Methods

VirusesRepresentative viruses NY26 [A/New York/26/2003 (H3N2)] and NY33 [A/New York/33/2004 (H3N2)] (clade A); NY32 [A/New York/32/2003 (H3N2)] and NY198 [A/New York/198/2003 (H3N2)] (clade B); and NY470 [A/New York/470/2004 (H3N2)] (clade B′) were provided by Kirsten St. George at the New York State Department of Health (Albany, New York). Viruses were passaged in MDCK cells (ATCC CCL-34) in Dulbecco’s Modified Eagle’s Medium (DMEM; Quality Biological) containing 1 μg/mL TPCK-Trypsin (Sigma-Aldrich). Viral titers were determined by plaque assay and were expressed as plaque forming units per milliliter (pfu/mL) using standard protocols [25]. A hemagglutination inhibition assay was performed on all viruses using standard methods [25]. Anti-A/Fujian/411/02/H3N2 [Fujian411] ferret reference serum was supplied by Maryna Eichelberger at the US Food and Drug Administration, and the A/NY/312/01/H1N1 [NY312] virus was used as control

Viral growth curvesMDCK cells were infected with 0.01 multiplicity of infection. Plates were incubated for 1 h at 37°C. Cells were washed, and DMEM containing 1 μg/mL TPCK-Trypsin was added. Cells were incubated at 37°C. Every 12 h, 500 μL of supernatant was collected. Titers were determined using standard plaque assays [26]. Mean titers are reported as pfu/mL. Growth curve titrations were performed 3 times for each virus

Animal experimentsAll animal experiments were performed following National Institutes of Health Institutional Animal Care and Use Committee–approved protocols and guidelines in an Animal Biosafety Level 2 facility. Sixteen 5-month-old male ferrets were purchased from Triple F Farms. Animals with no evidence of neutralizing antibodies to A/New Caledonia/20/00/H1N1, A/NY/55/04/H3N2, and B/Jiangsu/10/03 were used. After a 5-day acclimatization period, animals were anesthetized using Ketamine (100 mg/mL; Vedco), Atropine (0.4 mg/mL; Baxter Healthcare), and Xylazine (100 mg/mL; Akom) and were inoculated intranasally with 10⁶ pfu in 1 mL of DMEM. Four ferrets were inoculated with each virus, and 4 were inoculated with DMEM

Body temperature and weight were assessed and a physical examination was performed daily. A clinical score was assigned from 0–5 by a blinded investigator. One point was assigned for each of the following: decreased activity, decreased eating habits or diarrhea, nasal crusting or discharge, active sneezing, and respiratory distress. The score is a sum of the points assigned

On day 4 and day 14 after inoculation, 2 ferrets from each group were humanely euthanized. Whole lungs were harvested, one-half of the lung was frozen for virus titration, and the rest was fixed in 10% neutral-buffered formalin for histopathological analysis. Nasal turbinate tissue was stored

Tissue virus titrationsFrozen tissues were thawed, weighed, and homogenized in sterile cold Leibovitz medium (Gibco L-15; Invitrogen) containing 1 × penicillin, streptomycin, and amphotericin B (Gibco Anti-Anti, Invitrogen). Standard plaque assays were performed [26]. Titers are expressed as pfu per gram of tissue

Histopathological analysisSlides were prepared by American HistoLabs (Gaithersburg, MD), and 5-μm sections of lung tissue were placed on positively charged slides. Slides were deparaffinized and stained with hematoxylin and eosin. The histopathological characteristics of the slides were reviewed by a single pathologist

PlasmidsThe complementary DNA was synthesized as described [27]. Full-length complementary DNA of the PB1, PB2, PA, and NP segments of all viruses were individually cloned into the pCAGGS/MCS protein expression vector [28], supplied by Adolfo García-Sastre at the Mt. Sinai School of Medicine (New York, NY). Three mutant clones were constructed from the NY470 PA using the QuikChange Multi Site Directed Mutagenesis Kit (Stratagene), with 1 clone containing the M62V mutation, 1 clone containing the Q256K mutation, and 1 clone containing both mutations

The green fluorescent protein (GFP) reporter construct contained the GFP open reading frame flanked by a human polymerase I promoter and mouse RNA polymerase I terminator in the pHH21 vector [29]. This construct was obtained from Andrew Mehle at the University of California (Berkeley, CA) [30]

