Identification of Amino Acid Substitutions Supporting Antigenic Change of Influenza A(H1N1)pdm09 Viruses

ABSTRACT

The majority of currently circulating influenza A(H1N1) viruses are antigenically similar to the virus that caused the 2009 influenza pandemic. However, antigenic variants are expected to emerge as population immunity increases. Amino acid substitutions in the hemagglutinin protein can result in escape from neutralizing antibodies, affect viral fitness, and change receptor preference. In this study, we constructed mutants with substitutions in the hemagglutinin of A/Netherlands/602/09 in an attenuated backbone to explore amino acid changes that may contribute to emergence of antigenic variants in the human population. Our analysis revealed that single substitutions affecting the loop that consists of amino acid positions 151 to 159 located adjacent to the receptor binding site caused escape from ferret and human antibodies elicited after primary A(H1N1)pdm09 virus infection. The majority of these substitutions resulted in similar or increased replication efficiency in vitro compared to that of the virus carrying the wild-type hemagglutinin and did not result in a change of receptor preference. However, none of the substitutions was sufficient for escape from the antibodies in sera from individuals that experienced both seasonal and pandemic A(H1N1) virus infections. These results suggest that antibodies directed against epitopes on seasonal A(H1N1) viruses contribute to neutralization of A(H1N1)pdm09 antigenic variants, thereby limiting the number of possible substitutions that could lead to escape from population immunity.

IMPORTANCE Influenza A viruses can cause significant morbidity and mortality in humans. Amino acid substitutions in the hemagglutinin protein can result in escape from antibody-mediated neutralization. This allows the virus to reinfect individuals that have acquired immunity to previously circulating strains through infection or vaccination. To date, the vast majority of A(H1N1)pdm09 strains remain antigenically similar to the virus that caused the 2009 influenza pandemic. However, antigenic variants are expected to emerge as a result of increasing population immunity. We show that single amino acid substitutions near the receptor binding site were sufficient to escape from antibodies specific for A(H1N1)pdm09 viruses but not from antibodies elicited in response to infections with seasonal A(H1N1) and A(H1N1)pdm09 viruses. This study identified substitutions in A(H1N1)pdm09 viruses that support escape from population immunity but also suggested that the number of potential escape variants is limited by previous exposure to seasonal A(H1N1) viruses.

INTRODUCTION

Influenza pandemics occur when a novel influenza A virus is introduced in the human population and spreads around the globe. Since existing antibody responses are typically not cross-reactive with the antigenically novel virus, the virus encounters little preexisting humoral immunity and can cause severe outbreaks. Three influenza pandemics occurred during the 20th century: A(H1N1) virus in 1918, A(H2N2) virus in 1957, and A(H3N2) virus in 1968 (1). In each case the newly introduced subtype replaced the previous subtype. In 1977, an A(H1N1) virus that caused epidemics in the early 1950s was reintroduced in the human population (2), and it continued to cocirculate with A(H3N2) until 2009. In April 2009, a swine origin A(H1N1) virus [A(H1N1)pdm09] caused the first influenza A virus pandemic of the 21th century (3). It replaced the previously circulating seasonal A(H1N1) virus, but it continues to cocirculate with seasonal A(H3N2) virus (4).

A prerequisite for the influenza virus to infect the host cell is the binding of the hemagglutinin (HA) surface protein to sialylated glycan receptors on the host cell through its receptor binding site (RBS). HA is the main target of neutralizing antibodies and is therefore a critical component of influenza vaccines (5). Influenza viruses continually escape antibody-mediated neutralization by variation of the amino acids in the HA protein. This process is referred to as antigenic drift, and it allows the virus to infect individuals that are immune to contemporary or previously circulating antigenic variants.

Studies from the 1980s identified four immunodominant antigenic regions within the HA of A(H1N1) virus (6, 7). Similar antigenic regions were identified for A(H3N2) (8) and A(H5N1) (9, 10) viruses. Amino acid substitutions in these so-called antigenic sites, which cover much of the HA globular head, can result in escape from antibody recognition. More recently, it was shown that major antigenic change during evolution of A(H3N2) and A(H5N1) viruses and recent antigenic change of seasonal A(H1N1) and influenza B viruses were predominantly caused by single substitutions that occurred near the RBS (11, 12).

Antigenic change may also be a secondary effect of substitutions in HA that facilitate more efficient replication in the human host. HA is pivotal in adaptation of zoonotic influenza A viruses to a new host because of its function in receptor binding (13). Human influenza viruses bind to sialic acids (SAs) linked to the galactose in an α2,6 linkage, avian influenza viruses have a preference for α2,3-linked SAs, while swine viruses bind either α2,6- or both α2,3- and α2,6-linked SAs (14). We hypothesized that substitutions that modify or fine-tune receptor specificity, thereby altering host range and tissue tropism, may result in escape from antibodies directed at the RBS area. Finally, addition or removal of carbohydrate side chains on HA has been associated with changes in the antigenic properties of influenza viruses (15, 16).

The HA of A(H1N1)pdm09 viruses is antigenically most similar to that of recent classical and triple-reassortant swine A(H1N1) viruses (17, 18). The HA of these swine viruses descended from the 1918 pandemic influenza virus but, in contrast to the human lineage, did not undergo extensive antigenic drift, as was reported for A(H3N2) swine viruses (19). Structural analyses suggested high antigenic similarity between the A(H1N1)pdm09 virus and A(H1N1) viruses that circulated in the first decades after the 1918 pandemic (20, 21). Accordingly, age groups that experienced A(H1N1) virus infection before 1950 were partially immune to the A(H1N1)pdm09 virus (22).

