Structural and Functional Characterization of K339T Substitution Identified in the PB2 Subunit Cap-binding Pocket of Influenza A Virus*

Background: Amino acid changes in PB2 are associated with evolution of influenza virus. Results: K339T substitution in PB2cap reduces the cap binding affinity, polymerase activity, RNA synthesis activity, and murine mortality. Conclusion: Substitution in PB2cap modulates the polymerase activity and virulence by regulating the cap binding activity. Significance: We identified and characterized an emerging K339T substitution in PB2cap. Influenza virus RNA-dependent RNA polymerase is a heterotrimer composed of PA, PB1, and PB2 subunits. RNA-dependent RNA polymerase is required for both transcription and replication of influenza viral RNA taking place in the nucleus of infected cells. A “cap-snatching” mechanism is used to generate a 5′-capped primer for transcription in which the cap-binding domain of PB2 (PB2cap) captures the 5′ cap of the host pre-mRNA. Our statistical analysis of PB2 sequences showed that residue Lys339 located in the cap-binding pocket of H5N1 PB2cap was gradually replaced by Thr339 over the past decade. To understand the role of this amino acid polymorphism, we solved the crystal structures of PB2cap with or without a pre-mRNA cap analog, m7GTP, in the presence of Lys339 or Thr339. The structures showed that Lys339 contributes to binding the γ-phosphate group of m7GTP, and the replacement of Lys339 by Thr eliminates this interaction. Isothermal titration calorimetry analysis showed that Thr339 attenuated the PB2cap cap binding activity in vitro compared with Lys339. Further functional studies confirmed that Thr339-PB2-containing ribonucleoprotein complex has a reduced influenza polymerase activity and RNA synthesis activity, and a reconstituted H5N1 virus containing the Thr339 substitution exhibited a lower virulence to mice but more active replication in Madin-Darby canine kidney cells. The K339T substitution in the cap-binding pocket of PB2 modulates the polymerase activity and virulence by regulating the cap binding activity. It is informative to track variations in the cap-binding pocket of PB2 in surveillance of the evolution and spread of influenza virus.

Both seasonal and pandemic influenza can cause severe illness and death in humans and farm animals. Virus variants emerge frequently and pose a constant health threat to humans. What changes in viral genes will result in viral evolutionary advantages, such as change of virulence and more efficient replication and transmissibility, is an intensive topic. Human infections by avian influenza A virus (H5N1) and the recent pandemic H1N1 influenza A virus 2009 (pH1N1) originating from swine are examples for transmission of animal influenza viruses to humans. Recent reports have shown that an adapted H5N1 virus with mutations in HA and PB2 can be transmitted between ferrets (1,2). And the new substitution S590G/R591Q in PB2 of pH1N1 with the avian signature Glu 627 could rescue the polymerase activity and viral replication (3).
The genome of influenza virus consists of eight ribonucleoprotein complexes that encapsidate eight viral genomic RNA segments. Like other negative-stranded RNA viruses, the viral RNA polymerase of influenza virus is always packaged in the infectious virion as a complex with the nucleoprotein (4 -7). The RNA-dependent RNA polymerase of influenza virus is composed of PA, PB1, and PB2 subunits. The heterotrimeric polymerase is required for RNA transcription and replication that take place in the nucleus during influenza virus infection. Influenza virus accomplishes its transcription by using a capsnatching mechanism (8). After the ribonucleoprotein complexes enter the nucleus, the 5Ј cap of the host pre-mRNA in the nucleus is captured by the cap-binding domain of PB2 (PB2 cap ). The cap, together with 10 -13 nucleotides downstream of the cap, is cleaved off by the N-terminal cap-dependent endonuclease of PA (9,10). This 5Ј-capped oligonucleotide is then used as the primer for initiation of viral transcription by PB1. This process is termed "cap snatching." Residues 318 -483 of PB2 are defined as the cap-binding domain that binds the cap of the host pre-mRNA (11). The C terminus of PB2 (amino acids 535-759), on the other hand, is involved in nucleoprotein binding and nuclear import (12,13), whereas residues 1-37 at the N-terminal region were found to interact with PB1 (14).
