Articles

Origin and diversity of novel avian inuenza A H7N9 viruses causing human infection: phylogenetic, structural, and coalescent analyses
Di Liu*, Weifeng Shi*, Yi Shi, Dayan Wang, Haixia Xiao, Wei Li, Yuhai Bi, Ying Wu, Xianbin Li, Jinghua Yan, Wenjun Liu, Guoping Zhao, Weizhong Yang, Yu Wang, Juncai Ma, Yuelong Shu, Fumin Lei, George F Gao

Summary
Background On March 30, 2013, a novel avian inuenza A H7N9 virus that infects human beings was identied. This virus had been detected in six provinces and municipal cities in China as of April 18, 2013. We correlated genomic sequences from avian inuenza viruses with ecological information and did phylogenetic and coalescent analyses to extrapolate the potential origins of the virus and possible routes of reassortment events. Methods We downloaded H7N9 virus genome sequences from the Global Initiative on Sharing Avian Inuenza Data (GISAID) database and public sequences used from the Inuenza Virus Resource. We constructed phylogenetic trees and did 1000 bootstrap replicates for each tree. Two rounds of phylogenetic analyses were done. We used at least 100 closely related sequences for each gene to infer the overall topology, removed suspicious sequences from the trees, and focused on the closest clades to the novel H7N9 viruses. We compared our tree topologies with those from a bayesian evolutionary analysis by sampling trees (BEAST) analysis. We used the bayesian Markov chain Monte Carlo method to jointly estimate phylogenies, divergence times, and other evolutionary parameters for all eight gene fragments. We used sequence alignment and homology-modelling methods to study specic mutations regarding phenotypes, specically addressing the human receptor binding properties. Findings The novel avian inuenza A H7N9 virus originated from multiple reassortment events. The HA gene might have originated from avian inuenza viruses of duck origin, and the NA gene might have transferred from migratory birds infected with avian inuenza viruses along the east Asian yway. The six internal genes of this virus probably originated from two dierent groups of H9N2 avian inuenza viruses, which were isolated from chickens. Detailed analyses also showed that ducks and chickens probably acted as the intermediate hosts leading to the emergence of this virulent H7N9 virus. Genotypic and potential phenotypic dierences imply that the isolates causing this outbreak form two separate subclades. Interpretation The novel avian inuenza A H7N9 virus might have evolved from at least four origins. Diversity among isolates implies that the H7N9 virus has evolved into at least two dierent lineages. Unknown intermediate hosts involved might be implicated, extensive global surveillance is needed, and domestic-poultry-to-person transmission should be closely watched in the future. Funding China Ministry of Science and Technology Project 973, National Natural Science Foundation of China, China Health and Family Planning Commission, Chinese Academy of Sciences.
Published Online May 1, 2013 http://dx.doi.org/10.1016/ S0140-6736(13)60938-1 *These authors contributed equally to this work These authors contributed equally to this work CAS Key Laboratory of Pathogenic Microbiology and Immunology (D Liu PhD, Y Shi PhD, Y Bi PhD, Y Wu PhD, Prof J Yan PhD, Prof W Liu PhD, Prof G F Gao DPhil), and Network Information Center (D Liu, W Li MSc, Prof J Ma PhD), Institute of Microbiology, Chinese Academy of Sciences, Beijing, China; School of Basic Medical Sciences, Taishan Medical College, Shandong Province, China (W Shi PhD); Key Laboratory of Zoological Systematics and Evolution, Institute of Zoology, Chinese Academy of Sciences, Beijing, China (W Shi, Prof F Lei PhD); Beijing Institutes of Life Science, Chinese Academy of Sciences, Beijing, China (Y Shi, Prof G F Gao); National Institute for Viral Disease Control and Prevention, Chinese Center for Disease Control and Prevention, Beijing, China (D Wang PhD, Prof Y Shu PhD, Prof G F Gao); Tianjin Institute of Industrial Biotechnology, Chinese Academy of Sciences, Tianjin, China (H Xiao); Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China (X Li BSc); University of Chinese Academy of Sciences, Beijing, China (X Li, Prof G F Gao); Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China (Prof G Zhao PhD); and Chinese Center for Disease Control and Prevention, Beijing, China (Prof W Yang MD, Prof Y Wang MD, Prof G F Gao)