TransfectionsTransfections were performed in human non–small cell lung carcinoma H1299 cells (ATCC CRL-5803) propagated in Roswell Park Memorial Institute 1640 media and 10% fetal bovine serum (Invitrogen). A total of 400 ng of the pHH21-GFP construct and each of the pCAGGS constructs containing 1 of the 4 RNP segments were transfected using the Effectene transfection reagent (Qiagen). Transfections were incubated at 37°C for 48 h. All transfections were performed a minimum of 3 times

Measurement of GFPCells were treated with TrypLE Express (Invitrogen) and analyzed using a FACS Canto II (BD Biosciences). The fluorescein isothiocyanate channel was used to record fluorescence intensity. Data were analyzed using FlowJo, version 8.7.1 (TreeStar)

Statistical analysisDifferences in tissue viral titers and clinical parameters were tested for statistical significance with Student’s t test, and 95% confidence intervals were generated to evaluate differences in mean fluorescence intensity and in vitro wild-type viral titers. Graphpad Prism software was used for analysis

Phylogenetic analysisSequences were downloaded from the National Center for Biotechnology Information Influenza Virus Resource [15]. Alignments and neighbor-joining phylogenetic trees bootstrapped 1000 times were generated using Molecular Evolutionary Genetics Analysis software, version 3.1 [31]

Results

Phylogenetic analysisA phylogenetic tree derived from the concatenated major open reading frames of the 6 “internal” genes of a representative subset of 35 H3N2 influenza A viruses from New York is shown (Figure 1). The genome constellation of the clade A viruses is distinct from the clade B and derivative clade B′ viruses. A small number of amino acid changes in each gene segment distinguish the clade B and B′ viruses (Figure 2). The isolates in each clade share high levels of identity at the amino acid level. Alignments of the concatenated major “internal” open frames (containing 3212 amino acids) show an intraclade percentage identity range of 99.7%–100%. The range of interclade identity between clade A and clade B is 98.6%–98.9%

Evaluation of the coding sequences from the RNP of each virus reveals 10 amino acid changes between NY32 (clade B) and NY470 (clade B′). Only 6 changes are unique to the clade B′ viruses, compared with clade A. The largest number of unique changes was found in PA. One amino acid change in each of the PB1, PB2, and the NP segments, and 2 of the 3 changes (V62M and K256Q) found in the PA segment of the clade B′ viruses are conserved in progeny H3N2 viruses occurring after the 2004–2005 season. The 2 changes found in the PA segment are unique to clade B′ viruses and their descendants. In an alignment of 929 PA sequences from human H3N2 viruses isolated from 1968 through 2002, none contained these changes. In addition, these changes are not observed in the PA of 748 human H1N1 viruses from 1918–2008 or in the 69 available PA sequences from human H2N2 viruses from 1957–1968. In an alignment of 1565 avian influenza PA sequences, a methionine at position 62 was seen in only 1 sequence, and a glutamine at position 256 was not found. The remaining 3 conserved amino acid changes from the PB1, PB2, and NP segments are found as preexisting polymorphisms in multiple human influenza viruses before 2003

Despite the NY470 virus HA bearing changes in the antigenic domains associated with the A/California/7/2004-like drift variant, all 3 viruses showed a strong Fujian-like reactivity, with NY470 and NY26 exhibiting a titer of 1:640, and NY32 exhibiting a titer of 1:1280. The A/NY/312/2001 (H1N1) virus showed no inhibition of hemagglutination

Viral growth curvesViral growth was evaluated. By 12 h, the NY470 (clade B′) virus had grown to a significantly higher titer than the clade A or clade B viruses, maintaining a 1–2 log higher titer at each time point (Figure 3)

Ferret clinical diseaseOne representative clade A and clade B virus was selected along with the clade B′ virus for evaluation in vivo. Signs of clinical illness were observed after 24 h in all ferrets that were inoculated with virus. Weight loss, from 2%-5% by the end of 14 days, was similar in all 3 virus-inoculated groups. Control ferrets showed consistent weight gain (Figure 4A). Body temperatures varied, and no statistically significant difference was observed (data not shown)

