INTRODUCTION
Influenza A virus (IAV) causes a serious public health problem. The virus is endemic in the human population and circulates globally, causing seasonal epidemics that result in the deaths of up to half a million people each year (
1). Occasionally, a large antigenic shift occurs in IAV, resulting in an influenza pandemic, which can threaten the lives of millions (
2). The first pandemic of the 21st century was caused by an H1N1 subtype of IAV in 2009 (2009 H1N1) (
3,
4). The hemagglutinin (HA) of this virus was antigenically distinct from the previously circulating seasonal human H1N1 (hH1N1) viruses (
5–7) and resembled that of swine-lineage H1N1 viruses (
8). Based on a structural analysis, the HA epitopes of the 2009 H1N1 virus appeared to be similar to those of the 1918 H1N1 “Spanish Flu” virus, which caused one of the largest known pandemics and is the evolutionary ancestor of human- and swine-lineage H1N1 viruses (
9–11). Another similarity between the two pandemic viruses is the absence of N-linked glycosylation (NLG) at the globular head of HA (
12,
13). These features of the 2009 H1N1 virus explained why antibodies that protected against infection with the 2009 H1N1 virus were detected in the elderly who had experienced the 1918 pandemic and in people who had been vaccinated against the “swine flu” in 1976 (
5,
6,
14).
During evolution in humans, IAV accumulates genetic mutations at the antigenic sites of HA that circumvent preexisting immunity, a process known as antigenic drift (
15). Some of these mutations result in variations in the glycosylation state of the HA associated with changes in viral antigenicity (
16) and with immune evasion (
17). Since 1918, hH1N1 viruses have acquired multiple NLGs, which occur at asparagine (N) residues in accessible N-Xaa-S/T (Xaa, any amino acid except proline) sequons, in the top, side, and stem regions of HA (
12,
18). However, HA is neither hyperglycosylated (
19) nor glycosylated at random sites (
20). Using the HA amino acid sequences of hH1N1 viruses that were available from influenza virus database of the National Centers for Biotechnology Information (NCBI), we compared the NLG patterns in the head region of these HAs. As noted previously (
12), we found that various combinations of NLG using only five glycosites (asparagines at 142, 144, 172, 177, and 179 of HA; H1 numbering was applied in the present study unless otherwise specified) have been utilized by hH1N1 viruses (
Table 1). In the 1930s, a single NLG at residues 142, 144, or 179 was placed in the HA of hH1N1 viruses. From 1942 to 1985, hH1N1 viruses maintained several combinations of NLGs at residues 144, 172, 177, and/or 179; residue 179 was utilized mostly until 1948. In 1986, hH1N1 viruses exhibited a double mutation, K142N and N144S/T, which allowed a glycan shift from residue 144 to residue 142. The NLGs at residues 142 and 177 then became the NLG signature of hH1N1 viruses until the sugar-free HA head was observed in the 2009 H1N1 viruses (
Table 1) (
21).
A recent report showed that the presence of lysine (K) at HA residue 147 (K147; referred to as 133a in H3 numbering) of the A/California/04/2009 (Ca/04; a 2009 H1N1 virus) is associated with the stabilization of the interaction between HA and sialic acid (SA) (
22). K147 is also present in the HA of the 1918 H1N1 virus; however, this signature was not present in all of the subsequent hH1N1 descendants. Some of the 1918 H1N1 descendants had an arginine (R) or isoleucine (I) at HA residue 147 instead of a K (referred to as HA-147-bearing hH1N1 viruses), whereas in others this amino acid was even absent (referred to as HA-Δ147 hH1N1 viruses) (
Table 1).
