A Y161F Hemagglutinin Substitution Increases Thermostability and Improves Yields of 2009 H1N1 Influenza A Virus in Cells

ABSTRACT

Vaccination is the primary strategy for influenza prevention and control. However, egg-based vaccines, the predominant production platform, have several disadvantages, including the emergence of viral antigenic variants that can be induced during egg passage. These limitations have prompted the development of cell-based vaccines, which themselves are not without issue. Most importantly, vaccine seed viruses often do not grow efficiently in mammalian cell lines. Here we aimed to identify novel high-yield signatures for influenza viruses in continuous Madin-Darby canine kidney (MDCK) and Vero cells. Using influenza A(H1N1)pdm09 virus as the testing platform and an integrating error-prone PCR-based mutagenesis strategy, we identified a Y161F mutation in hemagglutinin (HA) that not only enhanced the infectivity of the resultant virus by more than 300-fold but also increased its thermostability without changing its original antigenic properties. The vaccine produced from the Y161F mutant fully protected mice against lethal challenge with wild-type A(H1N1)pdm09. Compared with A(H1N1)pdm09, the Y161F mutant had significantly higher avidity for avian-like and human-like receptor analogs. Of note, the introduction of the Y161F mutation into HA of seasonal H3N2 influenza A virus (IAV) and canine H3N8 IAV also increased yields and thermostability in MDCK cells and chicken embryotic eggs. Thus, residue F161 plays an important role in determining viral growth and thermostability, which could be harnessed to optimize IAV vaccine seed viruses.

IMPORTANCE Although a promising complement to current egg-based influenza vaccines, cell-based vaccines have one large challenge: high-yield vaccine seeds for production. In this study, we identified a molecular signature, Y161F, in hemagglutinin (HA) that resulted in increased virus growth in Madin-Darby canine kidney and Vero cells, two cell lines commonly used for influenza vaccine manufacturing. This Y161F mutation not only increased HA thermostability but also enhanced its binding affinity for α2,6- and α2,3-linked Neu5Ac. These results suggest that a vaccine strain bearing the Y161F mutation in HA could potentially increase vaccine yields in mammalian cell culture systems.

INTRODUCTION

Influenza A viruses (IAVs) cause seasonal outbreaks and occasional pandemic outbreaks among humans and pose challenges to public health. The viruses responsible for four pandemics have been characterized: one each in 1918 and 2009 caused by H1N1 IAVs, one in 1957 caused by an H2N2 IAV, and one in 1968 caused by an H3N2 IAV (1). The impacts of these pandemic outbreaks varied, but each outbreak resulted in substantial mortality in a short time. Compared with pandemic outbreaks, seasonal influenza outbreaks are typically mild but still cause approximately 200,000 hospitalizations and 36,000 deaths each year in the United States alone (2).

Vaccination has been the most efficient and economic strategy for preventing influenza virus infection and controlling the spread of disease (3). Three types of virus-based influenza vaccines, inactivated vaccines, live-attenuated vaccines, and recombinant hemagglutinin (HA), are licensed in the United States, with egg-produced vaccine being the dominant source (4, 5). The egg-based platform for vaccine production has been used since the 1950s (6), but it has several disadvantages: first, the passage of seed viruses in eggs can result in undesired egg-adapting mutations in HA that can lead to changes in viral antigenicity (7–10); second, due to reactogenicity concerns, egg-grown vaccines are contraindicated for those with egg allergies (11); and third, rapidly scaling up egg production is not easily achievable. Cell-based vaccine production platforms do not have the same limitations (12). All continuous cell lines, including Madin-Darby canine kidney (MDCK) cells and African green monkey kidney-derived Vero cells, must be certified before being approved by regulatory authorities for use in the production of influenza vaccines (13–15).

A high-yield vaccine seed strain is required for timely vaccine manufacture and is thus a critical component of a successful influenza vaccination program. Unfortunately, it is not uncommon that the vaccine strains recommended by the World Health Organization (WHO) have less-than-desirable yields in eggs, cells, or both (16, 17). For example, the 2009 H1N1 pandemic seed strain was a low-yield strain, and it required almost 3 months for WHO collaborative laboratories and vaccine companies to engineer the selected strain to meet the criteria required for vaccine production. Because of this delay, vaccine-derived immunity among the population arose after the peak of the second wave of the 2009 H1N1 pandemic (18). Therefore, the quick generation of a high-yield vaccine seed virus is critical for rapid vaccine production and, thus, for effective influenza prevention and control. Adaptation of viruses to cells by multiple passages or the development of high-yield reassortant seeds using reverse genetics has been shown to be an effective way to increase the yield of vaccine seed viruses (19–22). In addition, other studies have been performed to improve virus yields in cells by modifying the virus or the cell line. For example, Suphaphiphat et al. (23) showed that mutations S186P and L194I in the receptor binding site (RBS) of A/California/04/09 (CA/04) H1N1 HA increased the growth of the virus by more than 10-fold in MDCK cells, and Hamamoto et al. (24) reported that MDCK cells engineered with a stable knockdown of interferon regulatory factor 7 increased IAV yields.

Thermostability is also important for vaccine quality, especially in low-income countries that lack the infrastructure to maintain a low and stable temperature during vaccine transportation (25). The reduced thermostability of the live attenuated vaccine for the 2009 H1N1 pandemic virus may have been responsible for its restricted replication in vaccinated persons (26). Mutations in the HA protein can improve thermostability; for example, mutations S133N, T189A, N198D, and L226Q in the RBS of influenza virus HA were reported to be associated with a significant increase in the thermostability of an H9 IAV (27).

The objective of this study was to randomly introduce mutations into HA and screen for those that improved the thermostability and yields of IAV in MDCK and Vero cells. Such studies are important to optimize the preparation of a cell-based influenza vaccine.

RESULTS

Generation of RBS variants of CA/04 and assessment of their growth characteristics in cells.

To identify mutations associated with high yields of CA/04-influenza A/Puerto Rico/8/34 (H1N1) (PR8) reassortant viruses, a cDNA library carrying random mutations in the CA/04 HA RBS was generated by error-prone PCR (epPCR) and screened by Sanger sequencing. Here, we use the term rg to refer to mutants from epPCR. For example, rg-Y161F denotes a reassortant virus from epPCR with a Y161F mutation. Each mutated plasmid, together with the neuraminidase plasmid of CA/04 and internal gene (PB2, PB1, PA, NP, M, and NS) plasmids of PR8, was used to rescue virus by a reverse-genetics approach. As a result, we obtained a total of eight mutants: rg-D130E, rg-P140T, rg-L154F-K156Q, rg-S160T, rg-Y161F, rg-K174E, rg-S188I, and rg-Y201H (Table 1).

