This document is downloaded from DR‑NTU (https://dr.ntu.edu.sg) Nanyang Technological University, .

Molecular insights into evolution, mutations and receptor‑binding specificity of influenza A and B viruses from outpatients and hospitalized patients in Singapore

Ivan, Fransiskus Xaverius; Zhou, Xinrui; Lau, Suk Hiang; Rashid, Shamima; Teo, Jasmine S. M.; Lee, Hong Kai; Koay, Evelyn S.; Chan, Kwai Peng; Leo, Yee Sin; Chen, Mark I. Cheng; Kwoh, Chee Keong; Chow, Vincent T.

2019

Ivan, F. X., Zhou, X., Lau, S. H., Rashid, S., Teo, J. S. M., Lee, H. K., . . . Chow, V. T. (2020). Molecular insights into evolution, mutations and receptor‑binding specificity of influenza A and B viruses from outpatients and hospitalized patients in Singapore. International Journal of Infectious Diseases, 90, 84‑96. doi:10.1016/j.ijid.2019.10.024 https://hdl.handle.net/10356/142170 https://doi.org/10.1016/j.ijid.2019.10.024

© 2019 The Authors (Published by Elsevier Ltd on behalf of International Society for Infectious Diseases). This is an open access article under the CC BY‑NC‑ND license (http://creativecommons.org/licenses/by‑nc‑nd/4.0/).

Downloaded on 25 Sep 2021 02:27:31 SGT

International Journal of Infectious Diseases 90 (2020) 84–96

Contents lists available at ScienceDirect

International Journal of Infectious Diseases

journal homepage: www.elsevier.com/locate/ijid

Molecular insights into evolution, mutations and receptor-binding

specificity of influenza A and B viruses from outpatients and

hospitalized patients in Singapore

a,1 a,1 b,1 a

Fransiskus X. Ivan , Xinrui Zhou , Suk Hiang Lau , Shamima Rashid ,

b c,d c,e f g

Jasmine S.M. Teo , Hong Kai Lee , Evelyn S. Koay , Kwai Peng Chan , Yee Sin Leo ,

g,h a, b,

Mark I.C. Chen , Chee Keong Kwoh *, Vincent T. Chow *

a

School of Computer Science and Engineering, Nanyang Technological University, Singapore

b

Department of Microbiology and Immunology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore

c

Molecular Diagnosis Centre, National University Hospital, Singapore

d

Singapore Immunology Network, Agency for Science, Technology and Research, Singapore

e

Department of Pathology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore

f

Department of Pathology, Singapore General Hospital, Singapore

g

National Centre for Infectious Diseases, Singapore

h

Saw Swee Hock School of Public Health, National University of Singapore, Singapore

A R T I C L E I N F O A B S T R A C T

Article history: Background: This study compared the genomes of influenza viruses that caused mild infections among

Received 4 July 2019

outpatients and severe infections among hospitalized patients in Singapore, and characterized their

Received in revised form 16 October 2019

molecular evolution and receptor-binding specificity.

Accepted 18 October 2019

Methods: The complete genomes of influenza A/H1N1, A/H3N2 and B viruses that caused mild infections

Corresponding Editor: Eskild Petersen, Aar-

among outpatients and severe infections among inpatients in Singapore during 2012–2015 were

hus, Denmark

sequenced and characterized. Using various bioinformatics approaches, we elucidated their evolutionary,

mutational and structural patterns against the background of global and vaccine strains.

Keywords:

Results: The phylogenetic trees of the 8 gene segments revealed that the outpatient and inpatient strains

Influenza

overlapped with representative global and vaccine strains. We observed a cluster of inpatients with A/H3N2

A/H1N1 viruses

A/H3N2 viruses strains that were closely related to vaccine strain A/Texas/50/2012(H3N2). Several protein sites could

Influenza B accurately discriminate between outpatient versus inpatient strains, with site 221 in neuraminidase (NA)

Evolution achieving the highestaccuracyfor A/H3N2.Interestingly, amino acid residues ofinpatient but notoutpatient

Mutations isolates at those sites generally matched the corresponding residues in vaccine strains, except at site 145 of

Receptor binding

hemagglutinin (HA).Thiswould beespecially relevantforfuturesurveillanceof A/H3N2strains inrelationto

Severity

their antigenicity and virulence. Furthermore, we observed a trend in which the HA proteins of influenza A/

Singapore

H3N2 and A/H1N1 exhibited enhanced ability to bind both avian and human host cell receptors. In contrast,

the binding ability to each receptor was relatively stable for the HA of influenza B.

Conclusions: Overall, our findings extend our understanding of the molecular and structural evolution of

influenza virus strains in Singapore within the global context of these dynamic viruses.

© 2019 The Authors. Published by Elsevier Ltd on behalf of International Society for Infectious Diseases.

This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-

nd/4.0/).

Introduction a city-state with a high population density in , has

also been affected by influenza pandemics and epidemics. In

Influenza pandemics and seasonal epidemics have resulted in Singapore, Lee et al. (2008) estimated 3500 deaths during the 1918

substantial public health, social and economic impacts. Singapore, A/H1N1 outbreak; 680 deaths and more than 77,000 outpatient

attendances during the 1957 A/H2N2 outbreak; and an increase in

outpatient attendances of over 65% during the 1968 A/H3N2

* Corresponding authors. outbreak. Cutter et al. (2010) reported 18 A/H1N1-related deaths

E-mail addresses: [email protected] (C.K. Kwoh), [email protected] and estimated over 270,000 infected persons during the 2009

(V.T. Chow).

outbreak in Singapore. Influenza burden in non-epidemic years has

1

Authors with equal contribution.

https://doi.org/10.1016/j.ijid.2019.10.024

1201-9712/© 2019 The Authors. Published by Elsevier Ltd on behalf of International Society for Infectious Diseases. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

F.X. Ivan et al. / International Journal of Infectious Diseases 90 (2020) 84–96 85

also been assessed. Ng et al. (2002) estimated 630,000 influenza fluid using QIAamp Viral RNA Mini kit (Qiagen, Valencia, CA, USA).

cases (>20% of the Singapore population), which gave rise to Viral cDNAs were then generated using SuperScript III reverse

520,000 doctor visits, 315,000 days of sick absence from work, and transcriptase (Invitrogen, Carlsbad, CA, USA) and universal primer

about 1,500 deaths among 4,200 infected elderly persons. Ang et al. for influenza A or B (Lee et al., 2013). Reverse transcription (RT) was



(2014) estimated that hospitalizations due to influenza were 28– carried out at 50 C for 30 min, followed by enzyme inactivation at



30 per 100,000 person-years during 2004–2008 and 2010–2012, 95 C for 1 min. Subsequently, 40 cycles of polymerase chain

with the very young and elderly at higher risk for hospitalization. reaction (PCR) were performed using a T-Personal thermal cycler

Overall, these data warrant an effective vaccination program in the (Biometra, Gottingen, Germany) or ABI 2400 thermal cycler

population. (Applied Biosystems, Foster City, CA, USA), each consisting of



Vaccination targeting particularly the hemagglutinin (HA) has denaturation at 95 C for 15 s, followed by annealing and extension



been considered to be a cost-beneficial strategy to reduce influenza at 72 C. One-step qPCR (Lee et al., 2013) was performed to

burden (Duncan et al., 2012). Nonetheless, vaccination seems to be differentiate A/H1N1 and A/H3N2 strains, while conventional two-

more effective against A/H1N1 and B than against A/H3N2, as step RT-PCR was performed using lineage-specific primers to

observed in a study involving Singapore military personnel in 2010– differentiate B/Victoria and B/Yamagata strains.