Based on these phenomena, we hypothesized that the NLG-free headed and K147-bearing HA of 2009 H1N1 virus might evolve into a similar genetic flow as previously seen in 1918 H1N1 descendant viruses. We thus investigated how the attached N-linked glycans and the polymorphism at HA residue 147 affected viral characteristics. Using mutant viruses generated by reverse genetics, we show here the importance of the polymorphism at HA residue 147. When the globular head region of HA acquired specific N-linked glycans, HA residue 147 was indispensable for viral fitness, and it allowed NLG-containing viruses to be transmissible in an animal model. In addition, the HA residue 147 affected the immunogenicity of the H1N1 viruses. Considered together, we propose that the polymorphism at HA residue 147 cooperates with NLG acquisition at the HA head of hH1N1 viruses to evade preexisting immunity without compromising viral fitness to the point of losing their transmissibility.
MATERIALS AND METHODS
Viruses and cells.
Viruses were propagated in 10-day-old embryonated chicken eggs. To determine the virus titers, the viruses were hemagglutinated with turkey red blood cells (tRBCs), or plaque assays were carried out in Madin-Darby canine kidney (MDCK) cells. All of the viruses were plaque purified and sequenced before use in the experiments. MDCK cells were obtained from the American Type Culture Collection (ATCC), maintained in Eagle minimal essential medium, and used for the cell-based assay. Human embryonic kidney (293T) cells and human lung epithelial (A549) cells were obtained from the ATCC and maintained in Dulbecco modified Eagle medium. All of the media were supplemented with 10% fetal bovine serum and antibiotics.
Animal experiments.
To minimize animal suffering, all animal procedures performed in the present study were conducted in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the Animal, Plant, and Fisheries Quarantine and Inspection Agency of Korea. The protocol was approved by the Institutional Animal Care and Use Committee of Hallym University (permit number Hallym 2012-24).
To determine the 50% lethal dose (LD
50) for mice and their body weight changes, five C57BL/6 mice (female, 5 weeks old; NaraBiotech, Korea) per group were anesthetized and intranasally infected with each virus. Body weight changes and survival rates were recorded until 14 days postinfection (dpi). Mice that experienced more than a 25% loss in body weight were considered experimentally dead and were humanely euthanized. The LD
50 was determined using the Reed and Muench method (
23). To determine virus titers in the lungs, six C57BL/6 mice per group were anesthetized and intranasally infected with 10
5 PFU of each virus. At 3 and 6 dpi, three mice per group were sacrificed for lung collection. The collected lung samples were homogenized with TissueLyzer II (Qiagen, Germany) and were titrated to assess the presence of virus by plaque assay in MDCK cells.
For a direct contact transmission model, guinea pigs (Hartley strain, female, 5 to 6 weeks old; Charles River Laboratories, Wilmington, MA) were anesthetized and intranasally infected with 105 PFU of each virus. The following day, a naive guinea pig was co-caged with each infected guinea pig. To collect nasal wash samples, the anesthetized guinea pigs were intranasally lavaged with 1 ml of phosphate-buffered saline supplemented with penicillin-streptomycin and 0.3% bovine serum albumin (PBS-PS-BSA), and the lavage fluid was allowed to drain onto a sterilized petri dish. The virus that was present in the nasal wash samples was titrated by plaque assay in MDCK cells.
Antibodies.
To obtain guinea pig polyclonal antibodies specific for each virus, naive (influenza virus-negative) guinea pigs were intranasally primed and boosted with 105 PFU of virus at 2-week intervals. Antisera were then prepared from whole blood and treated with RDE (neuraminidase from Vibrio cholerae; Denka Seiken, Japan) before use. Sheep polyclonal antibodies specific for 2009 H1N1 HA (influenza anti-A/California/7/2009 H1N1 HA serum, NIBSC code 11/110) were obtained from the NIBSC (Hertfordshire, United Kingdom) and were used in Western blots to detect HA proteins.
Rescue of rK/09 and HA mutant viruses.