TABLE 1.

Characterization of MDCK cell-grown receptor binding-site mutants generated by an error-prone PCR-based mutagenesis strategy

Mutationa	HI titerb	Mean TCID50 ± SDc	HA titerd
			Guinea pig	Chicken	Horse	Turkey	Doge
rg-wt	640	5.749	4	8	<2	16	8
P140T	640	5.91 ± 0.29	4	4	<2	32	4
S188I	640	5.91 ± 0.29	4	16	<2	16	4
S160T	320	6.08 ± 0.29	4	2	<2	16	8
Y201H	320	6.08 ± 0.29	16	8	<2	64	16
D130E	320	6.249	16	16	<2	128	32
K174E	320	6.249	16	16	<2	64	32
L154F-K156Q	640	6.249	16	4	<2	128	64
Y161F	640	8.249	256	1,024	<2	512	128

^aViruses that carry mutations at the receptor binding site of wild-type influenza A/California/04/09 (H1N1) virus (CA/04) were generated by using an error-prone-based reverse genetic system. rg-wt, CA/04 mutant.

^bHI, hemagglutination inhibition. Titers were determined by using ferret serum (anti-CA/04).

^cThe virus titers were determined by a 50% tissue culture infectious dose (TCID₅₀) assay in MDCK cells.

^dHA, hemagglutination assays against types of red blood cells were performed by using standard procedures.

^eBeagle.

Analysis of viral growth kinetics showed that the replication efficiencies of these 8 mutants varied greatly in MDCK cells. Among the mutants, rg-D130E, rg-K174E, rg-L154F-K156Q, and rg-Y161F increased virus titers compared with that of the wild-type reassortant virus rg-wt. The rg-S160T, rg-P140T, rg-S188I, and rgY201H mutants grew to titers similar to those of rg-wt (Table 1). Among all mutants, rg-Y161F had the highest virus titer at 8.249 log units (50% tissue culture infectious doses [TCID₅₀]), which was >300-fold higher than that of the wild-type virus in MDCK cells. To assess the stability of the mutation, the rg-Y161F mutant was passaged on MDCK cells three times. The rg-Y161F mutant at the third passage was sequenced to confirm no additional mutations across HA, neuraminidase (NA), and six internal gene segments.

To determine if the mutations at the RBS altered HA antigenicity, we subjected the panel of eight RBS CA/04 mutants to a hemagglutination inhibition (HI) assay, using ferret antisera against CA/04. Both turkey and guinea pig erythrocytes were used in the HI assays, and the results were identical. In summary, the HI titers of four reverse genetic variants, rg-Y161F, rg-L154F-K156Q, rg-S188I, and rg-P140T, were equivalent to that of the rg-wt virus. The remaining four mutants, rg-D130E, rg-K174E, rg-S160T, and rg-Y201H, had 2-fold-lower HI titers than that of rg-wt. Thus, all of the RBS mutants were antigenically similar to the parental CA/04 virus despite the presence of altered virus growth properties. Due to its preferred growth and unaltered antigenic characteristics, rg-Y161F was selected as a candidate vaccine virus for further studies.

In addition, we compared the HI titers of the vaccine candidate rg-Y161F and wild-type CA/04 against a panel of 18 human sera. The results showed that there were no more than 2-fold changes between the HI titers of rg-Y161F and those of the wild-type CA/04 virus (Table 2), confirming that mutation Y161F did not alter the antigenicity of the CA/04 virus.

TABLE 2.

Serological responses of the wild type and the 161F mutant against a panel of human sera using HI assays

Seruma	HI titerb
	rg-wt	rg-Y161F
AJP125	80	40
AJP126	80	80
AJP128	40	40
AJP225	160	160
AJP226	160	160
AJP228	20	20
CAJP251	160	160
CAJP252	160	160
CAJP261	80	80
CAJP262	80	80
CAJP271	80	80
CAJP272	80	40
CAJP281	80	80
CAJP282	160	80
CAJP291	160	160
CAJP292	160	160
CAJP301	80	80
CAJP302	160	80

^aPanel of 18 human sera randomly collected in a vaccine efficacy study.

^bTiters were determined by using turkey red blood cells according to standard procedures.

Growth properties of rg-Y161F.

To evaluate the replication efficiencies of rg-Y161F, we characterized its growth kinetics alongside rg-wt in MDCK and Vero cells. We infected cells with viruses at a multiplicity of infection (MOI) of 0.001 (MDCK cell infection) or 0.01 (Vero cell infection) and determined the growth kinetics of the viruses for up to 96 h in MDCK cells and 120 h in Vero cells. In MDCK cells, the virus titers of rg-Y161F reached 10^8.66 TCID₅₀/ml 72 h after infection, >300-fold higher than the highest virus titer from rg-wt (Fig. 1A). In Vero cells, the titers of rg-Y161F reached 10^8.25 TCID₅₀/ml 96 h after infection, a titer which is >100-fold higher than the highest virus titer from rg-wt (Fig. 1B). The total viral protein of rg-Y161F in cells reached a mean titer of 1,236.2 μg/ml, which is 2.07-fold higher than that of wild-type CA/04 (P = 0.008) (Fig. 1C). These results suggest that mutation Y161F facilitates the replication efficiency of CA/04 in MDCK and Vero cells.

FIG 1.

(A and B) Growth properties of wild-type (WT) and Y161F mutant viruses in Madin-Darby canine kidney cells (A) and Vero cells (B). Each data point represents the mean virus yields (log₁₀ TCID₅₀ per milliliter) from three individually infected wells ± standard deviations. (C and D) Total protein of viruses propagated in Madin-Darby canine kidney cells (C) and 10-day-old embryonated chicken eggs (D). Viruses were purified from the cell supernatant or allantoic fluid by low-speed clarification and then subjected to sucrose density gradient centrifugation. The virus band was collected and purified through a cushion of 30% sucrose. The virus pellet was resuspended in 200 μl of phosphate-buffered saline, and the total amount of purified virion proteins was determined by using a Pierce BCA protein assay kit (Thermo Scientific, Rockford, IL).

Impact of HA RBS mutations on virus binding to erythrocytes.

