Search tips
Search criteria 


Logo of jvirolPermissionsJournals.ASM.orgJournalJV ArticleJournal InfoAuthorsReviewers
J Virol. 2011 July; 85(14): 7048–7058.
PMCID: PMC3126612

The Virulence of 1997 H5N1 Influenza Viruses in the Mouse Model Is Increased by Correcting a Defect in Their NS1 Proteins [down-pointing small open triangle]


The NS1 protein of human influenza A viruses binds the 30-kDa subunit of the cleavage and polyadenylation specificity factor (CPSF30), a protein required for 3′ end processing of cellular pre-mRNAs, thereby inhibiting production of beta interferon (IFN-β) mRNA. The NS1 proteins of pathogenic 1997 H5N1 viruses contain the CPSF30-binding site but lack the consensus amino acids at positions 103 and 106, F and M, respectively, that are required for the stabilization of CPSF30 binding, resulting in nonoptimal CPSF30 binding in infected cells. Here we have demonstrated that strengthening CPSF30 binding, by changing positions 103 and 106 in the 1997 H5N1 NS1 protein to the consensus amino acids, results in a remarkable 300-fold increase in the lethality of the virus in mice. Unexpectedly, this increase in virulence is not associated with increased lung pathology but rather is characterized by faster systemic spread of the virus, particularly to the brain, where increased replication and severe pathology occur. This increased spread is associated with increased cytokine and chemokine levels in extrapulmonary tissues. We conclude that strengthening CPSF30 binding by the NS1 protein of 1997 H5N1 viruses enhances virulence in mice by increasing the systemic spread of the virus from the lungs, particularly to the brain.


Highly pathogenic H5N1 influenza A viruses were first transmitted from chickens to humans in 1997 in Hong Kong (5, 52). Following culling of the poultry in Hong Kong, subsequent transmission of H5N1 viruses to humans was not documented until 2003 (13, 45). Since then, H5N1 influenza viruses have spread from Asia to Africa and Europe, resulting in more than 500 human infections, with a case fatality rate of approximately 60% (4, 22, 42, 48, 63). Fortunately, H5N1 viruses have not yet acquired the ability for efficient poultry-to-human or human-to-human transmission (58, 59), but the possibility of efficient human-to-human transmission coupled with high virulence makes these viruses a major public health concern.

The molecular basis for the high virulence of H5N1 viruses that have infected humans remains unclear (9, 16). The presence of multiple basic amino acids adjacent to the hemagglutinin (HA) cleavage site is crucial for virulence because these amino acids allow the HA to be cleaved by ubiquitous intracellular furin-like proteases (16, 19). However, the presence of such a HA cleavage site is not sufficient for lethality in mammalian hosts. Therefore, it is of great importance to identify specific functions of other viral proteins that play critical roles in virulence for humans. Both structural and nonstructural gene products have been reported to contribute to the enhanced replication and virulence of H5N1 in mammalian animal models (3, 39, 49).

The NS1 protein of influenza A viruses is comprised of two functional domains: an N-terminal RNA-binding domain (amino acids 1 to 73), which binds double-stranded RNA, and a C-terminal effector domain (amino acid 74 to the C terminus), which binds several host proteins (14). The NS1 protein plays a major role in countering the innate immune response to influenza viral infection, largely by preventing the interferon (IFN) response. The NS1 protein inhibits the IFN response through at least two mechanisms. In one mechanism, the NS1 protein inhibits the activation of the IRF3 transcription factor, thereby inhibiting the activation of beta IFN (IFN-β) gene transcription and hence the synthesis of IFN-β pre-mRNA (8, 38, 43, 46, 54, 61). Influenza A virus strains that circulate in humans differ markedly in the abilities of their NS1 proteins to inhibit the activation of IRF3 and IFN-β transcription. For example, the NS1 proteins of currently circulating H3N2 strains do not inhibit the activation of IRF3 and IFN-β transcription, whereas the NS1 proteins of currently circulating H1N1 strains do inhibit these activations (28). The NS1 proteins of H5N1 viruses were also found to inhibit these activations. In a second mechanism, the NS1 protein binds the 30-kDa subunit of the cellular cleavage and polyadenylation specificity factor (CPSF30), a protein that is required for 3′ end processing of cellular pre-mRNAs (40). As a consequence of sequestration of CPSF30 by the NS1 protein, most of the large amount of the IFN-β pre-mRNA that is synthesized in cells infected with H3N2 viruses is not processed to form mature IFN-β mRNA, thereby suppressing the IFN response (6, 24, 28, 41, 57). The production of other host antiviral mRNAs would also be expected to be similarly suppressed.

X-ray crystallography identified the CPSF30-binding pocket on the NS1 protein and showed that two highly conserved (>99% in all human influenza A viruses) NS1 amino acids that are outside the binding pocket, F at 103 and M at 106, are needed to stabilize the NS1-CPSF30 complex (6). However, the pathogenic H5N1 viruses isolated in 1997, including influenza A/Hong Kong/483/97 (HK97), which is lethal to chickens, mice, and ferrets and was isolated from a fatal human case, contain nonconsensus amino acids at these two positions, specifically L at 103 and I at 106 (6, 56). Although the HK97 NS1 protein does bind CPSF30 to some extent (6, 27), it does so only weakly, leading to attenuated replication in tissue culture cells (56). Thus, changing L103 to F and I106 to M in the HK97 NS1 protein results in a 20-fold increase in the rate of virus replication in MDCK cells, coupled with a 9-fold decrease in the production of IFN-β mRNA, indicating that the wild-type (wt) HK97 virus is impaired in its ability to suppress IFN-β and presumably other antiviral effectors in the host cell. Consequently, even though the H5N1 NS1 protein apparently inhibits the activation of IRF3 and IFN-β transcription (28), strong CPSF30 binding to the NS1 protein is still required for optimum replication and suppression of the IFN response (56).

Variations in NS1 protein function have previously been linked to altered influenza virus virulence in laboratory animals (20, 23, 31, 33, 50, 51). Because over 98% of highly pathogenic H5N1 viruses analyzed since 2003 possess the consensus F103 and M106 amino acids in the NS1 protein, we sought to understand the role of this motif, and therefore of high-affinity CPSF30 binding, in the virulence of H5N1 viruses. Here we have demonstrated that changing these two amino acids in the HK97 NS1 protein to the consensus amino acids results in a remarkable increase in the virulence of the 1997 H5N1 virus in mice, and we have identified salient features underlying this enhanced virulent phenotype, specifically faster systemic spread of the virus, particularly to the brain, where increased replication and severe pathology occur.