Ferrets inoculated with NY470 (clade B′) showed significant illness and on day 4 had a statistically significantly higher clinical score than NY32 (clade B) or NY26 (clade A) virus-inoculated ferrets (P=.049 and P=.003, respectively. These animals had the highest clinical scores after day 2 and exhibited clinical symptoms throughout the 14 days, whereas the other ferrets began to recover by day 8 (Figure 4B). They also exhibited more-severe symptoms, consisting of more-active sneezing and nasal discharge, as well as greater lethargy and dehydration, which prompted the facility veterinarian to provide supportive care consisting of special nutrition and monitoring. NY32 (clade B) virus–inoculated ferrets had a higher clinical score than did NY26 (clade A) virus–inoculated ferrets on day 4 (P=.002); throughout the study, their scores continued to be higher

Virus growth in ferret tissuesThe NY470 (clade B′) virus grew to a 10-fold and 100-fold higher titer than did NY32 (clade B) or NY26 (clade A), respectively (P=.034 and P=.019, respectively) in nasal turbinate tissue. The NY32 virus grew to a statistically significantly higher titer than did NY26 (P=.012) (Figure 5). No virus was detected on day 14 in the nasal turbinates of any animal (data not shown)

NY470 (clade B′) grew from the lung tissue at a 2 log–higher titer than did NY32 (clade B), a difference that was not statistically significant (P=.187). Both NY470 and NY32 grew to a significantly higher titer than the NY26 (clade A) virus, because it failed to grow from the lung tissue obtained on day 4 (Figure 5). No virus was recovered from the lungs of any animal on day 14 (data not shown)

Ferret lung pathologyHistopathological examination of lung tissue on day 4 revealed a similar pattern of pathology in the NY470 (clade B′) and NY32 (clade B) inoculated ferrets. There were focal areas throughout the lung of bronchiolitis, alveolitis, and a mixed inflammatory infiltrate consisting of many neutrophils. The pathology was more widespread and more severe among the NY470-inoculated ferrets than it was among those inoculated with NY32 (Figure 6). The lungs of NY26 (clade A)–inoculated ferrets also showed histopathologic changes, but the pathology was more focal, less widespread, and different in nature, consisting of a mild focal bronchiolitis with no alveolitis and a focal interstitial mixed inflammatory infiltrate; predominantly lymphocytes and histiocytes (Figure 6). The control ferrets displayed no histopathologic abnormalities (Figure 6)

On day 14, the NY470-infected ferret lung tissue showed widespread interstitial fibrosis and remodeling consistent with healing (data not shown). There were areas of vascular thrombosis and alveolar edema. The NY32-infected ferret lung tissues showed a similar pathology of interstitial fibrosis and remodeling but no evidence of thrombosis or alveolar edema. The NY26-infected ferrets exhibited only very small areas of interstitial fibrosis and repair. No histopathologic abnormalities were seen in controls

In vitro RNP assayThe NY470 (clade B′) RNP produced a statistically significantly higher mean fluorescence, compared with that of the clade B or clade A RNP plasmid sets (Figure 7A). The NY470 PA substituted into the NY32 RNP set (with the PB2, PB1, and NP plasmids being derived from NY32) exhibited a marked increase in fluorescence, similar to that for the NY470 RNP (Figure 7B)

Introducing the 2 identified back mutations (M62V and Q256K) in the PA evaluated with wild-type NY470 PB2, PB1, and NP plasmids generated a marked reduction in fluorescence. This was statistically significantly lower than that for the wild-type NY470 RNP (Figure 7B) and overlapped the results from the NY32 (clade B) RNP. Both single back mutation–containing PA segments were evaluated in the assay with the wild-type NY470 PB2, PB1, and NP plasmids, and both generated a decrease in fluorescence, compared with the wild type NY470

Discussion

In this study, 3 contemporary, human, antigenically similar H3N2 influenza viruses varied significantly in their replicative fitness and pathogenicity. The different glycan specificities of the reassortant clade A viruses HA and neuraminidase may have contributed to a reduced fitness [32], but it did not impede this clade from becoming dominant during the 2003–2004 influenza season. The genome-wide selective sweep associated with emergence of the California-like clade B′ viruses in 2004–2005 was associated with the acquisition of a small number of mutations in proteins making up the RNP complex, 2 of which were unique PA gene mutations that have never been identified in avian or human influenza viruses before the 2004–2005 season and were maintained in progeny viruses. In this study, higher growth in MDCK cells in vitro and in H3N2-naive ferrets in vivo suggests that enhanced replicative fitness, in addition to the acquisition of key antigenic changes in HA, may have played a significant role in the selection of these viruses as the dominant clade in 2004–2005 and in future influenza seasons