We next wanted to explore possible mechanisms for the increased yields of rg-Y161F by examining its interaction with host receptors. Due to their unique glycan receptor profiles (i.e., types and distributions of alpha-2,3-linked sialic acid on galactose [SA2,3GA] and alpha-2,3-linked sialic acid on galactose [SA2,6GA]), erythrocytes from various hosts have often been used to characterize receptor binding properties for influenza viruses through hemagglutination assays (HA assays). We used erythrocytes from guinea pig, chicken, horse, turkey, and dog (beagle) to compare the glycan profiles of the full panel of eight mutants. As shown in Table 1, all mutants and the wild-type (rg-wt) virus agglutinated erythrocytes from guinea pig, chicken, turkey, and beagle to different extents, but they did not agglutinate those from horse. The eight mutants could be separated into three groups: (i) those with increased HA titers against guinea pig, chicken, turkey, and beagle erythrocytes (rg-D130E, rg-K174E, and rg-Y161F mutants); (ii) those with a hemagglutination pattern similar to that for wild-type virus (rg-S160T, rg-P140T, and rgS188I mutants); and (iii) those that had increased HA titers against guinea pig, turkey, and beagle erythrocytes but no change in HA titers against chicken erythrocytes (rg-L154F-K156Q and rg-Y201H mutants). Among the eight mutants, it was striking that rg-Y161F had the highest HA titers (range, 128 to 1,024 HA units [HAU]) for erythrocytes from guinea pig, chicken, turkey, and beagle.

Effect of the Y161F mutation on the receptor binding.

To further explore the molecular mechanisms of the rg-Y161F high-growth-yield phenotype, we characterized its receptor binding profile by using an N-linked glycan isoform microarray. The microarray consisted of 66 chemoenzymatically synthesized and purified N-glycans (28) (see Fig. S1 and Table S1 in the supplemental material). As shown in Fig. 2, both rg-wt and rg-Y161F bound predominantly to N-glycans terminating with Neu5Ac. The rg-wt virus showed binding to α2,6-Neu5Ac-linked glycans (N0x3, N113, NN144, N213, N223, and N244), whereas the Y161F mutant showed a preference for binding to α2,3-Neu5Ac-linked glycans (N0x2, N112, N122, N134, N212, N222, N234, N0x5, N115, N125, N155, N215, N225, and N255) and relatively weaker binding to α2,6-Neu5Ac-linked glycans. Neither rg-wt or rg-Y161F was observed with noticeable binding toward α2,6-Neu5Gc-terminated glycans (N013G and N023G) or α2,3-Neu5Gc-terminated glycans (N012G, N022G, N015G, and N025G).

FIG 2.

Receptor binding specificity of wild-type and Y161F hemagglutinin 1 mutant viruses analyzed by glycan microarray analysis. A to D represent different categories of glycans: A, alpha-2,3-linked sialic acid glycans; B, alpha-2,6-linked sialic acid glycans; C, alpha-2,3- and alpha-2,6-linked sialic acid glycans; D, non-sialic-acid glycans. Vertical bars denote the fluorescence binding signal intensity. The signal for the marker on the array was derived from the reaction between the anti-human IgG antibody Cy3 (0.01 mg/ml) and the anti-human IgG antibody Alexa 647 (0.01 mg/ml), which served as a positive control for the glycan microarray.

To confirm the binding profiles revealed by a glycan array, we performed glycan binding assays to characterize the dynamics and avidity of virus binding to two glycan analogs, 3′-sialyl-N-acetyllactosamine (3′SLN) and 6′-sialyl-N-acetyllactosamine (6′SLN). The representative binding plots for each glycan analog are shown in Fig. 3A and B. Both rg-wt and rg-Y161F showed strong binding to 6′SLN, although rg-Y161F exhibited a relatively weaker binding affinity than that of rg-wt (<1.2-fold) (Fig. 3B). Conversely, rg-Y161F showed much stronger binding to 3′SLN, whereas rg-wt had no detectable binding with the same 3′SLN concentration range tested (Fig. 3D). These results confirmed the glycan array profiles showing that the Y161F mutation in HA dramatically increased the binding affinity for 3′SLN while retaining a strong binding affinity for 6′SLN.

FIG 3.

Glycan binding specificity of virus by biolayer interferometry (FortéBio, Menlo Park, CA). (A and B) Representative binding curve of wild-type H1N1 CA/04 and the Y161F mutant toward the alpha-2,6-linked sialic acid (SA2,6GA) and alpha-2,3-linked sialic acid (SA2,3GA) receptors, respectively. The streptavidin-coated biosensors were first preloaded with biotin-labeled sialic acid receptors, followed by virus binding at 1 pM each for 1,200 s in a standard kinetic buffer with neuraminidase inhibitors (zanamivir hydrate and oseltamivir phosphate). (C and D) Sialic acid receptor concentrations were titrated from 0.1 to 5 μg/ml (SA2,6GA) or 0.2 to 5 μg/ml (SA2,3GA) during loading with the biotin-labeled receptors. The experiments under each individual concentration were performed three times independently, and the binding response unit (nanometers) was recorded at the 1,196-s time point (4 s before the start of dissociation). (E and F) Three-dimensional structures of hemagglutinin of the wild-type (HA-WT) influenza A/California/04/09 (H1N1) virus and the mutant virus (HA-161F) in contact with the avian-like receptor analog carbohydrate 3′-sialyl-N-acetyllactosamine (3′SLN) (E) and the human-like receptor avian 6′-sialyl-N-acetyllactosamine (6′SLN) (F). (G) Calculated PoseScores for viruses.

Structural mechanism of increased rg-Y161F binding to 3′SLN and 6′SLN.

Crystal structure modeling was performed to characterize the effect of the Y161F mutation on the binding affinity between HA and two testing glycan analogs, 3′SLN (Fig. 3E) and 6′SLN (Fig. 3F). For 3′SLN, the PoseScore for wild-type CA/04 HA was 5.99, whereas that for Y161F mutant HA was −6.63. For 6′SLN, the PoseScores for wild-type CA/04 HA and Y161F mutant HA were 2.06 and −11.89, respectively (Fig. 3G). These results suggest that the Y161F mutation leads to increased binding of CA/04 to 3′SLN and 6′SLN.

Effect of the Y161F mutation on the replication of other IAVs.