The reported complex structure of PB2 cap (A/Victoria/3/ 75(H3N2)) with bound cap analog m 7 GTP shows a 13 residuecontaining pocket that is essential for cap binding (11). His 357 , Phe 404 , Glu 361 , and Lys 376 in the pocket are essential for binding to the 7-methylated guanine base of m 7 GTP. The phosphate groups of m 7 GTP are fixed by Asn 429 , His 432 , Lys 339 , and Arg 355 , and the latter two residues form a positively charged surface on the pocket edge.
It has been reported that amino acid changes in PB2, such as T271A, K627E, and D701N, are related to adaptation of avian influenza virus to humans and virulence (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25). In this study, we found that the residue at position 339 in the cap-binding pocket displays a significant and widespread amino acid polymorphism. Our structural comparison showed that PB2 Lys 339 has a significant conformational change during cap binding. Surface charge analysis showed that substitution with Thr 339 eliminates the interactions with the ␥-phosphate group in m 7 GTP originally made by Lys 339 . Measurements of cap binding by isothermal titration calorimetry (ITC), 3 overall polymerase activity, and specific RNA synthesis activity in mammalian cells confirmed that residue Thr 339 attenuates the cap binding activity. Furthermore, a recombinant H5N1 virus containing Thr 339 showed a lower virulence to mice but more active replication in MDCK cells.
Protein Preparation-All the His tag fusion proteins were overexpressed and purified in Escherichia coli strain Rosetta (DE3). The expressions of the proteins were induced with 0.5 mM isopropyl ␤-D-1-thiogalactopyranoside (Sigma) at 18°C for 16 h. After purification with a nickel affinity HiTrap chelating HP column (GE Healthcare), the proteins were digested with thrombin at room temperature for 0.5 h and then applied to a nickel affinity column again to remove the His tag and undigested proteins. The proteins were further purified by Superdex 75 gel filtration chromatography (GE Healthcare) and desalted in buffer containing 10 mM Tris-HCl, pH 8.0 and 200 mM NaCl with a HiTrap desalting column (GE Healthcare).
Crystallization-The purified proteins were concentrated to 10 mg ml Ϫ1 and incubated with 5 mM m 7 GTP (Sigma) on ice for 1 h. The initial crystallization screening for both native proteins and m 7 GTP-bound proteins from the four subtypes was performed by the sitting drop vapor diffusion method at 20°C using kits including PEG-Ion Screen, Crystal Screen, Crystal Screen 2, Index, and Natrix (Hampton Research). Further crystal optimizations were carried out using the Additive Screen kit (Hampton Research). Finally, appropriate crystals that yielded better diffraction qualities were obtained by the hanging drop vapor diffusion method at 20°C in the following conditions. Native PR8-H1N1 PB2 cap was from 0. Data Collection and Structure Determination-X-ray diffraction data were collected on beamline BL17U at Shanghai Synchrotron Radiation Facility (China) with a wavelength of 0.9795 Å. The crystals of 2005-H5N1 PB2 cap with m 7 GTP and native PR8-H1N1 PB2 cap were flash cooled and maintained at 100 K in cooled nitrogen gas during data collection. In the case of 1968-H3N2 PB2 cap with m 7 GTP, 20% (v/v) glycerol was used as the cryoprotectant. The data were processed with HKL2000 and solved by the molecular replacement method using the CCP4 suite with the x-ray structure of influenza A/Victoria/3/ 1975(H3N2) PB2 cap (Protein Data Bank code 2VQZ) as a search model. The 424-loop of native PR8-H1N1 PB2 cap was rebuilt by ARP/wARP. Structure refinements of the three structures were performed by Coot (28), REFMAC5 (29), and TLS motion (30). The detailed data collection statistics are presented in supplemental Table S1.
ITC Analysis-ITC measurements were performed at 20°C with an ITC200 titration calorimeter (MicroCal Inc.). The wildtype and mutant PB2 cap protein samples were purified as described above and dialyzed with buffer containing 150 mM NaCl and 10 mM HEPES, pH 7.4. 180 -200 M protein in the cell was titrated with 2 mM m 7 GTP in the syringe, and 1 l was injected followed by 29 injections of 2 l. Data were fitted to a single binding site model and analyzed to obtain the parameter K a using the Origin 7.0 program.