During the 5 years after the emergence of A(H1N1)pdm09 virus multiple antigenic variants have been detected, but the vast majority of recently isolated viruses remain antigenically similar to the A/California/7/2009 vaccine virus (4). However, as population immunity to A(H1N1)pdm09 virus builds up, it becomes beneficial for the virus to be antigenically different from the pandemic strain. Therefore, the goal of the present study was to explore molecular changes that contribute to antibody escape of A(H1N1)pdm09 virus.

We selected 25 single and 5 double substitutions based on substitutions that were shown to be important for antigenic change in other subtypes, changes in receptor specificity, or genetic differences between A(H1N1)pdm09, swine A(H1N1), and seasonal A(H1N1) viruses. The substitutions were introduced into the HA gene of influenza virus A/Netherlands/602/09 in an attenuated virus backbone, and their antigenic effect was tested in hemagglutination inhibition (HI) assays using a ferret antiserum prepared against A/Netherlands/602/09. Mutants that displayed altered antigenic properties were further tested against a larger panel of ferret antisera and human sera. In addition, the impact of these substitutions on replication kinetics and receptor specificity was evaluated.

MATERIALS AND METHODS

Cells.

293T cells were cultured in Dulbecco modified Eagle medium (DMEM; Lonza, Breda, The Netherlands) supplemented with 10% fetal calf serum (FCS; Sigma-Aldrich, Zwijndrecht, The Netherlands), 100 IU/ml of penicillin, 100 μg/ml of streptomycin, 2 mM glutamine, 1 mM sodium pyruvate, nonessential amino acids (Lonza), and 500 μg/ml of Geneticin (Life Technologies, Bleiswijk, The Netherlands). Madin-Darby canine kidney (MDCK) cells were cultured in Eagle minimal essential medium (EMEM; Lonza) supplemented with 10% FCS, 100 IU/ml of penicillin, 100 μg/ml of streptomycin, 2 mM glutamine, 1.5 mg/ml of sodium bicarbonate (Lonza), 10 mM HEPES (Lonza), and nonessential amino acids.

Plasmid construction.

A/Netherlands/602/09 was isolated from a patient in The Netherlands during the 2009 influenza pandemic (23) and was used in this study to represent the antigenic properties of A(H1N1)pdm09 viruses. The full HA gene was cloned in the modified pHW2000 expression plasmid as previously described (24–26). Mutations were introduced with the minimal number of nucleotide substitutions necessary to change the amino acid. When more than one single nucleotide change could lead to the desired substitution, the codon change observed in naturally occurring amino acid substitutions was selected [e.g., genetic differences between human A(H1N1) viruses]. If the desired amino acid substitution did not occur previously in A(H1N1) viruses, the mutation was introduced using a codon observed in A(H3N2) viruses. Mutations were introduced in the HA gene using the QuikChange multisite-directed mutagenesis kit (Agilent Technologies, Amstelveen, The Netherlands) according to the manufacturer's instructions. The presence of introduced mutations and absence of undesired additional mutations were confirmed by sequence analysis of the modified HA gene.

Construction of recombinant virus stocks.

Plasmids containing wild-type or modified A/Netherlands/602/09 HA genes were used to generate recombinant viruses consisting of the (modified) HA gene and seven remaining gene segments of A/Puerto Rico/8/34, as described previously (25). Briefly, 293T cells were seeded in 100-mm dishes 1 day prior to transfection. Cells were transfected overnight with 40 μg of plasmid DNA. Transfection medium was subsequently replaced with medium containing 2% FCS. Cells were incubated for 72 h at 37°C and 5% CO₂ before harvesting of the supernatant. The virus stocks were propagated by inoculation of MDCK cells with 2 ml of supernatant of transfected cells. After 2 h, the inoculum was replaced with MDCK infection medium consisting of EMEM, 100 IU/ml of penicillin, 100 μg/ml of streptomycin, 2 mM glutamine, 1.5 mg/ml of sodium bicarbonate, 10 mM HEPES, nonessential amino acids, and 25 μg/ml of trypsin. Viruses were harvested after incubation for 72 h at 37°C and 5% CO₂. The culture supernatants were subjected to ultracentrifugation to increase the viral particle concentration if hemagglutination titers were below 12 hemagglutinating units (HAU). The presence of introduced mutations and absence of unwanted additional mutations were confirmed by sequencing of the HA gene. Work with recombinant viruses was performed under biosafety level 2 conditions under a permit from the Ministry of Infrastructure and the Environment.

Antisera.

Ferret antisera were prepared by intranasal inoculation with 500 μl of virus stock. Antisera were collected 14 days after inoculation. Ferrets were housed and experiments were conducted in strict compliance with European guidelines (46) and Dutch legislation (47). The protocol was approved by an independent animal experimentation ethical review committee, “stichting DEC consult.” Animal welfare was monitored daily, and all animal handling was performed under light anesthesia (ketamine) to minimize animal discomfort.

Human serum samples were selected from the serum bank of the viroscience department. Sera from patients in whom the possibility of non-naturally obtained antibody responses to A(H1N1)pdm09 virus (e.g., vaccination or intravenous immunoglobulin administration) existed or sera from patients with immune deficiencies (e.g., autoimmune disease, HIV-positive status, or use of immunosuppressive medication) were excluded from use in this study. In addition, only sera of patients who, or for whom the caregivers, did not object to scientific use of leftover materials were included in this study. The study protocol was reviewed and approved by the medical ethics board of the Erasmus University Medical Center (study number MEC-2012-181). Informed consent was waived because patient inclusion was performed retrospectively and anonymously.

HI assay.

