs (Fig. (Fig.9A).9A). We then compared the PR8 background viruses to the TX WT background viruses in DC-mediated activation of naïve T cells. As shown in Fig. Fig.9B,9B, PR8 induces more IFN-γ production in DC and naïve CD4 cocultures than TX WT while deltaNS1 induces levels of IFN-γ in DCs comparable to those of TX 1-99 and TX 1-126. We show here for the first time that a wild-type human isolate of influenza A is a better inhibitor of human DC stimulation of adaptive immunity than a wild-type strain of mouse-adapted influenza A virus. These data support the idea that TX WT influenza virus is a better inhibitor of immune responses than PR8, both at the innate and adaptive arms of immunity.
In generating NS1 C-terminal truncation mutants of a human influenza A strain, we were able to ablate the functions of NS1 in that strain, demonstrating its potent inhibition potential. Interestingly, using these human viruses, we were also able to compare the NS1 functions in a laboratory strain of influenza virus, and we have characterized key differences in their manners of inhibition of antiviral responses. Importantly, we have demonstrated from our current study that the human influenza A NS1 protein has multiple capacities to inhibit the innate immune responses to influenza A virus infection, which gives it an advantage to human influenza virus. However, this strong virulence factor is rendered incompetent with C-terminal truncations as these deletions lead to a gross destabilization of NS1 expression. This suggests that the C terminus of human influenza A virus NS1 is important for the stability of NS1 leading to viral pathogenesis, and mutants that contain this type of truncation would be virtual NS1 knockouts and thus potential options for an efficacious vaccine.
We have investigated, for the first time, an ex vivo model of human influenza A virus infection using a human isolate of influenza A virus and human primary cells to better understand the role of the NS1 protein of influenza virus in its pathogenesis and in its efficient replication and evasion of host immunity. Previous studies have demonstrated the importance of the C-terminal region of the influenza A virus NS1 protein during infection by showing that truncating the C-terminal domain of the NS1 results in a decreased viral growth rate and decreased viral NS1, HA, and M protein expression (12, 54, 63, 66). However, our model is unique in its ability to study human NS1 mutant viruses using primary human cells. Using human TX WT virus, we found that this virus efficiently expressed viral proteins in a human lung epithelial cell line (A549), in primary HTBE cells, and in DCs. As DCs support influenza virus infection but do not release infectious viral particles (for mice, see reference 35; data not shown for humans), viral replication was demonstrated in these cells by analyzing Western blots for viral proteins in the present study. Our comparison of TX WT viral replication with the NS1-truncated TX 1-99 and TX 1-126 viruses supports a role for the NS1 protein in enhancing the expression of other viral proteins in relevant human cells. Although TX 1-99 and TX 1-126 appeared to replicate well in the low-IFN environment of 7-day-old eggs, they exhibited diminished viral replication in A549 cells and DCs, as demonstrated by the poor expression of the HA, M1, and NS1 proteins. In HTBE cell infections, most of the infected cells were nonciliated, consistent with the known tropism of human influenza viruses (40). There were no significant differences in the numbers of infected cells during early TX WT or TX 1-126 infections (9.5 h postinfection). However, the number of infected cells was decreased at later times in HTBE infections with TX 1-126 virus (25 h postinfection). These results are consistent with the lack of differences in viral titers between these two viruses at 9 h postinfection. However, at 25 h, more than 50% of the cells on the culture were infected with TX WT, and only approximately 30% were infected with TX 1-126. Poor viral replication can be explained by a possible impairment in the ability of the mutant NS1 protein to dimerize (70), resulting in lost augmentation of viral replication and aberrant IFN antagonistic functions. Nevertheless, the comparable virus titers at early time points of HTBE cell infections and limited viral replication in HTBE cells, A549 cells, and DCs suggest that NS1 C-terminal truncation mutants will replicate enough to provide antigen for immunity but not replicate efficiently to become pathogenic, both of which traits are prerequisites for vaccine design.