Polymerase Activity Assay-Polymerase activity of the ribonucleoprotein complex was detected and quantified using Gaussia luciferase (Gluc) system as described previously (31). The reporter plasmid polI-Gluc consisting of the Gluc ORF flanked by the noncoding regions of influenza NS segment was co-transfected in human A549 cells with PB1, PB2, PA, and NP that were cloned, respectively, into the bidirectional expression plasmid pHW2000. Cells transfected with the reporter plasmid and pHW2000 empty vector were used as the negative control. The polymerase activity of 2005-H5N1 ribonucleoprotein complex was also detected in DF-1 cells at 39°C. Gluc activity in supernatants was analyzed 24 h post-transfection using the Bio-Lux Gaussia luciferase assay kit (New England Biolabs). All the experiments were performed in triplicate.
Primer Extension Assay-The recombinant plasmids pcDNA-PA, pcDNA-PB1, pcDNA-NP, and pPOLI-NA-RT were co-transfected with pcDNA-PB2 containing Thr 339 or Lys 339 into human kidney 293T cells for RNA synthesis. The total RNA was isolated using TRIzol reagent (Kangwei) at 12 and 24 h post-transfection. Two NA gene-specific primers (32) and a 5S rRNA primer (33) were labeled with [␥-32 P]ATP (PerkinElmer Life Sciences) for reverse transcription. The primer extension assays were performed at 42°C for 1.5 h using the Gold-Script cDNA synthesis kit (Invitrogen). The products were analyzed in 7% polyacrylamide gels containing 7 M urea and detected by autoradiography. The lanes corresponding to vRNA, mRNA, cRNA, and 5 S rRNA were quantified with ImageQuant software (GE Healthcare).

Generation of Recombinant Viruses and Virus
Titration-Recombinant 2005-H5N1 viruses containing Lys 339 and Thr 339 were generated by reverse genetics as described previously (26). pHW2000 plasmids containing eight influenza gene segments were co-transfected into a 293T/MDCK co-culture monolayer. The propagated viruses were ascertained by sequencing, and viral titers were determined by TCID 50 assay on MDCK cells. All experiments associated with live virus were performed in an approved biosafety level 3 facility of the Beijing Institute of Microbiology and Epidemiology.
Virulence in Mice-Six-week-old female BALB/c mice were purchased from the Beijing Animal Center (Beijing, China). The mice were anesthetized with isoflurane and intranasally inoculated with PBS (25 l) containing 10-fold serial dilutions of recombinant viruses (1 ϫ 10 2 , 1 ϫ 10 3 , and 1 ϫ 10 4 TCID 50 ) using four mice per dilution. 25 l of PBS was used to infect the negative control mice. The mortality of mice was monitored continuously for 14 days. All experiments associated with live virus were performed in an approved biosafety level 3 facility of the Beijing Institute of Microbiology and Epidemiology.
Viral Growth Kinetics-MDCK cells were infected with the indicated viruses at a multiplicity of infection of 0.0001 for 1 h and then incubated at 37°C in 5% CO 2 . Supernatants were harvested at 12, 24, 36, and 48 h postinoculation for virus titer determination by plaque assay in MDCK cells.
Ethics statement-The mouse studies were approved by the Institutional Animal Care and Use Committee (approval num-ber 20120017) of the Beijing Institute of Microbiology and Epidemiology. Studies were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the State Key Laboratory of Pathogen and Biosecurity, Beijing Institute of Microbiology and Epidemiology (protocol number LAC008), which acknowledged and accepted both the legal and ethical responsibility for the animals, as specified in the Regulations on the Administration of Laboratory Animals under the jurisdiction of the Ministry of Science and Technology of China.

RESULTS
Statistical Analysis of Residues in the Cap-binding Pocket of PB2-The previously reported structure of the PB2 cap bound with m 7 GTP shows that 13 key residues are involved in binding m 7 GTP (11) (supplemental Fig. S1). We performed a statistical analysis of the 13 residues using a total of 9246 sequences with unambiguous host annotation ( In H1N1, H3N2, H9N2, and influenza B virus, residue 339 is predominantly a Lys. However, this location frequently has a Gly (57.1%) in avian H7N2 ( Table 1). The most noticeable variation at position 339 was found in avian and human-isolated H5N1. Lys 339 was changed to Thr in 57.3 and 48.9% isolates, respectively, over the past decade (Fig. 1A). When the occupancy of Lys and Thr was calculated for a total of 1017 sequences of avian H5N1, the rate of Lys 339 presence decreased, whereas that of Thr 339 increased from years 2003 to 2011 (Fig.  1B). In 2003, Thr 339 was detected for the first time in avian H5N1. Rapidly, the rate of Thr 339 reached 86.7% in the isolates of 2006. And from 2006 to 2011, Thr 339 stably remained at the rate of ϳ80% of the avian H5N1 isolates. Correlated changes were also observed in the human-isolated H5N1 sequences (Fig. 1C). A similar pattern was found in H7N2, another avian influenza A virus (supplemental Fig. S2A). In this case, Lys 339 was gradually replaced by Gly 339 from 2004 to 2007. Other amino acid substitutions were identified in minor cases.