HI assays were performed using standard procedures (27). Briefly, ferret antisera or human sera were pretreated overnight with the receptor-destroying enzyme Vibrio cholerae neuraminidase (VCNA) at 37°C, followed by inactivation for 1 h at 56°C. Twofold serial dilutions of the pretreated sera, starting at a 1:20 dilution, were mixed with 25 μl of virus stock containing 4 HAU, and the mixture was incubated at 37°C for 30 min. Subsequently, 25 μl of 1% turkey erythrocytes (TRBCs) was added, and hemagglutination patterns were read after a 1-h incubation at 4°C. The HI titer is expressed as the reciprocal value of the highest serum dilution that completely inhibited agglutination of TRBCs.

Plaque assay.

The assay was performed as described previously (28). In brief, MDCK cells were seeded in a 6-well plate to reach 90% confluence the following day. One hour after inoculation with virus, the inoculum was replaced with a 1:1 mixture of 2.4% Avicel (FMC Biopolymers, Brussels, Belgium) with 2× EMEM infection medium. After 36 h, cells were washed with phosphate-buffered saline (PBS) and incubated with 80% acetone for at least 30 min at −20°C. Fixed and permeabilized cells were washed 3 times with PBS and incubated for 1 h at 37°C with mouse-anti-NP monoclonal antibody (1 mg/ml of HB65; ATCC). Following three washes with PBS, cells were incubated for 1 h at 37°C with rabbit anti-mouse fluorescein isothiocyanate (FITC) (Life Technologies, Bleiswijk, The Netherlands). Cells were washed with PBS and allowed to air dry. Plaques were scanned on a Typhoon 9410 variable-mode imager (GE Healthcare, Diegem, Belgium). The data were analyzed with an ImageQuant TL colony counter and image feature measurement (Amersham Biosciences, Freiburg, Germany). Plaque size was plotted as the radius of the plaques.

Modified TRBC hemagglutination assay.

The modified TRBC assay was performed as described previously (29), with modifications. Briefly, SAs were removed from the TRBC surface by incubating 20% TRBCs in a total volume of 62.5 μl of phosphate-buffered saline supplemented with 50 mU of VCNA (Roche, Almere, The Netherlands) in 8 mM calcium chloride at 37°C for 1 h. The removal of SAs was confirmed by absence of hemagglutination of the treated TRBCs by control viruses. TRBCs stripped of SAs were resialylated to contain only α2,3- or α2,6-linked SAs using either 0.5 mU of α2,3-N-sialyltransferase (Calbiochem, CA) or 2 mU of α2,6-N-sialyltransferase (Japan Tobacco, Inc., Shizuoka, Japan) and 1.5 mM cytidine monophospho-N-acetylneuraminic (CMP) sialic acid (Sigma-Aldrich, Zwijndrecht, The Netherlands) at 37°C for 2 h in a total volume of 75 μl. Subsequently, resialylated TRBCs were washed with PBS and were resuspended to a final concentration of 0.5% in PBS containing 1% bovine serum albumin. Standard hemagglutination assays were performed to confirm correct resialylation using control viruses with known receptor specificities for α2,3- and α2,6-linked SAs. The receptor specificities of mutant viruses were tested by conventional hemagglutination assay with the modified TRBCs using cell culture supernatants.

Structural analysis.

Amino acid positions were plotted on the HA crystal structure of the A/California/04/09 virus (PDB accession code 3LZG [20]) using MacPyMOL (PyMOL molecular graphics system, version 1.3; Schrödinger, LLC) to visualize the trimer.

RESULTS

Selection of substitutions.

The antigenic regions of seasonal A(H1N1) virus have previously been mapped (6, 7), but the substitutions that caused antigenic change during A(H1N1) virus evolution remain almost entirely unknown. Additionally, fine-grain differences in receptor specificity beyond the coarse distinction for α2,3-SA or α2,6-SA preference that allow more efficient replication in the human host and that may change the virus antigenically are currently incompletely understood (13, 30). Because of the limited insight on the molecular basis for previous antigenic change of A(H1N1) viruses, we selected substitutions to introduce into the A/Netherlands/602/09 representative virus HA gene based on the following three approaches.

First, we aligned the HA sequences of 28 viruses representing pre-1957 and post-1977 seasonal A(H1N1) viruses, swine A(H1N1) viruses, and A(H1N1)pdm09 viruses and selected substitutions based on genetic differences between these viruses (Table 1). We hypothesized that part of the substitutions that became fixed during A(H1N1) virus evolution had a selective advantage because of their role in immune escape. In addition, amino acid differences between swine or A(H1N1)pdm09 viruses and pre-2009 human A(H1N1) viruses may contribute to adaptation to the human host and potentially change the virus antigenically as a secondary effect. Previous work indicated a critical role for antibodies targeting the RBS area in virus neutralization (11, 12), and we therefore focused primarily on amino acid differences in this region of the HA. Substitution 84SN is located away from the RBS but was consistently different between A(H1N1)pdm09 viruses and all swine and human A(H1N1) viruses (H1 numbering is used throughout, unless indicated otherwise).

TABLE 1.