Introduction
On March 30, 2013, a novel avian inuenza A H7N9 virus causing human infections was identied in China.1 As of April 18, 2013, the virus had spread to six provinces and municipal citiesie, Shanghai, Anhui, Jiangsu, Zhejiang, Beijing, and Henan. To our knowledge, this outbreak represents the rst time that the H7N9 subtype has infected people and caused fatal cases (as of April 18, 2013, 87 people have been infected and 17 have died). In modern times, H7 subtype avian inuenza viruses, including H7N1, H7N2, H7N3 and H7N7, have caused more than 100 human infections, including a fatal case in the Netherlands.24 In wild birds, H7 and N9 avian inuenza viruses have evolved to American, Oceanian, and Eurasian lineages.5,6 Preliminary analyses1 have shown that the H7N9 viruses causing the 2013 outbreak in China are

novel reassortants, the HA gene originated from avian inuenza viruses circulating in ducks in Zhejiang Province, the NA gene is related to avian inuenza viruses isolated from wild birds, and the internal genes probably orinigated from an earlier H9N2 lineage. Kageyama and colleagues7 have postulated that the internal genes originated from poultry viruses. In this study, we integrated data from phylogenetic analyses, coalescent analyses, and host ecology to infer the potential origins and genetic diversity of the novel avian inuenza A H7N9 viruses.

Methods
Sequence alignment
We downloaded the H7N9 virus genome sequences from the Global Initiative on Sharing Avian Inuenza Data (GISAID) databasespecically, A/Anhui/1/2013

www.thelancet.com Published online May 1, 2013 http://dx.doi.org/10.1016/S0140-6736(13)60938-1

Articles

Correspondence to: Prof George F Gao, Institute of Microbiology, Chinese Academy of Sciences, 1 West Beichen Road, Chaoyang, Beijing 100101, China. gaof@im.ac.cn

(Anhui/1), EPI439503~EPI439510; A/Shanghai/1/2013 (Shanghai/1), EPI439486~EPI439491, EPI439493, EPI439494; A/Shanghai/2/2013 (Shanghai/2), EPI439495~ EPI439502; and A/Hangzhou/1/2013 (Hangzhou/1), EPI440095~EPI440097. We downloaded all the public sequences used from the Inuenza Virus Resource.8 CLC

Main Workbench (CLC Bio, Aarhus, Denmark) was used for alignment and editing of sequences.

Phylogenetic analysis
Phylogenetic trees were inferred on the basis of maximumlikelihood methods, the general time-reversible model9 of

C A
Wild bird(W) /Duck(D)

E E E E

H7

E C C Eu H9N2, Chicken (Ck) Am PB2, PB1, PA, NP, M Oc E M E,C E

NS

H7N9, Ck/D

H7N9, H

D
H9N2, Ck M W W M M N9 M M D Mediterranean yway East Asian yway Southeast China C Eu Oc Am H7 N9 PA PB1 PB2 E E M C

E,C

2010

2011 Year

2012

2013

Figure 1: Spatial and temporal model of origin of novel avian inuenza A H7N9 virus In panel A, each circle represents an inuenza virus. The eight gene segments (horizontal bars) are, from top to bottom,PB2, PB1, PA, HA, NP, NA, MP, and NS. The question mark indicates the uncertainty of internal gene reassortment. In panel B, red circles represent the estimated times of most recent common ancestors. Green bars represent the times when the most closely related sequences (identied from phylogenetic analyses) of the novel H7N9 virus were collected (appendix). Panel C shows schematic phylogenetic trees of HA of the H7 subtype, and panel D shows schematic phylogenetic trees of NA of the N9 subtype, constructed on the basis of the maximum likelihood method and with 1000 bootstrap replicates. Figure 2 shows detailed bootstrap values. Coloured boxes and letters show clades in dierent yways. E=east Asian yway. M=Mediterranean yway. C=China. Am=American clade. Eu=Eurasian clade.

www.thelancet.com Published online May 1, 2013 http://dx.doi.org/10.1016/S0140-6736(13)60938-1

Articles

nucleotide substitution, and -distributed rates among sites (appendix). We did 1000 bootstrap replicates for each tree. All phylogenetic trees were constructed with RAxML.10 Two rounds of phylogenetic analyses were done. In the rst round, we used at least 100 closely related sequences for each gene to infer the overall topology. In the second round, on the basis of phylogenetic analysis, we removed a few suspicious sequences from the trees and focused on the closest clades to the novel H7N9 viruses. Tree topologies obtained from the two rounds

were compared with those from bayesian evolutionary analysis by sampling trees (BEAST) analysis.11