We previously showed that the NS1 protein of a mouse-adapted laboratory strain of influenza virus (PR8) has the ability to inhibit the production of IFN and the activation of human DCs, and therefore an influenza PR8 virus lacking the NS1 protein (deltaNS1) is an efficient stimulant of DC activation (15). We now analyzed human DC infections with PR8 influenza virus and human influenza virus strain TX WT. These two viruses displayed very different abilities to replicate and stimulate innate and adaptive immune responses. TX WT virus infections demonstrated a marked increased in production of NP and NS1 mRNA and protein in human DCs compared with PR8, suggesting its enhanced ability of replication in human DCs. There was some activation of the type I IFN induction pathway, secretion of type I IFN, and stimulation of IFN signaling with minimal increases in IFN-responsive gene expression, all evidence for the initiation of an antiviral state in human DCs. However, during later stages of TX WT infections, the expression of IFN-responsive genes was diminished, whereas PR8 infections of human DCs resulted in a gradual increase in IFN-responsive gene expression, which surpassed that in TX WT infections. And this respective inability of the PR8 virus to completely block human DC activation resulted in increased naïve CD4 stimulation. A likely explanation is reflected in a recent study (32) that showed that the NS1 protein of TX WT is more potent at inducing a general host shutoff of gene expression than the NS1 of PR8 virus. The amino acid residues F103 and M106 of NS1 mediated interaction with the cellular factor CPSF, resulting in inhibition of cellular mRNA processing. Since most human influenza strains, like TX WT, bear F103 and M106 residues in their NS1 proteins (Influenza Virus Resource, NCBI), it is possible that NS1-CPSF interactions potentiate the ability of human influenza viruses to overcome the type I IFN system in humans, as demonstrated in this study with reduced IP10, MxA, STAT1, and IFI56K expression in human DCs infected with human influenza A virus compared to human DCs infected with PR8. Nevertheless, it remains to be determined how influenza viruses expressing NS1 proteins that do not interact with CPSF, such as the PR8 virus, are able to replicate to high levels and cause disease in their hosts. Nevertheless, our results underscore the need to study human influenza viruses in human systems to understand the impact during natural infection.
RIG-I is the major sensor for detection of influenza virus infection by DCs and the resultant IFN and cytokine expression in mouse systems (30). Furthermore, NS1 is known to bind to RIG-I-containing complexes, decreasing RIG-I-mediated NF-κB and IRF3 activation (48), which leads to decreased expression of type I IFN and IFN-related genes (22, 42). These lines of evidence taken together suggest that infections with TX WT would show greater inhibition of RIG-I-mediated activation of downstream transcription factors such as NF-κB, IRF3, and c-Jun than infections with attenuated TX 1-99 and TX 1-126 viruses (60). Activation of these proteins was determined by phosphorylation at specific serine residues. Indeed, an increase was observed in phosphorylation of NF-κB/RelA, c-Jun, and IRF3 at S396 in DCs infected with TX 1-99 and TX 1-126 (data not shown) versus those infected with TX WT. There was also inhibition of RelA, IRF3, and c-Jun activation in PR8-infected human DCs. Nevertheless, there was a gradual increase in the production of antiviral genes in those cells. Here, we further demonstrate and support the hypothesis that the NS1 of TX WT inhibits the establishment of an antiviral state not only by expression of diminishing IFN-responsive genes but also by blocking RIG-I signaling in human DC infections, making TX WT more efficient at the evasion of innate and adaptive immunity. These results conflict with those obtained from experiments studying infections with the A/Udorn/72 laboratory influenza virus strain (43), underscoring again the need to conduct studies with human viruses in human cells to gain insights into the mechanisms of regulation of host gene expression by influenza virus in humans.