Rationale of selected substitutions

Substitution	Subtype no. conversion	Genetic difference betweena:			Position involved in antigenic change of other influenza virusesb
		A(H1N1)pdm09 viruses and swine A(H1N1) viruses	A(H1N1)pdm09 viruses and seasonal A(H1N1) viruses	Swine A(H1N1) viruses and seasonal A(H1N1) viruses	A(H3N2) virus	A(H5N1) virus	Influenza B virus	Substitutions associated with changes in receptor binding specificity
84SN		√	√
127DE	126c	√				√f
127DN	126c	√				√f
127DT	126c		√			√f
142KN	145d				23
142KS	145d		√		23
152VT	151,c 155d			√	23	24
153KE	156d			√	23
153KH	156d				23
153KQ	156d				23
155GE	158d, 165e				23		23
156ND	159d				23
156NG	159d		√	√	23
156NS	159d			√	23
156NY	159d				23
186AK	185,c 189d				23	24
186AQ	185,c 189d				23	24
190SD	189,c 193d				23	24
190SN	189,c 193d				23	24
222DE								9
222DG								9
223QL	222c							10
223QR	222c							8
224EA		√	√	√				9
225GS	224c							10
152VT156NS				√
187DE222DG								9
155GE224EA								SP
222DG224EA								9
223QL225GS								10

^aGenetic differences were identified from an amino acid alignment of A(H1N1) viruses that circulated between 1918 and 2009, including 11 pre-1957 human and 5 swine A(H1N1) viruses (data not shown). Subtype number conversion is indicated where appropriate.

^bNumbers refer to previous studies listed in the References. SP, substitution predicted to change receptor binding specificity based on structural modeling (data not shown).

^cH5.

^dH3.

^eInfluenza B.

^fUnpublished results.

A second approach was based on information regarding the molecular basis for antigenic change during evolution of other influenza virus subtypes. Substitutions at seven positions (145, 155, 156, 158, 159, 189, and 193; H3 numbering) were entirely responsible for the major antigenic changes during evolution of A(H3N2) virus (11), three of which (151, 185, and 189; H5 numbering) were also identified as key positions for antigenic change of A(H5N1) clade 2.1 virus (12). The corresponding A(H1N1)pdm09 virus positions were identified, and where possible, the substitution responsible for the observed antigenic change was introduced (Table 1). Multiple mutants were made if it was not possible to introduce the desired substitution. For example, 156QH was responsible for a major antigenic change during A(H3N2) virus evolution, and we therefore generated mutants with 153KQ and 153KH.

Third, we selected substitutions shown or predicted to affect receptor binding specificities of influenza viruses of the A(H1N1), A(H3N2), and A(H5N1) subtypes (Table 1). A(H1N1)pdm09 viruses containing the substitutions 222DE and 222DG were previously tested for their effects on receptor binding and antigenic properties by Chutinimitkul et al. (24). That work additionally suggested that amino acid differences at positions 142, 187, 222, and 224 were responsible for differences in receptor binding between a 1918 A(H1N1) virus and A/Netherlands/602/09 (24, 31). Mutants with these substitutions, or combinations thereof, were included in the current more extensive analyses. Substitutions 222QL and 224GS (223QL and 225GS in H1 numbering) were previously shown to affect the receptor specificity of A(H5N1) virus (32).

Recombinant viruses with the modified A/Netherlands/602/09 HA gene and the seven remaining gene segments of A/Puerto Rico/8/34 were rescued by reverse genetics. We were unable to rescue the 127DN and 186AK mutants. The mutant with 187DE did not agglutinate TRBCs even upon concentration of the virus and was therefore omitted from the antigenic analyses. After MDCK cell passaging, we found an addition of substitution 153KE to the 222DG224EA mutant, and this mutant was therefore excluded from this study. Figure 1A indicates the positions of introduced substitutions on an A/California/04/09 HA crystal structure.

FIG 1.

Position of amino acid substitutions indicated on an A/California/04/09 HA crystal structure. (A) The three HA monomers are indicated in white, gray, and black; the RBS is in yellow. Amino acid positions that were mutated in this study are indicated in orange. (B) Zoom image of the globular head of HA. Amino acid substitutions in mutants that were substantially antigenically different from A/Netherlands/602/09 (escape mutants) are indicated in red; substitutions in mutants that were antigenically similar to A/Netherlands/602/09 and that were included in further antigenic analyses (nonescape mutants) are indicated in blue.

Analysis of antigenic properties using ferret antisera.

HI assays with a ferret antiserum prepared against A/Netherlands/602/09 were performed to test if the mutant viruses could escape recognition by antibodies against the wild-type virus. When comparing HI titers obtained with an individual antiserum, viruses were considered substantially antigenically different from the reference virus if the HI titer was at least 4-fold (2 log₂) lower. Ten of the 27 mutants were substantially antigenically different from A/Netherlands/602/09 (Fig. 2): the 224EA, 127DT, 155GE, 156ND, 156NG, 153KE, 156NY, 156NS, 152VT156NS, and 155GE224EA mutants.

FIG 2.

HI titer differences between viruses with wild-type or mutant HAs against an A/Netherlands/602/09 ferret antiserum. Viruses with either wild-type or mutant A/Netherlands/602/09 HAs were tested in HI assays with a ferret antiserum prepared against the A/Netherlands/602/09 wild-type virus. Each point represents the log₂ HI titer difference between a mutant and A/Netherlands/602/09. Mutants with HI titers at least 4-fold (2 log₂) lower than that of A/Netherlands/602/09 (dashed line) were considered substantially antigenically different. The viruses are ordered by the log₂ HI titer difference from A/Netherlands/602/09.

Two groups of mutants were further tested in HI assays with a panel of ferret A(H1N1)pdm09 antisera. One group contained the 10 viruses that were substantially antigenically different from A/Netherlands/602/09; the second group contained seven mutants, the 84SN, 152VT, 190SN, 222DE, 222DG, 223QL, and 223QR mutants, that were antigenically similar to A/Netherlands/602/09 (Fig. 1B and 2). These groups are referred to here as escape mutants and nonescape mutants, respectively. The panel of 14 ferret antisera was prepared against seven A(H1N1)pdm09 viruses isolated between 2009 and 2011. Of these, six wild-type viruses were antigenically similar to A/Netherlands/602/09, in agreement with the fact that these viruses did not have any of the substitutions present in the mutants. However, A/Netherlands/219/11 contains the 155GE substitution and had a >8-fold (3-log₂)-lower mean HI titer (data not shown).