Generation of mutant viruses by reverse genetics.

Wild-type (wt) A/Hong Kong/483/1997 (HK97) (subtype H5N1) and the HK97G2+ mutant were generated using plasmid-based reverse genetics as described previously (3). To generate the HK97G2+ virus, codons 103 and 106 of the NS1 open reading frame were changed from L (TTA) and I (ATT) to F (TTC) and M (ATG), respectively, using standard oligonucleotide mutagenesis methods (1). Plasmid DNA was transfected into cocultured 293T/MDCK cells (18), and recombinant viruses were amplified and titrated by plaque assay on MDCK cells as described previously. The eight genomic RNA segments of the recombinant viruses were sequenced to confirm the presence of the appropriate wild-type or mutant base at each position. All experiments with the recombinant HK97 viruses were performed in compliance with the Institutional Biosafety Committee and NIH Guidelines for Research Involving Recombinant DNA Molecules. Viruses were handled in biosafety level 3 containment at CDC, including enhancements required by the U.S. Department of Agriculture and the Select Agents program (


Eight- to nine-week-old female BALB/c mice (Jackson Laboratories, Bar Harbor, ME) were used in this study. Mice were anesthetized by isoflurane inhalation and inoculated intranasally with 50 μl of virus diluted in phosphate-buffered saline (PBS), pH 7.4, or PBS, pH 7.4, for uninfected controls. All animal studies were conducted according to protocols approved by the CDC Institutional Animal Care and Use Committee.

Determination of MLD50.

Groups of 5 animals were inoculated with 0.01 to 1,000 PFU of virus in 10-fold infectious dose increments (repeated two more times for 10 and 100 PFU and one more time for 0.1, 1, and 1,000 PFU). Mice in each group were weighed and monitored daily for survival (Fig. 1) and clinical signs for 14 days after infection. Animals with neurological signs or with severe weight loss (≥25%) were euthanized, and the event was considered a lethal endpoint. Fifty percent mouse lethal dose (MLD50) values were calculated as described previously (47) using cumulative survival data and expressed in PFU.

Fig. 1.
Survival of mice infected with wt HK97 or mutant HK97G2+ virus. BALB/c mice were infected in groups of 5 with 100 PFU (A), 10 PFU (B), 1 PFU (C), or 0.1 PFU (D) of indicated virus and monitored daily. Values of P were calculated by log-rank (Mantel-Cox) ...

Virus replication in vivo.

To study the kinetics of virus replication in vivo, mice were intranasally inoculated with wt HK97 or HK97G2+ virus (Fig. 2). Lungs, spleen, and brain were collected after infection, frozen on dry ice, and stored at −80°C until further processing. Organs were thawed, homogenized in 1 ml of cold PBS, pH 7.4, using a Magnalyzer system and polystyrene beads, and clarified by centrifugation (2,200 × g) at 4°C. Virus titers were determined by plaque assay in MDCK cells; the limit of detection of the assay was 5 PFU/ml. Whole blood was collected after infection, and coagulation was prevented with EDTA. The blood was fractionated into plasma, white blood cells, and red blood cells by centrifugation over an 18.23% Histodenz cushion at 225 × g. The 50% egg infectious dose (EID50) was determined by inoculating groups of 6 eggs with 10-fold dilutions of each fraction and testing for infection by hemagglutinin assay as previously described (13).

Fig. 2.
Replication kinetics of wt HK97 and HK97G2+ viruses in mice. BALB/c mice were infected intranasally with the indicated dose of virus. *, P < 0.05; **, P < 0.01; ***, P < 0.001. For panels A (n = 4), C (n = 3) and D (n = 3), tissue ...


Lungs and brains from uninfected (n = 4 per group) and infected (n = 5 per group) animals were fixed in 10% formalin, and paraffin-embedded sections were stained with hemotoxylin and eosin. Immunohistochemistry for caspase-3 and terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) staining of lung sections were performed by Cureline Biopathology (San Francisco, CA).

Flow cytometry.

Lung cells were collected and stained for flow cytometry as previously described (30). Mice were euthanized by isoflurane inhalation. To obtain leukocytes and other (small) infiltrated cells, lungs were perfused through the left ventricle with 3 ml of PBS. Lungs were minced and incubated for 30 min at 37°C with Hanks balanced salt solution (HBSS) containing 5% fetal bovine serum 9FBS), 10 mM HEPES, 1 mg/ml collagenase (crude, type 1A; Sigma), and 0.2 mg/ml DNase I (from bovine pancreas; Sigma). The digestion was stopped by addition of EDTA. Cells were then dissociated through a 70-μm-mesh strainer and centrifuged at 225 × g for 20 min over an 18.23% Histodenz (Sigma) cushion. Low-density cells were collected, washed with HBSS with 5% FBS, 10 mM EDTA, and 10 mM HEPES (HBSS-5), and suspended in red blood cell lysing medium (Sigma) for 5 min. Cells were washed, resuspended in HBSS-5, passed through a 35-μm-mesh strainer to remove DNA aggregates, and counted prior to staining. All antibodies were used at the staining saturation concentration. Cells were washed twice with HBSS-5 and then resuspended in flow buffer containing 5% normal mouse serum, 5% normal rat serum, 5% normal hamster serum (Jackson ImmunoResearch), and 1% anti-CD16/CD32 (BD Biosciences). The cells were incubated on ice for 20 min, and then antibodies were added and the cells were incubated for 30 min. Cells were washed twice with HBSS-5 and resuspended in HBSS-5 containing 4% formaldehyde. Cells were analyzed using a BD FACS Canto II flow cytometer (BD Biosciences). Unstained cells and cells stained with each antibody alone were used to compensate the data. The data were analyzed using the FlowJo software program (Tree Star). Cells were gated on forward scatter area/forward scatter height (FSC-A/FSC-H) to remove doublets and forward scatter area/side scatter area (FSC-A/SSC-A) to remove debris (data not shown). The following antibodies were used to stain cells: CD4-phycoerythrin (PE)-Cy7 (RM4-5), CD8α-V450 (53-6.7), CD11b-peridinin chlorophyll protein (PerCP)-Cy5.5 (M1/70), CD11c-allophycocyanin (APC) (HL3), CD19-APC (1D3), CD49b-PE (DX5), Ly6C-PE-Cy7 (AL-21), and Ly6G-PE (1A8) from BD Biosciences and I-A/I-E-eFluor 450 (MS/114.15.2) from eBioscience.