Besides residue 339, an amino acid polymorphism at residue 355 was recognized in human H1N1 ( Table 1). The PB2 sequences of H1N1 were first grouped into five periods: years 1918 -1977, 1978 -1991, 1995-1999, 2000 -2008, and 2009 -2011. During the first four periods, residue 355 was changed successively from Arg to Lys, Asp, or Thr (supplemental Fig.  S2B). Between 2009 and 2011, the majority of the sequences in the database corresponded to pH1N1, and this position reverts to Arg 355 . This observation is consistent with the notion that PB2 of pH1N1 is derived from a relatively ancient avian-isolated H1N1 strain containing Arg 355 during the reassortment event (34 -36).
Conformational Differences between the Apo-and m 7 (Fig. 2B), the 1968-H3N2 PB2 cap with m 7 GTP at 1.95-Å resolution ( Fig. 2A), and the 2005-H5N1 PB2 cap with m 7 GTP at 1.80-Å resolution (Fig. 3A). All the structures were solved by the molecular replacement method using the 2.3-Å x-ray structure of influenza A/Victoria/3/ 1975(H3N2) PB2 cap with m 7 GTP (Protein Data Bank code 2VQZ) as a search model (11). Data collection, phasing, and refinement statistics are listed in supplemental Table S1.
In 2005-H5N1 PB2 cap , the residue at position 339 is Thr, which replaced the positively charged basic residue Lys 339 that interacts with the ␥-phosphate group found in other PB2s (Fig.  3, A and B). The side chain of Thr is shorter than that of Lys and uncharged, resulting in disorder of the ␥-phosphate group. The positively charged electrostatic surface formed by Lys 339 and Arg 355 is also disrupted by Thr 339 (Fig. 3C). In the structure of 2005-H5N1 PB2 cap with m 7 GTP, two conformational forms of m 7 GTP are present in four independent molecules in the crystallographic asymmetric unit. In molecule A, the ␥-phosphate group of m 7 GTP is rotated by 90°and interacts with residue Asn 429 with which ␣and ␤-phosphate groups also interact (Fig. 3, A and B). In the other three molecules, the ␥-phosphate groups assume the same conformation as that of m 7 GTP bound in 1968-H3N2 PB2 cap that has Lys 339 .
A structural model of 1968-H3N2 PB2 cap with Thr 355 was built from the structure of 1968-H3N2 PB2 cap with m 7 GTP ( Fig. 3C). Similar to the disruption by Thr 339 in 2005-H5N1 PB2 cap , the positively charged surface formed by Lys 339 and Arg 355 is disrupted by Thr 355 . Interestingly, when Arg is gradually replaced by Lys, Asn, and Thr at position 355, the side chain becomes shorter and shorter and lesser charged (supplemental Fig. S2B). It is worth mentioning that in influenza B virus a conserved residue equivalent to residue 355 in PB2 of influenza A virus is Gly 357 , which has no side chain (Table 1). Similar to replacing Lys 339 by Thr 339 in 2005-H5N1 PB2 cap , substitutions of Arg 355 by other residues may reduce interactions with the ␥-phosphate group of m 7 GTP.