The mean HI titers of the nonescape mutants and the 224EA mutant were less than 2-fold lower than that of A/Netherlands/602/09 with the panel of ferret antisera (Fig. 3). The 156ND, 127DT, 155GE, and 156NG escape mutants had 3- to 4-fold-lower mean HI titers than A/Netherlands/602/09 (4-fold for the 155GE mutant when antisera prepared against A/Netherlands/219/11 were not included). The mean HI titers of the 156NS, 152VT156NS, 156NY, 155GE224EA, and 153KE mutants were up to 14-fold (3.8 log₂) lower than A/Netherlands/602/09 HI titers. These results indicate that the amino acid substitutions in all escape mutants except the 224EA mutant caused evasion of recognition by antisera raised against this panel of A(H1N1)pdm09 viruses. The 156NS and 152VT156NS mutants were antigenically similar, as were the 152VT mutant and A/Netherlands/602/09, thus indicating that 156NS was solely responsible for the antigenic change of this double mutant. Introduction of 224EA as single substitution had only minor effects on HI titers with most antisera. The 155GE224EA double mutant displayed a mean 12-fold (3.6 log₂) decrease in HI titer, indicating that these substitutions had a cumulative antigenic effect.

FIG 3.

HI titer differences between viruses with wild-type or mutant HAs in ferret antisera. Each point in panel A represents the log₂ HI titer difference between a mutant and A/Netherlands/602/09 for an individual ferret antiserum. The viruses are ordered by the mean log₂ HI titer difference from A/Netherlands/602/09, which is indicated as red horizontal lines. Names of escape and nonescape mutants are shown in black and gray, respectively. Ferret antisera are indicated in the leftmost column of panel B and are ordered from top to bottom by a decreasing ability of the serum to inhibit the test viruses in the HI assay. Two antisera (labeled A and B) were prepared against each virus. HI titers are color-coded for the difference from A/Netherlands/602/09 (NL602): orange, equal to or higher than that of A/Netherlands/602/09; yellow, up to 2-fold lower; green, 2- to 4-fold lower; cyan, 4- to 8-fold lower; blue, 8- to 16-fold lower; purple, 16- to 32-fold lower; and magenta, at least 32-fold lower.

Analysis of antigenic properties using human infant sera.

Antibodies in sera obtained from infants that experienced a primary A(H1N1)pdm09 virus infection were elicited in response to a single antigenic variant of influenza virus, as was the case for the antibody repertoire of inoculated ferrets. We next examined if the substitutions that led to antigenic variation as tested by ferret antisera were sufficient to escape recognition by human antibodies. Forty-nine surplus sera from unvaccinated infants were tested in HI assays for the presence of antibodies to A(H1N1)pdm09 virus. Six sera from infants born in 2009 or 2010 that were between 6 and 11 months of age at the time of sampling had detectable HI titers for A/Netherlands/602/09. HI titers for A/Brisbane/59/07 [seasonal A(H1N1)] were below the detection limit of the HI assay at a starting dilution of 1/40. This suggested that the infants experienced a primary infection and that maternal antibodies potentially present in the sera did not influence our results. Escape and nonescape mutants were tested in HI assays with sera 6, 11, 14, and 16. Nonescape mutants were not included in HI assays with sera 7 and 25 because of insufficient material available for these sera.

Escape and nonescape mutants had 3.5- to 14-fold-lower mean HI titers than A/Netherlands/602/09 (Fig. 4A). The variation in HI titers between the different infant sera was larger than for ferret antisera, and contradicting results were obtained with different sera. For example, the 222DE mutant was antigenically similar to A/Netherlands/602/09 using ferret sera and infant serum 14, yet the HI titer with serum 16 was more than 10-fold lower. Sera 6, 7, and 16 poorly discriminated between the mutants. Serum 7 yielded a low HI titer, 160, for A/Netherlands/602/09, and the HI titers of the mutants tested with this serum were similarly low or up to 4-fold higher. The HI titers with sera 6 and 16 were at least 8-fold lower than that of A/Netherlands/602/09 for all mutants.

FIG 4.

HI titer differences between viruses with wild-type or mutant HAs against human infant sera. (A) Symbols, order, and nomenclature are as in Fig. 3. (B) HI titers are color-coded for the difference from A/Netherlands/602/09: orange, equal to or higher than that of A/Netherlands/602/09; yellow, up to 2-fold lower; green, 2- to 4-fold lower; cyan, 4- to 8-fold lower; blue, 8- to 16-fold lower; purple, 16- to 32-fold lower; and magenta, at least 32-fold lower. NT, the virus-serum combination was not tested. (C) The analysis was repeated with inclusion of only the sera that differentiated between the mutants (sera 6, 7, and 16 were omitted). Nonescape mutants were not tested with serum 25 because of insufficient material available for this serum. Gray horizontal lines indicate the mean log₂ HI titer difference from A/Netherlands/602/09 when this serum was also omitted for the other mutants.