The augmented DC activation by NS1 mutant viruses was corroborated by increased proinflammatory cytokine gene transcription and protein secretion including type I IFN following TX 1-99 and TX 1-126 infections, which led to the activation of the IFN-responsive genes IP10, IFI56K, and ISG54 as well as RIG-I and STAT1 (data not shown). Additionally, the NS1 protein of a human influenza virus isolate can efficiently block IFN-related genes (antiviral state) in human DCs, contributing to the establishment of infection in humans. Moreover, although the NS1 of TX WT uses multiple mechanisms to inhibit establishment of an antiviral state (32), truncating the C terminus of NS1 efficiently enhances the ability of the virus to activate DCs and allows increased upregulation and secretion of IFN, IFN-related genes, and proinflammatory cytokines (15, 17, 36, 37).
The efficient activation of DCs by the TX 1-99 and TX 1-126 NS1 C-terminal deletion mutants leads to potent stimulation of T cells. This may explain the strong induction of B- and T-cell responses of TX 1-126 virus in nonhuman primates (5). For an efficient induction of adaptive immune responses, DCs must upregulate MHC-I and MHC-II, as well as costimulatory molecules, to the membrane surface (2, 3) for proper antigen presentation and stimulation of NK, T, and B cells. We observed increased MHC-II (data not shown) and CD86 expression in DCs infected with TX 1-99 and TX 1-126 over that of TX WT infections. And this increased CD86 surface expression was primarily in Flu NP+ cells, suggesting that CD86 upregulation is mostly dependent on infection. Furthermore TX 1-99- and TX 1-126-infected DCs stimulated increased IFN-γ secretion from T cells compared to conventional DCs infected with TX WT. The increased IFN-γ suggests a Th1 polarity, which is believed to be beneficial for clearance of influenza virus infections (11, 19, 44, 45). These observations correlate well with the efficacy in protection achieved using NS1 C-terminal truncation mutants from other influenza virus strains in mice and pigs (55, 63, 66).
In this study, we show that NS1 C-terminal deletions that disrupt NS1 functions in a human strain of influenza decrease the rate and efficiency of viral replication, which is concomitant with an impaired ability to prevent molecular and cellular antiviral mechanisms. This strongly supports the idea that such mutant viruses might be used as live attenuated vaccine viruses for human influenza. Live attenuated vaccine viruses must replicate sufficiently to produce viral antigen and stimulate immune response and memory but not inflict serious pathology and disease (6, 45, 49, 50). Strong supporting evidence for demonstrations from our system that NS1 C-terminal deletion mutants are possible live attenuated human influenza vaccines comes from a recent study using macaques (5). In this study, macaques were mock infected, vaccinated with inactivated TX WT virus, or infected with live TX 1-126 virus. Macaques infected with live TX 1-126 acquired humoral and cellular immunity to influenza A virus that was more robust than that acquired with inactivated virus, as demonstrated by higher levels of specific HA antibody production and an increase in the percentages of influenza virus-specific CD4 and CD8 T cells. An added benefit to the use of these mutants as human influenza virus vaccines is that they are generated by reverse genetics techniques, which is predicted to curtail the length of production from the current time of months to weeks (28, 49). Availability of efficient live attenuated influenza virus vaccines would be advantageous in efforts to protect the public against the increasing resistance of influenza A virus to current antiviral treatments, as well as potential influenza A virus pandemics (26, 28, 67).
We thank Richard Cadagan for excellent technical assistance; Alica Solorzano for helpful reagents, suggestions, and discussions; Mike Pazos for technical assistance with flow cytometry; and Sharon Czelusniak for her help in manuscript editing.
This work was partly supported by the NIAID-funded Center for Investigating Viral Immunity and Antagonism (grant U19 AI62623 to T.M., A.F.-S., and A.G.-S.), the Center for Research on Influenza Pathogenesis (grant HHSN266200700010C to A.F.-S. and A.G.-S.), by NIAID grants R01 AI46954 and u01AI070469 to A.G.-S., and by an NCI fellowship to K.H. (1F31CA1132701PI).
Published ahead of print on 29 April 2009.
†Supplemental material for this article may be found at http://jvi.asm.org/.