Molecular evolution of influenza B virus during 2011–2017 in Chaoyang, Beijing, suggesting the free influenza vaccine policy

Two influenza B virus lineages, B/Victoria and B/Yamagata, are co-circulating in human population. While the two lineages are serologically distinct and TIV only contain one lineage. It is important to investigate the epidemiological and evolutionary dynamics of two influenza B virus lineages in Beijing after the free influenza vaccine policy from 2007. Here, we collected the nasopharyngeal swabs of 12657 outpatients of influenza-like illness and subtyped by real-time RT-PCR during 2011–2017. The HA and NA genes of influenza B were fully sequenced. The prevalence is the highest in the 6–17 years old group among people infected with influenza B. Yamagata-lineage virus evolved to two inter-clade from 2011–2014 to 2014–2017. The amino acids substitutions of HA1 region were R279K in strains of 2011–2014 and L173Q, M252V in strains of 2014–2017. Substitutions L58P, I146V were observed in HA1 region of Victoria-lineage virus in 2011–2012 and I117V, N129D were showed in 2015–2017. Phylogenetic analysis of NA showed Yamagata-Victoria inter-lineage reassortant occurred in 2013–2014. Influenza B mainly infect the school-aged children in Beijing and the free influenza vaccine inoculation does not seem to block school-age children from infection with influenza B. The antigen characteristics of circulating influenza B were different to the recommended vaccine strains. We concluded that the Victoria-lineage vaccine strain should been changed and the free influenza vaccine should be revalued.

Influenza virus poses a serious threat to public health by causing the annual epidemics and occasional pandemics. Influenza viruses are divided into influenza A, B, C and D virus based on the antigenic specificity of the nucleoprotein and matrix protein and influenza A and B virus co-circulated globally in a typical seasonal pattern [1][2][3] . Unlike influenza A which has a broad host range and caused pandemics by antigenic shift, influenza B usually causes local epidemic with no natural animal host (other than seals) and a slower mutation rate [4][5][6] . However, influenza B still contributes about one third of the global influenza disease burden. Many recent reports are indicated that influenza B is associated with serious illness, such as acute myocardial infarction and so on 7 . More seriously, influenza B caused 0.058% deaths for all-cause death in Southern China, which was higher than influenza A 8 . Despite the significance of influenza B to human health, their epidemiological characteristics and antigenic dynamics are far less studied compared to influenza A.
Influenza B was first isolated in 1940, designated as influenza B/Lee/40 and then was classified into two distinct lineages by the genetically relationship of the HA gene, represented by the prototype viruses B/Victoria/2/87 (Victoria-lineage) and B/Yamagata/16/88 (Yamagata-lineage) since 1983 [9][10][11] . The Yamagata-and Victoria-lineage have been co-circulating globally and often alternating in regional dominance. Two lineages undergo a slower antigenic variation with genetic evolution of the HA and NA genes including nucleotide mutations, and reassortment 12,13 .
The primary measure to prevent influenza virus infection is vaccination. Trivalent vaccine (TIV) has been produced containing influenza A/H1N1 and A/H3N2 antigens and a single influenza B antigen since 1977, despite (2019) 9:2432 | https://doi.org/10.1038/s41598-018-38105-1 www.nature.com/scientificreports www.nature.com/scientificreports/ some studies said low level of cross-protection provided by immunization with vaccine containing antigen from a single influenza B lineage 14,15 . Since 2007, the municipal government of Beijing provided free influenza vaccination for adult residents more than 60 years old and 6-17 years old elementary and high school students (Beijing influenza vaccination program for middle school in 2007, http://zfxxgk.beijing.gov.cn/110088/qt33/2015-10/20/ content_625915.shtml) 16,17 . However, the overall estimate of influenza vaccine effectiveness (VE) was 46.9% for the 2013-2014 season and 5.0% for the 2014-2015 season 18 . After this policy of free vaccine was implemented in Beijing, there was no continuous monitoring of the restrictions on influenza B infection and the influence to the pathogenic characteristics of influenza B. So, the continuous surveillance for influenza B in Beijing reflected the relationship between influenza vaccination rate and immune protective effect which imperative and has economic significance. The aim of this retrospective surveillance during the 2011-2017 seasons in Beijing was to assess the epidemiological and phylogenetic characteristics of influenza B from outpatients with influenza-like illness (ILI), and analyze the mismatch ratio between the circulating and vaccine strains.