To test if escape mutants had an antigenic advantage over nonescape mutants in evasion of the infant sera that discriminated between the mutants, the results from serum 7 (which yielded similarly high titers for all viruses) and sera 6 and 16 (which yielded low titers for all mutants) were omitted and the analysis was repeated. Nonescape mutants and the 156ND mutant had mean HI titers that ranged from similar to 2-fold lower than that of A/Netherlands/602/09; individual HI titers with the different sera were up to 3-fold lower (Fig. 4B and C). The remaining mutants, which were all escape mutants, had mean HI titers 2- to 10-fold lower than A/Netherlands/602/09 HI titers and had one or more individual HI titers that were at least 4-fold lower than HI titers of A/Netherlands/602/09. Therefore, substitutions responsible for antigenic change as measured by ferret antisera can also mediate escape from recognition by human antibodies elicited in response to a primary infection with an A(H1N1)pdm09-like virus. However, in contrast to the ferret sera, the infant sera were sometimes nondiscriminating (serum 7) or reacted low across the board with all mutants (sera 6 and 16). Thus, the 127DT, 153KE, 155GE, 156ND, 156NG, 156NS, 156NY, 152VT156NS, and 155GE224EA mutants escaped recognition by antibodies in ferret sera and human infant sera (Fig. 3 and 4).

Analysis of virus replication.

Next, plaque assays were performed to test the effects of the introduced substitutions on replicative fitness. Plaque sizes were determined at a fixed time point after inoculation and were used as a proxy for replication efficiency; i.e., larger plaques indicated more efficient replication.

The majority of the mutants displayed plaque sizes similar to those of A/Netherlands/602/09 (Fig. 5). The 187DE, 156NY, 224EA, 84SN, and 222DG single mutants displayed reduced plaque sizes. The 127DT, 153KH, 153KE, 155GE, 153KQ, and 156ND mutants had predominantly larger plaques. Of the 10 mutants that had an antigenic effect in HI assays with ferret or infant sera, only the 156NY and 224EA mutants had substantially smaller plaques. The 155GE224EA double mutant displayed plaque sizes similar to those of A/Netherlands/602/09, indicating that 155GE compensated for the adverse effect of 224EA. Thus, eight mutants (the 127DT, 155GE, 156ND, 156NG, 153KE, 156NS, 152VT156NS, and 155GE224EA mutants) were antigenically different from A/Netherlands/602/09, with no apparent loss of replication efficiency (Fig. 3, 4, and 5).

FIG 5.

Effects of substitutions on virus replication. MDCK cells were inoculated with viruses containing wild-type or mutant A/Netherlands/602/09 HAs. After 36 h, the plaque sizes were determined as a measure of replication efficiency. Each point indicates the size of a single plaque. The mutants are ordered by increasing median plaque size, which is indicated by red horizontal lines. Escape mutants are in bold.

Effects of substitutions on receptor binding specificity.

To test if substitutions that caused antigenic change in the HI assays had altered receptor binding specificity, the escape mutants were tested in hemagglutination assays using normal TRBCs or TRBCs resialylated to contain either α2,3- or α2,6-linked SAs. Removal of SAs from the TRBC surface and correct resialylation were confirmed in hemagglutination assays using the avian A/Vietnam/1194/2004 (H5N1) and A/Netherlands/213/03 (H3N2) viruses. A/Netherlands/602/09 and all mutants yielded hemagglutination titers with TRBCs reconstituted with α2,6-linked SA (Table 2). Only the 153KE, 155GE, and 156ND mutants showed a very weak binding to TRBCs with α2,3-linked SAs. The 156NG and 156NS mutants showed markedly reduced binding to α2,6-reconstituted TRBCs compared to unmodified TRBCs. Although some mutations that affect antigenic properties appeared to influence receptor specificity to some extent, a clear tendency toward reduced affinity for α2,6-linked SAs was not demonstrated.

TABLE 2.

Agglutination of TRBCs by viruses with wild-type or mutant HAs^a

Antigen	HA titer
	TRBC	VCNA	α2,3-TRBC	α2,6-TRBC
A/Netherlands/602/09	512	0	0	32
127DT mutant	128	0	0	32
153KE mutant	128	0	1	64
155GE mutant	512	0	1	16
156ND mutant	512	0	2	128
156NG mutant	256	0	0	4
156NS mutant	256	0	0	2
156NY mutant	64	0	0	32
224EA mutant	128	0	0	64
152VT156NS mutant	128	0	0	8
155GE224EA mutant	1,024	0	0	1,024
A/Vietnam/1194/2004	128	0	256	0
A/Netherlands/213/2003	256	0	0	256

^aHemagglutination titers are expressed as the HAU with unmodified TRBCs, TRBCs stripped from SAs using VCNA, or TRBCs resialylated to contain either α2,3- or α2,6-SA. A/Vietnam/1194/2004 and A/Netherlands/213/2003 served as typical avian and human viruses with α2,3- and α2,6-SA preferences, respectively.

Frequency of substitutions in natural isolates.

We analyzed the frequencies of the tested substitutions in 10,422 A(H1N1)pdm09 sequences from GenBank submitted between April 2009 and February 2014 (data not shown). Four substitutions were detected in more than 1% of the sequences: those in the 222DE, 222DG, and 223QR nonescape mutants (4.80, 1.94, and 1.60%, respectively) and that in the 155GE escape mutant (1.46%). Of the mutants with substantially increased plaque sizes, substitutions 155GE, 156ND, 153KE were found in 1.46, 0.36, and 0.17% of the sequences, respectively. Substitutions 127DT and 153KH were not detected. Substitution 156NS, which had a large antigenic effect but did not affect plaque size, was detected in 0.18% of the sequences. Other substitutions in escape mutants that did not affect plaque size or resulted in substantially smaller plaques were not detected.

Analysis of antigenic properties using human sera.