Cytokine measurement.

Lungs, spleen, and brain were homogenized in 1 ml PBS, pH 7.4, using a Magnalyzer system with polystyrene beads. The supernatants were clarified by centrifugation (2,200 × g) at 4°C, aliquoted, and frozen at −80°C. Blood was centrifuged at 2,200 × g, and serum was aliquoted and frozen at −80°C. Samples were thawed and diluted 1:5 with assay diluent, and cytokine levels were determined by mouse alpha IFN (IFN-α) and IFN-β enzyme-linked immunosorbent assay (ELISA) (R&D Systems) and by using a mouse Bio-plex 23-plex cytokine kit (Bio-Rad) according to the manufacturers' instructions.

Expression microarray analysis.

Groups of 4 mice were infected with 100 PFU of wt HK97 or HK97G2+ virus by the intranasal route. Animals were euthanized at 1, 2, 4, and 6 days after infection, and brain, lung, and spleen tissues were collected and immediately frozen at −80°C. Total RNA was isolated using Trizol reagent and the PureLink, Micro-to-Midi total RNA purification system (Invitrogen) according to the manufacturer's manual. The RNAs were pooled from the 4 mice for each virus and for each time point and were used for comparative microarray analysis. We used Nimblegen Mus musculus 12 plex microarrays; duplicate arrays were used for each RNA sample (for each virus, for each time point, and for each organ). Ten micrograms of pooled total RNA was processed and labeled, following the standard NimbleGen protocol. Briefly, RNA was converted into cDNA using the SuperScript II cDNA conversion kit (Invitrogen). Double-stranded cDNA was random-prime labeled with Cy3 nonamers and hybridized to the microarrays for 16 h at 42°C. The arrays were washed, dried, and scanned at 5 μM resolution using a GenePix 4000B microarray scanner (Molecular Devices, Sunnyvale, CA). Data were extracted from scanned images using the Nimblescan 2.2, version 5, software program (Roche Nimblegen). Briefly, TIFF images of the hybridized chips were analyzed, and the data were normalized using the quantile normalization and robust multiarray averaging (RMA) analysis tool in the Nimblescan 2.2, version 5, software package. The normalized data were then imported into the Arraystar software program for further analysis. The mean of the duplicate data was determined for each pair of arrays. The fold change in wt HK97 and HK97G2+ gene expression was then determined relative to that of the mock-infected animal (moderated t test). The data from the replicate arrays for each RNA sample showed minimal differences (r2 > 0.95). Consequently the fold changes for individual mRNAs showed minimal differences between the two replicate arrays. Messenger RNAs from the data set that met the 2-fold change cutoff (P ≤ 0.05) were loaded into the DAVID web server (, and search parameters were set to include all available categories from the three gene ontologies (GOs): biological process, cellular component, and molecular function. This program uses the Fisher exact test to determine significance.

Statistical analysis.

Data were analyzed using the Prism software program (GraphPad). Differences in survival were determined by the log-rank (Mantel-Cox) test. Differences in cell number and cytokine level were tested with two-tailed, unpaired Student's t tests on matched pairs. Differences in the kinetics of virus replication were tested using two-way analysis of variance (ANOVA) with Bonferroni posttests.


HK97G2+ mutant virus is more virulent than wt HK97 in the mouse model.

We used reverse genetics to construct both wild-type A/Hong Kong/483/1997 (HK97) (subtype H5N1) (“wt HK97”) and a mutated version (“HK97G2+”) differing only in the NS1 protein, in which amino acids at positions 103 and 106 in wt HK97 virus were converted from L and I to the consensus F and M, respectively (56). The relative virulences of these viruses were evaluated in mice infected intranasally. Following doses of 0.1 to 100 PFU, HK97G2+-infected mice lost weight more quickly than wt-HK97-infected mice, and most had to be euthanized 6 to 8 days after infection (Fig. 1). The difference in virulence between wt HK97 and HK97G2+ was most obvious at a dose of 1.0 PFU. With the 1.0-PFU dose, all of the HK97G2+-infected mice had to be euthanized by day 8, whereas none of the wt-HK97-infected mice succumbed (Fig. 1C). The median mouse lethal dose (MLD50) was 300-fold lower for HK97G2+ than for wt HK97 (10 PFU for wt HK97 and 0.032 PFU for HK97G2+).

HK97G2+ rapidly disseminates systemically.

We measured the titers of these two viruses in lung, peripheral blood, spleen, and brain (Fig. 2). After intranasal infection with 100 PFU of virus, both wt HK97 and HK97G2+ replicated well in the respiratory tract. However, HK97G2+ reached somewhat higher levels in the lungs than did wt HK97 (ANOVA, P = 0.01), particularly at early times after infection, when there was as much as a 5-fold difference (Fig. 2A). By 2 days after infection, HK97G2+ titers in the spleen were more than 10-fold higher than those of wt HK97 (Fig. 2C), indicating that HK97G2+ spreads systemically much more efficiently and earlier than wt HK97. Similarly, by day 2, HK97G2+ was detected in the brain while wt HK97 was not, and by 4 days postinfection, HK97G2+ titers in the brain were approximately 10-fold higher than those of wt HK97 (Fig. 2D). Regardless of the challenge dose, titers in brain at 4 days postinfection were higher for mice infected with HK97G2+ (Fig. 2E). Consequently, the total amount of HK97G2+ virus at days 2 and 4 in the lung, spleen, and brain exceeds the total amount of the wt HK97 virus, showing that the HK97G2+ virus replicated considerably more rapidly than the wt HK97 virus.

We also measured virus titers in plasma, red blood cells, and white blood cells 24 h after infection. Neither virus was detected in the peripheral white or red blood cell populations (data not shown). However, HK97G2+ was readily detectable in plasma 24 h postinfection at a much higher titer than that of wt HK97 (Fig. 2B), demonstrating increased viremia during HK97G2+ infection. These results indicate that enhancing the ability of wt HK97 NS1 to bind CPSF30 enables the HK97G2+ virus to rapidly and efficiently spread systemically through early, increased viremia, leading to more rapid and efficient infection of the spleen and brain.

HK97G2+ spares the lungs but causes severe brain damage.