2013 (Ho et al., 2014). The relative inefficacy of the vaccine against A/ Illumina MiSeq Next Generation Sequencer was employed to

H3N2 may be partially explained by the study of Lee et al. (2015), sequence the full genomes of influenza A, while FluSeq v1.0 was

whichfound that 84%ofclinicalisolatescollectedin2009–2013were used for genome assembly (Lee et al., 2016). Sanger sequencing

mismatched to the vaccine strain A/Perth/16/2009(H3N2) recom- was employed to validate the HA1 segments of influenza A, and to

mended by WHO. Interestingly, this study also observed different sequence the genomes of influenza B. A/H3N2 primers (Lee et al.,

patterns of A/H3N2 dominance in Singapore and regions within the 2013), A/H1N1 primers (Deng et al., 2015) and 18 sets of B primers

Northern and Southern hemispheres. This finding highlights the (Tewawong et al., 2015a) were used for sequencing. An additional

importance of local or regional vaccine strategy and warrants set of primers (Chi et al., 2005) was also used for sequencing the

influenza virus surveillance. Indeed, surveillance studies in HA1 genes of B viruses. The sequence data were analyzed using

Singapore have been carried out regularly by the Singapore National MEGA software (Tamura et al., 2013) and the NCBI Influenza Virus

Surveillance Program for Influenza that is part of the WHO Sequence Annotation Tool (Bao et al., 2007).

international laboratory-based surveillance network (Ang et al.,

2016a),inaddition to otherstudiesundervarioussettings(Seahetal., Phylogenetic analyses

2010; Yap et al., 2012; Virk et al., 2017). However, only a few of these

surveillance studies were linked to genomic information. For each viral segment of A/H1N1, A/H3N2 and B viruses, a

Considering that genomic surveillance may improve vaccine phylogenetic tree was reconstructed using relevant sequences

design and unravel molecular patterns of influenza evolution and from genomes of viruses collected in this study, WHO-recom-

circulation, such activity needs to be conducted more systematically mended vaccine strains and other representative viruses in the

and regularly. As part of this effort, we present analyses of two period from 2009 to early 2018. Supplementary Table S1 shows

influenza genomedatasets that were derived in Singapore from 2012 metadata for GenBank and GISAID sequences of vaccine strains and

to 2015, which were obtained from subjects who sought outpatient representative viruses, while Supplementary Table S2 provides

treatment for mild influenza at a university health center and from acknowledgement of the source of the GISAID sequences.

inpatients admitted for severe influenza into hospitals across Representative genomes were selected by clustering the HA

Singapore. Molecular insights of these viral genomes were revealed sequences of viruses collected in each year in the period and

in comparison with vaccine strains and other genomes in public whose full genomes are available in NCBI Influenza Virus Resource

databases. These findings may help to identify viral factors for (Bao et al., 2008) or GISAID (Shu and McCauley, 2017). The

disease severity in a broader sense, and also provide insights about clustering was performed with CD-HIT (Li and Godzik, 2006) using

Singapore strains in the context of global influenza circulation. a similarity threshold of 0.97, generating 1–5 clusters along with

their representative virus. If multiple sequences exist for a segment

Methods of a viral isolate, only one of them was used for tree reconstruction.

To reconstruct the trees, nucleotide sequences of each of the eight

Clinical sample collection viral segments of A/H3N2, A/H1N1, and B viruses were aligned to

codon position using MUSCLE (Edgar, 2004), and columns with

Clinical specimens (e.g. nasal swabs) were collected from 27 many gaps at both ends of the alignment were trimmed. A time

outpatient subjects (between 20 to 55 years old) attending the tree was then inferred using the Bayesian Markov Chain Monte

University Health Center at the National University of Singapore Carlo (MCMC) analysis implemented in Beast 2 v2.4.4 (Bouckaert

(NUS-UHC) from2013 to 2014, and from20inpatients (between15 to et al., 2014) under the HKY substitution model, a Bayesian skyline

77 years of age) at local Singapore hospitals from 2012 to 2015, coalescence model, and a strict molecular clock model to produce

including the Communicable Disease Center (CDC), Tan Tock Seng 5,000 tree samples logged every 2,000 generations. Finally,

Hospital (TTSH), National University Hospital (NUH), and Singapore maximum clade credibility (MCC) time trees were produced with

General Hospital (SGH). QuickNavi-Flu rapid diagnostic test kit 10% burn-in, and the GGTREE package (Yu et al., 2017) was

(Denka Seiken, Tokyo, Japan) was used to screen for influenza among employed to visualize the MCC time trees.

outpatients. Although there are no specified criteria for hospitaliza-

tion, admission rates are generally greater in high-risk groups with Protein site mutation and discrimination analyses

influenza such as the elderly, individuals with underlying chronic

cardio-respiratory diseases, and patients with respiratory distress or Each sampled viral genome was compared against the closest

chest X-ray abnormalities (Ang et al., 2014). reference among vaccine strains using FluSurver (https://flusurver.

bii.a-star.edu.sg/), except for G2-27.1 which used automatic best

Virus propagation, RNA extraction and genome sequencing hits among vaccine strains to identify phenotypically or epidemi-

ologically interesting mutations. Furthermore, alignments of PB2,

Viruses from each specimen were propagated in MDCK cells at PB1, PA, HA, NP, NA, M1 and NS1 proteins were used for

low passage for 72 h. Viral RNAs were extracted from the culture discrimination analysis between inpatient and outpatient strains.

86 F.X. Ivan et al. / International Journal of Infectious Diseases 90 (2020) 84–96

In each column, the alignments were preprocessed by substituting from different influenza strains in the GenBank database, except

each non-specific amino acid symbol or gap (which was for samples G2-10.1 and G2-44.1.

considered due to sequencing “doubt”) with the amino acid that The phylogenetic trees for HA and NA of influenza A/H1N1

was the majority in the corresponding column. For each site in the (Figure 1), A/H3N2 (Figure 2), and B strains (Figure 3), as well as the

alignment, a rule was generated for each symbol at the site by: (a) trees for the other segments of each virus subtype (Supplementary

counting how often the symbol belongs to each of the outpatient Figures S1–S3) generally displayed overlaps between outpatient,

and inpatient cases; (b) finding the most frequent case; (c) making inpatient, vaccine and global representative strains. In particular,

the rule assign the symbol to that case. The accuracy of the set of the inpatient strains of A/H1N1, A/H3N2 and B viruses were not

rules in classifying the outpatient and inpatient cases, which could uniquely clustered into a single clade, but were quite closely

be considered as the power of a site in discriminating the two related to different outpatient strains and global representative

cases, was then calculated. The accuracy for a protein site that has strains. However, a group of A/H3N2 strains from hospitalized

the same amino acid across the samples, in which accuracy is patients was clustered together and closely related to the 2014–

calculated when all records are assigned to the class value that has 2015 vaccine strain A/Texas/50/2012(H3N2) as shown by the

the most observations, constitutes a point of reference for the phylogenetic tree of each A/H3N2 segment. In addition, the

increase in discriminating power. This procedure follows the One phylogenetic tree of A/H1N1 HA confirmed that both outpatient

Rule (OneR) classification approach (Holte, 1993), except that we and inpatient strains were 2009 pandemic (pdm09) strains since

considered all sites and did not select the site with the highest they “rooted” to A/California/7/2009(H1N1). For B viruses, we

accuracy as the final output. observed that all segments of 2012 hospital isolate TTSH-69 could

be “rooted” to the 2013–2015 vaccine strain B/Massachusetts/2/

Structural analyses of receptor-binding specificity 2012(Yamagata). Furthermore, the HA and NA of outpatient B/

Yamagata strains were mainly closely related to the 2015–2018

We analyzed the binding specificity of A/H1N1, A/H3N2 and B vaccine strain B/Phuket/3073/2013(Yamagata). The segments of

HA proteins with host cell receptors. Representative influenza the outpatient B/Victoria strains were mainly “rooted” to 2010–

virus strains were selected and their HA (specifically HA1) 2012 and 2016–2018 vaccine strain B/Brisbane/60/2008(Victoria),

structures were predicted from their primary sequences using except for PA of isolate G2-7.1 isolate.