See Online for appendix

Bayesian Markov chain Monte Carlo evolutionary analyses and homology modelling
We used the bayesian Markov chain Monte Carlo method, implemented in BEAST11 (version 1.7.2) to jointly estimate phylogenies, divergence times, and other evolutionary parameters for all eight gene fragments. For all analyses we used the uncorrelated log-normal-distributed model12

A
79 100 83 93

B
A/duck/Korea/BC10/2007_H7N3 A/northern pintail/Aomori/1001/2008_H7N7 A/northern pintail/Akita/1367/2008_H7N7 A/duck/Tsukuba/922/2008_H7N7 A/duck/Tsukuba/30/2007_H7N7 A/mallard/Korea/GG2/2007_H7N7 D A/duck/Vietnam/G32/2008_H11N9 A/duck/Vietnam/G30/2008_H11N9 100 99 A/wild duck/Korea/SH20-27/2008_H7N9 A/duck/Vietnam/LBM81/2012_H11N9 A/northern pintail/Aomori/1192/2008_H5N9 A/duck/Tsukuba/239/2005_H11N9 A/duck/Tsukuba/164/2005_H11N9 A/sharp-tailed sandpiper/Australia/10/2004_H11N9 D 93 A/shore bird/Korea/S8/2006_H11N9 A/sharp-tailed sandpiper/Australia/6/2004_H11N9 A/duck/Hokkaido/W245/2004_H11N9 A/duck/Tsukuba/441/2005_H11N9 A/duck/Chiba/7/2006_H11N9 78 A/duck/Chiba/11/2006_H11N9 A/duck/Chiba/23/2006_H11N9 100 A/duck/Chiba/21/2006_H11N9 A/Baikal teal/Hongze/14/2005_H11N9 A/mallard/Czech Republic/13438-29K/2010_H11N9 W Hangzhou/1 Anhui/1 100 89 Shanghai/2 H Shanghai/1 100 A/wild bird/Korea/A14/2011_H7N9 100 A/wild bird/Korea/A3/2011_H7N9 100 A/spot-billed duck/Korea/47/2011_H7N9 A/wild bird/Korea/A9/2011_H7N9 93 A/duck/Mongolia/119/2008_H7N9 99 A/mallard/Korea/LBM616/2007_H11N9 D A/muscovy duck/Thailand/CU-LM1984/2009_H4N9 A/duck/Hunan/1590/2007_H6N9 A/Anas crecca/Spain/1460/2008_H7N9 W 83 A/mallard/PT/26153/2007_H11N9 A/mallard/Switzerland/WV1071028/2007_H11N9 100 A/mallard/Czech Republic/15962-1T/2010_H6N9 A/mallard/Czech Republic/15902-18K/2009_H11N9 W,G A/goose/Czech Republic/1848-T14/2009_H7N9 A/goose/Czech Republic/1848-K9/2009_H7N9 100 A/goose/Zambia/2009_H11N9 D,G A/duck/Zambia/11/2009_H11N9 A/mallard/Sweden/65/2002_H10N9 75 A/mallard/Sweden/36/2003_H11N9 A/mallard/Sweden/103/2005_H2N9 95 A/mallard/Netherlands/42/2006_H11N9 W A/mallard/Sweden/49/2005_H2N9 80 A/mallard/Sweden/90/2005_H11N9 A/mallard/Sweden/72/2005_H2N9 A/mallard/Sweden/49/2002_H5N9 A/mallard/Sweden/48/2002_H11N9 A/mallard/Netherlands/13/2001_H2N9 100 A/mallard/Netherlands/12/1999_H2N9 W A/mallard/Netherlands/11/1999_H2N9 A/mallard/Netherlands/11/1999_mixed 100 A/duck/Nanchang/4-184/2000_H2N9 A/duck/Nanchang/2-0486/2000_H2N9 D A/duck/Nanchang/2-0485/2000_H2N9 A/pintail/Shimane/324/1998_H1N9 A/swan/Shimane/227/2001_H3N9 A/swan/Shimane/190/2001_H6N9 100 W A/wild duck/Shantou/311/2001_H6N9 100 A/mallard/New Zealand/1365-350/2005_H6N9 A/mallard/New Zealand/1440-365/2005_H11N9 A/common teal/Netherlands/9/2000_H11N9 A/duck/Siberia/700/1996_H11N9 A/duck/Vietnam/OIE-2386/2009_H11N9
93