The observation that the NY470 clade B′ virus grew to a significantly higher peak titer in vitro, and that it reached a high titer earlier and maintained a higher titer throughout the course of the 60-h culture suggests that replicative fitness may have been an important factor in viral evolution. Having acquired a small number of mutations in the RNP complex encoding genes, increased polymerase activity might account for the observed enhanced replicative fitness of the NY470 clade B′ virus

An enhanced replicative fitness of the NY470 clade B′ virus was also demonstrated in vivo, because NY470 clade B′ virus, compared with the other viruses, had higher titers in the upper and lower ferret respiratory tract. The NY470 virus produced more-severe clinical illness and more-severe lung pathology. The added supportive veterinary care required to maintain the nutrition of the NY470-inoculated ferrets due to the severity of their illness may have lessened the difference noted in weight loss and clinical illness. The clade A NY26 virus was only recovered from the upper respiratory tract, correlating with the statistically significantly lower clinical scores seen for these animals

The different and more-severe pathology observed in ferrets that were infected with clade B and B’ virus, compared with that observed in ferrets infected with clade A virus, was unexpected, given that these were co-circulating human influenza A/H3N2 viruses with antigenically similar HA genes. The difference in histopathology directly correlates with the differences in clinical illness observed. Because the ferrets were seronegative for influenza virus and inoculated with viruses that expressed antigenically similar HA proteins, differences in other segments beyond HA must have contributed to the significant differences in pathogenesis and clinical outcomes for these ferrets. The basis for how different genome constellations could induce such a different inflammatory response needs further investigation

Although the clade B–inoculated and clade B′–inoculated ferrets exhibited similar pathology, their clinical course was different, with the NY470 clade B′–inoculated ferrets exhibiting a greater number of significant episodes of active sneezing and more nasal discharge for a longer time period, possibly making this virus more transmissible. This could be one mechanism that allowed the phenotypic differences observed to drive the evolutionary selection of the clade B′ virus in humans (a possibility to be addressed in future studies)

The clade B′ H3N2 viruses arose from clade B viruses without reassortment; thus, a small number of amino acid differences in each of the 8 gene segments made clade B′ unique. Other than the key antigenic changes in HA, the 4 genes of the viral RNP contained the greatest number of unique genotypic differences between the NY470 clade B′ and the NY32 clade B viruses (Figure 2). The role of the 4 gene segments of the RNP complex is highlighted by the NY470 RNP clearly producing higher mean fluorescence than either the clade A or clade B virus RNP’s, demonstrating the enhanced activity of the clade B′ RNP in vitro. We therefore sought to find amino acid changes that could be directly correlated with the enhanced viral clade B′ replication and phenotypic differences. Of the coding differences in the RNP between the clade B and clade B′ viruses (Figure 1), the biological importance of 2 amino acid changes (V62M and K256Q in the PA segment) is emphasized by the fact that these 2 changes were conserved and unique to the H3N2 strains of this clade and its progeny after the 2004–2005 influenza season. V62M was fixed through 2006, whereas K256Q has remained fixed through 2008. The significance of these changes is further suggested by the observation that the NY470 clade B′ PA segment, when substituted into the NY32 clade B RNP in an in vitro reporter assay, produced a mean fluorescence similar to that of the wild-type NY470 RNP. Although amino acid changes in other segments could have contributed, our data suggest that the 2 mutations may be largely responsible for the differential activity between the clade B and clade B′ RNP. This enhanced RNP function may have given the NY470 clade B′ a replicative fitness advantage and may have played a role in evolutionary selection

Influenza viruses have caused human pandemic and interpandemic outbreaks for centuries [33, 34]. It is clear that influenza viral evolution in humans is not solely influenced by antigenic drift of the HA protein [13]. Further study of how genotypic changes in the internal segments can influence the phenotypic properties of influenza A viruses is therefore important to elucidate the numerous genetic mechanisms by which interpandemic influenza viruses evolve. As we have done in this study, providing biological support and developing a better understanding of new mechanisms of influenza strain evolution and selection that occur, in addition to the long-understood mechanism of HA antigenic drift, may allow for the development of better strategies to mitigate the seasonal burden of influenza A and better prepare for or prevent another pandemic