To understand the naturally occurring molecular polymorphisms at residue 161 of HA, we compared a total of 59,016 HA sequences covering all 18 documented HA subtypes (H1 to H18) of IAVs. These results showed that Y161 is conserved in the H1 to H5, H8, H9, H11, H13, H14, and H16 IAV subtypes, whereas F161 is conserved in the H7, H10, H12, and H15 IAV subtypes (Fig. 4A). Phylogenetically, H7, H10, and H15 are group 2 HAs, and H12 is a group 1 HA (Fig. 4B).

FIG 4.

(A) Sequence alignment of hemagglutinin 1 (HA1) from 18 different influenza A virus HA subtypes (H1 to H18). Residues 161 (H3 numbering) are indicated by a vertical rectangle. (B) Y161 is conserved in subtypes H1 to H5, H8, H9, H11, H13, H14, and H16. F161 is conserved in subtypes H7, H10, H12, and H15. Each residue is numbered according to the H3 HA numbering. (C and D) Growth properties of canine subtype H3N8 influenza virus (cH3N8) and its hemagglutinin 1 (HA1) F161 mutant virus (cH3N8 Y161F) (C) and of the influenza A/Puerto Rico/8/34 (H1N1) virus (PR8) and its HA1 F161 mutant virus (PR8 Y161F) (D) in Madin-Darby canine kidney cells. Each data point represents the mean virus yield (log₁₀ TCID₅₀ per milliliter) from three individually infected wells ± standard deviations. (E) Effect of the 161F mutation (H1N1 161) on the thermostability of the influenza A/California/04/09 (H1N1) virus (H1N1 WT). (F) Effect of the 161F mutation (H3N8 161F) on the thermostability of the influenza A/canine/Iowa/13628/2005 (H3N8) virus. The viruses with equal HA titers were incubated at the indicated temperatures for 40 min, and titers were then determined.

To test whether the Y161F mutation would increase growth yields in IAVs other than CA/04, we generated 161F mutants for two additional strains: A/Texas/50/2012(H3N2) (TX/50) and A/canine/Iowa/13628/2005 (H3N8) (canine-H3N8). Analysis of the growth kinetics in MDCK cells at an MOI of 0.001 showed that the TX/50 161F mutant generated viral titers of 10^3.17, 10^5.92, 10^7.20, and 10^7.00 TCID₅₀/ml at 12, 24, 48, and 72 h, respectively. This finding compared with 10^2.33, 10^5.25, 10^7.08, and 10^7.00 TCID₅₀/ml at 12, 24, 48, and 72 h, respectively, for the wild-type TX/50 virus. The TX/50 161F mutant in MDCK cells generated a mean total viral protein titer of 882.2 μg/ml, which is 1.15-fold higher than that for the TX/50 wild-type virus (Fig. 1C). When subjected to ferret antisera in an HI assay, both the TX/50 wild-type virus and the 161F mutant had mean HI titers of 1:1,280.

Analysis of growth kinetics in MDCK cells at an MOI of 0.001 showed that the 161F mutant of canine-H3N8 had the highest titer (10^7.249 TCID₅₀/ml) 72 h after infection; this titer was about 10-fold higher than that generated by the canine-H3N8 wild-type virus (Fig. 4C). The mean HI titers of wt and mutant canine-H3N8 viruses in ferret antisera were 640 and 533, respectively. The total viral protein of the canine-H3N8 161F mutant reached a mean titer of 777.2 μg/ml, which was 1.85-fold higher than that of the wild-type canine-H3N8 virus (P = 0.02) (Fig. 1C). We further compared influenza virus-specific protein yields by Western blotting using an NP-specific monoclonal antibody, and the results showed that the Y161F mutation increased the influenza virus-specific NP protein yields for rg-CA/04, rg-Tex/50, and rg-H3N8 by 42%, 18%, and 20%, respectively. By using an H1-specific monoclonal antibody, the results showed that the Y161F mutation increased the HA protein yields for rg-CA/04 by 39% (Fig. 5).

FIG 5.

Western blot showing the NP and HA protein expression levels for the wild-type and mutant viruses. The bands were analyzed by using ImageJ software.

These results suggest that the Y161F mutant can enhance the replication efficiency but does not change the antigenicity of the TX/12 and canine-H3N8 viruses.

Effect of the Y161F mutation on replication efficiency of IAVs in eggs.

While we were able to show an elevated-growth phenotype of Y161F mutants in cells, we next sought to determine its impact on egg growth. To do this, we quantified the total protein yields of wt and Y161F mutant H1N1, H3N2, and H3N8 IAVs in eggs. The total viral protein of the rg-Y161F mutant reached a mean titer of 1,460.2 μg/ml, which is 1.22-fold higher than that of wild-type CA/04. Similarly, for TX/50, the Y161F mutation conferred a 1.23-fold increase in the total amount of viral protein (1,589.2 μg/ml). The total viral protein of the canine-H3N8 Y161F mutant reached a mean titer of 1,824.2 μg/ml, which was 1.45-fold higher than that of wild-type canine-H3N8 (Fig. 1D). Thus, the Y161F mutation also increased the viral replication efficiencies of three H1N1, H3N2, and H3N8 IAVs tested in eggs.

Impact of the Y161F mutation on viral thermostability.

Another desirable property of an influenza vaccine virus is increased stability. To determine whether the Y161F mutation correlated with changes in viral thermostability, purified viruses were diluted to 128 HAU/50 μl and incubated at a series of high temperatures (51.5°C to 65°C) for 40 min, and the integrity of the HA protein was then determined by an HA assay using 0.5% turkey erythrocytes. The CA/04 rg-wt virus showed a precipitous drop in the HA titer (from 128 to 2 HAUs) after 40 min of incubation at 55.7°C. In contrast, rg-Y161F maintained an HA titer of 64 HAU at 55.7°C, which did not drop until 59.5°C. The mutant virus maintained an HA titer of 2 HAU even at 61.4°C. The rg-wt virus completely lost its hemagglutination ability at 55.7°C (Fig. 4E). Similar increased stability was observed for the 161F mutants for canine-H3N8. The canine-H3N8 Y161F mutant had an HA titer of 16 HAU when incubated at 57.6°C for 40 min; this titer was 8-fold higher than that of the wild-type H3N8 virus (Fig. 4F). For TX/50, the wild-type and Y161F viruses maintained titers of 8 HAU and 16 HAU, respectively, at 61.4°C (data not shown). Taken together, these results show that the Y161F mutation conferred higher viral temperature stability on viruses.

The high-yield vaccine candidate protects mice against lethal challenge.

