Genetic and antigenic characterisation of influenza A(H3N2) viruses isolated in Yokohama during the 2016/17 and 2017/18 influenza seasons

Background: Influenza A(H3N2) virus rapidly evolves to evade human immune responses, resulting in changes in the antigenicity of haemagglutinin (HA). Therefore, continuous genetic and antigenic analyses of A(H3N2) virus are necessary to detect antigenic mutants as quickly as possible. Aim: We attempted to phylogenetically and antigenically capture the epidemic trend of A(H3N2) virus infection in Yokohama, Japan during the 2016/17 and 2017/18 influenza seasons. Methods: We determined the HA sequences of A(H3N2) viruses detected in Yokohama, Japan during the 2016/17 and 2017/18 influenza seasons to identify amino acid substitutions and the loss or gain of potential N-glycosylation sites in HA, both of which potentially affect the antigenicity of HA. We also examined the antigenicity of isolates using ferret antisera obtained from experimentally infected ferrets. Results: Influenza A(H3N2) viruses belonging to six clades (clades 3C.2A1, 3C.2A1a, 3C.2A1b, 3C.2A2, 3C.2A3 and 3C.2A4) were detected during the 2016/17 influenza season, whereas viruses belonging to two clades (clades 3C.2A1b and 3C.2A2) dominated during the 2017/18 influenza season. The isolates in clades 3C.2A1a and 3C.2A3 lost one N-linked glycosylation site in HA relative to other clades. Antigenic analysis revealed antigenic differences among clades, especially clade 3C.2A2 and 3C.2A4 viruses, which showed distinct antigenic differences from each other and from other clades in the antigenic map. Conclusion: Multiple clades, some of which differed antigenically from others, co-circulated in Yokohama, Japan during the 2016/17 and 2017/18 influenza seasons.


Introduction
Influenza A(H3N2) virus has continued to infect humans since its emergence as a pandemic virus in 1968, resulting in considerable economic burden, hospitalisations and deaths [1]. After half a century of circulating in humans, A(H3N2) virus has accumulated numerous amino acid substitutions in its haemagglutinin (HA) to escape from human antibodies against this protein. Mouse monoclonal antibodies identified five major antigenic sites, A through E, on HA [2,3] and amino acid substitutions at these major antigenic sites are associated with antigenic drift. Based on the antigenicity of HA, A(H3N2) viruses form antigenic clusters [4]. Seven positions i.e. 145 at antigenic site A and 155, 156, 158, 159, 189 and 193 at antigenic site B, are mainly responsible for antigenic cluster transitions [5]. In addition, modification of HA with N-linked glycans also affects the antigenicity of HA via steric hindrance at these antigenic sites [6,7].
During the 2014/15 influenza season, the majority of reported influenza infections in Japan were caused by A(H3N2) viruses of clade 3C.2A [11], whereas in the 2015/16 influenza season only a few infections caused by A(H3N2) virus were reported [12]. Therefore, the 2016/17 and 2017/18 influenza vaccines contained antigens from a virus of clade 3C.2A [13]. Here, we analysed the HA sequences of A(H3N2) viruses detected in Yokohama, Japan during the 2016/17 and 2017/18 influenza seasons to capture the epidemic trend of A(H3N2) virus infection.

Study samples
Clinical specimens were collected in sentinel clinics and hospitals as part of the national epidemiological surveillance of infectious diseases in Japan during the 2016/17 and 2017/18 influenza seasons. These specimens were tested by reverse transcription (RT)quantitative PCR (RT-qPCR) targeting H3-HA gene [14] and virus isolation was achieved by using AX4 cells.

Sequence analysis
Viral RNA was extracted from the isolated viruses by using an RNeasy Mini Kit (QIAGEN, Tokyo, Japan). The viral RNA was subjected to one step RT-PCR to amplify the HA gene by PCR using the AccessQuick RT-PCR system (Promega, Madison, Wisconsin, United States of America (USA)) as follows: after 45 minutes of cDNA synthesis at 48 °C and 2 minutes of denaturation at

Phylogenetic analysis
The phylogenetic tree was constructed by using the neighbour-joining method with Kimura distances and the bootstrap procedure (n = 1,000) using ClustalW 2.1 on the DDBJ (DNA Data Bank of Japan) website (http:// clustalw.ddbj.nig.ac.jp/) and was visualised by using the MEGA 7.0.26 software. A/Perth/16/2009 is defined as an outgroup. HA sequences of reference isolates were obtained from the Global Initiative on Sharing Avian Influenza Data (GISAID) EpiFlu database (https:// www.gisaid.org/). The sequencing dataset used in this study is available upon request.