76

A/wild bird faeces/Korea/HDR22/2006_H7N7

93

A/wild bird faeces/Shihwa/21/2006_H7N3 A/duck/Mongolia/119/2008_H7N9

100

A/duck/Tsukuba/664/2007_H7N7 A/duck/Tsukuba/700/2007_H7N7 A/Northern shoveler/Seongdong/175/2008_H7N3 A/mallard/Korea/GH170/2007_H7N7 A/mallard/Korea/21-9/2008_H7N8

100 96

99

89

98

100 88 100

A/duck/Korea/A349/2009_H7N2 A/duck/Korea/A79/2010_H7N7 A/duck/Chiba/20/2009_H7N7 A/duck/Hokkaido/1/2010_H7N7 A/wild bird/Korea/A72/2010_H7N7 A/wild bird/Korea/A330/2009_H7N7

77

79

83 83 89

A/duck/Shimane/18/2006_H7N7 A/duck/Shiga/B149/2007_H7N7 A/duck/Korea/GJ56/2007_H7N8 A/quail/Thailand/CU-J2882/2009_H7N1

90

A/duck/Thailand/CU-LM7297T/2010_H7N6

93

78

A/duck/Shimane/83/2006_H7N3 A/mallard/Korea/5-80/2005_H7N8 W A/wild bird/Korea/5-77/2005_H7N8

100

100

Shanghai/1 Hangzhou/1 100 Anhui/1 Shanghai/2

100

A/duck/Zhejiang/12/2011_H7N3 A/duck/Zhejiang/2/2011_H7N3 A/duck/Zhejiang/10/2011_H7N3

85

100

A/wild bird faeces/Korea/ESD07/2005_H7N7 A/duck/Jiangxi/1814/2003_H7N7 D A/duck/Taiwan/4201/1999_H7N7 100 A/duck/Hokkaido/143/2003_H7N1

100

A/wild bird faeces/Nakdonggang/2-14/2004_H7N2 W A/wild bird faeces/Korea/HDR16/2003_H7N2

100

A/duck/Hokkaido/Vac-2/2004_H7N7 D A/duck/Mongolia/867/2002_H7N1 A/turkey/Italy/1265/1999_H7N1 A/mallard/Netherlands/12/2000_H7N3 99 W,T A/turkey/Germany/R11/2001_H7N7

100

001

002

Figure 2: Phylogenetic trees of H7 (A) and N9 (B) Dierent lineages are shown by dierent-coloured boxes (each colour corresponds to that in gure 1C, 1D). H7N9 viruses are shown by red lines and text. D=duck. W=wild bird. H=human being. T=turkey. G=goose.

www.thelancet.com Published online May 1, 2013 http://dx.doi.org/10.1016/S0140-6736(13)60938-1

Articles

For more on the SWISS MODEL see http://swissmodel.expasy.org

and the general time-reversible nucleotide substitution model. To investigate the extent to which dating estimates are aected by the demographic model chosen,13 we reanalysed on the basis of the constant size and exponential growth models (appendix). Bayesian Markov chain Monte Carlo analysis was run for dierent steps, 10% of which were removed as burn-in and sampled every 10 000 steps. We used the online structure homology-modelling server SWISS MODEL in automated model mode for our homology modelling of the ectodomain protein sequences of the H7 HAs and NAs.

Results
The phylogeny of the H7 gene sequences available showed three main independent lineagesie, American, Oceanian, and Eurasian lineages. The novel H7N9 virus fell within the Eurasian lineage, and was genetically close to sequences isolated from ducks in Zhejiang Province in 2011 (gures 1, 2). The H7 phylogenetic tree also showed that varied H7 viruses were circulating in wild ducks along the east Asian yway, which covers eastern China, South Korea, and Japan. Genetic exchange between wild birds and ducks
A