Although we showed that the Y161F mutation had no impact on HAI titers, we wanted to directly confirm that there was not an associated loss in vaccine efficacy. To do this, we prepared inactivated whole-virus vaccines from rg-wt and rg-Y161F CA/04 viruses and evaluated their efficacy in a mouse model. Mice were administered the vaccine or phosphate-buffered saline (PBS) (as a mock vaccine), and 2 weeks later, we collected blood samples for testing. All vaccinated mice had seroconverted, and their HI titers were substantially higher than those in mock-vaccinated mice (Table 3); mice vaccinated with the rg-wt vaccine and the rg-Y161F vaccine had HI titers of 7.65 ± 0.57 and 7.32 ± 0 log₂ units, respectively. The heterologous HI titers were indistinguishable (P = 0.3739) from homologous titers, again demonstrating the antigenic similarity of wt and mutant viruses.

TABLE 3.

Immunologic and pathogenic responses in mice challenged with mouse-adapted influenza A/California/04/09 (H1N1)

Vaccine group	Mean log10 TCID50/ml ± SDa	Mean log2 HI titer ± SDb	Mean log2 HI titer ± SDc
wt	Below detection limit	7.65 ± 0.57	7.32 ± 0
Mutant	Below detection limit	7.32 ± 0	7.32 ± 0
PBSd	5.28 ± 0.14	Below detection limit	Below detection limit

^aGroups of BALB/c mice were inoculated intranasally with 10× the 50% lethal dose of the mouse-adapted CA/04 virus under light anesthesia. Three mice from each group were euthanized on day 4 after virus challenge, and virus titers in lungs were determined by a TCID₅₀ assay in MDCK cells.

^bSerum samples were collected before challenge, and antibody response levels against the wild-type virus were measured by using an HI assay.

^cSerum samples were collected before challenge, and antibody response levels against the immunogen mutant were measured by using the HI assay.

^dMock infection.

Following challenge, a high level of virus replication (up to 10^5.45 TCID₅₀) was observed in mock-vaccinated mice, but no virus was detected in mice vaccinated with the rg-wt- or rg-Y161F-derived vaccine. Mock-vaccinated mice exhibited signs of inactivity and lethargy, had ruffled hair, and rapidly lost weight following challenge (Fig. 6A). In contrast, mice vaccinated with rg-wt- or rg-Y161F-derived vaccines did not exhibit any detectable clinical signs. All vaccinated mice survived, but by postchallenge day 6, all mock-vaccinated mice (n = 5) had lost 25% of their preexperiment body weight and were euthanized (Fig. 6B).

FIG 6.

Weight loss and survival among vaccinated mice challenged with a lethal dose of the influenza A/California/04/09 (H1N1) virus (CA/04). Mice were vaccinated with the wild-type (WT) or Y161F mutant virus vaccine or mock vaccinated with PBS, challenged with 10 LD₅₀ of mouse-adapted CA/04 by intranasal inoculation, and monitored daily for 2 weeks. (A) Percent change in body weight. Each point represents the mean body weight of 5 mice per group. (B) Percentage of surviving mice after challenge. ***, P < 0.0001 for the PBS group versus the wild-type group, as calculated by using GraphPad Prism 5 software. (C to F) Histopathologic changes in hematoxylin- and eosin-stained lung samples from groups of mice vaccinated with the wild-type rg-CA/04 (C) or rg-CA/04 161F mutant (D) virus vaccine, mice mock vaccinated with phosphate-buffered saline (E), or mice serving as controls (F). Samples were collected 4 days after vaccinated and mock-vaccinated mice were challenged with 10 LD₅₀ of the mouse-adapted influenza A/California/04/09 (H1N1) virus. Magnification, ×200.

The results from histopathological analyses showed that mice immunized with the rg-wt- or rg-Y161F-derived vaccine had no apparent pathological changes (Fig. 6C and D). However, the mock-vaccinated mice exhibited severe bronchiolitis, and their bronchioles showed necrosis and some attenuated regenerative epithelial cells along the basement membrane (Fig. 6E).

In summary, our results from experiments in mice suggest that the Y161F mutation in HA did not alter the viral antigenicity of CA/04 or the efficiency of the vaccine in mice.

DISCUSSION

Effective influenza vaccination is dependent on a number of factors, not the least of which are high-yielding and stable vaccine viruses. A high yield of virus is critical for vaccine manufacturing, and the thermostability of viral antigens is critical for the vaccine shelf life, which can be compromised during transportation and storage. In this study, we screened a group of randomly generated CA/04 mutants carrying substitutions at the HA RBS. While a number of these mutants displayed enhanced features, one, containing a Y161F change, had increased thermostability and the highest viral yields. This mutation was also able to impart these properties on a seasonal H3N2 virus and a canine H3N8 virus, showing that the effect was not subtype specific. Compared to those on CA/04, the effects of Y161F on the seasonal H3N2 and canine H3N8 viruses that we tested were relatively less effective. One possible explanation is that, unlike the wild-type CA/04 virus, which replicates poorly in MDCK cells (10^5.75 TCID₅₀/ml at 72 h), the wild-type TX/50 and canine-H3N8 viruses have high replication efficiencies, with 10^7.20 TCID₅₀/ml and 10^6.50 TCID₅₀/ml at 72 h, respectively. Furthermore, the Y161F mutation did not change the antigenicity of the H1N1, H3N2, and H3N8 viruses tested. Animal experiments further showed that the Y161F change in CA/04 did not have a measurable impact on the efficacy of inactivated whole-virus vaccines containing it. These results highlight the application potential of the HA 161F signature in influenza vaccine manufacture.

The initial step in viral infection is the binding of HA to the sialic acid receptors on the epithelial cell surface (29). This interaction is mediated by the RBS, which is located at the globular head of HA and consists of the loops at positions 130 and 220 and the helix at position 190. Mutations at the RBS have long been known to affect the yield of a vaccine strain, which was our rationale for targeting it. For example, the L194P mutation increased the yield of an A/England/611/07 (H3N2) 6+2 reassortant virus (30), and single or double mutations at positions 191 (194 in H3), 197 (200 in H3), 222 (225 in H3), and 223 (226 in H3) increased the replication of A/California/7/09 (H1N1) in eggs (31). Mutations at residues 186 and 194 in HA of an A(H1N1)pdm09 virus have also been shown to improve viral titers in MDCK cells and eggs (23). Avian virus yields can also be improved by targeting the RBS, and the double mutation of N133D-G198E in HA has been reported to increase H7N9 virus yields (32). The challenge of targeting the RBS for improving virus yields is that some mutations that lead to improved growth also alter antigenicity. For example, mutation G144E was shown to increase the yield of B/Victoria/504/2000, but the antigenic properties of the virus were also changed (33). Single-amino-acid changes at positions 119 (122 in H3), 153 (156 in H3), 154 (157 in H3), and 186 (189 in H3) could increase the yield of A/California/7/09 (H1N1) in eggs, but mutations at residues 153 (156 in H3), 154 (157 in H3), and 155 (158 in H3) drastically altered viral antigenicity (34).