The majority of the human population has a more broadly reactive antibody repertoire than influenza virus-inoculated ferrets or infants after primary infection due to previous infections or vaccination with seasonal A(H1N1) viruses (33, 34). Therefore, we also tested the mutants in HI assays with human sera obtained from individuals anticipated to have experienced both seasonal A(H1N1) virus and A(H1N1)pdm09 virus infections. We first tested the reactivity of a panel of pre-2009 swine (origin) viruses (A/swine/Shope/56 and A/New Jersey/8/76) and human vaccine strains (A/USSR/92/77, A/Chile/1/83, A/Taiwan/1/86, A/New Caledonia/20/99, and A/Brisbane/59/07) to a ferret antiserum prepared against A/Netherlands/602/09. HI titers were below the detection limit of the HI assay at a starting dilution of 1/20 (data not shown), indicating that these viruses were antigenically distinct from A(H1N1)pdm09 viruses. Twenty-one sera obtained from individuals aged 7 to 85 at the time of serum collection (October 2009 to May 2011) were selected based on their ability to inhibit A/Netherlands/602/09 in HI assays. The sera were tested in HI assays with 10 seasonal A(H1N1) viruses covering the period from 1977 to 2009 (Fig. 6A). All sera had detectable HI titers with five or more seasonal A(H1N1) viruses, demonstrating a broadly reactive antibody repertoire. Subsequently, the escape and nonescape mutants and A/Netherlands/602/09 were tested in HI assays using the selected sera (Fig. 6B and C). The mean HI titers of escape as well as nonescape mutants were less than 1.5-fold (0.54 log₂) lower than A/Netherlands/602/09 HI titers, suggesting that none of the mutants escaped recognition by this panel of sera. The majority of individual HI titers with the different sera were also less than 2-fold lower than A/Netherlands/602/09 HI titers. Six mutants had a single titer that was more than 2-fold lower than the A/Netherlands/602/09 HI titer: the 155GE, 155GE224EA, 223QL, and 222DG mutants displayed a 4-fold-lower HI titer, and the 224EA and 127DT mutants had 8- and 12-fold-lower HI titers, respectively. The antigenic effect of the substitutions in neither nonescape nor escape mutants was therefore sufficient to escape recognition by this panel of human sera.

Fig 6.

HI titer differences between viruses with wild-type or mutant HAs against human sera. (A) HI titers of human sera against antigenic variants of seasonal A(H1N1) viruses isolated between 1977 and 2009 (USSR77, A/USSR/90/77; NL78, A/Netherlands/3075/78; TA89, A/Taiwan/1/89; NL87, A/Netherlands/414/87; NC99, A/New Caledonia/20/99; NL99, A/Netherlands/271/99; NL03, A/Netherlands/02/03; NL06, A/Netherlands/364/06; SS06, A/Solomon Islands/03/06; and NL09, A/Netherlands/1005/09). (B) HI titer differences between viruses with wild-type or mutant NL602 HAs against human sera. Symbols, order, and nomenclature are as in Fig. 3. (C) HI titers are color-coded for the difference from A/Netherlands/602/09: orange, equal to or higher than A/Netherlands/602/09; yellow, up to 2-fold lower; green, 2- to 4-fold lower; cyan, 4- to 8-fold lower; and blue, 8- to 16-fold lower. The first two digits of the serum number indicate the age of the individual at the time of sampling.

DISCUSSION

In this study, we attempted to identify amino acid substitutions that contribute to antibody escape of A(H1N1)pdm09 viruses. Substitutions introduced into the HA of A/Netherlands/602/09 were selected based on genetic and antigenic changes of influenza A viruses that circulated in the past and on substitutions associated with changes in receptor binding specificity. We show that at least nine mutants were antigenically distinct from A/Netherlands/602/09 in HI assays using ferret antisera and found that the substitutions that caused escape from ferret antibodies also allowed evasion of antibodies in some human sera collected after primary infection.

Substitution 127DT introduces a glycosylation pattern (N-X-S/T-X) that potentially adds a carbohydrate side chain at position 125. Carbohydrate side chains can mask antibody epitopes and have been shown to change the antigenic properties of influenza viruses (15, 16). Seasonal A(H1N1) viruses that circulated from the 1930s until 1986 had a glycosylation site starting at position 127, while viruses that circulated from 1986 onwards had a glycosylation site starting at position 125. Both glycosylation sites were absent from the 1918 A(H1N1) and A(H1N1)pdm09 viruses. The absence of glycosylation at these positions has been suggested to contribute to the antigenic difference between seasonal A(H1N1) viruses and A(H1N1)pdm09 viruses (35). Our results indicate that reintroduction of a potential glycosylation site in this region was sufficient for escape from antibodies generated in response to A(H1N1)pdm09 virus.

Substitutions 153KE and 155GE in A(H1N1)pdm09 virus are analogous to substitutions 156KE and 158GE, which were responsible for major antigenic changes during evolution of A(H3N2) viruses (11). The observation that these substitutions changed the antigenic properties of an A(H1N1)pdm09 virus is also in agreement with previous studies that reported escape of A(H1N1)pdm09 virus from monoclonal antibodies (36, 37) and from polyclonal antibody responses in ferrets (35, 38, 39).

The four different substitutions introduced at position 156 (ND, NG, NY, and NS) all had a large antigenic effect. Two additional substitutions at this position, 156NK and 156NE, were previously reported to allow escape of antibody neutralization of A(H1N1)pdm09 virus (37, 38). Interestingly, the large changes in biophysical properties of the substituted amino acids introduced by 156ND or 156NY (charge and volume differences, respectively) had an antigenic effect similar to that of the small difference in biophysical properties introduced by 156NS. This finding emphasizes the potential importance of this position for antigenic change of A(H1N1)pdm09 viruses.