SWISS-MODEL (Arnold et al., 2006; Biasini et al., 2014). Then,

sialotrisaccharide 3’SLN and 6’SLN isolated from co-crystallized Differentiation between virus strains by protein site mutation and

HA-ligand structures in Protein Data Bank (PDB) were respectively discriminant analyses

used as the avian and human receptor analogs to assess receptor-

binding specificity of selected strains (Berman et al., 2000). All Mutations with high-interest level known to enhance virulence

rotatable bonds were enabled to conduct molecular docking with were identified by FluSurver (data not shown); however, most of

the HA proteins (Xu et al., 2012). Subsequently, we conducted them did not differentiate between inpatient and outpatient

molecular docking by QuickVina 2 (Alhossary et al., 2015) to isolates. In this regard, we adopted the OneR approach which

compare their binding specificity. A grid box covering the binding provided more informative results. For A/H1N1, one site displayed

sites was predefined individually for each HA protein (Wang et al., notable discriminating power (OneR accuracy >80%), i.e. NS1-111

2007). (Figure 4A). Interestingly, while all outpatient isolates and the

The models generated from molecular docking studies were recommended vaccine strain for the same year harbored isoleucine

further analyzed to identify the structural characteristics of the at this site, the inpatient isolates had varied amino acids (threonine

binding affinities and effects of mutations. The solvent accessible for SGH-C, isoleucine for SGH-F, and methionine for SGH-G). For A/

surface areas (ASA) of ligand (L), target (T), docked complex (C), H3N2, 11 sites were discriminatory (OneR accuracy >70%),

and putative hydrogen bonds were determined using PyMOL v1.8.4 including PB2-588, PA-272, PA-668, PA-675, HA-145, HA-159,

(Schrodinger, 2015). The buried surface area of the complex was HA-225, HA-489, NA-221, NA-267, and NS1-26 (Figure 4B).

estimated by L + T À C. Parameters such as distance between donor Supplementary Table S4 shows the comparison between amino

and acceptor atoms, and the ASA probe radius were set to default acids at these sites for outpatient, inpatient and vaccine strains.

values (3.6 Å and 1.4 Å, respectively). For comparison of bound Notably, inpatient isolates were dominated by the same amino acid

ligands, all HA protein structures were first aligned to C-alpha that appeared in the vaccine strains recommended during the

atoms and displayed using PyMOL. period of the study (2012–2015) at most sites (except HA-145),

whereas outpatient isolates harbored different amino acids. At HA-

Results 145, serine was observed in all outpatient isolates, whereas about

equal proportions of serine and asparagine were present in

Genetic similarity and evolutionary analyses of influenza strains inpatient isolates. For B/Yamagata, the relatively small number of

samples restricted our interpretation, but sites within HA and NA

In Singapore, influenza viruses generally circulate all year may have potential to discriminate (Figure 4C).

round, with co-circulation of A/H1N1/pdm09, A/H3N2, B/Victoria,

and B/Yamagata strains. In the outpatients, we identified 6 A/H1N1, Molecular docking analyses reveal differential binding to human and

12 A/H3N2, 4 B/Victoria and 4 B/Yamagata viruses. Interestingly, avian receptor analogs

outpatient isolate G1-27.1 from 2014 was designated as A/H1N2

since its viral genome consisted of a combination of HA, PB2, PB1, The HA proteins of representative strains and their optimal

PA and M segments of A/H1N1, together with NA, NP and NS matching templates (along with GMQE scores) predicted from the

segments of A/H3N2. In the hospitalized patients, 3 A/H1N1, 16 A/ SWISS-MODEL are presented in Table 2. The results of binding

H3N2, and 1 B/Yamagata viruses were identified. Table 1 lists the affinities of representative HA proteins of A/H1N1, A/H3N2 and B

identification (ID) numbers, strain names, collection dates and with 3’SLN and 6’SLN (independently docked 1,000 times) are

GenBank accession numbers of these 47 viruses examined in this presented in Figure 5. Student’s t-test results for the mean

study. By BLAST analyses, the top similarity hits of the HA and NA differences between binding of each HA with 3’SLN and 6’SLN are

nucleotide segments are presented in Supplementary Table S3. The provided in Supplementary Table S5. Representative A/H1N1 HA

highest similarity hits of a particular queried genome were usually proteins acquired enhanced ability to bind both 3’SLN and 6’SLN, Table 1 List of identification (ID) numbers, strain names, collection dates, and GenBank accession numbers of outpatient and inpatient influenza virus isolates from Singapore. 627K refers to the variant in PB2 associated with enhanced replication, while 31N in the M2 protein denotes the mutation conferring amantadine resistance.

No. Isolate ID Strain name Collection Institution GenBank ID GenBank ID GenBank ID GenBank ID GenBank ID GenBank ID GenBank ID GenBank ID 627K 31N date Segment 1 Segment 2 Segment 3 Segment 4 Segment 5 Segment 6 Segment 7 Segment 8 (PB2) (M) 1 G2-5.1 A/Singapore/G2-5.1/ 02-12-2013 NUS-UHC MN493833 MN493834 MN493835 MN493836 MN493837 MN493838 MN493839 MN493840 Yes Yes 2013(H3N2) 2 G2-6.1 A/Singapore/G2-6.1/ 02-12-2013 NUS-UHC MN394115 MN394116 MN394117 MN394118 MN394119 MN394120 MN394121 MN394122 Yes Yes 2013(H3N2) 3 G2-7.1 B/Singapore/G2-7.1/ 26-11-2013 NUS-UHC MN480784 MN480785 MN480786 MN480787 MN480788 MN480789 MN480790 MN480791 No No 2013 4 G2-8.1 A/Singapore/G2-8.1/ 09-12-2013 NUS-UHC MN387230 MN387231 MN387232 MN387233 MN387234 MN387235 MN387236 MN387237 Yes Yes 2013(H3N2) 5 G2-9.1 A/Singapore/G2-9.1/ 09-12-2013 NUS-UHC MN386991 MN386992 MN386993 MN386994 MN386995 MN386996 MN386997 MN386998 Yes Yes 2013(H3N2) 6 G2-10.1 A/Singapore/G2-10.1/ 23-12-2013 NUS-UHC MN493862 MN493863 MN493864 MN493865 MN493866 MN493867 MN493868 MN493869 Yes Yes F.X.