The natural plasticity of the RBS for accepting substitutions was highlighted previously by Yasugi et al. (35). Those authors used Roche 454 sequencing to directly sequence nasal specimens from three patients infected with A(H1N1)pdm09 virus. They found that the virus' HAs showed high levels of amino acid diversity, with frequencies of polymorphisms ranging from 3.45 to 8.59% for a K119N substitution, 1.01 to 4.99% for an N125D substitution, 0.74 to 21.49% for a D222G substitution, and 2.39 to 4.64% for a Q223R substitution (35). The percentages of the K119N, N125D, D222G, and Q223R mutations reached up to 60.4%, 96.7%, 85%, and 95.8%, respectively, after egg adaptation of the primary specimens (35). Similarly, mutations K119N and D222G were also found in the high-yield, egg-adapted A(H1N1)pdm09 virus vaccine strain NIBRG-121xp, and a Q223R mutation was found in another high-yield, egg-adapted A(H1N1)pdm09 strain, NYMC-181A (34). These findings demonstrate that the selection or generation of an HA variant, especially in the RBS of HA, can be rapidly achieved in some cases. However, the genetic features of the high-yield properties are still not fully understood, and the generation of high-yielding viruses using classical virological techniques is sometimes more challenging. In this study, we therefore opted for a random-mutagenesis approach, which generated eight mutants, four of which (D130E, K174E, L154F-K156Q, and Y161F) increased viral yields in cells without changing antigenic properties (Table 1). Among these mutants, Y161F had the largest increase in the viral replication efficiency in both MDCK and Vero cells (Fig. 1).

Binding to the host cell is the first step of influenza virus infection. Thus, the presence of favorable receptors on a specific cell is one of the key factors determining host and tissue tropisms of IAV. Most studies related to influenza virus receptors have classified the sialic acid receptors into two groups on the basis of positions of the sialic acid-galactose linkage: SA2,3GA or SA2,6GA. Both SA2,3GA and SA2,6GA are present in any single cell type, but their distributions vary based on the types of cells. For example, SA2,3GA and SA2,6GA are present in both MDCK and Vero cells, but SA2,3GA is considerably more abundant than SA2,6GA in both cell lines (36–38). However, in chicken erythrocytes, SA2,6GA is more abundant than SA2,3GA (39). Thus, an ideal high-yield vaccine candidate would have high binding affinities for both SA2,3GA and SA2,6GA (40). This double binding was the phenotype of the Y161F mutant in multiple assays, providing a plausible explanation for the increased yields in MDCK and Vero cells.

The HA RBS is a member of the lectin superfamily, and the specificity of the RBS contributes to the host range of IAVs. The results of our structure modeling showed that residue 161 is located at the top of the RBS (Fig. 3). Others previously reported that a Y161A substitution in H5N1 HA changed the receptor binding preference from Neu5Ac to N-glycolylneuraminic acid (Neu5Gc), however, and the Y161A mutation did not increase the virus replication efficiency on MDCK cells (41). In addition, the results of that study showed that the Y161A mutants had the best viral replication and plaque-forming abilities. Our findings suggest that the binding preference of the virus is changed by introducing the F substitution at residue 161. This substitution in CA/04 accommodates virus binding to both SA2,3GA and SA2,6GA and is responsible for the acquisition of specificity to SA2,3GA receptors. The Phe side chain lacks the O⁴ hydroxyl group present on the Tyr residue, shortening the distance between the oxygen atom of a water molecule and residue 161, and this potentially facilitates the acquisition of virus specificity for SA2,3GA.

Environmental factors such as temperature and pH have been reported to affect the airborne transmission of IAV, possibly by affecting viral thermostability (42). Thus, viral thermostability could affect the quality and transmissibility of influenza viruses. A T318I substitution in HA increased its stability and affected the binding properties of a reassortant H5 HA/H1N1 influenza virus (43). Thus, amino acid substitutions that increase viral thermostability would also be important risk assessment factors for emerging IAVs, such as those of subtypes H5 and H7 (44). Moreover, in agreement with our observations, Watanabe et al. (45) reported that HA thermostability was correlated with the viral replication and glycan receptor binding of H5N1 viruses. However, further studies are needed to interpret the molecular mechanism of increased HA thermostability by mutations.

In conclusion, our study shows that mutation Y161F in the RBS of A(H1N1)pdm09 HA significantly increased the viral yield in MDCK and Vero cells by promoting virus binding to both SA2,3GA and SA2,6GA without altering the original antigenicity. These results suggest that the Y161F mutation in HA might have potential for generating seed viruses of high yields for cell-based influenza vaccine development and production.

MATERIALS AND METHODS

Cells and viruses.

Human embryonic kidney (293T) cells, MDCK cells, and Vero cells were purchased from the American Type Culture Collection (Manassas, VA). Cells were maintained at 37°C with 5% CO₂ in Dulbecco's modified Eagle medium (Gibco/BRL, Grand Island, NY) supplemented with 5% fetal bovine serum (Atlanta Biologicals, Lawrenceville, GA) and penicillin-streptomycin (Invitrogen, Carlsbad, CA). The HA gene of CA/04 was cloned into the vector pHW2000 and used as the template for the construction of the mutant library; mouse-adapted CA/04 (46) was used for challenge experiments in mice. Influenza A/Texas/50/2012(H3N2) (TX/50) and A/canine/Iowa/13628/2005 (H3N8) (canine-H3N8) viruses were used to validate the identified molecular markers.

The viruses generated by reverse genetics were propagated in MDCK cells and cultured at 37°C with 5% CO₂ in Opti-MEM (Gibco/BRL, Grand Island, NY) supplemented with 1 μg/ml of TPCK (N-tosyl-l-phenylalanine chloromethyl ketone)-trypsin (Sigma-Aldrich, St. Louis, MO) and penicillin-streptomycin (Invitrogen, Carlsbad, CA). Virus titers were determined by a TCID₅₀ assay in MDCK cells.