Histopathological assessment of lungs from mice infected with 100 PFU of wt HK97 or HK97G2+ virus showed that infection with either virus led to moderate lung inflammation by 2 days postinfection, with some accumulation of neutrophils and exudate in the alveolar spaces. In mice infected with wt HK97, this damage progressed to bronchiolitis and bronchitis with epithelial necrosis and lumen debris accumulation by 4 days in all animals and to bronchopneumonia by 6 days in 4 out of 5 animals (Fig. 3A, left panel). In contrast, lung damage did not progress in mice infected with HK97G2+. Only 3 out of 5 mice displayed evidence of bronchiole or bronchus involvement 6 days after infection (Fig. 3A, right panel), with no indication of bronchopneumonia in any of the mice.

Fig. 3.
Histopathology after infection of mice with wt HK97 and HK97G2+. (A) Hemotoxylin-and-eosin staining of lung sections 6 days after infection with 100 PFU of indicated virus. Magnification, ×10. (B) Caspase-3 staining of lung sections 4 days after ...

Similarly, analysis of apoptosis by caspase-3 staining indicated that while both viruses led to an increase in apoptotic and necrotic cells in the lungs in both the alveolar and airway-associated cells, cell death was less marked in the airways of mice infected with HK97G2+, most notably at 4 days after infection in the terminal bronchioles, bronchioles, and bronchi (Fig. 3B). There were no differences in the proportions of stained type I pneumocytes, macrophages, and lymphocyte aggregates between mice infected with wt HK97 and those infected with HK97G2+.

We also assessed lung inflammation by using flow cytometry to analyze the numbers of innate and adaptive immune cells present in the lung. This analysis was carried out at 1, 4, and 6 days after infection with wt HK97 or HK97G2+ virus. Cells were gated and identified as summarized in Fig. 4 (30). There was no difference in the numbers of immune cells detected 1 day after infection with either virus relative to uninfected controls, except that the number of alveolar macrophages was reduced (Fig. 5F). However, by 4 days after infection with wt HK97 virus, there was a >10-fold increase in the numbers of inflammatory monocytes, exudate macrophages, and neutrophils and an almost 10-fold increase in the number of dendritic cells (Fig. 5A to D). The increase in these cell types after HK97G2+ infection was significantly lower. Notably, by 6 days after infection, there were significantly fewer dendritic cells (Fig. 5B) and exudate macrophages (Fig. 5C) in the lungs of mice infected with HK97G2+. Other cell populations, including constitutive monocytes, natural killer cells, B cells, CD4+ T cells, and CD8+ T cells, increased by at most 3-fold after infection compared to results for uninfected controls, with those found in the lungs of HK97G2+-infected mice generally lower than those for infection with wt HK97 (Fig. 5). In particular, at days 4 and/or 6, alveolar macrophages, B cells, constitutive macrophages, and CD8+ T cells were less abundant in the lungs of HK97G2+-infected mice than in those of wt-HK97-infected mice (Fig. 5). There was no difference in the numbers of NK cells at any time (not shown). The overall smaller amounts of these innate and adaptive immune cells in the lungs of HK97G2+-infected mice are further evidence of decreased lung inflammation after infection with the HK97G2+ virus.

Fig. 4.
Identification of cells in lungs by flow cytometry. Lung cells were isolated, stained, and analyzed by flow cytometry. Using the FlowJo software program, data were gated to remove doublets and debris smaller than cells (not shown). Surface marker staining ...
Fig. 5.
Inflammatory cells recruited to the lungs of mice after infection with wt HK97 and HK97G2+. BALB/c mice in groups of 5 were infected with 100 PFU of virus, including a group of uninfected mice as a control. Cell populations in the lung were identified ...

In contrast, analysis of hemotoxylin-and-eosin-stained brain sections showed that substantially more brain pathology was caused by the HK97G2+ virus than by the wt HK97 virus (Fig. 3C). In wt-HK97-infected mice, there was no evidence of changes to brain tissue until 4 days after infection, and most mice (3 out of 5) showed only mild changes with minimal cell death and multifocal gliosis by 6 days after infection (Fig. 3C, left panel). In contrast, mild changes were already present in the brain by 2 days after infection with HK97G2+, which rapidly progressed to perivascular cuffing, moderate cell death, periventricular infiltration, and meningitis in most mice by 6 days after infection (Fig. 3C, right panel), at which time they had to be euthanized. The extent of histopathological damage in the brain corresponded well to the amount of virus detected in the brain.

Spread of the HK97G2+ virus is associated with increased cytokine and chemokine levels in extrapulmonary tissues.

We measured cytokine and chemokine levels in lung, spleen, serum, and brain after infection of mice with 100 PFU of wt HK97 or HK97G2+, using ELISA and cytometric bead array assays. The levels of IFN-α and IFN-β and cytokines with more than a 5-fold difference between wt HK97 and HK97G2+ are shown in Table 1. In the lungs, both viruses induced (relative to findings for uninfected controls) similar levels of type I and II IFNs and cytokines, including granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF), interleukin 1α (IL-1α), IL-1β, IL-2, IL-3, IL-5, IL-6, IL-9, IL-10, IL-12 (p40), IL-13, IL-17, keratinocyte-derived chemokine (KC), MCP-1, MIP-1α, MIP-1β, and tumor necrosis factor (TNF), reaching peak levels by 2 days after infection (Table 1 and data not shown). In the serum, higher levels of MCP-1, G-CSF, KC, and IL-6 were found at 1 and 2 days after HK97G2+ infection, most of which remained elevated until at least 4 days after infection compared to levels for uninfected animals and mice infected with wt HK97 (Table 1). After infection with either virus, there were higher levels of G-CSF, GM-CSF, IFN-β, gamma IFN (IFN-γ), IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-6, IL-9, IL-12(p70), KC, MCP-1, MIP-1α, MIP-1β, RANTES, and TNF in the spleen relative to levels for uninfected mice (Table 1 and data not shown). However, in the spleen at 2 days postinfection, levels of almost all of these cytokines were markedly higher in mice infected with HK97G2+ than in those infected with wt HK97, consistent with the earlier spread of the HK97G2+ virus to the spleen. By 4 days postinfection, this difference was no longer apparent, and the levels of increased cytokines and chemokines were similar in the spleens of mice infected by the two viruses.

