Background
Although influenza A viruses (IAVs) infect many avian and mammalian species, only three subtypes (H1N1, H2N2, and H3N2) infect and transmit efficiently among humans [
1]. Viral hemagglutinin (HA) is the major antigenic glycoprotein responsible for binding the virus to cells, and it is, therefore, the major target of neutralizing antibodies. Indeed, the accumulation of amino acid substitutions, known as “antigenic drift,” which allows the virus to escape from the neutralizing antibody response, frequently occurs in HA [
2‐
4]. The attachment of an oligosaccharide to the N-glycosylation sites (NGSs) in the globular head region of HA via N-glycosylation sequons (i.e., Asn-X-Ser/Thr, where X is any amino acid, except Pro) contributes to the escape of viruses from host immune response [
5]. The number of NGSs in the globular head region of H3N2 virus HA has increased during circulation in the human population [
6‐
9], and most of the currently circulating H3N2 viruses have six (Asn residues 63, 122, 126, 133, 165, and 246) or seven (Asn residues 63, 122, 126, 133, 144, 165, and 246) NGSs. When the H3N2 virus first emerged in the human population in 1968, there were only two NGSs, at amino acid positions 81 and 165, in the globular head region of HA. Thereafter, the appearance or disappearance of glycosyl chains in the globular head has been reported to occur naturally during the antigenic drift of H3N2 from 1968 to 1975 [
2,
6,
10]. Glycosylation of HA has been shown to modulate the sensitivity of H3N2 viruses to innate proteins in airway secretion, the virulence properties of the strain [
11], and the replication of viruses in the respiratory tract of ferrets [
12].
Meanwhile, it is well known that H3N2 is one of the major subtypes of IAV that circulate in humans after the Hong Kong flu pandemic of 1968 [
1]. According to the National Institute of Infectious Diseases in Japan (
http://www.nih.go.jp/niid/en/), although the traditional seasonal influenza H1N1 virus caused the epidemic in 2008, this virus disappeared after the 2009 pandemic of H1N1 virus (H1N1pdm09), and H1N1pdm09 appears to have replaced the traditional H1N1 virus. This observation is supported by the fact that the rates of resistance to oseltamivir was 100 % among the H1N1 viruses circulating in Japan in the 2008–09 season, but suddenly dropped to 0.5 % in the 2010–11 season after the 2009 pandemic. However, the H1N1pdm09 lacks additional NGSs in the globular head region of HA. In contrast to the traditional H1N1, H3N2 became predominant again in Japan in the seasons (2010–11 and 2011–12) immediately following the 2009 pandemic of H1N1. These findings prompted us to investigate whether the re-emergence of H3N2 is due to a change in the HA within this subtype that escapes immunity from the prior H3N2 virus.
In this study, we therefore conducted a retrospective analysis of the amino acid substitutions of the HA globular head observed in seasonal H3N2 viruses during two consecutive seasons (2010–11 and 2011–12) immediately after the 2009 pandemic. Furthermore, using multiple approaches, we examined whether the presence and/or absence of N-linked glycans in HA affects viral neutralization, and in this manner, the re-emergence of H3N2.
Methods
Retrospective database analysis
Viruses and cells
Clinical specimens from patients with confirmed IAV H3N2 infection were obtained during two consecutive seasons (2010–11 and 2011–12) immediately after the 2009 pandemic of H1N1pdm09. Each specimen was propagated once in Madin-Darby canine kidney cells overexpressing α-2,6-sialyltransferase (MDCK-SIAT1 cells; DS Pharma Biomedical, Japan) in Eagle’s minimum essential medium (MEM) supplemented with 10 % fetal calf serum and 1 μg/ml acetylated trypsin (Sigma, Japan). For subsequent experiments, we chose four stored H3N2 isolates (virus stocks) from each season, in the order in which they were obtained, while also making sure that sufficient quantities of purified virus were available. All samples were obtained from patients who had not received any medication prior to sample collection. Four samples (A/Okayama/2/11, A/Okayama/3/11, A/Okayama/4/11, and A/Okayama/5/11) were isolated in the Okayama prefecture in the 2010–11 season and four samples (A/Shizuoka/10/12, A/Shizuoka/23/12, A/Shizuoka/24/12, and A/Shizuoka/26/12) were isolated in the Shizuoka prefecture in the 2011–12 season. All samples were stored in liquid nitrogen until use.