Results
Pathogen spectrum of influenza cases from October 2011 to September 2017. Total 12657 outpatients in the surveillance seasons from October 2011 to September 2017 were included in this study. Among them 446 (3.52%, from 0 to 7.8% per year) were positive for influenza A/H1N1; 1085 (8.57%, from 2.1% to 16.9% per year) were positive for influenza A/H3N2; 395 (3.12%, from 0 to 8.8% per year) were positive for influenza B/Yamagata-lineage; 246 (1.94%, from 0 to 7.2% per year) were positive for influenza B/Victoria-lineage. 3 (0.02%) were co-infected with two influenza viruses (Fig. 1). In total, 29.47% of influenza cases during the six years were infected by influenza B, which was higher than influenza A/H1N1, but lower than influenza A/H3N2. Notably, the positive percentage of influenza Yamagata-lineage in 2014-2015 was 7.2% that was higher than the other three influenza virus subtypes.
Age distribution of influenza B. The ILI cases were divided to four age groups (≤5 years old, 6-17 years old, 18-59 years old, ≥60 years old) in order to analyze the age distribution of influenza B. The positive rates of influenza B in four age groups were 17.24%, 4.99%, 4.82% and 6.71% (Table S1). Comparing with subtypes of influenza A, the prevalence of influenza B, especially for Yamagata-lineage, was higher in age group ≤5 years old (χ 2 test, p = 0.000). Although the prevalence of influenza B was lower than influenza A/H3N2, it was higher than influenza A/H1N1 in all age groups excepting ≤5 years old group. For all outpatients caused by influenza B, the positive percentage of Yamagata-lineage was the highest in ≤5 years old group. While the prevalence of Victoria-lineage virus was higher in age group 6-17 years old group than the other age groups (Fig. 2). Therefore, influenza B infection people concentrated in preschool children and school-aged children.

Matching ratio of influenza B between seasonal viruses and vaccine strains. During 2011-2017
surveillance seasons, Yamagata-and Victoria-lineage were infected synchronously or one lineage was in the ascendancy. But one lineage was recommended vaccine strain (Table S2)   The period from October to September next year is defined as one influenza seasons. The age group ≤5 years old was shown by a yellow lines; the age group 6-17 years old was shown by a green lines; the age group 18-59 years old was shown by a blue lines and the age group ≥60 years old was shown by a violet lines. www.nature.com/scientificreports www.nature.com/scientificreports/ Among 83 HA sequences, 22 (26.51%) of influenza B strains belonged to Victoria-lineage and 61 (73.49%) virus belonged to Yamagata-lineage based on the phylogenetic analysis of the HA sequence. Influenza B has been separated previously into three major antigenically distinct clades (clade 1, the B/Brisbane/60/2008 clade, clade 2, the B/Massachusetts/2/2012 clade, and clade 3, the B/Wisconsin/1/2010 clade), based on phylogenetic analysis of HA and NA genes 19 . Phylogenetic analysis showed that all HA sequences of Yamagata-lineage belonged to clade 3 (Fig. 4) (Table 3).