Thr 339 -PB2 cap Has a Reduced Cap Binding Affinity in Comparison with Lys 339 -PB2 cap -The structural analyses suggest that the change from Lys to Thr at position 339 weakens interactions with the cap by PB2. To investigate whether the change indeed has an effect on the cap binding activity, the m 7 GTP binding activity by Thr 339 -PB2 cap and Lys 339 -PB2 cap as represented by 2005-H5N1, PR8-H1N1, and 2009-pH1N1, respectively, was measured by ITC assays at 20°C at pH 7.4. The results showed that the change of Lys 339 to Thr reduced the binding affinity to m 7 GTP. The derived association constant (K a ) of m 7 GTP to Lys 339 -PB2 cap is at least 2-fold higher than that to Thr 339 -PB2 cap (Fig. 4). These observations confirmed that the polymorphism at position 339 correlates with the PB2 cap binding affinity and that Thr 339 reduces the in vitro m 7 GTP binding affinity of PB2 cap .
The binding affinity of Arg 355 -PB2 cap to m 7 GTP was also compared with Thr 355 -PB2 cap within the human H1N1 sub-type. As expected, the results showed that Arg 355 -PB2 cap also has a 2-fold higher m 7 GTP binding affinity than Thr355-PB2 cap (supplemental Fig. S3).
Thr 339 -PB2 cap Shows a Reduced Polymerase Activity and RNA Synthesis Activity in Comparison with Lys 339 -PB2 cap in Mammalian Cells-The H5N1 virus A/bar-headed goose/ Qinghai/15c/2005 is the first highly pathogenic avian influenza virus isolated from migratory waterfowl in Qinghai Lake, West China in 2005. This strain contains Thr at PB2 position 339 and a mammalian adaptive marker of Lys at PB2 position 627. An A/bar-headed goose/Qinghai/15c/2005-like virus caused an outbreak in aquatic birds initiated in 2005 and then was transmitted into Africa, resulting in large scale outbreaks in poultry in Nigeria. As a species-specific signature, Glu 627 generally predominates in avian viruses, whereas Lys 627 predominates in human viruses, and the avian signature Glu 627 in PB2 attenuates polymerase activity (37,38). However, the 2009 pandemic H1N1 isolates and approximately two-thirds of the human-isolated H5N1 viruses contain the avian-like Glu 627 (3,39).
The polymerase activity of both wild-type Thr 339 -PB2 and mutant Lys 339 -PB2 from A/bar-headed goose/Qinghai/15c/ 2005 (H5N1) was quantitated using a Gluc system in human A549 cells (31). With either wild-type Lys 627 or avian-like Glu 627 , the polymerase activity of Thr 339 -PB2 was only ϳ50% of that of the Lys 339 mutant (Fig. 5A). The polymerase activity was also measured for PR8-H1N1, a human subtype with Lys 339 and Lys 627 in wild-type PB2. Similarly, the mutant Thr 339 -PB2 reduced the polymerase activity by 65% (Fig. 5B). These results demonstrate that replacement of Lys by Thr at position 339 indeed decreases the polymerase activity of influenza virus. We also determined the polymerase activity of 2005-H5N1 in DF-1 cells. Unexpectedly, with either Lys 627 or Glu 627 , the polymerase activity of Thr 339 -PB2 was more active than that of the Lys 339 mutant (supplemental Fig. S4).
Primer extension assays were performed at 12 and 24 h posttransfection in 293T cells to demonstrate whether K339T in PB2 influences viral transcription (Fig. 5, C and D). Quantification of mRNA over vRNA showed that replacement of Lys 339 by Thr in PR8-H1N1 decreased the mRNA level at both time points, which is consistent with the result of polymerase activity analysis.
Thr 339 -virus Has a Reduced Mortality in Mice but More Active Replication in MDCK Cells-To determine the contribution of the polymorphism at amino acid 339 to viral virulence in mammals, recombinant 2005-H5N1 viruses containing Thr 339 or Lys 339 were generated by reverse genetics as described previously (26). Six-week-old female BALB/c mice were anesthetized with isoflurane and intranasally inoculated with 25 l of PBS containing 10-fold serial dilutions of recom-binant viruses (1 ϫ 10 2 , 1 ϫ 10 3 , and 1 ϫ 10 4 TCID 50 ) using four mice per dilution. The negative control was infected with 25 l of PBS. The mortality was monitored continuously for 14 days. The survival rate of the group of mice inoculated with wild-type Thr 339 -virus is significantly higher compared with the Lys 339virus-infected group especially with lower virus titers in initial inoculums (Fig. 6A). This result indicates that wild-type Thr 339 -containing virus is less virulent and induces a lower mortality in mice.