Table 1.
Cytokine protein levels in mice infected with wt HK97 or mutant HK97G2+ virus

The cytokine and chemokine levels in the brain also differed between the two viruses. Only a minor increase in these levels was observed in the brains of wt HK97-infected mice. In contrast, HK97G2+ began to induce an inflammatory response in the brain by 4 days postinfection, at which time increased levels (more than 2-fold compared to those for uninfected mice) of KC, MCP-1, and RANTES were observed in the brains of HK97G2+-infected mice (Table 1 and data not shown).

We also performed microarray analyses of cellular mRNAs in the lungs, spleens, and brains of mice infected with the two viruses to determine whether cellular mRNAs encoding other cytokines and chemokines or for apoptotic proteins were upregulated during infection. Groups of four mice were infected with 100 PFU of either wt HK97 or HK97G2+ virus, and tissues were collected at 1, 2, 4, and 6 days after infection. Another group of four mice was mock infected, and tissues were collected 2 days later. Total RNA was extracted, and the RNAs of each group of four mice were pooled and used for microarray analysis. We compared the mRNA profile of wt HK97-infected mice and HK97G2+-infected mice to the mRNA profile in the mock-infected mice as described in Materials and Methods. Messenger RNAs that increased by 2-fold or greater (P ≤ 0.05) in the tissues of the mice infected with either virus relative to levels for the mock-infected mice were categorized using the DAVID program (7).

Analysis of cellular gene expression in the lungs showed that the levels of many cytokine and chemokine mRNAs were substantially increased in mice infected with either wt HK97 or HK97G2+ virus at 1, 2, 4, and 6 days postinfection relative to levels in mock-infected mice. This analysis detected increases in several (1622) cytokine and chemokine mRNAs at 1 day postinfection. Of these, 5 were at higher levels in the HK97G2+-infected mice (CCL7, CXCL9, CCL12, IL-6, and TNF-α) and 4 were at higher levels in the wt-HK97-infected mice (IL-23p19, IL-24, IL-4, and IL-31) (Table 2). At the other days postinfection, there was no statistically significant difference in the levels of lung cytokine and chemokine mRNAs between the wt HK97- and HK97G2+-infected mice, which is consistent with the results of the direct protein assays for cytokines and chemokines shown in Table 1. In contrast, in the spleen the levels of cytokine and chemokine mRNAs at day 2 postinfection were significantly higher for the HK97G2+-infected mice, also in agreement with the direct protein analysis. Representatives of such mRNAs are shown in Table 2. These mRNAs encode MCP-1 and IFN-γ, which were analyzed in the direct protein assay, and also several cytokines/chemokines that were not analyzed in the direct protein assay (lambda IFN [IFN-λ], CXCL11, CXCL2, CXCL9, CCL4, and CCL12). The difference in the levels of these mRNAs between the two viruses decreased at later days postinfection, consistent with the protein assays of cytokines and chemokines. In addition, the microarray analysis revealed that increased levels of several mRNAs encoding apoptosis-related proteins were induced in the spleens of mice infected by either virus (GzmA, GzmE, Prf1, GzmC, and GzmB).

Table 2.
Activation of cytokine and chemokine mRNAs in various organs of wt HK97 and HK97G2+ virus-infected micea

Even more dramatic differences in the levels of cytokine and chemokine mRNAs between the two viruses occurred in the brain (Table 2). Thus, at 4 days postinfection, the levels of a large group of these mRNAs in the brain were substantially higher in HK97G2+-infected mice than in wt-HK97-infected mice. By 6 days postinfection, a time that was not analyzed in the protein assays, the difference in the levels of these mRNAs increased further, ranging from a 3- to 10-fold difference. Again, these results are consistent with the much earlier and more extensive infection of the brain with the HK97G2+ virus and with the increased brain pathology caused by the HK97G2+ virus.


Pathogenic 1997 H5N1 viruses are lethal for chickens and humans in nature, and in laboratory experiments they are lethal for mice and ferrets (5, 9, 16, 31, 45, 49, 65; this study). These H5N1 viruses were highly virulent despite the fact that their encoded NS1 proteins contain a defect in one of their major functions directed at countering the host antiviral (IFN) response, namely, the binding of CPSF30, which causes the suppression of the production of mature cytoplasmic cellular mRNAs, including IFN-β mRNA and presumably other antiviral mRNAs (6, 24, 28, 41, 57). Strong binding of CPSF30 requires the presence of F and M at positions 103 and 106, respectively, in the NS1 protein (56). Instead of these two amino acids, the NS1 proteins of pathogenic 1997 H5N1 viruses contain L and I, respectively, at positions 103 and 106 and as a consequence bind CPSF30 nonoptimally. Changing these two amino acids in the HK97 NS1 protein to the post-2003 consensus residues in the NS1 proteins of H5N1 viruses (F and M at 103 and 106, respectively) strengthens CPSF30 binding and enhances virus replication in tissue culture (56). In the present study, we demonstrated that changing these two NS1 amino acids to the consensus amino acids leads to a very dramatic (300-fold) increase in the lethality of the virus in mice.

Our results indicate that this enhanced virulence of HK97G2+ is likely due to its earlier and more efficient replication and systemic spread. After intranasal inoculation with HK97G2+ but not wt HK97, virus was readily detected in blood within 24 h of infection. Consistent with this rapid establishment of viremia, HK97G2+ also reached much higher titers in the spleen by 2 days and in the brain by 4 days. The total amount of HK97G2+ virus at days 2 and 4 in the lung, spleen, and brain substantially exceeds the total amount of the wt HK97 virus (Fig. 2), showing that the HK97G2+ virus replicated considerably more rapidly than the wt HK97 virus. However, the titer of the HK97G2+ virus in the lung was only slightly higher than that of the wt HK97 virus. It is likely that this relatively smaller difference in titers reflects the fact that the HK97G2+ virus not only replicates more rapidly in the lungs but also spreads more rapidly from the lungs. Such a rapid dissemination from the lung would explain why the HK97G2+ virus caused less damage and cell death in the lungs and attracted fewer infiltrating inflammatory cells into the lungs than did the wt HK97 virus. In addition, the host cytokine response in the lung was essentially the same as the cytokine response in the lung to wt HK97 virus infection from 2 to 6 days postinfection.