The A/Korea/01/2009 (K/09; NCBI taxonomy ID 644289) virus is a 2009 H1N1 strain isolated in Korea. This virus was used as the backbone virus for reverse genetics in the present study. All eight gene segments of the K/09 virus were cloned into ambisense pDZ plasmids (kindly provided by Peter Palese, Mount Sinai School of Medicine, New York, NY). After sequence analysis of these plasmids, HA mutant plasmids were constructed to reflect various NLG statuses and the genetic polymorphism at residue 147. The rK/09 and HA mutant viruses were then rescued by reverse genetics as previously described (
24). Briefly, cocultured 293T/MDCK cells were transfected with eight pDZ plasmids encoding the genes and proteins of wild-type K/09 or HA mutant viruses. At 48 to 72 h after transfection, supernatants were inoculated into 10-day-old embryonated chicken eggs and maintained at 37°C. The allantoic fluids were harvested 48 to 72 h later and titrated in HA and plaque assays to determine the viral presence. The rescued viruses were purified with a plaque assay and then repropagated in 10-day-old embryonated chicken eggs. All of the rescued viruses were sequence confirmed before being used in the experiments.
RBC binding assay.
To evaluate the RBC binding avidities of the HAs in each virus, a protocol modified from the work of Hensley et al. (
25) was used. Briefly, turkey RBCs (tRBCs) or human RBCs (hRBCs) (turkey or human blood in Alsever's solution, Rockland, PA) were treated with 0.025 to 3.2 μg of sialidase (α2-3,6,8 neuraminidase from
Vibrio cholerae; Roche, Germany)/ml at 37°C. After 1 h, the sialidase-treated RBCs were washed and diluted to 0.5% (vol/vol) in the RBC solution with PBS. A total of four HA units of virus in 50 μl were then allowed to agglutinate with RBCs for another 30 min. The HA-RBC agglutination was read, and the RBC binding avidities of viruses were determined in three independent experiments.
Growth kinetics in cultured cells.
For the growth property analysis, the MDCK and A549 cell monolayers were inoculated with each virus at a multiplicity of infection (MOI) of 0.001. After 1 h, the cells were washed five times and maintained in the appropriate medium supplemented with 0.3% BSA in the presence of TPCK (tolylsulfonyl phenylalanyl chloromethyl ketone)-treated trypsin (1 μg/ml). At the indicated time points, the supernatants were collected and titrated for viral presence by plaque assay in the MDCK cells. The results are presented as the mean titers at each time point of more than three independent experiments.
HI assay.
In 96-well plates, the hemagglutination inhibition (HI) titers of guinea pig antisera were determined against homologous and heterologous viruses using tRBCs. Briefly, 8 HA units of viruses in 25 μl were incubated with the same amount of guinea pig antisera at 37°C. After 1 h of incubation, the virus and each antiserum mixture were agglutinated with 50 μl of 0.5% tRBC for 30 min. The HI titers of antisera against homologous and heterologous viruses were determined in three independent experiments.
Plaque reduction neutralization test (PRNT).
A total of 100 PFU of each virus was incubated with the same amount of guinea pig antisera at 37°C. After 1 h of incubation, the virus and each antiserum mixture were plaque assayed in the MDCK cells. The control wells of MDCK cells were infected only with virus. The neutralization titers were determined by the dilution of each antiserum that resulted in a 50% reduction in the plaque titers compared to the controls. The PRNT titers of each antiserum were determined in three independent experiments.
Western blotting and PNGase F treatment.
To confirm the biochemical utilization of NLG in the HA of the rK/09 and HA mutant viruses, the monolayers of MDCK cells were infected with an MOI of 3. After 16 h of infection, the cell lysates were prepared for SDS-PAGE. The HA in the cell lysates was primarily detected by sheep anti-2009 pH1N1 HA-specific polyclonal antibodies and then secondarily detected by anti-sheep IgG-HRP antibodies. To confirm the mobility identities in the HAs of the rK/09 and HA mutant viruses (after introduced N-linked glycans were removed), the prepared cell lysates were incubated with PNGase F (from Flavobacterium meningosepticum, recombinant from E. coli; Roche, Germany) at 37°C for 1 h, and the HAs were then detected using the procedures described above.