The antigenic effect of 155GE224EA in HI assays with ferret antisera was larger than the sum of the effects of the single substitutions. A possible explanation for how substitutions bordering opposite sides of the RBS can amplify each other's effect on antibody escape is given by the greater surface area of an antibody footprint in relation to the size of the RBS (40), which allows more efficient escape from antibodies that cover the RBS in addition to the proportion of antibodies evaded by the individual substitutions. Interestingly, substitution 224EA had a small antigenic effect in HI assays with ferret antisera but was substantially different from A/Netherlands/602/09 when tested with infant sera. In tests with infant sera, the 155GE224EA and 224EA mutants were equally different from A/Netherlands/602/09, indicating that 155GE did not add to the antigenic effect of 224EA. These results suggest that position 224 plays a more prominent role in escape from human antibodies than in escape from ferret antibodies. Antigenic variants with a substitution at this position may therefore not be readily detected in conventional HI assays using ferret antisera.

The high mutation rate of influenza A viruses and the observation that single substitutions caused substantial antigenic change during evolution of A(H3N2), A(H1N1), and influenza B viruses contradict the relatively low rate at which influenza viruses have changed antigenically (11). One possible explanation for this paradox is that substitutions responsible for escape from antibodies targeting the RBS have an adverse effect on HA function, necessitating the co-occurrence of compensatory substitutions, which slows down the emergence of new antigenic variants. Substitutions in and surrounding the A(H1N1)pdm09 RBS may affect receptor binding and consequently change replication efficiency (41). Surprisingly, we found that most of the substitutions that caused antigenic change in HI assays with ferret and human sera did not result in less efficient virus replication. Substitutions at positions 153 and 156 often had large effects on replication efficiency. Interestingly, changes of the lysine at position 153 for three biophysically diverse amino acids all resulted in increased replication efficiency. Replacing the asparagine at position 156 with either an aspartic acid or tyrosine had opposite effects on replication. These results suggest that positions 153 and 156 are key determinants of replication efficiency. It should be noted that replication assays are a surrogate measure of viral replication in vivo and that results may vary with the model system used. In addition, differences in replication efficiency may in part be caused by the use of an A/Puerto Rico/8/34 backbone. However, other studies have also demonstrated that A(H1N1)pdm09 viruses with 153KE, 155GE, and 156ND had increased replication efficiency in eggs or MDCK cells (21, 35), and they were often associated with cell culture adaptation (21, 35, 38).

The substitutions that caused antigenic change in HI assays with ferret and human infant sera were, with the exception of 224EA, located in or near the loop that consists of amino acid positions 151 to 159 (151–159 loop). Substitutions in this region have previously been shown to affect the antigenic properties of A(H3N2) and A(H5N1) viruses (11, 12). In addition, these positions often affected the replication efficiencies of the mutants constructed in this study. Amino acid substitutions at other positions tested in this study that caused similarly large changes in biophysical properties did not substantially change the antigenic properties or replication efficiency. However, it should be noted that it is possible that other amino acid changes outside the 151–159 loop, that were not included in this study, can result in phenotypic changes (42). Our results are in agreement with previous studies that reported mouse or human monoclonal antibodies (36, 37) or antibodies in ferret antisera (38) that target the 151–159 loop. We show here that several substitutions in or affecting the 151–159 loop caused antigenic escape while retaining replicative fitness. These results substantiate the importance of this region for antigenic evolution of A(H1N1) viruses.

We found large differences in the reactivities of infant antisera. Surprisingly, three sera (from the youngest infants) had almost uniform HI titers with all mutants that were either the same as that of A/Netherlands/602/09 or much lower. Infection in the presence of maternally derived antibodies or mechanisms associated with the transformation of the neonatal immune system to a more mature immunological phenotype may explain the difference from the other infant sera (43, 44). Nevertheless, the different reactivities of these three sera warrant further investigation. However, the remaining three sera had reactivities similar to those observed with ferret antisera. This indicates that substitutions that caused escape from ferret antibodies also promote escape from human antibodies elicited in response to infection with an A(H1N1)pdm09-like virus. In contrast, the substitutions responsible for escape from ferret and infant sera were not sufficient to allow escape from recognition by antisera from individuals that experienced infection with seasonal and pandemic A(H1N1) viruses. Thus, the antibody repertoire of the individuals that had a more experienced immune system included antibodies that were absent in the sera of infants. These results suggest that antibodies directed to epitopes on seasonal A(H1N1) viruses are cross-reactive with epitopes on the mutants tested in this study and thereby complement the antibody repertoire elicited in response to A(H1N1)pdm09 infection alone. The presence of such antibodies in much of the population may also explain why mutations that gave an in vitro replication advantage have not been detected at higher frequencies. Carter et al. showed that sera from ferrets sequentially infected with antigenically diverse seasonal A(H1N1) viruses can efficiently neutralize A(H1N1)pdm09 virus, whereas sera from ferrets infected with a single seasonal A(H1N1) virus were not cross-reactive (45). Additionally, a recent study by Miller et al. indicated that antibodies against previously encountered influenza A viruses are periodically boosted upon natural exposure to drift variants of the same subtype (33), possibly resulting in the more broadly reactive antibody repertoire observed in this study. This study and other studies (39, 42) suggest that the complexity of the repertoire of antibodies to A(H1N1) viruses in much of the human population cannot be represented by antigenic analyses based on single-infection ferret antisera. This limits the use of ferret antisera to test the ability of an antigenic variant to escape from population immunity.

In conclusion, substitutions in or near the RBS can influence the antigenic properties of A(H1N1)pdm09 viruses. Based on the current and previous studies of antigenic change of influenza A viruses (11, 12), it is probable that emerging antigenic variants of A(H1N1)pdm09 viruses will escape from population immunity because of substitutions in or near the RBS. However, our results also suggest that the presence of antibodies directed to epitopes on seasonal A(H1N1) and A(H1N1)pdm09 viruses in much of the population limits the number of antigenic variants that can emerge to cause new epidemics.