We monitored the growth kinetics of the recombinant 2005-H5N1 viruses containing Thr 339 or Lys 339 in MDCK cells for 48 h. The result showed that wild-type Thr 339 -virus replicated more actively than Lys 339 -virus in MDCK cells especially after 36 h (Fig. 6B).

DISCUSSION
Both natural selection and human intervention can cause mutation in influenza A virus, and an influenza pandemic usually occurs when new virus signatures emerge (3,39). Changes in influenza virus polymerase will alter the polymerase activity and influence viral fitness in certain hosts. Species-specific res- idue 627 in PB2 has been studied extensively and was shown to be related to the polymerase activity, virus replication, transmission, and mortality in mammals. More importantly, the E627K mutation has been frequently observed in avian H5N1 viruses crossing the species barrier and causing human death (37,39). Recent studies found that substituting Glu 627 in PB2 with Lys changes its interaction with importin-␣1 and importin-␣7 as well as the viral nucleoprotein in a species-specific manner (13, 18, 40 -42). Other host range determinants in PB2 such as amino acids at positions 271 (23) and 701 (24,25) are also hypothesized to relieve the restriction and recover the polymerase activity by interacting with cellular factors in new host cells (43).
Here we analyzed the amino acid frequencies by using the 9246 full-length PB2 sequences annotated with human and avian hosts from the NCBI Influenza Virus Resource database. The frequency of the PB2 sequence with Lys or Thr at the 339 position showed a significant trend in isolates from years 2004 to 2007 that have a similar sample size per year. The increase of Thr 339 -possessing isolates may imply that the overall fitness of Thr 339 -H5N1 virus could have been strengthened.
The parallel increase of Thr 339 -PB2-containing H5N1 virus isolated from human infections suggests that the K339T substitution in PB2 does not reduce the probability of human infection by avian H5N1 virus. However, different from the observation in human A549 cells, Thr 339 polymerase showed an increased activity in avian DF-1 cells (Fig. 5A and supplemental  Fig. S4). Whether some host factors function together with the residue at 339 in PB2 remains to be examined.
The residue at 339 in PB2, which is involved in cap snatching, is directly related to transcription. Here we showed that Thr 339 in PB2 reduced mRNA and protein synthesis in mammalian cells. Unexpectedly, Thr 339 -H5N1 virus showed more efficient replication than Lys 339 -virus in MDCK cells. How this residue influences viral replication also needs to be investigated further.
The relative levels of different viral functional groups need to be balanced for optimal fitness in corresponding hosts. On the premise of survival and reproduction, parasites tend to be less virulent to the host for long term coexistence (44,45). For example, T271A in PB2 that was believed to contribute to the 2009 H1N1 influenza pandemic has been shown to enhance viral replication in mammalian cells but does not significantly increase viral pathogenicity in mice (23). The retained substitution K339T in PB2 cap has made the influenza A virus evolve toward lower mortality and more active replication, which actually could benefit viral fitness. In contrast, virulence-enhancing mutations usually show low popularity, for example I504V and H357N in PB2, the latter of which is located in the PB2 cap-binding pocket (21,46).
Previous reports show that Lys 355 and Arg 355 in PB2 are correlated with high pathogenicity in mice, and Gln 355 is correlated with low pathogenicity (47)(48)(49). Here we found that the residue at position 355 in H1N1 PB2 also changed from the basic residue arginine to threonine (supplemental Fig. S2B). Our ITC assays showed that the cap binding activity of R355T also decreased similar to that of K339T (supplemental Fig. S3). This observation is supported by the structure model that showed a similar reduction of the positive charge in the K339T change (Fig. 3C). By analogy, Gly 339 in H7N2 PB2 and Gly 357 in influenza B virus may affect the polymerase activity through the same mechanism. These amino acid changes that occurred in PB2 may also be advantageous to the fitness of viruses in their respective hosts.
In conclusion, we identified that substitution of Lys 339 by Thr attenuates the cap binding affinity, polymerase activity, and mRNA production level in mammalian cells and leads to a lower mortality rate in mice but increases the virus replication. The polymorphism at this position was unveiled with viruses that were isolated from both avian and human hosts without host bias in our statistics. The retained K339T substitution suggests that influenza virus might fit better with this substitution, and K339T substitution could be a new strategy utilized by influenza virus to increase its viral fitness in different hosts.