Genetic divergence of Influenza A(H3N2) amino acid substitutions mark the beginning of the 2016–2017 winter season in Israel

Highlights • A(H3N2) dominated the early stages of the 2016–2017 influenza season.• 36% of hospitalized infected patients received the influenza vaccine.• Circulating A(H3N2) viruses were different from the vaccine strain.


Background
Influenza vaccine composition is evaluated each year due to the frequency and accumulation of genetic changes that influenza viruses undergo. The beginning of the 2016-2017 influenza surveillance period in Israel has been marked by the dominance of influenza A(H3N2) virus infection both among community and hospitalized patients. Influenza A(H3N2) is considered the most common cause of seasonal influenza [1], and has been associated with most influenza-related deaths [2].
Hemagglutinin (HA) amino acid (AA) substitutions are considered to occur more frequently for influenza A(H3N2) than influenza A(H1N1) viruses [3], and vaccine effectiveness (VE) against influenza A(H3N2) has been overall lower than VE against influenza A(H1N1) or against influenza B [4]. Thus, it is of paramount importance to detect AA substitutions in the circulating influenza A(H3N2) viruses relative to the recommended vaccine strain as early in the influenza season as possible. In this work, we analyzed influenza A(H3N2) viruses detected early in the season in respiratory samples of influenza patients that were either vaccinated or not vaccinated with the 2016-2017 influenza vaccine.

Sample collection
Respiratory samples were collected from patients hospitalized with respiratory illness at the Chaim Sheba Medical Center, Israel, and from patients with influenza-like illness (ILI) from 26 outpatient sentinel clinics throughout Israel. The latter were collected as a part of the Israel Center for Disease Control (ICDC) surveillance program for respiratory viruses morbidity.

Laboratory techniques
Viral genomic RNA was extracted from respiratory samples using the NucliSENS easyMAG (BioMerieux, France). Influenza A, Influenza B and A(H1N1)pdm09 infection was detected by using a panel of realtime reverse transcription-PCR (rRT-PCR), as previously described [5,6]. For influenza A(H3N2) detection and sequencing specific primers were used as previously described [7].
RT-PCR products were sequenced using ABI PRISM Dye Deoxy Terminator cycle sequencing kit (Life Technology, Foster City, CA, USA). Reaction mixtures were analyzed on Applied Biosystems 3500 sequence analyzer.
Nucleic acid sequences were aligned and compared using the Sequencher ® 5.0 program (Gencodes Corporation, Ann Arbor, MI). To infer the evolutionary relationships and the most recent common ancestor (MRCA) for the influenza A(H3N2) virus HA sequences, a Bayesian Markov chain Monte Carlo (MCMC) method was applied using a relaxed molecular clock, as implemented in the BEAST program (version 1.7.5

Cell culture and hemagglutination inhibition analysis
Influenza viruses were grown in Madin-Darby canine kidney (MDCK) cells stable transfected with the cDNA of human 2,6-sialtransferase (SIAT1) as previously described [9].
Hemmagglutination assays were performed using fifty microliters of 0.75% guinea pig red blood cells in the presence of 20 nM oseltamivir , were added. The assay was performed with guinea pig red blood cells in the presence of 20 nM oseltamivir. The mixture was mixed by shaking the plates on a mechanical vibrator and then incubated at room temperature for 1 h. Agglutination patterns were read after 60 min and the hemagglutination inhibition (HI) titer was defined as the reciprocal of the last dilution of serum that fully inhibited hemagglutination [10].