2013(H3N2) Ivan 7 G2-13.2 B/Singapore/G2-13.2/ 15-01-2014 NUS-UHC MN485871 MN485872 MN485873 MN485874 MN485875 MN485876 MN485877 MN485878 No No

2014 et

al. 8 G2-14.1 B/Singapore/G2-14.1/ 15-01-2014 NUS-UHC MN480836 MN480837 MN480838 MN480839 MN480840 MN480841 MN480842 MN480843 No No

/

2014 International 9 G2-15.1 B/Singapore/G2-15.1/ 16-01-2014 NUS-UHC MN483442 MN483443 MN483444 MN483445 MN483446 MN483447 MN483448 MN483449 No No 2014 10 G2-19.1 A/Singapore/G2-19.1/ 13-02-2014 NUS-UHC MN384219 MN384220 MN384221 MN384222 MN384223 MN384224 MN384225 MN384226 No Yes

2014(H1N1) Journal 11 G2-20.1 A/Singapore/G2-20.1/ 13-02-2014 NUS-UHC MN382319 MN382320 MN382321 MN382322 MN382323 MN382324 MN382325 MN382326 No Yes 2014(H1N1)

12 G2-22.1 A/Singapore/G2-22.1/ 13-02-2014 NUS-UHC MN382307 MN382308 MN382309 MN382310 MN382311 MN382312 MN382313 MN382314 No Yes of

2014(H1N1) Infectious 13 G2-23.1 A/Singapore/G2-23.1/ 13-02-2014 NUS-UHC MN382298 MN382299 MN382300 MN382301 MN382302 MN382303 MN382304 MN382305 No Yes 2014(H1N1)

14 G2-24.1 B/Singapore/G2-24.1/ 18-02-2014 NUS-UHC MN493609 MN493610 MN493611 MN493612 MN493613 MN493614 MN493615 MN493616 No No Diseases 2014 15 G2-25.1 A/Singapore/G2-25.1/ 06-03-2014 NUS-UHC MN382289 MN382290 MN382291 MN382292 MN382293 MN382294 MN382295 MN382296 No Yes

2014(H1N1) 90

16 G2-26.1 A/Singapore/G2-26.1/ 06-03-2014 NUS-UHC MN372387 MN372388 MN372389 MN372390 MN372391 MN372392 MN372393 MN372394 Yes Yes (2020) 2014(H3N2) 17 G2-27.1 A/Singapore/G2-27.1/ 06-03-2014 NUS-UHC ÀÀÀÀÀÀÀÀNo Yes

2014(H1N2) 84 –

18 G2-29.1 A/Singapore/G2-29.1/ 21-01-2014 NUS-UHC MN498054 MN498055 MN498056 MN498057 MN498058 MN498059 MN498060 MN498061 Yes Yes 96

2014(H3N2) 19 G2-31.1 A/Singapore/G2-31.1/ 03-04-2014 NUS-UHC MN394140 MN394141 MN394142 MN394143 MN394144 MN394145 MN394146 MN394147 Yes Yes 2014(H3N2) 20 G2-34.1 B/Singapore/G2-34.1/ 21-04-2014 NUS-UHC MN493620 MN493621 MN493622 MN493623 MN493624 MN493625 MN493626 MN493627 No No 2014 21 G2-36.1 B/Singapore/G2-36.1/ 10-06-2014 NUS-UHC MN493745 MN493746 MN493747 MN493748 MN493749 MN493750 MN493751 MN493752 No No 2014 22 G2-43.1 B/Singapore/G2-43.1/ 18-06-2014 NUS-UHC MN493737 MN493738 MN493739 MN493740 MN493741 MN493742 MN493743 MN493744 No No 2014 23 G2-44.1 A/Singapore/G2-44.1/ 02-07-201408-07-2014 NUS-UHC NUS-UHC MN480506 MN394149 MN480507 MN394150 MN480508 MN394151 MN480509 MN394152 MN480510 MN394153 MN480511 MN394154 MN480512 MN394155 MN480513 MN394156 No Yes Yes Yes 2014(H1N1)2014(H3N2) 2425 G2-46.1 G2-51.1 A/Singapore/G2-46.1/ A/Singapore/G2-51.1/ 12-08-2014 NUS-UHC MN394170 MN394171 MN394172 MN394173 MN394174 MN394175 MN394176 MN394177 Yes Yes 2014(H3N2) 26 G2-52.1 A/Singapore/G2-52.1/ 27-08-2014 NUS-UHC MN400557 MN400558 MN400559 MN400560 MN400561 MN400562 MN400563 MN400564 Yes Yes 2014(H3N2) 27 G2-63.1 A/Singapore/G2-63.1/ 14-11-2014 NUS-UHC MN394383 MN394384 MN394385 MN394386 MN394387 MN394388 MN394389 MN394390 Yes Yes 2014(H3N2) 87 88

Table 1 (Continued) No. Isolate ID Strain name Collection Institution GenBank ID GenBank ID GenBank ID GenBank ID GenBank ID GenBank ID GenBank ID GenBank ID 627K 31N date Segment 1 Segment 2 Segment 3 Segment 4 Segment 5 Segment 6 Segment 7 Segment 8 (PB2) (M) 28 CDC-64 A/Singapore/CDC-64/ 2012 CDC MN449466 MN449467 MN449468 MN449469 MN449470 MN449471 MN449472 MN449473 Yes Yes 2012(H3N2) 29 CDC-73 A/Singapore/CDC-73/ 2012 CDC MN449433 MN449434 MN449435 MN449436 MN449437 MN449438 MN449439 MN449440 Yes Yes 2012(H3N2) 30 CDC-85 A/Singapore/CDC-85/ 2012 CDC MN449348 MN449349 MN449350 MN449351 MN449352 MN449353 MN449354 MN449355 Yes Yes 2012(H3N2) 31 CDC-90 A/Singapore/CDC-90/ 2012 CDC MN473262 MN473263 MN473264 MN473265 MN473266 MN473267 MN473268 MN473269 Yes Yes 2012(H3N2) 32 CDC-91 A/Singapore/CDC-91/ 2012 CDC MN473270 MN473271 MN473272 MN473273 MN473274 MN473275 MN473276 MN473277 Yes Yes 2012(H3N2) 33 CDC-109 A/Singapore/CDC-109/ 2012 CDC MN448518 MN448519 MN448520 MN448521 MN448522 MN448523 MN448524 MN448525 Yes Yes 2012(H3N2) 34 CDC-126 A/Singapore/CDC-126/ 2012 CDC MN443922 MN443923 MN443924 MN443925 MN443926 MN443927 MN443928 MN443929 Yes Yes F.X.

2012(H3N2) Ivan 35 CDC-148 A/Singapore/CDC-148/ 2012 CDC MN454513 MN454514 MN454515 MN454516 MN454517 MN454518 MN454519 MN454520 Yes Yes

et

2012(H3N2) al.

36 CDC-149 A/Singapore/CDC-149/ 2012 CDC MN456813 MN456814 MN456815 MN456816 MN456817 MN456818 MN456819 MN456820 Yes Yes /

2012(H3N2) International 37 CDC-204 A/Singapore/CDC-204/ 2012 CDC MN400666 MN400667 MN400668 MN400669 MN400670 MN400671 MN400672 MN400673 Yes Yes 2012(H3N2) 38 NUH-6 A/Singapore/NUH-6/ 2012 NUH MN473251 MN473252 MN473253 MN473254 MN473255 MN473256 MN473257 MN473258 Yes Yes

2012(H3N2) Journal 39 SGH-A A/Singapore/SGH-A/ 30-07-2012 SGH MN394448 MN394449 MN394450 MN394451 MN394452 MN394453 MN394454 MN394455 Yes Yes 2012(H3N2)

40 SGH-B A/Singapore/SGH-B/ 02-03-2012 SGH MN394462 MN394463 MN394464 MN394465 MN394466 MN394467 MN394468 MN394469 Yes Yes of

2012(H3N2) Infectious 41 SGH-C A/Singapore/SGH-C/ 06-12-2012 SGH MN394621 MN394622 MN394623 MN394624 MN394625 MN394626 MN394627 MN394628 No Yes 2012(H1N1)