RT-PCR and sequence analysis
Viral suspensions propagated in MDCK-SIAT1 cells were centrifuged at 10,000 × g for 90 min at 4 °C, and viral RNA was extracted from the precipitate by using ISOGEN (Wako Chemicals, Japan). Subsequently, cDNA was synthesized from RNA by using the Omniscript RT Kit (QIAGEN, Germany) and reverse transcription-polymerase chain reaction (RT-PCR) was performed using Pyrobest polymerase (Takara, Japan) with the following primers: forward, 5’-TAA TTC TAT TAA CCA TGA AG-3’; reverse, 5’-TTT TTA ATT AAT GCA CTC AAA TGC-3’. The PCR products were subjected to agarose gel electrophoresis, and the specific bands were excised from the gel and purified using a QIAquick Gel Extraction Kit (QIAGEN, Germany). The purified PCR products were subjected to direct sequencing.
Generation of recombinant viruses
RNA polymerase I-driven expression plasmid (pPolI) expressing each gene segment of WSN and pCAGGS plasmids expressing the WSN viral proteins, PA, PB1, PB2, and NP, were kindly provided by Prof. Yoshihiro Kawaoka (University of Wisconsin). The cDNA of HA gene of A/Okayama/6/01 (H3N2) was prepared by RT-PCR and cloned into the pPolI vector designated as pPolI-Oka/6/01-wt (code named H3-0) in our previous study [
14]. For constructing HA mutant plasmid lacking glycosylation of Lys144 residue, which is designated as pPolI-Oka/6/01-mutant (code named H3-1), a single amino acid substitution, from Ser to Ala, at residue 146 was introduced into the pPolI-Oka/6/01-wt (H3-0) plasmid by using the following primers: 5’-AGA TCT AAT AAA GCT TTC TTT AGT AGA-3’ and 5’- TCT ACT AAA GAA AGC TTT ATT AGA TCT-3’. 293T cells were prepared as half-to-three-fourth confluence on the wells of 6-well cell culture plate for plasmid transfection. pPolI-Oka/6/01 (H3-0 or H3-1) and other pPolI plasmids encoding the vRNA of seven internal genes derived from WSN were transfected together with the pCAGGS plasmid into 293T cells by TransIT-293 Transfection Reagent (Minus Bio, USA), according to the manufacturer’s instructions. Transfected 293T was incubated at 37 °C in OPTI-MEM and the supernatant was harvested at 48 h post transfection. MDCK cells were inoculated with the collected supernatant to amplify the rescued viruses.
Preparation of antisera
After sequence analysis of the clinically isolated viruses, we chose three types of viruses for the production of polyclonal antisera in guinea pigs: A/Okayama/2/11 (Oka/2) as the 144K type, A/Shizuoka/23/12 (Sk/23) as the 144N type, and A/Shizuoka/26/12 (Sk/26) as the 144N/45N type (Table
1). These viruses were propagated in MDCK-SIAT1 cells and concentrated. Subsequently, 4-week-old naïve female guinea pigs (Hartley strain; Japan SLC, Hamamatsu, Japan) were intraperitoneally primed and boosted with each concentrated virus suspension mixed with an adjuvant (TiterMax Gold; CytRx Co., USA) at 2-week intervals. Finally, the antisera were prepared from the whole blood and stored at −80 °C until use. The sera from recovered virus-infected mice were also stocked at −80 °C. All animal experiments were approved by the Institutional Animal Care and Research Advisory Committee of Kawasaki Medical School, prior to initiation of the study.