The total viral protein yield in eggs was tested as described previously by Adamo et al. (47). Briefly, 10-day-old chicken embryotic eggs were infected with an influenza virus and then incubated at 37°C for 72 h. The allantoic fluid of infected eggs was collected for virus purification and quantification of protein concentrations as described below.

Extractions of RNA and plasmids.

RNA was extracted by using an RNeasy minikit (Qiagen, Valencia, CA); the plasmids used for transfection were prepared by using the GeneJET Plasmid Miniprep kit (Thermo Scientific, Waltham, MA).

Generation of mutants using an epPCR-based reverse-genetics strategy.

The mutant library with random mutations in HA RBSs was generated by epPCR as previously described (48). In brief, the randomly mutated short sequences (about 200 nucleotides) were used as primers for site-directed mutagenesis with HA-pHW2000, leading to plasmids with random mutations in the HA RBSs. This strategy can avoid the need for a labor-intensive gene-cloning process, and HA-pHW2000 with mutations can be used directly for generating vaccine candidates (48). One day before transfection, 293T cells and MDCK cells were cocultured in 24-well plates, using a 20:1 ratio of 293T cells to MDCK cells. The cell cultures were transfected with 125 ng of each of eight plasmids, harboring a mutant HA gene, the NA gene of CA/04, and 6 internal genes (PB2, PB1, PA, NP, M, and NS) of influenza A/PR/8/1934(H1N1) virus. Transfection was done by using TransIT-LT (Mirus, Madison, WI) according to the manufacturer's instructions. In brief, the TransIT-LT transfection reagent was mixed with DNA at 2.5 μl/1 μg, incubated at room temperature for 20 min, and then added to the cells. After 24 h, Opti-MEM (Gibco/BRL) supplemented with 1 μg/ml of TPCK-trypsin (Sigma-Aldrich, St. Louis, MO) was added to the cells. After 72 h of incubation, supernatants were collected and titrated in MDCK cells.

For phenotype comparison, we generated the wild-type reassortant virus (rg-wt) containing the wild-type HA and NA genes from CA/04 and six internal genes from A/Puerto Rico/8/34 (H1N1). The reassortant rg-TX/50, with HA and NA from TX/50 and six internal genes from the A/Puerto Rico/8/34 (H1N1) virus, and the reassortant rg-H3N8, with HA and NA from canine-H3N8 and six internal genes from the A/Puerto Rico/8/34 (H1N1) virus, were also rescued by reverse genetics.

Site-directed mutagenesis.

The QuikChange II site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA) was used to create specific mutations in the HA gene. We used forward primer 5′-CCACTTAAACTTCAAATTCCCAGCATTGAACGTG-3′ and reverse primer 5′-CACGTTCAATGCTGGGAATTTGAAGTTTAAGTGG-3′ to generate the Y161F mutation in HA of TX/50, and we used forward primer 5′-CAAAATCTGGAAGCTCTTTCCCCACATTGAATGTGAC-3′ and reverse primer 5′-GTCACATTCAATGTGGGGAAAGAGCTTCCAGATTTTG-3′ to generate the Y161F mutation in HA of canine-H3N8. To ensure the absence of unwanted mutations, Eurofins (Louisville, KY) used Sanger sequencing to completely sequence all constructs.

Growth kinetics.

To determine the growth kinetics of viruses, we inoculated MDCK cells with a testing virus at an MOI of 0.001 and then incubated the cells in 5% CO₂ at 37°C for 1 h. The inocula were then removed, and cells were washed twice with PBS. Opti-MEM I (Gibco, Grand Island, NY) containing TPCK-trypsin (1 μg/ml) was added to the cells, which were then incubated in 5% CO₂ at 37°C. At the specified time points after inoculation, 200 μl of supernatant was collected from the incubated cells, aliquoted, and stored at −70°C until use. Virus titers in the supernatants collected at different time points were determined by a TCID₅₀ assay in MDCK cells.

Viral protein purification and quantification of protein concentrations.

Viruses were purified from the cell supernatant or allantoic fluid by low-speed clarification (2,482 × g for 20 min at 4°C) to remove debris and then ultracentrifuged through a gradient of 30% to 60% sucrose in a 70Ti rotor (Beckman Coulter, Fullerton, CA) (100,000 × g for 3 h at 4°C). The virus band was collected and purified through a cushion of 30% sucrose in a 70Ti rotor (100,000 × g for 3 h at 4°C). The virus pellet was resuspended in 200 μl of PBS, and the total amount of purified virion protein was determined by using the Pierce bicinchoninic acid (BCA) protein assay kit (Thermo Scientific, Rockford, IL).

HA and HI assays.

HA and HI assays were performed by using 0.5% turkey erythrocytes as described in the Manual for the Laboratory Diagnosis and Virological Surveillance of Influenza by the WHO Global Influenza Surveillance Network (49). Guinea pig, chicken, horse, turkey, and dog (beagle) erythrocytes were obtained from Lampire Biological Products (Everett, PA). The erythrocytes were washed three times with 1× PBS (pH 7.2) and diluted to 0.5% for chicken, beagle, and turkey erythrocytes; 0.75% for guinea pig erythrocytes; and 1% for horse erythrocytes.

Western blot analysis.

Western blot analysis was used to compare the influenza virus-specific protein yields of a wild-type virus and those of a testing mutant. The monoclonal antibody specific for influenza A virus nucleoprotein (NR-43899) and H1 (NR-42019) were obtained through the BEI Resources Repository (https://www.beiresources.org/). The Western blots were developed by using horseradish peroxidase (HRP)-conjugated anti-mouse IgG antibody (Sigma-Aldrich, St. Louis, MO) and a 3,3′-diaminobenzidine tetrahydrochloride (DAB) substrate kit (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer's instructions. Bands were visualized by chemiluminescence using a ChemiDoc imaging system (Bio-Rad, Hercules, CA) and analyzed by using ImageJ (National Institutes of Health, Bethesda, MD).

Glycan microarray and data analyses.