42 SGH-D A/Singapore/SGH-D/ 29-01-2013 SGH MN508793 MN508794 MN508795 MN508796 MN508797 MN508798 MN508799 MN508800 Yes Yes Diseases 2013(H3N2) 44 SGH-F A/Singapore/SGH-F/ 05-07-2014 SGH MN444136 MN444137 MN444138 MN444139 MN444140 MN444141 MN444142 MN444143 No Yes

2014(H1N1) 90

45 SGH-G A/Singapore/SGH-G/ 10-09-2014 SGH MN400338 MN400339 MN400340 MN400341 MN400342 MN400343 MN400344 MN400345 No Yes (2020) 2014(H1N1) 43 SGH-H A/Singapore/SGH-H/ 28-03-2015 SGH MN498065 MN498066 MN498067 MN498068 MN498069 MN498070 MN498071 MN498072 Yes Yes

2015(H3N2) 84 –

46 TTSH- A/Singapore/TTSH- 12-07-2012 TTSH MN396699 MN396700 MN396701 MN396702 MN396703 MN396704 MN396705 MN396706 Yes Yes 96 13A 13A/2012(H3N2) 47 TTSH-69 B/Singapore/TTSH-69/ 28-09-2012 TTSH MN396687 MN396688 MN396689 MN396690 MN396691 MN396692 MN396693 MN396694 No No 2012

F.X. Ivan et al. / International Journal of Infectious Diseases 90 (2020) 84–96 89

Figure 1. Phylogenetic trees of the (A) HA and (B) NA segments of influenza A/H1N1 from outpatient (blue circle) and inpatient (red circle) strains in this study against the

background of representative (grey diamond) and vaccine (purple diamond) strains from 2009 to early 2018. A representative in the database was selected from each cluster

that contained sequences from the same year and with pairwise similarity of at least 97%.

especially 6’SLN from 1918 to 2014 (p < 0.001). The 2014 G2-25.1 revealed preference for binding with 6’SLN, albeit the B(Victoria)

HA showed binding preference for the human receptor analog, HAs showing stronger binding ability with both 3’SLN and 6’SLN

whereas the 2012 SGH-C, CA09 and SC18 HAs demonstrated than B(Yamagata) HAs.

binding preference for the avian receptor analog. HA proteins of

representative A/H3N2 strains from 2009 to 2014 exhibited Interesting receptor-binding modes of selected A/H1N1 strains from

significant binding preference with 6’SLN. It is noteworthy that Singapore

the representative HAs displayed more significant variation in

binding with 6’SLN than 3’SLN. The 2012 CDC-73 HA exhibited In view of the preference of the G2-25.1 strain for 6’SLN, and

slightly stronger binding ability with both avian and human the SGH-C isolate for 3’SLN, we proceeded to analyze their

receptor analogs compared with the 2012 CDC-204 and 2014 G2- receptor-binding modes to better understand this differential

26.1 HAs. In general, representative influenza B HA proteins binding. Figure 6 compares the ligand-binding poses of SGH-C

90 F.X. Ivan et al. / International Journal of Infectious Diseases 90 (2020) 84–96

Figure 2. Phylogenetic trees of the (A) HA and (B) NA segments of influenza A/H3N2 from outpatient (blue circle) and inpatient (red circle) strains in this study against the

background of representative (grey diamond) and vaccine (purple diamond) strains from 2009 to early 2018. A representative in the database was selected from each cluster

that contained sequences from the same year and with pairwise similarity of at least 97%.

F.X. Ivan et al. / International Journal of Infectious Diseases 90 (2020) 84–96 91

Figure 3. Phylogenetic trees of the (A) HA and (B) NA segments of influenza B from outpatient (blue circle) and inpatient (red circle) strains in this study against the

background of representative (grey diamond) and vaccine (purple diamond) strains from 2009 to early 2018. A representative in the database was selected from each cluster

that contained sequences from the same year and with pairwise similarity of at least 97%.

and G2-25.1 HAs with those of A/California/04/2009 and 6’SLN which binds in the cis conformation (Figure 7C). We

A/Brevig Mission/1/1918 (abbreviated CA09 and BM18, respec- detected a potentially novel ligand pose for SGH-C and G2-25.1 in

tively), bound to the avian analog. In contrast, Figure 7 illustrates which the SIA faces the binding groove near the 220 loop, while

binding with the human receptor analog. Typically, 3’SLN the NAG portion faces downwards to the 130 loop, compared to an

binds in an extended conformation, with sialic acid (SIA) facing upward orientation towards the 190 helix in CA09 and BM18

inwards to the binding groove (Figure 6), compared to (Figure 7).

92 F.X. Ivan et al. / International Journal of Infectious Diseases 90 (2020) 84–96

Figure 4. Accuracy of each protein site within PB2 (red), PB1 (orange), PA (yellow), HA (green), NP (blue), NA (cyan), M1 (purple) and NS1 (black) in discriminating inpatient

and outpatient influenza strains of (A) A/H1N1, (B) A/H3N2 and (C) B/Yamagata in this study based on the OneR approach. The accuracy of the Zero Rule, which assigns all

isolates into the class with the higher number of records, is the baseline of OneR performance, i.e. 66.7% for A/H1N1, 57.1% for A/H3N2, and 80.0% for B/Yamagata. Please refer

to Supplementary Table S4 for further details.

Table 2

The HA proteins of representative strains of influenza A/H1N1, A/H3N2 and B with their optimal matching templates (along with GMQE scores) predicted from the SWISS-

MODEL. Outpatient and inpatient strains were compared with several reference strains, i.e. CA09: A/California/04/2009(H1N1); SC18: A/South Carolina/1/1918(H1N1); Vic11:

A/Victoria/361/2011(H3N2); Per09: A/Perth/16/2009(H3N2); Aichi68: A/Aichi68/2/1968(H3N2); Phu13: B/Phuket/3073/2013-like virus; Col17: B/Colorado/06/2017-like

virus; Wis10: B/Wisconsin/01/2010; Bris08: B/Brisbane/60/2008.

Subtype/ Isolates and their predicted optimal matching templates Lineage Outpatient Template Inpatient Template Vaccine Template Other Template

isolates (GMQE) isolates (GMQE) strains (GMQE) strains (GMQE)

A/H1N1 G2-25.1 4LXV (0.99) SGH-C 4LXV (0.99) CA09 3ZTN (0.99) SC18 6D8W (0.95)

A/H3N2 G2-26.1 4WE8 (0.94) CDC-204 4WE8 (0.94) Vic11 4WE8 (0.95) Aichi68 1HA0 (0.97)

CDC-73 4WE8 (0.95) Per09 4WE8 (0.93) – 1HA0 (0.97)

B/Yamagata G2-36.1 4M40 (0.98) TTSH-69 4M40 (0.99) Phu13 4M40 (0.98) Wis10 4M40 (0.98)

B/Victoria G2-14.1 4FQK (0.99) – – Col17 4FQK (0.98) Bris08 4FQK (0.99)

In addition, we compared the ASA of the docked complexes high percentage of hydrogen bonds to the GAL portion of 3’SLN,

bound with their respective ligands (Supplementary Table S6), and whereas such bonds were lacking for CA09 and BM18. Moreover,

determined the number of hydrogen-bonded contacts made by the absence of contact between SGH-C to the GAL portion of 6’SLN

each of the three receptor moieties (Supplementary Table S7). For may reiterate its preference for 3’SLN over 6’SLN.

the first analysis, G2-25.1 HA binding with both 3’SLN and 6’SLN

revealed a similar fraction of buried surface area; SGH-C HA Discussion

binding with 6’SLN showed a slightly larger buried surface area

than with 3’SLN; while binding of both CA09 and BM18 HAs with This study had a number of limitations including different years

6’SLN exhibited a greater buried surface area than with 3’SLN. For of collection of patient samples, lack of influenza vaccination

the second analysis, both SGH-C and G2-25.1 displayed a relatively history of the subjects, and the relatively small sample size.