In contrast, the cytokine and chemokine levels in the spleen and brain, as measured by both direct protein assays and microarray analysis, were much higher in HK97G2+-infected mice than in wt-HK97-infected mice. These differences largely mirrored the time course of accumulation of the two viruses in these two organs. On day 2 postinfection, the HK97G2+ virus achieved a considerably higher titer than the wt HK97 virus in the spleen, and the HK97G2+ virus induced higher chemokine and cytokine levels in the spleen on this day. Similarly, the HK97G2+ virus spread to the brain faster than the wt HK97 virus, where it replicated faster, and the chemokine and cytokine levels in the brain at 4 and 6 days postinfection were much larger after HK97G2+ virus infection. We interpret these results to indicate that the increased levels of the cytokines and chemokines produced in the spleen and brain after HK97G2+ virus infection represent mostly the responses of local uninfected cells, including trafficking immune cells, to increased viral loads in these two organs.

Consistent with the more rapid spread and replication of the HK97G2+ virus in the brain, mice infected with HK97G2+ but not with wt HK97 exhibited extensive brain damage by 6 days postinfection, at which point most mice infected with HK97G2+ succumbed to infection. Our results fit a model in which the lung is the site of initial enhanced replication by the HK97G2+ virus, but lethality results from rapid dissemination to other organs, particularly the brain, where severe pathology occurred. We conclude that changing the NS1 amino acids at positions 103 and 106 to F and M, respectively, enables the 1997 H5N1 virus to replicate more rapidly and to spread throughout the body more efficiently, particularly to the brain, dramatically increasing its virulence. wt HK97 also spreads to the brain, albeit much more slowly. Spread of wt HK97 virus to the brain was also observed in previous studies (44, 55). However, a recent study has provided evidence that the wt HK97 virus kills mice by rapid replication in the lungs that overcomes the host immune response (17).

In contrast to the H5N1 HK97 NS1 protein, CPSF30 binding by the NS1 protein of H1N1 viruses appears to be less critical for optimal suppression of the host antiviral response. For example, the NS1 protein of the 2009 H1N1 virus does not bind CPSF30, because the consensus binding site is blocked by other NS1 amino acids. Removal of this block, leading to the establishment of CPSF30 binding, has only a minimal effect on IFN production, virus replication, and mouse virulence (14, 15). An important issue is therefore why strong CPSF30 binding by the NS1 protein of the H5N1 HK97 virus is required for optimal suppression of the host antiviral response. One possibility is that the suppression of the activation of IRF3 and IFN-β transcription by the NS1 protein of H5N1 viruses is actually not as effective as the suppression mediated by the NS1 protein of H1N1 viruses and that this difference has not yet been detected by the methods that have so far been employed. It has already been established that the NS1 proteins of different influenza A virus subtypes differ in their ability to suppress the activation of IRF3 and IFN-β transcription (28). The NS1 proteins of human H2N2 and H3N2 strains do not inhibit the activation of IRF3 and IFN-β transcription, whereas the NS1 proteins of currently circulating H1N1 strains do inhibit these activations (28). Perhaps the NS1 proteins of H1N1 and H5N1 viruses also differ, specifically in the extent to which they inhibit the activation of IRF3 and IFN-β transcription.

It will be important to elucidate the mechanism by which the HK97G2+ virus rapidly disseminates from the lung and enters the brain. For some H5N1 viruses, neurotropic spread via the vagus nerve to the brain has been demonstrated (21, 55). In the present case, an alternate possibility is suggested by our finding that the HK97G2+ virus but not the wt HK97 virus was readily detected in blood within 24 h of infection, indicating that HK97G2+ might increase pulmonary vascular permeability, thereby establishing viremia at very early times of infection. At day 1 postinfection, MCP-1 and TNF-α were increased in HK97G2+-infected lungs compared to levels in wt-HK97-infected lungs. These two cytokines are known to increase pulmonary vascular permeability (29, 32, 37), which has previously been implicated in influenza virus pathogenesis (60). This early viremia would enable HK97G2+ to rapidly seed peripheral organs, such as the spleen, and subsequently spread to the brain, where cytokinemia may also increase the permeability of the blood-brain barrier, whereas the wild-type virus may be restricted to slower neurotropic spread through peripheral nerves.

Previous studies have shown that humans (10, 12, 26), birds (53, 66), mice (2, 34, 36), and ferrets and martens (11, 25, 62) infected with certain strains of H5N1 influenza viruses develop severe brain infection, whereas other viral strains lead to more striking pulmonary damage (26, 34, 35, 53, 64). Our results with the HK97G2+ virus, coupled with a recent study of the wt HK97 virus (17), indicate that a two-amino-acid change in the NS1 protein likely leads to a dramatic change in the site of severe pathology induced by the HK97 virus, from the lung to the brain.


This research was supported in part by NIH grant AI11772 to R.M.K.

We acknowledge Wilina Lim, Government Virus Unit, Queen Mary Hospital, Hong Kong, China, for providing the original H5N1 virus used for reverse genetics and Laura Zambutto for regulatory compliance. We thank the Animal Resources Branch for excellent animal care.

The findings and conclusions in this report are ours and do not necessarily represent the views of the Centers for Disease Control and Prevention or the Agency for Toxic Substances and Disease Registry.


[down-pointing small open triangle]Published ahead of print on 18 May 2011.