has also been recorded (gure 2). The estimated time to most recent common ancestor of the novel H7N9 strains was roughly January, 2012 (mean Jan 4, 2012, 95% highest posterior density March 5, 2011Oct 12, 2012). Similar to the H7 phylogeny, the N9 phylogenetic tree showed that the NA gene of the novel H7N9 virus also belonged to the Eurasian lineage (gure 1). The most closely related gene sequences were from H7N9 viruses isolated from wild ducks in South Korea in February and April of 2011,14 implying a similar virus circulating in this area. At that time, these wild ducks colonised for breeding during northward migration from southeast China along the east Asian yway, and the estimated time to most recent common ancestor of the NA gene was roughly mid-June, 2011 (mean June 18, 2011, 95% highest posterior density March 1, 2010July 9, 2012; gure 1). Without exception, the six internal genes of the H7N9 virus were clustered together with H9N2 viruses isolated from China.1,7 Nevertheless, phylogenetic analyses suggested at least two dierent origins of H9N2 avian inuenza viruses, one for the NS gene and one for the remaining sets of internal genes (gure 3; appendix).
B
A/chicken/Hebei/0329/2007_H9N2 A/chicken/Shandong/LY/2006_H9N2 A/chicken/Hunan/5260/2005_H9N2 A/chicken/Hunan/6108/2005_H9N2 A/chicken/Hunan/4444/2005_H9N2 A/chicken/Henan/2/2004_H9N2 A/chicken/Henan/1/2005_H9N2 A/chicken/Henan/1/2004_H9N2 A/chicken/Shanghai/1/2002_H9N2 A/chicken/Shanghai/2/2002_H9N2 A/duck/Zhejiang/3/2002_H9N2 A/chicken/Zhejiang/611/2011_H9N2 A/chicken/Zhejiang/607/2011_H9N2 A/chicken/Shanghai/C1/2012_H9N2 A/chicken/Zhejiang/329/2011_H9N2 A/chicken/Shanghai/C2/2012_H9N2 A/chicken/Shanghai/C3/2012_H9N2 A/chicken/Zibo/B1/2008_H9N2 A/chicken/Jiangxi/12/2007_H9N2 A/chicken/Guangdong/A1/2003_H9N2 A/brambling/Beijing/16/2012_H9N2 A/chicken/Vietnam/OIE-1611/2012_H9N2 A/chicken/Taixing/10/2010_H9N2 A/chicken/Yangzhou/11/2010_H9N2 A/chicken/Shuanggou/1/2011_H9N2 A/chicken/Xigou/1/2011_H9N2 A/pigeon/Xuzou/1/2011_H9N2 Anhui/1 Shanghai/2 Shanghai/1 A/chicken/Dawang/1/2011_H9N2 A/chicken/Shandong/01/2009_H9N2 A/chicken/Shandong/02/2009_H9N2 A/chicken/Shandong/01/2010_H9N2 A/chicken/Shandong/1231/2008_H9N2 A/chicken/Shandong/SG2/2009_H9N2 A/duck/Tibet/S2/2009_H9N2 A/chicken/China/AH-10-01/2010_H9N2 A/chicken/Tibet/S4/2009_H9N2 A/environment/Xinjiang/6/2009_H5N1 A/chicken/Zhejiang/HJ/2007_H9N2 A/chicken/Shandong/KD/2009_H9N2 A/chicken/Shandong/12/2008_H9N2 A/environment/ChangSha/6/2009_H9N2 A/chicken/Hubei/10/2009_H9N2

100

100

A/chicken/Zhejiang/329/2011_H9N2 A/brambling/Beijing/16/2012_H9N2 Shanghai/1 Anhui/1 Shanghai/2 A/chicken/Shanghai/C1/2012_H9N2 A/chicken/Hebei/1102/2010_H5N2 A/chicken/Shanghai/C2/2012_H9N2 A/chicken/Shanghai/C3/2012_H9N2 A/chicken/Jiangsu/Q3/2010_H9N2 A/chicken/Anhui/HF/2010_H9N2 A/chicken/Zhejiang/607/2011_H9N2 A/chicken/Wuxi/7/2010_H9N2 A/chicken/Tongshan/1/2011_H9N2 A/chicken/Yongcheng/1/2011_H9N2 A/chicken/Jiawang/2/2011_H9N2 A/chicken/Shandong/03/2010_H9N2 A/chicken/Shandong/H/2009_H9N2 A/chicken/Shandong/BD/2010_H9N2 A/chicken/Shandong/02/2010_H9N2 A/chicken/Qianzhou/12/2010_H9N2 A/chicken/Zhejiang/Q1D4/2010_H9N2 A/chicken/Guangdong/ZCY/2011_H9N2 A/chicken/Guangdong/ZHJ/2011_H9N2 A/equine/Guangxi/3/2011_H9N2 A/canine/Guangxi/1/2011_H9N2 A/chicken/Shuanggou/1/2011_H9N2 A/chicken/Dawang/1/2011_H9N2 A/chicken/Zhejiang/611/2011_H9N2 A/chicken/China/AH-10-01/2010_H9N2 A/pigeon/Xuzou/1/2011_H9N2 A/chicken/Xigou/1/2011_H9N2
100