Table 1
Glycosylation sites in the hemagglutinin (HA) of H3N2 viruses isolated from clinical specimens
2010–11 season |
A/Okayama/2/11 (Oka/2) | 4 | F | SSS | NCT | NES | NWT | NGT | KNS | NVT | NST |
A/Okayama/3/11 | 5 | M | SSS | NCT | NES | NWT | NGT | NNS | NVT | NST |
A/Okayama/4/11 | 6 | M | SSS | NCT | NES | NWT | NGT | KNS | NVT | NST |
A/Okayama/5/11 | 12 | F | SSS | NCT | NES | NWT | NGT | KNS | NVT | NST |
2011–12 season |
A/Shizuoka/10/12 | 4 | M | NSS | NCT | NES | NWT | NGT | NNS | NVT | NST |
A/Shizuoka/23/12 (Sk/23) | 2 | F | SPS | NCT | NES | NWT | NGT | NNS | NVT | NST |
A/Shizuoka/24/12 | 5 | M | NSS | NCT | NES | NWT | NGT | NNS | NVT | NST |
A/Shizuoka/26/12 (Sk/26) | 5 | M | NSS | NCT | NES | NWT | NGT | NNS | NVT | NST |
Hemagglutination inhibition (HI) assay
HA inhibition (HI) tests were performed for antigenic characterization of the isolates. Two-fold serial dilution lines of receptor-destroying enzyme (RDE)-treated antiserum in a volume of 25 μl were prepared with phosphate-buffered saline (PBS), and 25 μl of the virus antigen adjusted to 8 HA units/50 μl was added to each well. After 60-min incubation at room temperature, 50 μl of 0.5 % chicken erythrocytes in PBS was added to each well. After additional incubation for 45 min, the wells showing a block of HA were considered HI positive; the HI value was determined as the reciprocal of the dilution that was effective for HI.
Microneutralization assay
The procedure for influenza virus microneutralization assay was modified from that listed in the WHO Manual on Animal Influenza Diagnosis and Surveillance 2002 version (
http://www.who.int/csr/resources/publications/influenza/whocdscsrncs20025.pdf). Namely, three-fold serial dilution lines of the receptor-destroying enzyme-treated antiserum in a volume of 25 μl were prepared with virus growth medium (MEM containing 2 mM L-glutamine, 1× MEM amino acids, 1× MEM vitamin, 10 mM HEPES, and 0.2 % bovine albumin). Next, 20 μl of the virus antigen adjusted to 8 HA units/50 μl was added to each well. After 30-min incubation at room temperature, the antigen-virus mixtures were transferred into MDCK cells seeded in 96-well plates containing 2.5 μg/ml of tolylsulfonyl phenylalanyl chloromethyl ketone-trypsin (Sigma, Japan). After four days of incubation, the viral cytopathic effect was observed under an inverted microscope or by cell staining using amide black 10B.
Plaque reduction neutralization test
A total of 100 plaque-forming units (PFU) of each viral suspension were incubated with the same volume of 10-fold serial dilutions of antiserum in PBS at 37 °C. After 1-h incubation, each sample was subjected to plaque assay in MDCK-SIAT1 cells and indirect immunostaining as previously described [
15].
Statistical analysis
The Mann-Whitney U test was used to compare data between two groups. Values of p < 0.05 were considered statistically significant.
Discussion
IAV escapes host immune response by changing the antigenicity of HA and neuraminidase, both gradually (antigenic drift) and abruptly (antigenic shift) [
16]. Antigenic drift is achieved via changes in the amino acids at the antigenic sites that are recognized by antibodies [
17‐
19]. Since the efficacy of vaccines requires a close antigenic match between circulating and vaccine strains, and since mismatches result in increased disease burden, it is important to identify the mutations that affect significantly the neutralizing antibody response mounted against natural infection or vaccination. Especially, the antigenic sites of HA, which affect the receptor binding properties [
20] and virulence [
11] are crucial for understanding antigenic drift and vaccine strain selection. Moreover, it has been proposed that the attachment of a glycosyl chain to an NGS (Asn-X-Ser/Thr, where X is any amino acid except Pro) in the globular head of HA contributes to immune escape [
7,
20‐
24].