1. Ausubel F., et al. 2006. Current protocols in molecular biology. Wiley, Hoboken, NJ
2. Bright R. A., Cho D. S., Rowe T., Katz J. M. 2003. Mechanisms of pathogenicity of influenza A (H5N1) viruses in mice. Avian Dis. 47:1131–1134 [PubMed]
3. Chen H., et al. 2007. Polygenic virulence factors involved in pathogenesis of 1997 Hong Kong H5N1 influenza viruses in mice. Virus Res. 128:159–163 [PubMed]
4. Chen H., et al. 2005. Avian flu: H5N1 virus outbreak in migratory waterfowl. Nature 436:191–192 [PubMed]
5. Claas E. C., et al. 1998. Human influenza A H5N1 virus related to a highly pathogenic avian influenza virus. Lancet 351:472–477 [PubMed]
6. Das K., Ma L.-C., Xiao R., Aramini J., Marklund J., Kuo R.-L., Arnold E., Krug R. M., Montelione G. T. 2008. Structural basis for suppression by influenza A virus of a host antiviral response. Proc. Natl. Acad. Sci. U. S. A. 105:13093–13098 [PubMed]
7. Dennis G., Jr., et al. 2003. DAVID: Database for Annotation, Visualization, and Integrated Discovery. Genome Biol. 4:P3. [PubMed]
8. Gack M. U., et al. 2009. Influenza A virus NS1 targets the ubiquitin ligase TRIM25 to evade recognition by the host viral RNA sensor RIG-I. Cell Host Microbe 5:439–449 [PMC free article] [PubMed]
9. Gambotto A., Barratt-Boyes S. M., de Jong M. D., Neumann G., Kawaoka Y. 2008. Human infection with highly pathogenic H5N1 influenza virus. Lancet 371:1464–1475 [PubMed]
10. Gao R., et al. 2010. A systematic molecular pathology study of a laboratory confirmed H5N1 human case. PLoS One 5:e13315. [PMC free article] [PubMed]
11. Govorkova E. A., et al. 2005. Lethality to ferrets of H5N1 influenza viruses isolated from humans and poultry in 2004. J. Virol. 79:2191–2198 [PMC free article] [PubMed]
12. Gu J., et al. 2007. H5N1 infection of the respiratory tract and beyond: a molecular pathology study. Lancet 370:1137–1145 [PubMed]
13. Guan Y., et al. 2004. H5N1 influenza: a protean pandemic threat. Proc. Natl. Acad. Sci. U. S. A. 101:8156–8161 [PubMed]
14. Hale B. G., Randall R. E., Ortin J., Jackson D. 2008. The multifunctional NS1 protein of influenza A viruses. J. Gen. Virol. 89:2359–2376 [PubMed]
15. Hale B. G., et al. 2010. Inefficient control of host gene expression by the 2009 pandemic H1N1 influenza A virus NS1 protein. J. Virol. 84:6909–6922 [PMC free article] [PubMed]
16. Hatta M., Gao P., Halfmann P., Kawaoka Y. 2001. Molecular basis for high virulence of Hong Kong H5N1 influenza A viruses. Science 293:1840–1842 [PubMed]
17. Hatta Y., et al. 2010. Viral replication rate regulates clinical outcome and CD8 T cell responses during highly pathogenic H5N1 influenza virus infection in mice. PLoS Pathog. 6:e1001139. [PMC free article] [PubMed]
18. Hoffmann E., Neumann G., Kawaoka Y., Hobom G., Webster R. G. 2000. A DNA transfection system for generation of influenza A virus from eight plasmids. Proc. Natl. Acad. Sci. U. S. A. 97:6108–6113 [PubMed]
19. Horimoto T., Kawaoka Y. 1994. Reverse genetics provides direct evidence for a correlation of hemagglutinin cleavability and virulence of an avian influenza A virus. J. Virol. 68:3120–3128 [PMC free article] [PubMed]
20. Imai H., et al. 2010. The HA and NS genes of human H5N1 influenza A virus contribute to high virulence in ferrets. PLoS Pathog. 6:e1001106. [PMC free article] [PubMed]
21. Jang H., et al. 2009. Highly pathogenic H5N1 influenza virus can enter the central nervous system and induce neuroinflammation and neurodegeneration. Proc. Natl. Acad. Sci. U. S. A. 106:14063–14068 [PubMed]
22. Kandun I. N., et al. 2008. Factors associated with case fatality of human H5N1 virus infections in Indonesia: a case series. Lancet 372:744–749 [PubMed]
23. Keiner B., et al. 2010. Intracellular distribution of NS1 correlates with the infectivity and interferon antagonism of an avian influenza virus (H7N1). J. Virol. 84:11858–11865 [PMC free article] [PubMed]
24. Kim M. J., Latham A. G., Krug R. M. 2002. Human influenza viruses activate an interferon-independent transcription of cellular antiviral genes: outcome with influenza A virus is unique. Proc. Natl. Acad. Sci. U. S. A. 99:10096–10101 [PubMed]
25. Klopfleisch R., et al. 2007. Encephalitis in a stone marten (Martes foina) after natural infection with highly pathogenic avian influenza virus subtype H5N1. J. Comp. Pathol. 137:155–159 [PubMed]
26. Korteweg C., Gu J. 2008. Pathology, molecular biology, and pathogenesis of avian influenza A (H5N1) infection in humans. Am. J. Pathol. 172:1155–1170 [PubMed]
27. Kuo R. L., Krug R. M. 2009. Influenza A virus polymerase is an integral component of the CPSF30-NS1A protein complex in infected cells. J. Virol. 83:1611–1616 [PMC free article] [PubMed]
28. Kuo R. L., Zhao C., Malur M., Krug R. M. 2010. Influenza A virus strains that circulate in humans differ in the ability of their NS1 proteins to block the activation of IRF3 and interferon-beta transcription. Virology 408:146–158 [PMC free article] [PubMed]
29. Lee Y. R., et al. 2006. MCP-1, a highly expressed chemokine in dengue haemorrhagic fever/dengue shock syndrome patients, may cause permeability change, possibly through reduced tight junctions of vascular endothelium cells. J. Gen. Virol. 87:3623–3630 [PubMed]
30. Lin K. L., Suzuki Y., Nakano H., Ramsburg E., Gunn M. D. 2008. CCR2+ monocyte-derived dendritic cells and exudate macrophages produce influenza-induced pulmonary immune pathology and mortality. J. Immunol. 180:2562–2572 [PubMed]
31. Lipatov A. S., et al. 2005. Pathogenesis of Hong Kong H5N1 influenza virus NS gene reassortants in mice: the role of cytokines and B- and T-cell responses. J. Gen. Virol. 86:1121–1130 [PubMed]
32. Lv S., et al. 2010. Tumour necrosis factor-alpha affects blood-brain barrier permeability and tight junction-associated occludin in acute liver failure. Liver Int. 30:1198–1210 [PubMed]
33. Ma W., et al. 2010. The NS segment of an H5N1 highly pathogenic avian influenza virus (HPAIV) is sufficient to alter replication efficiency, cell tropism, and host range of an H7N1 HPAIV. J. Virol. 84:2122–2133 [PMC free article] [PubMed]