100

0015

0015

Figure 3: Phylogenetic trees of PB2 (A) and NS (B) Avian inuenza A H7N9 viruses are shown in red, brambling viruses in pink, viruses isolated from Shanghai in 2012 in blue, and viruses isolated from Jiangsu province in green.

www.thelancet.com Published online May 1, 2013 http://dx.doi.org/10.1016/S0140-6736(13)60938-1

Articles

The most closely related genesPB2, PB1, and PA were all from one strain, A/brambling/Beijing/ 16/2012(H9N2), which was collected on Nov 7, 2012. The brambling is a migratory bird and might y south for wintering. The phylogenetic trees also showed that the polymerase genes of this virus fell within a clade of chicken H9N2 viruses in the vicinity of Shanghai in 2012 and Zhejiang in 2011; viruses from Beijing or other northern provinces were not identied in this clade (gure 3; appendix). The long branch between the brambling virus and the novel H7N9 viruses shows that few surveillance data are available. Furthermore, the estimated time to most recent common ancestor to the
A
Shanghai/1 180 helix

polymerase genes was before June, 2012 (gure 1), suggesting that the brambling H9N2 virus is a more recent strain than is the ancestor of the novel H7N9 avian inuenza viruses. Further examination of the H7 phylogeny showed a long branch separating Shanghai/1 from Anhui/1, Shanghai/2, and Hangzhou/1 in the the H7N9 viruses that infected people (gure 2). Sequence alignment of the HA protein sequences identied nine aminoacid mutations that distinguished Shanghai/1 from the other three strains (gure 4). The diversication between Shanghai/1 and other human isolates (shown by the long branch) was also noted in the NA and internal genes (gures 2, 3;

Anhui/1 180 helix

Hangzhou/1 180 helix

Gln217

Leu217

Ile217

210 loop

120 loop

210 loop

120 loop

210 loop

120 loop

Hydrophilic interaction

Hydrophobic interaction

Hydrophobic interaction

B
HA1 128 Ala . . . . . Ser 163 Lys . . Arg Arg Arg Arg 165 Asp . . Ser Ser Ser Asn 170 Ile . . Val Val Val Val 177 Gly . . Val Val Val . 193 Ile . . Val Val Val Val 212 Pro . . . . . Thr 217 Gln . . Leu Leu Ile . 267 Asp . . Asn Asn Asn . 274 His . . . . . Tyr 280 Val . . Ile Ile Ile Ile 317 Thr . . Ile Ile Ile Ile 392 Thr . . Asn Asn Asn . HA2 437 Asn . . Asp Asp Asp Asp 523 Ala . . Val Val Val .

A/duck/Zhejiang/2/2011(H7N3) A/duck/Zhejiang/10/2011(H7N3) A/duck/Zhejiang/12/2011(H7N3) A/Anhui/1/2013(H7N9) A/Shanghai/2/2013(H7N9) A/Hangzhou/1/2013(H7N9) A/Shanghai/1/2013(H7N9)

C
Deletion 22 Ile . . . Val Val Val Val 26 Ile . . . . Met . . 27 Thr . . . Ala Ala Ala Ala 40 Ser . . . Gly Gly Gly . 69 Gln . . . ~ Ser . . . 73 Thr 81 Ala . Ser Ser Thr Thr Thr Thr 84 Gly . . . Asn Asn Asn Asn 111 Lys . . . Glu Glu Glu Glu 117 Ile . . . Val Val Val Val 294 Arg . . . . . . Lys 401 Thr . . . Ala Ala Ala Ala 415 Asp Gly Gly Gly Ala Ala Ala Ala

A/wildbird/Korea/A9/2011(H7N9) A/wildbird/Korea/A14/2011(H7N9) A/wildbird/Korea/A3/2011(H7N9) A/spot-billed duck/447/2011(H7N9) A/Anhui/1/2013(H7N9) A/Shanghai/2/2013(H7N9) A/Hangzhou/1/2013(H7N9) A/Shanghai/1/2013(H7N9)