Three-dimensional mapping
Amino acid positions were plotted on an HA trimer of A/Victoria/361/2011 (PDB ID code 4O5N) by using the Python-enhanced molecular (PyMOL) molecular graphics system version 1.3. The N-glycans on the HA trimer were added to the N-linked glycosylation sites by using the GlyProt web server (http://www.glycosciences.de/ modeling/glyprot/php/main.php) and visualised by using the PyMOL software.

Ferret antisera
The ferret antisera were produced at an animal production and supply company (Japan SLC, Inc. [17]), with the approval of Japan's Institutional Animal Care and Use Committee of the National Institute of Infectious Diseases. Adult male ferrets that were serologically negative for influenza viruses were infected with the indicated A(H3N2) virus via the respiratory tract as a viral vapour generated by a nebuliser. At 2 weeks after infection, whole blood collected from the ferrets under anaesthesia was coagulated. The serum supernatants after centrifugation were collected as the antiserum. The serum samples were treated with receptor destroying enzyme (RDE) II (SEIKEN, Niigata, Japan), according to the manufacturer's instructions before use in the virus micro-neutralisation assay.

Antigenic analysis
An in vitro virus micro-neutralisation assay, based on the concept of plaque reduction, was conducted. Briefly, after 50 μl of twofold serial dilutions of the ferret antisera in DMEM containing 200 nM oseltamivir was added to a monolayer of MDCK-SIAT1 cells in a 96-well plate, 1,000 focus-forming units of the indicated virus (50 µl) was added to each well. After a 1 hour incubation, 1.6% Avicel RC-581 (FMC BioPolymer) in DMEM was overlaid. Oseltamivir was used to prevent receptor binding via NA, and trypsin was not added to avoid virus multiplication. At 18-20 hours after infection, the cells were stained with mouse antinucleoprotein (NP) monoclonal antibodies (MAB8257 and MAB8258; Millipore, Consett, US), followed by an anti-mouse IgG antibody conjugated with horseradish peroxidase. NP-positive cells were visualised with KPL TrueBlue substrate (Kirkegaard and Perry Laboratories) to count foci. The reciprocal number of the minimum dilution of sera needed to achieve 50% focus reduction was used as the neutralisation titre.

Antigenic cartography
The neutralisation data were analysed with antigenic cartography, which is a method to visualise and increase the resolution of neutralisation results, as detailed previously [4,18]. Diagnostic plots showed that three dimensions and a column basis titre of 1,280 were suitable for representing these neutralisation data.

Ethical statement
All experiments with ferrets were performed in accordance with the institutional Regulations for Animal Care and Use and were approved by the Japan's Institutional Animal Care and Use Committee of the National Institute of Infectious Diseases.

A(H3N2) viruses circulating in Yokohama, Japan
During the 2016/17 and 2017/18 influenza seasons, we detected 227 A(H3N2) viruses from clinical samples collected in Yokohama; 225 viruses were isolated using AX4 cells and two were positive by RT-qPCR ( Figure  1). The HA sequences were determined by Sanger sequencing and HA genetic clades were determined based on reference isolates available through EpiFlu To draw the phylogenetic tree using the neighbourjoining method with Kimura distances, we selected several isolates as representative of each clade. We found that A(H3N2) viruses isolated during the 2016/17 and 2017/18 influenza seasons in Yokohama could be classified into six clades: 3C.2A1, 3C.2A1a, 3C.2A1b, 3C.2A2, 3C.2A3, and 3C.2A4 (Figure 1 and Figure 2A)

Changes in the N-linked glycosylation sites on H3-HA
We next focused on the putative N-linked glycosylation sites on the H3-HA molecule because gradual accumulation of putative N-linked glycosylation sites has been linked to escape from human antibody pressure [19]. On a single HA molecule of A/Perth/16/2009, we found 11 putative N-linked glycosylation sites ( Figure  2B