In the present study, we observed that H3N2 became predominant in the seasons (2010–11 to 2011–12) immediately following the 2009 pandemic of H1N1, and almost all the isolates had an N residue that forms an additional NGS at residue 144 of HA in the 2011–12 season. Interestingly, there was an antigenic mismatch between the vaccine strain and the circulating viruses in the 2011–12 season. The vaccine strain used in the 2011–12 season had a K residue that cannot form NGS at position 144 of HA (Table
5 and Fig.
2). It may therefore be possible that the newly acquired NGS in the HA globular head is associated with the re-emergence of H3N2 through escape from humoral immunity within this subtype. There was re-emergence of H1N1 in the 2013–14 season in both Japan and United States (Fig.
1). However, the virus characterization data indicated that the re-emerged H1N1 was genetically and antigenically similar to H1N1pdm09 [
25]. Therefore, antigenic drift did not play an important role in the re-emergence. Other factors such as a waning of immunity in the population or vaccine effectiveness may be responsible for this re-emergence.
Table 5
The strains used in seasonal influenza vaccines from 2001 to 2015 and amino acid residue(s) of potential N-linked glycosylation sites on the globular head of HA in these strains
2014–15 | A/Texas/50/2012 (H3N2) | N | NNS | A/Switzerland/9715293/2013 (H3N2) | N | NSS |
2013–14 | A/Victoria/361/2011(H3N2) | N | NNS | A/Texas/50/2012 (H3N2) | N | NNS |
2012–13 | A/Victoria/361/2011(H3N2) | N | NNS | A/Victoria/361/2011(H3N2) | N | NNS |
2011–12 | A/Perth/16/2009 (H3N2) | K | KNS | A/Perth/16/2009 (H3N2) | K | KNS |
2010–11 | A/Perth/16/2009 (H3N2) | K | KNS | A/Perth/16/2009 (H3N2) | K | KNS |
2009–10 | A/Brisbane/10/2007 (H3N2) | N | NNS | A/Perth/16/2009 (H3N2) | K | KNS |
2008–09 | A/Brisbane/10/2007 (H3N2) | N | NNS | A/Brisbane/10/2007 (H3N2) | N | NNS |
2007–08 | A/Wisconsin/67/2005 (H3N2) | N | NNS | A/Brisbane/10/2007 (H3N2) | N | NNS |
2006–07 | A/Wisconsin/67/2005 (H3N2) | N | NNS | A/Wisconsin/67/2005 (H3N2) | N | NNS |
2005–06 | A/California/7/2004(H3N2) | N | NNS | A/California/7/2004(H3N2) | N | NNS |
2004–05 | A/Fujian/411/2002(H3N2) | N | NKS | A/Wellington/1/2004(H3N2) | N | NKS |
2003–04 | A/Moscow/10/99(H3N2) | I | INS | A/Fujian/411/2002(H3N2) | N | NKS |
2002–03 | A/Moscow/10/99(H3N2) | I | INS | A/Moscow/10/99(H3N2) | I | INS |
2001–02 | A/Moscow/10/99(H3N2) | I | INS | A/Moscow/10/99(H3N2) | I | INS |
We tested the effect of amino acid substitutions on the neutralizing activity of guinea pig antisera raised against H3N2 strains with or without additional NGSs. As expected, the results of both HI and microneutralization assays showed that the anti-Oka/2 (144K) serum showed lower neutralizing activity against both Sk/23 (144N) and Sk/26 (144N/45N) viruses compared to that of anti-Sk/23 and anti-Sk/26 sera against the Oka/2 (144K) virus; this suggests that substitution of the amino acid residue at position 144 might modulate the conformational fit of the antibody to the antigenic site of HA as a means of immune escape. However, in clear contrast, the plaque reduction neutralization assay showed that anti-Oka/2 serum can block the mutated Sk/23 and Sk/26 virus strains in a slightly better manner than the Oka/2 virus itself even at the lowest concentration (1:10000 dilution), whereas Oka/2 (144K) virus was barely neutralized with the anti-Sk/23 (144N) and anti-Sk/26 (144N/45N) sera. Perhaps most important information from this result is that the plaques can be detected at lower antiserum concentration. Namely, the antibody-mediated immune evasion becomes invalid at a “limiting concentration”, which mimics low antibody titers in patients after a certain period from the previous exposure or low vaccine effectiveness. If this is the case, added glycosylation site to a virus does not allow it to escape antibody directed against a less glycosylated variant, but a less glycosylated virus can escape immunity mounted against a more glycosylated variant. This may explain the observed fluctuations (i.e., gains and losses) in the N-glycosylation associated amino acid residue 144 in the HA globular head (Fig.