DISCUSSION
It was previously noted that the addition or removal of NLG from the globular head of HA may alter the viral characteristics (
39,
40). The introduction of sugar side chains at the tip of HA resulted in the reduced receptor binding ability of the H2 and H5 subtypes (
41,
42). However, these two studies presented different perspectives. In the study of H2 HA, two attached N-linked glycans at residues 131, 160, or 197 (H3 numbering; 144, 174, or 211 in H1 numbering, respectively) decreased the polyvalent HA binding avidity to SA, and three NLG sites abolished the fusogenic property of the H2 HA. The study of H5 HA provided somewhat similar results; however, the weak HA-SA binding that was mediated by NLG eventually benefited viral replication by enhancing viral release (
43). When considering viral entry to the cell, NLG may decrease viral infectivity by decreasing the strength of the virus-cell interaction. Conversely, when considering the release of viral progeny from the cell, a weak HA-SA bridge may result in the easy release of viral particles, even in cases where the virus retains a neuraminidase (NA) with low enzymatic activity. In our study, mutant viruses retained the same NA protein of wild-type K/09 virus, and no spontaneous mutations in other genes or reverted ones in the HAs were identified even after the multiple passages in eggs, which meant that the altered viral characteristics observed were mainly due to the HA attached N-linked glycans. However, K147-bearing viruses could maintain viral fitness even in the presence of the NLG-driven constraints, whereas ΔK147 viruses experienced reduced viability or lost transmissibility under the same conditions (
Fig. 3 and
4).
The biological activity of K147 of H1 HA was recently studied for a structural aspect in relation to interactions between HA and SA (
22). Here we suggest a new role of residue 147 as an evolution indicator of H1 HA. In the 1918 H1N1 descendant viruses, residue 147 was occupied by amino acids K (RNA codons AAA or AAG), R (AGA or AGG), or I (AUA), and sometimes it was even deleted (
Fig. 1 and Table S1 in the supplemental material). Based on the juxtaposition of residue 147 next to RBS (
Fig. 2), we speculated that residue 147 might be inevitably affected by the HA-SA interaction and its resultant polymorphic appearance controlled NLG attachment around the RBS. Indeed, we found that glycosites 144 and 172 were never utilized in the HA-Δ147 hH1N1 viruses (
Table 2).
The deletion of residue 147 was intermittently seen in the HA of hH1N1 viruses and then fixed from 1995 to 2009, right before the appearance of 2009 H1N1 viruses. From 1947 to 1986, the HA globular head region of hH1N1 viruses mostly utilized at least three glycosites including 144, 172, and 177 (
Table 1). However, no hH1N1 viruses have utilized glycosite 144 since a NLG shift from 144 to 142 in 1986, and glycosite 172 disappeared in 1988 (
Table 1). Since then, residues 142 and 177 had been signatures for N-linked glycans in the majority of hH1N1 viruses. This eliminated the need for amino acid 147 required for the glycosites 144 and 172, and the reappearance of the 147 deletion. These findings led us to investigate viral characteristics of mutant viruses having various combinations of NLG patterns within the head of HA and amino acids at residue 147. In our results, r142-ΔK147 and r142-ΔK147-177 viruses exhibited reduced RBC binding ability and replication kinetics in A549 cells (
Fig. 3C and
E), and they were not transmitted to exposed guinea pigs (
Fig. 4G and
H). These results indicate that viral fitness is dramatically influenced by N-linked glycans attached at glycosites 142 or 177 in the absence of residue 147. According to a previous study, hH1N1 viruses retaining these kinds of NLG patterns irrespective of residue 147 needed compensation mutation(s) to reduce NLG-driven fitness costs (
19).