The 83 N-glycans (28) (see Fig. S1 in the supplemental material) were printed on N-hydroxysuccinimide-derivatized slides as previously described (50). All glycans were printed in replicates of 6 in a subarray, and 8 subarrays were printed on each slide. All glycans were prepared at a concentration of 100 μM in phosphate buffer (100 mM sodium phosphate buffer, pH 8.5). The slides were fitted with an 8-chamber adapter (Grace Bio-Labs, Bend, OR) to separate the subarray into individual wells for assays. Before assays, slides were rehydrated for 5 min in TSMW buffer (20 mM Tris-HCl, 150 mM NaCl, 0.2 mM CaCl₂, 0.2 mM MgCl₂, and 0.05% Tween 20) and blocked for 30 min in TSMWB buffer (TSMW buffer with 1% bovine serum albumin [BSA]). Viruses were purified by sucrose density gradient ultracentrifugation and titrated to about 1.0 × 10⁵ HAU/ml. We added 15 μl of 1.0 M sodium bicarbonate (pH 9.0) to 150 μl of virus and then incubated the mixture with 25 μg of Molecular Probes Alexa 488 succinimidyl esters (NHS esters) (Thermo Fisher Scientific, Inc., Waltham, MA) for 1 h at 25°C. After overnight dialysis against a 7-kDa Slide-A-Lyzer Mini dialysis device (Thermo Fisher Scientific, Inc., Waltham, MA) to remove excess Alexa 488 dye, viruses were checked by an HA assay and then bound to a glycan array. Labeled viruses were incubated on a glycan microarray at 4°C for 1 h, washed, and centrifuged briefly before being scanned with an InnoScan 1100 AL microarray scanner (Inopsys, Toulouse, France).

Virus-glycan receptor binding assay.

Two biotinylated glycan analogs, the carbohydrates 3′-sialyl-N-acetyllactosamine (3′SLN), representing SA2,3GA, and 6′-sialyl-N-acetyllactosamine (6′SLN), representing SA2,6GA, were purchased from GlycoTech (Gaithersburg, MD). The glycan stocks were reconstituted at 1 mg/ml in a 50% (vol/vol) glycerol–PBS solution according to the manufacturer's instructions and were stored at 4°C until use. The viral particles in the wild-type reassortant virus bearing HA_161Y (rg-wt) and a mutant virus bearing HA_161F (rg-Y161F) were determined by using a ViroCyt 2100 virus counter (ViroCyt, Boulder, CO). Kinetics buffer (PBS [pH 7.4] with 0.01% bovine serum albumin and 0.002% Tween 20) containing neuraminidase inhibitors (10 μM zanamivir hydrate and 20 μM oseltamivir phosphate) was used to titrate the biotinylated glycan analogs and viruses during the binding assay (51). Binding of viruses (at 1 pM/virus) to the biotinylated glycan analogs was performed by using an Octet RED96 biolayer interferometer equipped with streptavidin biosensor tips (Pall FortéBio, Menlo Park, CA) according to the manufacturer's assay protocol: (i) biosensor coating with biotinylated glycan analogs for 300 s, (ii) virus association for 1,200 s, and (iii) dissociation in kinetics buffer with neuraminidase inhibitors for 1,000 s. The entire measurement cycle was maintained at 30°C with orbital shaking at 1,000 rpm.

Analyses of virus thermostability.

Purified viruses were diluted in PBS to 128 HAU and dispensed by 120 μl into 0.2-ml, thin-walled PCR tubes (USA Scientific, Ocala, FL). Tubes were placed into a Gradient Veriti 96-well thermal cycler (catalog number 9902; Life Technologies, Camarillo, CA). The temperature range was set at 51.5°C to 63.0°C. Tubes were heated for 40 min and then transferred to ice. Control samples containing 120 μl of virus were incubated for 40 min at 0°C. The virus content in each sample was determined by an HA assay using a 0.5% suspension of turkey erythrocytes. Each virus sample was analyzed three times for thermostability.

Animal experiments.

We intramuscularly inoculated two groups of 6-week-old female BALB/c mice (Harlan Laboratories, Indianapolis, IN) with 15 μg (in a 50-μl volume) of a formaldehyde-inactivated vaccine candidate or the wild-type virus (n = 10 mice/group). Specifically, the viruses used as vaccine candidates were inactivated in 0.025% formalin at 4°C for 3 days and then confirmed by three blind passages on MDCK cells. Two weeks later, we administered a booster vaccine with the same amount of immunogen. A group of mock-vaccinated mice (n = 10) received an equal volume of PBS. A group of mice serving as environmental controls (n = 5) was not vaccinated or challenged. Two weeks after the booster vaccination, mice were anesthetized and challenged by intranasal inoculation with mouse-adapted CA/04 (46) at 10 times the 50% lethal dose (LD₅₀). Serum samples were collected from mice before challenge and tested by HI assays. To determine lung virus titers, we euthanized five mice at day 4 after challenge. Lungs (n = 3) were homogenized and resuspended in 1 ml of sterile PBS, and virus titers were determined in MDCK cells. The lung samples were also fixed in formalin and stained with hematoxylin and eosin stain for pathological examination. We monitored the clinical signs, survival rate, and body weight of the remaining mice (n = 5) for 14 days after challenge. The mice were euthanized on 14 days after challenge.

Biosafety and animal handling.

All laboratory and animal experiments were conducted under biosafety level 2 (BSL-2) conditions, with investigators wearing appropriate protective equipment and in compliance with protocols approved by the Institutional Animal Care and Use Committee of Mississippi State University.

Structural modeling.

Crystal structures of the HA protein of the A(H1N1)pdm09 virus and the binding sites of 6′SLN and 3′SLN to this protein were obtained from the PDB (Protein Data Bank) (accession numbers 3LZG, 3UBN, and 3UBQ, respectively). Structural simulation of amino acid mutations was performed on HA by using the FoldX computer algorithm (http://foldxsuite.crg.eu/) with its empirical force field with crystal waters under the following conditions: temperature of 298 K, pH 7, and ion strength of 0.05. The binding structures and measure of contact distances were visualized by using Chimera (http://www.cgl.ucsf.edu/chimera/). PoseScore, which was designed for ranking near-native ligand-protein-interacting structures, was used to estimate the likeness of the protein-glycan binding avidities of the wild-type and mutant viruses to that of the native virus. PoseScore scores typically range from −100 to 100; the lower the score, the lower the binding affinity. Computational analysis of the effect of mutation on HA-glycan binding was focused on mutants with a location at position 161 (H3 numbering).

Genomic sequences, molecular characterization, and statistical analyses.

HA sequences of IAVs were downloaded from the database of the Influenza Virus Resource on 30 January 2017. Multiple-sequence alignments were conducted by using MUSCLE software (52). The mutations were identified by using BioEdit software (53). Survival curves were calculated by the Kaplan-Meier method, and significance was analyzed with the log rank test by using GraphPad Prism 5 software.