F.X. Ivan et al. / International Journal of Infectious Diseases 90 (2020) 84–96 93

Figure 5. Binding affinity of HA proteins of representative influenza strains of (A) A/H1N1, (B) A/H3N2, and (C) B with avian receptor analog 3’SLN (lighter colors) and with

human receptor analog 6’SLN (darker colors). The HA1 structures were predicted from their primary sequences using SWISS-MODEL. A grid box covering the binding sites was

defined specifically for each HA1 protein. Each HA-ligand docking experiment was conducted 1,000 times independently using QuickVina 2. Please refer to Supplementary

Table S5 for further details.

Notwithstanding this, our analyses offered interesting insights into virus strains from various regions of the world. Such sequence

the genetic evolution and mutations of circulating viruses in variants arise from frequent antigenic drift and/or re-assortments

Singapore during 2012 to 2015. Of particular interest was 2014 G2- within influenza A subtypes and B lineages (Nelson et al., 2008;

27.1 strain containing a combination of segments from both A/ Westgeest et al., 2014; Dudas et al., 2015). Together with the

H1N1 and A/H3N2. This strain may reflect a dual infection of A/ phylogenetic trees revealing that the Singapore strains were

H1N1(2009) and A/H3N2 which has been previously documented clustered around multiple geographically separated representative

(Lee et al., 2010), and may potentially generate novel A/H1N2 strains, it may indicate that Singapore is part of the global influenza

viruses that may spread globally (Komadina et al., 2014). circulation. Hence, being a major transportation hub, Singapore

We also revealed that the viral HA and NA sequences of A/H1N1, may regularly import and export A/H1N1, A/H3N2 and B viruses

A/H3N2 and B in this study were often closely related to different from and to the rest of the world, and contributes in driving the

94 F.X. Ivan et al. / International Journal of Infectious Diseases 90 (2020) 84–96

Figure 6. Binding of representative influenza A/H1N1 HA proteins with avian receptor analogs. (A) HA of SGH-C strain docked with 3’SLN. (B) HA of G2-25.1 strain docked

with 3’SLN. (C) HA of CA09 strain complexed with 3’SLN (PDB code 3UBQ). (D) HA of BM18 strain complexed with 3’SLN (PDB code 4JUH). The receptor is depicted as blue

sticks, while the respective HA protein is shown as a cartoon representation in the background. The receptor-binding site formed by the 190 helix, 220 loop and 130 loop is

indicated. Key HA residues forming potential hydrogen bonds with the avian receptor are depicted as sticks, where blue and red denote nitrogen and oxygen atoms,

respectively. Putative hydrogen-bonded contacts were identified using PyMOL polar contacts, and are indicated as black dashed lines. The modelled Singapore strains (A and

B) contain more hydrogen-bonded contacts, and the receptor binds deeper into the binding groove; both the sialic acid or SIA (innermost ring) and NAG (outermost ring)

moieties form potential hydrogen bonds with Asp190 (absent in 2009 and 1918 strains, C and D respectively). The galactose (GAL) moiety in SGH-C and G2-25.1 also makes

more hydrogen bonds with key residues Gln226 and Gly228. Please refer to Supplementary Table S7 for further details.

global influenza circulation patterns (Russell et al., 2008; Bedford In terms of the clinical significance of the sites discussed above,

et al., 2015). the residues in the recommended A/H3N2 vaccines during the

For protein site analyses, discrimination analysis using the study period matched the dominant residues in the inpatient

OneR approach mainly unravelled a number of sites in A/H3N2 isolates, except for site HA-145. Hence, the amino acid variant at

potentially linked to virulence. High accuracy was achieved at sites HA-145 (i.e. N145S) may potentially play a role in determining the

within PB2, PA, HA, NA and NS1. The site with the highest accuracy severity of A/H3N2 infection of those hospitalized patients. Indeed,

of 89.3% was position 221 in the NA loop, which may contribute to the mutation N145S has been acknowledged to be responsible for

the emergence of new antigenic variants (Gulati et al., 2002). For A/H3N2 antigenic drift during the 2014–2015 season (Chambers

the polymerase complex, a top site was position 668 in PA, which et al., 2015), and for A/H3N2 severity in Tunisia in 2013 (El Moussi

influences sensitivity to temperature and kinetics of viral et al., 2014). Moreover, our recent work in uncovering mutational

replication (Wei et al., 2018). The top sites in HA mainly resided patterns during A/H3N2 evolution revealed that this site is one of

at residues 145, 159 and 225 within epitopes A, B and D, the most frequently mutating sites positively correlated with

respectively. Sites 145 and 159 constitute two out of seven sites antigenic drift (data not shown). Nonetheless, it should be noted

where mutations are involved in antigenic transition during that all outpatient isolates also carried this mutation, with some

A/H3N2 evolution from 1968 to 2003 (Koel et al., 2013), while site strains harboring co-occurring mutations at other discriminating

225 is associated with reduced binding to the human alpha-2,6- sites (mainly I588T in PB2, T267K in NA, N272S, V668I and N675K

linked SIA receptor (Lin et al., 2012). Moreover, the most in PA). Future A/H3N2 surveillance could focus on HA-145 as well

discriminative HA site was at position 489, which was dominated as its associated co-mutating sites.

by asparagine in outpatient strains, but aspartic acid in inpatient Supplementary Table S8 presents the similarity between HAs of

strains. Overall, these results indicate that A/H3N2 strains harbor outpatient, inpatient and corresponding vaccine strains, and

more mutations that instigate antigenic and functional changes — indicates generally good vaccine match, e.g. for inpatient A/H3N2

which is unsurprising considering that the rate of evolution of A/ strains. However, mismatches between the recommended vaccine

H3N2 is higher than that of A/H1N1 and B (Chutinimitkul et al., strains and circulating viruses still remain a major challenge in

2010; Tewawong et al., 2015b). This also correlates with a greater preventing influenza outbreaks by immunization. Although we did

tendency of the A/H3N2 subtype to cause severe infections and not have the vaccination history of the study subjects, the seasonal

vaccine mismatches during epidemics, such as in the 2017–2018 influenza vaccine coverage in Singapore is generally low, e.g. only

season. about 15% among older adults (Ang et al., 2016b).

F.X. Ivan et al. / International Journal of Infectious Diseases 90 (2020) 84–96 95

Figure 7. Binding of representative influenza A/H1N1 HA proteins with human receptor analogs. (A) HA of SGH-C strain docked with 6’SLN. (B) HA of G2-25.1 strain docked

with 6’SLN. (C) Superimposed HA proteins of CA09 (dark purple; PDB code 3UBN) and BM18 (light pink; PDB code 4JUJ) strains complexed with 6’SLN (green) and LSTc (cyan),

respectively. The receptors are depicted as sticks, while the H1 HA protein is shown in cartoon representation. The orientation is cis for the pandemic strains, with sialic acid

(SIA) located near the 130 loop (C); but it is flipped for SGH-C (A) and G2-25.1 (B) with SIA facing the 220 loop instead.