For HA residue 147, K might be a better genetic signature than R. Although further investigation needs to define the exact effect of R side chain (guanidinium cation) on viral fitness compared to that of other amino acid side chains, the r144-R147-177 virus exhibited reduced RBC binding and replication kinetics in A549 cells in our study (
Fig. 3C and
E). This virus also less efficiently infected guinea pigs and transmitted to only one out of three exposed guinea pigs (
Fig. 4E). Low HI and PRNT titers were then observed in the case of the r144-R147-177 virus-infected guinea pig antiserum (
Fig. 5D and
Tables 5 and
6). In fact, HA R147-bearing hH1N1 viruses circulating in the past have also appeared to be less immunogenic than HA K147-bearing hH1N1 viruses (
44).
Differences in antigenicity were observed according to the presence or absence of K147 in vaccine antigens. For the 1986-1987 influenza season, the World Health Organization (WHO) recommended the A/Chile/1/83 (Ch/83; harboring three glycosylations at 144, 172, and 177) virus as an H1N1 antigen for use in the inactivated trivalent influenza vaccine (TIV) (
Table 7) (
45). However, since an increasing number of glycosite 142-harboring H1N1 variants were isolated in many geographical regions (
46,
47), the WHO had to add a new H1N1 antigen, namely, the A/Singapore/6/86 (Sg/86; glycosylations at 142, 172, and 177) virus, for the 1986–87 influenza season. Until they were replaced with the A/Bayern/7/95 (By/95) virus in 1997, Sg/86-like viruses (including A/Taiwan/1/86 and A/Texas/36/91) represented H1N1 antigens in TIVs (
48). One NLG shift from 144 to 142 could effectively cover the Sa antigenic site of Sg/86-like vaccine antigens and thus it may have induced HI reactive antibodies against subsequent hH1N1 variants even though the amino acid variability at the Sa antigenic site increased during the same period (
49). In 1998, a new virus, Bj/95, was selected for the H1N1 vaccine antigen. The HA head of the Bj/95 virus held only two NLGs at residues 142 and 177 (
Tables 1 and
7). Considering that the Sg/86 vaccine virus controlled many hH1N1 NLG variants during the 1986-1997 influenza seasons and that the difference in NLG patterns between the Sg/86 and Bj/95 viruses seemed to be minor (the Sg/86 virus had three NLGs at residues 142, 172, and 177 within the HA head region, and the Bj/95 virus had only two NLGs at residues 142 and 177; the amino acid difference between two viruses is calculated only as 5.81% within HA1 sequences), there might be another reason for the antigenic difference between the Sg/86 and Bj/95 viruses. Thus, it appears that the absence of K147 in Bj/95 may have contributed to antigenic differences. In the HI results derived from the use of postinfection ferret antisera, Bj/95 antiserum reacted only with the HA-ΔK147viruses and were less reactive to Sg/86- and By/95-like viruses, which retained K147 in their HAs (
50,
51). A similar result was also seen in our HI assays. The r142-177 antiserum induced more than 640 HI titers to homologous or heterologous viruses and even produced 80 HI titers to the r142-ΔK147-177 virus (
Table 5). However, the r142-ΔK147-177 antiserum resulted in only 10 to 40 HI titers to the HA-K147-bearing viruses (
Table 5). Considered together, the selected H1N1 vaccine antigens reflect how the HA of hH1N1 viruses has evolved against human immune responses and the genetic signature of HA residue 147 has likely impacted the immunogenicity of the selected H1N1 vaccine antigens.
Over the past several decades of circulation, hH1N1 viruses have responded against external pressures, such as neutralizing antibodies, and the HA of hH1N1 viruses has evolved through changing antigenicity. Similar to the evolution pathway of the 1918 virus, some descendants of the 2009 H1N1 virus already utilized HA glycosite 179 (
Table 8) (
12) and selected arginine, instead of lysine, as a genetic signature for HA residue 147 (
Table 9). Even though it is uncertain how future hH1N1 viruses may or not recapitulate the evolutionary pathway of the old H1N1s, we suggest that the genetic signature at HA residue 147 is an indicator for the evolution dynamics of hH1N1 viruses and that the presence of this amino acid allows for a broader possible use of additional NLGs within the globular head region of HA. Also, the presence of amino acid 147 increases the breadth of the antibody response induced by natural infection.