34. Maines T. R., et al. 2005. Avian influenza (H5N1) viruses isolated from humans in Asia in 2004 exhibit increased virulence in mammals. J. Virol. 79:11788–11800 [PMC free article] [PubMed]
35. Manz B., Matrosovich M., Bovin N., Schwemmle M. 2010. A polymorphism in the hemagglutinin of the human isolate of a highly pathogenic H5N1 influenza virus determines organ tropism in mice. J. Virol. 84:8316–8321 [PMC free article] [PubMed]
36. Mase M., et al. 2005. Characterization of H5N1 influenza A viruses isolated during the 2003–2004 influenza outbreaks in Japan. Virology 332:167–176 [PubMed]
37. Mazzon E., Cuzzocrea S. 2007. Role of TNF-alpha in lung tight junction alteration in mouse model of acute lung inflammation. Respir. Res. 8:75. [PMC free article] [PubMed]
38. Mibayashi M., et al. 2007. Inhibition of retinoic acid-inducible gene I-mediated induction of beta interferon by the NS1 protein of influenza A virus. J. Virol. 81:514–524 [PMC free article] [PubMed]
39. Naffakh N., Tomoiu A., Rameix-Welti M. A., van der Werf S. 2008. Host restriction of avian influenza viruses at the level of the ribonucleoproteins. Annu. Rev. Microbiol. 62:403–424 [PubMed]
40. Nemeroff M. E., Qian X., Krug R. M. 1995. The influenza virus NS1 protein forms multimers in vitro and in vivo. Virology 212:422–428 [PubMed]
41. Noah D. L., Twu K. Y., Krug R. M. 2003. Cellular antiviral responses against influenza A virus are countered at the posttranscriptional level by the viral NS1A protein via its binding to a cellular protein required for the 3′ end processing of cellular pre-mRNAS. Virol. 307:386–395 [PubMed]
42. Olsen B., et al. 2006. Global patterns of influenza A virus in wild birds. Science 312:384–388 [PubMed]
43. Opitz B., et al. 2007. IFNbeta induction by influenza A virus is mediated by RIG-I which is regulated by the viral NS1 protein. Cell Microbiol. 9:930–938 [PubMed]
44. Park C. H., et al. 2002. The invasion routes of neurovirulent A/Hong Kong/483/97 (H5N1) influenza virus into the central nervous system after respiratory infection in mice. Arch. Virol. 147:1425–1436 [PubMed]
45. Peiris J. S., et al. 2004. Re-emergence of fatal human influenza A subtype H5N1 disease. Lancet 363:617–619 [PubMed]
46. Pichlmair A., et al. 2006. RIG-I-mediated antiviral responses to single-stranded RNA bearing 5′-phosphates. Science 314:997–1001 [PubMed]
47. Reed L. J., Muench H. A. 1938. A simple method of estimating fifty per cent endpoints. Am. J. Hyg. (Lond.) 27:493–497
48. Sedyaningsih E. R., et al. 2007. Epidemiology of cases of H5N1 virus infection in Indonesia, July 2005–June 2006. J. Infect. Dis. 196:522–527 [PubMed]
49. Seo S. H., Hoffmann E., Webster R. G. 2002. Lethal H5N1 influenza viruses escape host anti-viral cytokine responses. Nat. Med. 8:950–954 [PubMed]
50. Soubies S. M., et al. 2010. Species-specific contribution of the four C-terminal amino acids of influenza A virus NS1 protein to virulence. J. Virol. 84:6733–6747 [PMC free article] [PubMed]
51. Steidle S., et al. 2010. Glycine 184 in nonstructural protein NS1 determines the virulence of influenza A virus strain PR8 without affecting the host interferon response. J. Virol. 84:12761–12770 [PMC free article] [PubMed]
52. Subbarao K., et al. 1998. Characterization of an avian influenza A (H5N1) virus isolated from a child with a fatal respiratory illness. Science 279:393–396 [PubMed]
53. Swayne D. E. 2007. Understanding the complex pathobiology of high pathogenicity avian influenza viruses in birds. Avian Dis. 51:242–249 [PubMed]
54. Talon J., et al. 2000. Activation of interferon regulatory factor 3 is inhibited by the influenza A virus NS1 protein. J. Virol. 74:7989–7996 [PMC free article] [PubMed]
55. Tanaka H., et al. 2003. Neurotropism of the 1997 Hong Kong H5N1 influenza virus in mice. Vet. Microbiol. 95:1–13 [PubMed]
56. Twu K. Y., Kuo R. L., Marklund J., Krug R. M. 2007. The H5N1 influenza virus NS genes selected after 1998 enhance virus replication in mammalian cells. J. Virol. 81:8112–8121 [PMC free article] [PubMed]
57. Twu K. Y., Noah D. L., Rao P., Kuo R.-L., Krug R. M. 2006. The CPSF30 binding site on the NS1A protein of influenza A virus is a potential antiviral target. J. Virol. 80:3957–3965 [PMC free article] [PubMed]
58. Uyeki T. M., Bresee J. S. 2007. Detecting human-to-human transmission of avian influenza A (H5N1). Emerg. Infect. Dis. 13:1969–1971 [PMC free article] [PubMed]
59. Vong S., et al. 2006. Low frequency of poultry-to-human H5NI virus transmission, southern Cambodia, 2005. Emerg. Infect. Dis. 12:1542–1547 [PMC free article] [PubMed]
60. Wang S., et al. 2010. Influenza virus-cytokine-protease cycle in the pathogenesis of vascular hyperpermeability in severe influenza. J. Infect. Dis. 202:991–1001 [PubMed]
61. Wang X., et al. 2000. Influenza A virus NS1 protein prevents activation of NF-kappaB and induction of alpha/beta interferon. J. Virol. 74:11566–11573 [PMC free article] [PubMed]
62. Wang X., Zhao J., Tang S., Ye Z., Hewlett I. 2010. Viremia associated with fatal outcomes in ferrets infected with avian H5N1 influenza virus. PLoS One 5:e12099. [PMC free article] [PubMed]
63. WHO/CSR 2009. Cumulative number of confirmed human cases of avian influenza A/(H5N1) reported to WHO (by 1 July 2009). WHO, Geneva, Switzerland. table_2009_07_01/en/index.html
64. Zhang Z., et al. 2009. Systemic infection of avian influenza A virus H5N1 subtype in humans. Hum. Pathol. 40:735–739 [PubMed]
65. Zitzow L. A., et al. 2002. Pathogenesis of avian influenza A (H5N1) viruses in ferrets. J. Virol. 76:4420–4429 [PMC free article] [PubMed]
66. Zou W., Yu Z., Zhou H., Tu J., Jin M. 2009. Genetic characterization of an H5N1 avian influenza virus with neurovirulence in ducks. Virus Genes 38:263–268 [PubMed]

Articles from Journal of Virology are provided here courtesy of American Society for Microbiology (ASM)