Here, we were also motivated to perform protein ligand docking lower respiratory tract (Chutinimitkul et al., 2010; Zhang et al., 2013;

experiments to model the optimal binding mode of the HA and its Iovine et al., 2015). Despite lacking the D225G mutation, each HA of

receptor. Previously, the HAs of strains causing pandemics were SGH-C and G2-25.1 makes many contacts with the galactose moiety

reported to have stronger binding with alpha-2,6 SIA linkage in of the 3’SLN receptor, thus increasing the number of putative

human receptors than alpha-2,3 SIA linkage in avian receptors (Xu hydrogen bonds and the buried surface area of complex formation,

et al., 2012; Zhang et al., 2013). In our study, A/H1N1 acquired compared to the 1918 and 2009 A/H1N1 HAs. The visualization of

enhanced ability to bind both avian and human receptor analogs. HA-binding modes of SGH-C and G2-25.1 with avian and human

Interestingly, this trend of enhanced dual binding was also observed receptor analogs revealed that the avian receptor binds in an

in our study on A/H7N9 (Zhou et al., 2018). In general, the analyzed B extended conformation with SIA facing inwards to the binding

strains exhibited binding preference for the human receptor analog, groove i.e. trans conformation (Liu et al., 2009), while the human

whichmayexplain the considerable humaninfectivityofinfluenzaB. receptor is in the cis conformation. These predicted ligand poses are

Interestingly, we observed relatively stronger dual binding ability of distinct from the determined structures of previous reference

strainsbelonging totheB/Victorialineagewhichismoreinfectiousto strains. Taken together, these findings computationally suggest

young adults, in contrast to the B/Yamagata lineage which is more viral evolution towards dual receptor binding.

likely to infect the elderly (Socan et al., 2014; Chen et al., 2019).

Furthermore, the docked ligand posesofreceptor binding to SGH- Funding source

C and G2-25.1 HAs indicated high affinity for both avian and human

receptors, suggesting that these two Singapore strains may possess This work was supported by grants from the Academic Research

dual receptor specificity. As shown by structural computation, the Fund (AcRF) Tier 2, Ministry of Education, Singapore (MOE2014-

binding is comparable to or better than the human receptor. The T2-2-023), and the National Medical Research Council, Singapore

three HA residues G228, Q226 and D225 are known to be important (PPG10-09).

in receptor recognition, and facilitate potential hydrogen bonds with

the galactose portion of the receptor. While the D225 site confers Conflict of interest

specificity for the human receptor, the D225G mutation is found in

viruses that prefer the 3’SLN receptor, and is thus considered an The authors declare no conflict of interest.

avian-like evolutionary adaptation by the virus. In the 2009 H1N1

pandemic, at least 1% of all reported strains carried the D225G Ethical approval

mutation and resulted in fatalities. The structural basis of this

specificity results from D225G facilitating the interaction of A/H1N1 This clinical study was approved by Nanyang Technological

HA with the avian receptor that is mainly expressed in the human University Institutional Review Board (IRB-2015-12-023). Written

96 F.X. Ivan et al. / International Journal of Infectious Diseases 90 (2020) 84–96

informed consent was obtained from the participants before infections among Singapore military personnel in 2010-2013. Influenza Other

Respir Viruses 2014;8(5):557–66.

sample collection.

Holte R. Very simple classification rules perform well on most commonly used

datasets. Mach Learn 1993;11(1):63–91.

Acknowledgements Iovine NM, Morris JG, Fredenburg K, Rand K, Alnuaimat H, Lipori G, et al. Severity of

influenza A(H1N1) illness and emergence of D225G variant, 2013-14 influenza

season, Florida, USA. Emerg Infect Dis 2015;21(4):664–7.

We thank KW Tham, Priscilla Lau, Waseem Ahamed, Douglas

Koel BF, Burke DF, Bestebroer TM, van der Vliet S, Zondag GC, Vervaet G, et al.

Tay, Ridzwan Abdullah, and Catherine Chua for their assistance. We Substitutions near the receptor binding site determine major antigenic change

during influenza virus evolution. Science 2013;342(6161):976–9.

are grateful to CK Mok for useful discussions.

Komadina N, McVernon J, Hall R, Leder K. A historical perspective of influenza A

(H1N2) virus. Emerg Infect Dis 2014;20(1):6–12.

Appendix A. Supplementary data Lee VJ, Wong CS, Tambyah PA, Cutter J, Chen MI, Goh KT. Twentieth century

influenza pandemics in Singapore. Ann Acad Med Singapore 2008;37(6):470–6.

Lee N, Chan PK, Lam WY, Szeto CC, Hui DS. Co-infection with pandemic H1N1 and

Supplementary material related to this article can be found, in

seasonal H3N2 influenza viruses. Ann Intern Med 2010;152(9):618–9.

the online version, at doi:https://doi.org/10.1016/j.ijid.2019.10.024.

Lee HK, Tang JW, Kong DH, Koay ES. Simplified large-scale Sanger genome

sequencing for influenza A/H3N2 virus. PLoS One 2013;8(5):e64785.

Lee HK, Tang JW, Loh TP, Oon LL, Koay ES. Predicting clinical severity based on

References substitutions near epitope A of influenza A/H3N2. Infect Genet Evol

2015;34:292–7.

Lee HK, Lee CK, Tang JW, Loh TP, Koay ES. Contamination-controlled high-

Alhossary A, Handoko SD, Mu Y, Kwoh CK. Fast, accurate, and reliable molecular

throughput whole genome sequencing for influenza A viruses using the MiSeq

docking with QuickVina 2. Bioinformatics 2015;31(13):2214–6.

sequencer. Sci Rep 2016;6:33318.

Ang LW, Lim C, Lee VJ, Ma S, Tiong WW, Ooi PL, et al. Influenza-associated

Li W, Godzik A. Cd-hit: a fast program for clustering and comparing large sets of

hospitalizations, Singapore, 2004-2008 and 2010-2012. Emerg Infect Dis

2014;20(10):1652–60. protein or nucleotide sequences. Bioinformatics 2006;22(13):1658 9.

Lin YP, Xiong X, Wharton SA, Martin SR, Coombs PJ, Vachieri SG, et al. Evolution of

Ang LW, Tien WS, Lin RT, Cui L, Cutter J, James L, et al. Characterization of influenza

the receptor binding properties of the influenza A(H3N2) hemagglutinin. Proc

activity based on virological surveillance of influenza-like illness in tropical

Natl Acad Sci U S A 2012;109(52):21474–9.

Singapore, 2010-2014. J Med Virol 2016a;88(12):2069–77.

Liu J, Stevens DJ, Haire LF, Walker PA, Coombs PJ, Russell RJ, et al. Structures of

Ang LW, Cutter J, James L, Goh KT. Factors associated with influenza vaccine uptake

receptor complexes formed by hemagglutinins from the Asian Influenza

in older adults living in the community in Singapore. Epidemiol Infect

2016b;145(4):775–86. pandemic of 1957. Proc Natl Acad Sci U S A 2009;106(40):17175 80.

Nelson MI, Viboud C, Simonsen L, Bennett RT, Griesemer SB, St George K, et al.

Arnold K, Bordoli L, Kopp J, Schwede T. The SWISS-MODEL workspace: a web-based

Multiple reassortment events in the evolutionary history of H1N1 influenza A

environment for protein structure homology modelling. Bioinformatics

2006;22(2):195–201. virus since 1918. PLoS Pathog 2008;4(2):e1000012.

Ng TP, Pwee KH, Niti M, Goh LG. Influenza in Singapore: assessing the burden of

Bao Y, Bolotov P, Dernovoy D, Kiryutin B, Tatusova T. FLAN: a web server for

illness in the community. Ann Acad Med Singapore 2002;31(2):182–8.

influenza virus genome annotation. Nucleic Acids Res 2007;35(Web Server

Russell CA, Jones TC, Barr IG, Cox NJ, Garten RJ, Gregory V, et al. The global

issue) W280-4.

circulation of seasonal influenza A (H3N2) viruses. Science 2008;320

Bao Y, Bolotov P, Dernovoy D, Kiryutin B, Zaslavsky L, Tatusova T, et al. The influenza

(5874):340–6.

virus resource at the National Center for Biotechnology Information. J Virol

2008;82(2):596–601. Schrodinger LLC. The PyMOL molecular graphics system, Version 1.8. 2015.