Antigenicity analysis
Since the phylogenetic tree revealed that A(H3N2) viruses that were distinct in terms of their HA sequence and number of HA glycosylation sites co-circulated in Yokohama during the 2016/17 and 2017/18 influenza seasons, we examined the antigenicity of viruses from different clades using sera obtained from infected ferrets. We prepared ferret antisera by infecting ferrets with each of five influenza A(H3N2) virus isolates:  (Figure 2A). For the antigenicity analysis, we selected 16 isolates (Table 1). We compared the HA sequence of the selected isolates with that of A/Hong Kong/7127/2014 (Hong Kong/7127,clade 3C.2A) at antigenic sites A through E [2,3] and mapped them onto the H3-HA molecule ( Figure 3 and Table 2). All six isolates possessed one or two substitutions at antigenic site D. An amino acid substitution at antigenic site A or E was found in five or four isolates, whereas that at antigenic site B or C was found in one or two isolates.  positions 135-137. All of the tested viruses were isolated and passaged using MDCK cells or derivative AX4 or MDCK-SIAT1 cells.
Using the five kinds of ferret antisera and the 16 isolates, we performed an in vitro virus neutralisation assay based on the concept of plaque reduction (Table  1). We then constructed an antigenic map to visualise these neutralisation data. This map revealed that clade 3C.2A2 (green circles) and 3C.2A4 (orange circles) viruses differed antigenically from other clades ( Figure  4)

Discussion
Here, we showed that genetically and antigenically distinct A(H3N2) viruses co-circulated in Yokohama, Japan during the 2016/17 and 2017/18 influenza seasons. In London and Thailand, two subclades of A(H3N2) viruses, clades 3C.2A1 (denoted these as groups I to III in their study [20]) and 3C.2A3 (denoted as clade 3C.2A2 or group IV in their study [21]), were mainly responsible for the epidemics in these locations. In some cases in Thailand, A(H3N2) viruses of clade 3C.2A4 (the authors denoted these as group V) were also detected [21]. These reports, together with our results, suggest that multiple clades of A(H3N2) viruses circulated during this recent influenza epidemic. Indeed, co-circulation of multiple subclades of A(H3N2) virus, including clades 3C.2A1a, 3C.2A1b, 3C.2A2, 3C.2A3 and 3C.2A4, was also found by sequence analysis of Australian samples during the 2017 season and was documented in an early report of the 2017/18 influenza season in Canada [22,23]. In Taiwan, clade 3C.2A3 viruses were more commonly isolated from patients with severe infections; clade 3C.2A4 viruses tended to be isolated in outbreaks [24]. Although the seven residues of HA responsible for major antigenic cluster transition [5] were not changed, the antigenicity of representative isolates of clades 3C.2A2 and 3C.2A4 differed from that of isolates of other clades. Furthermore, N-linked glycosylation sites of HA have frequently changed in recent years, which has likely affected the antigenicity of the viruses [25]. In particular, the N-glycan at positions 158-160 (NYT), which is conserved among isolates of clade 3C.2A, likely affects antigenicity because  An antigenic 3D map was generated from the neutralisation data in Table 1. Viruses are represented as circles and antisera as squares. Strains are coloured based on the clades. The grid indicates one unit of antigenic distance, a twofold dilution in neutralisation titre.
it sterically hides the residues at positions 156, 158 and 159 that are responsible for major antigenic cluster transition [5]. These reports, together with our results, show that multiple clades of A(H3N2) viruses possessing distinct antigenicity co-circulated during the 2016/17 and 2017/18 influenza seasons worldwide. In these influenza seasons, clade 3C.2A virus was used for vaccines (http://www.who.int/influenza/vaccines/ virus/recommendations/en/) and clade 3C.2A1b and 3C.2A2 viruses dominated during the 2017/18 influenza season. Although the antigenicity of clade 3C.2A viruses was similar to that of clade 3C.2A1b viruses, it differed from that of clade 3C.2A2 viruses, indicating that the vaccine based on the clade 3C.2A virus did not match the circulating clade 3C.2A2 viruses in the 2017/18 influenza season. The WHO recommended the clade 3C.2A1 virus as the vaccine seed virus for the 2018 southern and 2018/19 northern hemisphere influenza season. During the 2018 southern hemisphere influenza season, the clade 3C.2A1b and 3C.2A2 viruses continued to dominate (https://nextstrain.org/ flu/seasonal/h3n2/ha/3y). Whether the antigenicity of the viruses circulating in the 2018/19 northern hemisphere influenza season matches that of the clade 3C.2A1 viruses remains to be seen.