Discussion
In this study we described the epidemiology and phylodynamics of influenza B viruses during 2011-2017 surveillance seasons in Chaoyang District in Beijing. Epidemiological results of surveillance outpatients of ILI showed influenza B virus is more likely to infect people under 17 years old, who were preschool and school-aged children. The finding is consistent to the surveillance data of influenza B virus during 2009-2014 seasons in Shanghai and results from several community-based cohort studies that found that children were more frequently infected with B viruses than adults 22 . Influenza B contributed to a higher proportion (41.9%) among children and young teenagers and were associated with severe disease especially myocardial injury in children, which is observed in almost 70% of fatal cases [23][24][25] . Despite elementary and high school students can inoculate free influenza vaccines in Beijing, a recent report showed that the average coverage rate of TIV among students of the 43 schools in Beijing was 47.6% in 2014-2015 and children who received both 2013-2014 and 2014-2015 vaccinations had vaccine effectiveness of 29% 26 . VE against influenza B was higher among vaccinated individuals with no previous vaccination history was 75% compared with vaccinated individuals with a frequent vaccination history was 48% from 8 seasons 27 . Repeated previous vaccinations over multiple seasons had dose-dependent negative impacts on VE against both influenza A and B virus 28 . By this token, repeated vaccinations may low the VE of influenza vaccine. Contradictorily, the free influenza vaccine police may not the best protection to prevent infection of influenza B to school children. Further studies to confirm this finding are necessary.
Phylogenetic analysis of nucleotide and amino acids of influenza B virus is important in the assessment of antigenic evolvement. The evolution of the influenza B virus is driven by mutations and under positive selective pressure on the membrane-distal domain of HA1 in specific domains such as four epitopes, including loop-120, loop-150, loop-160 and helix 190 29 . Characterizing amino acid substitutions that occur along the trunk of Yamagata-and Victoria-lineage gene phylogenies, identifies changes that become fixed in the virus population across seasons. As expected, substitution N116K was key mutation in 2013-2015 seasons and caused Yamagata-lineage to be absolute prevalence. This mutation previously described to circulate in Italian in 2011-2015 30 . We also observed the 120-loop and 150-loop of HA1 of Victoria-lineage showed different mutations in 2011-2012 season and 2015-2017 seasons. The mutations of the antigenic site of HA1 make Victoria-lineage to be the absolute epidemic strain of these periods. Interestingly, we observed Yamagata/Victoria inter-lineage reassortment with HA of Yamagata-lineage and NA of Victoria-lineage, which was dominant in 2013-2014 seasons. The reassortment was reported in other researches and caused Yamagata-lineage to be dominated strain 31 . In conclusion, one or multiple mutation of four epitopes and subtypes reassortment may cause preponderant strain.
At the moment, TIV is still the effective measure to prevent infections and severe illnesses caused by influenza virus in the world 32,33 . However, the protective effect of vaccination is very weak 34 . The overall influenza VE for www.nature.com/scientificreports www.nature.com/scientificreports/ school children during the influenza season of 2016-2017 was 20.6% and estimated to 32% among risk groups and was 11% among the elderly 35 . The 3-seasons adjusted VE in preventing hospitalization as determined in a case-control study was 52% for influenza A and 28% for influenza B 36 . The reasons may consist three items for low protective rate. Firstly, TIV only include one lineage of influenza B and disparities from antigenic mismatches between the predominant circulating influenza B lineage in a given year and that year's seasonal influenza trivalent vaccine. Our data showed 31.20% of infected by influenza B virus was mismatched to recommended  26 . Secondly, pathogenic diversity generated by shift and drift prevent vaccine strains from producing protective antibodies against epidemic strains 4,21 . The constant genetic and antigenic changes of influenza virus render them the ability to evade host immune system, thus limiting the vaccine effectiveness. Thirdly, cross-lineage serum antibodies were not detected on ferrets who were challenged with Victoria-lineage or Yamagata-lineage virus, though a significantly lower level of heterologous challenge virus in the respiratory tract was observed 37 . Hemagglutination-inhibition antibody titers indicate antibody responses against HA, which is not cross-reactive, and do not protect against mismatching influenza strains 38 . So, the need for development of quadrivalent influenza vaccines (QIV) containing Yamagata-lineage and Victoria-lineage antigens or engineered synthetic vaccine to encode two novel and broadly cross-protective monoclonal antibodies targeting influenza A and B are imperative 33,39,40 . The first QIV (a LAIV formulation) entered the market in 2012, and several other QIV formulations based on inactivated vaccine formulations, such as split and subunit formulations, have been licensed. The VE of QIV showed an immunogenicity and safety profile of the vaccine comparable with the two licensed trivalent vaccines containing the same strains 41 .