Seah SG, Lim EA, Kok-Yong S, Liaw JC, Lee V, Kammerer P, et al. Viral agents

Bedford T, Riley S, Barr IG, Broor S, Chadha M, Cox NJ, et al. Global circulation

responsible for febrile respiratory illnesses among military recruits training in

patterns of seasonal influenza viruses vary with antigenic drift. Nature

2015;523(7559):217–20. tropical Singapore. J Clin Virol 2010;47(3):289 92.

Shu Y, McCauley J. GISAID: Global initiative on sharing all influenza data - from

Berman HM, Westbrook J, Feng Z, Gilliland G, Bhat TN, Weissig H, et al. The protein

vision to reality. Euro Surveill 2017;22(13) pii 30494.

data bank. Nucleic Acids Res 2000;28(1):235–42.

Socan M, Prosenc K, Ucakar V, Berginc N. A comparison of the demographic and

Biasini M, Bienert S, Waterhouse A, Arnold K, Studer G, Schmidt T, et al. SWISS-

clinical characteristics of laboratory-confirmed influenza B Yamagata and

MODEL: modelling protein tertiary and quaternary structure using evolutionary

Victoria lineage infection. J Clin Virol 2014;61(1):156–60.

information. Nucleic Acids Res 2014;42(Web Server issue):W252–8.

Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. MEGA6: molecular

Bouckaert R, Heled J, Kühnert D, Vaughan T, Wu CH, Xie D, et al. BEAST 2: a software

evolutionary genetics analysis version 6.0. Mol Biol Evol 2013;30(12):2725–9.

platform for Bayesian evolutionary analysis. PLoS Comput Biol 2014;10(4):

e1003537. Tewawong N, Suwannakarn K, Prachayangprecha S, Korkong S, Vichiwattana P,

Vongpunsawad S, et al. Molecular epidemiology and phylogenetic analyses of

Chambers BS, Parkhouse K, Ross TM, Alby K, Hensley SE. Identification of

influenza B virus in Thailand during 2010 to 2014. PLoS One 2015a;10(1):

hemagglutinin residues responsible for H3N2 antigenic drift during the

e0116302.

2014-2015 influenza season. Cell Rep 2015;12(1):1–6.

Tewawong N, Prachayangprecha S, Vichiwattana P, Korkong S, Klinfueng S,

Chen HJ, Su CP, Liu MT, Tsou TP. Comparative epidemiology of influenza B by lineage

Vongpunsawad S, et al. Assessing antigenic drift of seasonal influenza A

in intensive care unit-admitted patients with complications: a nationwide

(H3N2) and A(H1N1)pdm09 viruses. PLoS One 2015b;10(10):e0139958.

study in Taiwan, 2013-2017. Int J Infect Dis 2019;87:67–74.

Virk RK, Gunalan V, Lee HK, Inoue M, Chua C, Tan BH, et al. Molecular evidence of

Chi XS, Hu A, Bolar TV, Al-Rimawi W, Zhao P, Tam JS, et al. Detection and

transmission of influenza A/H1N1 2009 on a university campus. PLoS One

characterization of new influenza B virus variants in 2002. J Clin Microbiol

2005;43(5):2345–9. 2017;12(1):e0168596.

Wang Q, Tian X, Chen X, Ma J. Structural basis for receptor specificity of influenza B

Chutinimitkul S, Herfst S, Steel J, Lowen AC, Ye J, van Riel D, et al. Virulence-

virus hemagglutinin. Proc Natl Acad Sci U S A 2007;104(43):16874–9.

associated substitution D222G in the hemagglutinin of 2009 pandemic

Wei D, Yu DM, Wang MJ, Zhang DH, Cheng QJ, Qu JM, et al. Genome-wide

influenza A(H1N1) virus affects receptor binding. J Virol 2010;84(22):11802–13.

characterization of the seasonal H3N2 virus in Shanghai reveals natural

Cutter JL, Ang LW, Lai FY, Subramony H, Ma S, James L. Outbreak of pandemic

temperature-sensitive strains conferred by the I668V mutation in the PA

influenza A (H1N1-2009) in Singapore, May to September 2009. Ann Acad Med

subunit. Emerg Microbes Infect 2018;7(1):171.

Singapore 2010;39(4):273–310.

Westgeest KB, Russell CA, Lin X, Spronken MI, Bestebroer TM, Bahl J, et al.

Deng YM, Spirason N, Iannello P, Jelley L, Lau H, Barr IG. A simplified Sanger

Genomewide analysis of reassortment and evolution of human influenza A

sequencing method for full genome sequencing of multiple subtypes of human

(H3N2) viruses circulating between 1968 and 2011. J Virol 2014;88(5):2844–57.

influenza A viruses. J Clin Virol 2015;68:43–8.

Xu R, McBride R, Nycholat CM, Paulson JC, Wilson IA. Structural characterization of

Dudas G, Bedford T, Lycett S, Rambaut A. Reassortment between influenza B lineages

the hemagglutinin receptor specificity from the 2009 H1N1 influenza

and the emergence of a coadapted PB1-PB2-HA gene complex. Mol Biol Evol

2015;32(1):162–72. pandemic. J Virol 2012;86(2):982 90.

Yap J, Tan CH, Cook AR, Loh JP, Tambyah PA, Tan BH, et al. Differing clinical

Duncan IG, Taitel MS, Zhang J, Kirkham HS. Planning influenza vaccination

characteristics between influenza strains among young healthy adults in the

programs: a cost benefit model. Cost Eff Resour Alloc 2012;10(1):10.

tropics. BMC Infect Dis 2012;12:12.

Edgar RC. MUSCLE: multiple sequence alignment with high accuracy and high

Yu G, Smith DK, Zhu H, Guan Y, Lam TT. GGTREE: an R package for visualization and

throughput. Nucleic Acids Res 2004;32(5):1792–7.

annotation of phylogenetic trees with their covariates and other associated

El Moussi A, Ben Hadj Kacem MA, Slim A. Loss and gain of N-linked glycosylation

data. Methods Ecol Evol 2017;8(1):28–36.

sites in globular head and stem of HA found in A/H3N2 flu fatal and severe cases

Zhang W, Shi Y, Qi J, Gao F, Li Q, Fan Z, et al. Molecular basis of the receptor binding

during 2013 Tunisia flu seasonal survey. Virus Genes 2014;48(1):189–92.

specificity switch of the hemagglutinins from both the 1918 and 2009 pandemic

Gulati U, Hwang CC, Venkatramani L, Gulati S, Stray SJ, Lee JT, et al. Antibody

influenza A viruses by a D225G substitution. J Virol 2013;87(10):5949–58.

epitopes on the neuraminidase of a recent H3N2 influenza virus (A/Memphis/

Zhou X, Zheng J, Ivan FX, Yin R, Ranganathan S, Chow VT, et al. Computational

31/98). J Virol 2002;76(23):12274–80.

analysis of the receptor binding specificity of novel influenza A/H7N9 viruses.

Ho HP, Zhao X, Pang J, Chen MI, Lee VJ, Ang LW, et al. Effectiveness of seasonal

BMC Genomics 2018;19(Suppl. 2):88.

influenza vaccinations against laboratory-confirmed influenza-associated