2). Indeed, vaccine viruses used during the 2004–2007 seasons contained 144N (Table
5), which might have resulted in the emergence of 144K mutant, and then some natural viral fitness issue could have caused it to disappear again. It is conceivable that not only gains but also losses of N-linked glycan at the amino acid 144 of HA are effective strategies for persistence of circulating H3N2 virus even in the face of the humoral immune response. It is, of course, possible that these fluctuations are due to the non-neutralizing function of the antisera against emerged viruses at a limiting concentration. In other words, the influenza virus can survive at sufficiently low titers of antibody against the virus load in some cases.
To exclude the possibility that the observed neutralizing activity of serum is not due to the newly acquired N-glycosylation in the HA globular head, we generated mutant viruses with and without this particular NGS at residue 144 in the HA by a reverse genetics approach, and tested whether the substitution of the amino acid residue 144 significantly affected the neutralization activity. As expected, the neutralizing ability against a heterologous H3N2 strain was significantly reduced when the NGS at residue 144 was replaced, suggesting that the newly acquired NGS at residue 144 in the HA globular head may play an important role in the re-emergence of the endemic seasonal H3N2 within the subtype by helping it evade the humoral immunity. It is noteworthy that previous studies have also suggested the importance of amino acid changes in this region, entitled antigenic site A, for the generation of antigenically distinct viruses of epidemic significance [
26]. For instance, substitutions at the predicted NGS at position 144 in HA have been shown to contribute to increased infectivity of the reassorted H3N2 viruses of the 2003–2004 season, causing an epidemic in Denmark [
27]. Interestingly, our data showed that the substitution at residue 45 in the stem region of HA also had some effect on the plaque reduction neutralization assay.
As mentioned above, we found a new amino acid substitution at residue 45 of the stem of HA in 2011–12 season. Interestingly, not only in Japan, but also in Tunisia, 45N and 144N type H3N2 viruses (same as Sk/26 in this study) have been detected in one severe and one fatal case, whereas 45S and 144D type viruses were detected in one severe case and two mild cases, in 2013 [
28]. Since the stem of HA has been assumed to provide the main forces that stabilize HA trimer [
29], it is well known that the stem region of the viral HA tends to be conserved across different virus strains, whereas the globular head region shows considerable variation [
30]. Thus, the conserved stem domain of HA is considered to be a potential “universal” vaccine candidate with the potential to confer heterosubtypic protection. It has been reported that in H3N2 viruses, there are two NGSs in the stem of HA (nos. 22 and 38) [
30]. These two residues were indeed conserved in our samples throughout the study period, whereas a new amino acid substitution at residue 45 of the stem of HA was found in the 2011–12 season. It seems likely that although the addition of N-linked glycans may contribute to viral escape from neutralizing antibodies, the addition of glycans may also interfere with the receptor binding properties of HA, resulting in lower levels of viral entry and infection. It is therefore possible that the amino acid substitution observed at the NGS site within the stem of HA is a result of compensation, which increases binding affinity for efficient infection and viral replication. Further research is needed to confirm this hypothesis.
Acknowledgements
We are grateful to the staff and donors of Shizuoka Kosei Hospital, Kawasaki Medical School Hospital, and Medical Bioresource Research Unit of the Kawasaki Medical School. We also thank Ms Kumiko Matoba and Sachiyo Ohmori of Kawasaki Medical School for their excellent technical assistance.