In conclusion, influenza virus that circulated in Chaoyang District in Beijing during the influenza seasons 2011-2017 evolved multiple antigenic epitope mutations, reassortment and belonged to different inter-clade comparing to recommended vaccine of influenza B. The epidemic strain of influenza B was the Victoria lineage mismatching to the vaccine strain Yamagata lineage. Though the free influenza vaccine for school-age children in Beijing, pre-school and primary school students still was in high infection by influenza B virus. The reasons may explained the phenomenon including antigenic variation, mismatching strains and repeated vaccination. So, the introduction of the QIV and broad-immunity vaccine are needed and the free vaccine policy in Beijing need to be revalued.

Materials and Methods
Ethics statement. The study was approved by the Ethics Review Committee of Chaoyang District Center for Disease Prevention and Control. All patients signed informed consent before the samples were collected. If the participants were children, their guardians were informed the consent.
Sample collection. According to Influenza Surveillance Program for Beijing, the enrolment criteria for surveillance cases included: (a) outpatients were as ILI cases defined as fever with armpit temperature ≥38 °C accompanied by one of the general symptoms (i.e., cough, pharyngalgia or nasal congestion; (b) the clinical symptoms appeared at last 3 days; (c) patients have no antiviral treatment 42 . Forty nasopharyngeal or throat swabs  www.nature.com/scientificreports www.nature.com/scientificreports/ were collected from the enrolled patients by 2 sentinel hospitals including China-Japan Friendship Hospital and Beijing Chao-Yang Hospital every week. Specimens were kept in 3 mL virus transport medium at 4-8 °C and sent to the network laboratory of Chaoyang District Center for Disease Prevention and Control within 24 h. The epidemiological information about the patients was input into the Chinese influenza Surveillance Net.  (Table S4), referencing to the HA and NA gene of strain B/Yamagata/16/88. The RT-PCR was performed using the One Step RT-PCR Kit (Qiagen, Germany) according to the handbook. The PCR products were analyzed using QIAxcel capillary electrophoresis (Qiagen, Germany) and sequenced by ABI 3730 DNA sequencer (PE-Applied Biosystems, Foster City, CA).
Multiple sequence alignment was performed using Clustal W (version 1.83). Genome sequences included in this study were submitted in the Global Initiative on Sharing Avian Influenza Data (GISAID)/GISAID accession numbers: EPI1153147-EPI1153378 (Table S4). A total of 83 HA and 83 NA sequences derived ILI outpatients from 2011 to 2017 were analyzed and 60 strains were downloaded from NCBI and GISAID using as reference strains in this study. The phylogenetic trees were constructed using MEGA software (version 6.06) applying the neighbor-joining method with 1000 bootstrap replicates. Statistical analysis. Data was analyzed using SPSS (Version 18.0; SPSS Inc., Chicago, IL, USA) and graphs drawing were used with GraphPad Prism software (Version 5.01; GraphPad, La Jolla, CA, USA). Difference between groups was evaluated using Pearson χ 2 test or Fisher exact test. Table 3. Amino acid substitutions found on the HA protein of influenza B viruses. a Only amino acid substitutions shared by 50% of viruses or more in every season are reported in this table. Substitutions are compared with recommended vaccine strains in the same year. b The frequency was the number of influenza viruses having the substitution to all the strains of influenza B virus in surveillance seasons.