Results
Of the 388 samples obtained from community sentinel patients with ILI from October 2nd through December 9th, 2016, 59 (15.2%) were positive for influenza viruses. Of these, 56 (94.9%) were positive for influenza A(H3N2), one (1.7%) for influenza A(H1N1)pdm09 and two (3.4%) for influenza B virus. A total of 11 (20%) individuals who were infected with influenza A(H3N2) received the influenza vaccine 14 days or more prior to developing ILI.
Two AA substitutions, N122D and N246Y, resulted in a loss of a potential glycosylation site ( Table 2). Fig. 1 shows the protein structure of a representative virus from each group of AA substitutions. Phylogenetic analysis of 18 partially sequenced A(H3N2) HA genes (928 base pairs) using the BEAST program showed branching of the influenza A(H3N2) HA sequences (Fig. 2). One branch was related to the previously described 3C.2a1 subclade [11](marked with turquoise branches and designated as group 1) (Fig. 2). A second branch was related to the recently proposed 3C.2a2 subclade [12] (marked with green branches, designated as group 3) (Fig. 2).
To determine if these new clades are associated with an antigenic drift we performed HAI testing (as described in materials and methods), however we were unable to obtain hemagglutination using Guinea pigs red blood cells in the presence of oseltamivir carboxylate. Similarly, 90% of the viruses that were tested by the WHO did not hemagglutinate [14]. Additional influenza viruses from Israel were analyzed for HAI by the WHO Collaborating Centre for Reference and Research on Influenza. Of the six influenza viruses analyzed so far, four could not agglutinate red blood cells at all, and thus could not be analyzed for HAI. The other two influenza viruses that were analyzed by HAI were recognized by an antiserum raised against a cell culture-propagated cultivar of the currently recommended vaccine virus A/Hong Kong/4801/ 2014 at titers 4-fold lower and 8-fold lower than the titer of the antiserum for the homologous virus. An antiserum raised against the egg-propagated vaccine virus A/Hong Kong/4801/2014 also recognized the two viruses analyzed so far at titers 4-fold or greater lower than the titer of the antiserum for the homologous virus. These results are available in the report of the Vaccine Composition Meeting held in February [13].  Table 1). Colored branches represent different groups of amino acid substitutions, matching the colors of AA substitution groups in Fig. 1. The 3 groups are indicated.  Table 2. The 3 groups are indicated.

Discussion
Surveillance of influenza viruses at the beginning of the influenza season is of paramount importance for evaluation of current infecting viruses and preparation of the appropriate next season vaccine. Israel is located in the northern hemisphere in western Asia. Influenza A(H3N2) predominated in Israel during the preceding seasons of 2010/2011, 2011/2012, 2013/2014 and 2014/2015, in which it constituted 59.5%, 84.3%, 76.4% and 92.2% of the influenza-positive nasal-throat samples obtained from ILI sentinel patients, respectively [14][15][16][17]. During that period, the influenza A(H3N2) vaccine component was modified several times [18]. Amino acid substitutions in the HA protein are considered to occur more frequently in influenza A(H3N2) than in influenza A(H1N1) [3,19,20], and a significant drift was detected during the 2014-2015 season in the northern hemisphere, dominated by the 3C.2a clade [7,21]. Based on available vaccination data, 20% of ILI sentinel patients and at least 36% of hospitalized influenza A(H3N2)-positive cases became ill in the beginning of the 2016-2017 season despite being vaccinated.
In this regard, during the 2014-2015 influenza season that was dominated by a drifted influenza A(H3N2), approximately 20% of influenza A(H3N2)-positive influenza received the influenza vaccine [7].
Nucleotide sequencing revealed several differences as compared to the 2016-2017 influenza A(H3N2) vaccine strain. These changes demonstrated genetic divergence into three groups of AA substitution combinations. One of these combinations (Group 1) belonged to the newly described clade 3C.2a1 [11] and another (Group 3) was related to the recently proposed clade 3C.2a2. The remaining combination (Group 2) was most recently identified also in Europe [13].
It is interesting to note that influenza A(H3N2) was detected only in 1.7% of positive influenza samples received from outpatient sentinel patients with ILI during the 2015-2016 influenza season in Israel [22]. Thus, lack of a previous year exposure coupled with the mutations observed in the circulating viruses may contribute to the spread and dominance of influenza A(H3N2) during the present season.
Mutations that cause antigenic drift occur most frequently in the gene encoding the HA surface glycoprotein, which constitutes the target of neutralizing antibodies [23]. Influenza A(H3N2) has five neutralizing antigenic sites designated by the letters A to E, with 131 amino acid positions within them associated with antigenic changes [23]. However, seven of these positions (145, 155, 156, 158, 159, 189 and 193), found in antigenic sites A and B, have been more likely to be associated with antigenic changes [23,24]. None of the AA substitutions detected in our study occurred in these seven positions. However, the HAI results may provide some evidence for a possible antigenic drift. It is interesting to note that recent mid-season influenza A(H3N2) vaccine effectiveness studies estimates ranged mostly between 38% and 43% [25][26][27][28][29], which are consistent with modest protection overall.
The detection of three clear and different groups of AA substitutions, suggests that continued characterization of influenza A(H3N2) is necessary to see which group will prevail, prior to the decision regarding the future influenza vaccine composition.
In summary, characterization of early-season influenza viruses is important prior to the decision regarding future influenza vaccine composition. Due to the more rapid molecular changes that occur in influenza A(H3N2), detecting these changes is of paramount importance.