Ile

Asn . . . -

Figure 4: Homology-modelling structural analysis of haemagglutins derived from human avian inuenza A H7N9 virus isolates (A); and alignment of aminoacids of both haemagglutins and neuraminidases between H7N9 and viruses in adjacent clade (B, C) At position 217 (corresponding to position 226 of H3 numbering), the haemagglutinin of Shanghai/1 has a hydrophilic Gln, that of Anhui/1 and Shanghai/2 a hydrophobic Leu, and that of Hangzhou/1 a hydrophobic Ile. Gln is compatible with the avian receptor (which is hydrophilic), whereas Leu and Ile are compatible with the human receptor (which is hydrophobic). Numbering in (B) starts from the mature haemagglutin polypeptide, whereas that in (C) starts from the rst Met of the alignments. The Rasmol colouring scheme in CLC Main Workbench was used; it shows dierent properties of aminoacids.

www.thelancet.com Published online May 1, 2013 http://dx.doi.org/10.1016/S0140-6736(13)60938-1

Articles

Panel: Research in context Systematic review We did a basic local alignment search tool (BLAST) search of each H7N9 gene segment against the Inuenza Virus Resource on April 6, 2013. The most 100 similar sequences were then extracted for the phylogenetic and coalescent analyses. Recent reports have described clinical information about the rst patients infected with avian inuenza A H7N9 virus and provided preliminary information about the origins of the virus.1,7 The authors of a previous report proposed three origins of the H7N9 virus.1 Interpretation This study is the rst (to our knowledge) in which comprehensive methods were applied to infer the evolution of the H7N9 virus. We proposed that the virus might have evolved from at least four origins and discussed the potential roles of migratory birds and poultry in the H7N9 outbreak. Both the phylogenetic analysis and the phenotypic inference on some key aminoacid sites have emphasised that extensive surveillance of avian inuenza A H7N9 is necessary in humans, poultry, and wild birds.

Discussion
Tracing the origin of the novel H7N9 virus is of vital importance for formulation of eective prevention and surveillance policies. We propose at least four possible gene-segment origins are proposed for the novel H7N9 virus (panel). The estimated time to most recent common ancestor of the novel H7N9 strain was during the wintering period for wild birds. Thus, corresponding H7 strains might have been circulating in poultry (most likely in ducks) for at least 1 year. An earlier H11N9 strain of mallard (Anas platyrhynchos) origin from the Czech Republic in 2010, and an H7N9 strain of common teal (Anas crecca) origin from Spain in 2008, also had similar N9 genes to the H7N9 viruses, suggesting that the earlier N9 genes might have been introduced from Europe through bird migration (gure 2). Therefore, the NA gene fragment of the novel H7N9 virus possibly originated from avian inuenza viruses carried by wild birds. However, wild ducks are unlikely to transfer avian inuenza viruses directly to chickens. Most probably, the wild ducks rst transferred the viruses to domesticated ducks (wild and domesticated ducks have similar behaviours and share habitats in eastern China). Additionally, the long-branch between the novel H7N9 and avian inuenza viruses from wild birds in the phylogenetic tree raised the possibility of the existence of an intermediate host (gure 2). Do ducks obtain the HA and NA genes from migratory birds sequentially or simultaneously? We think that the two alternative possibilities coexist. Wild ducks, especially mallards, were frequently infected with or carried H7 viruses (according to sequence information available at the Inuenza Virus Resource), and mallards and spot-billed ducks (Anas poecilorhyncha) often mix together in a very large colony for molting and wintering.17 Furthermore, mallards and spot-billed ducks, along with the common teal, are the dominant wintering ducks in southeast China.17 These large mixed colonies of wild ducks that share habitats with domestic ducks might increase opportunities for genetic reassortment of viruses. The NS gene was clearly closely related to a group of H9N2 avian inuenza viruses in chickens in Jiangsu, China, whereas the remaining internal genes were closely related to those noted in avian inuenza viruses isolated from chickens in Shanghai and the vicinity. The distance between these locations where dierent groups of H9N2 avian inuenza viruses were collected is roughly 200 km, and hence virus transmission could have occurred via chicken transportation. Although the NP genes of all three H7N9 isolates were clustered together with avian inuenza viruses from Shanghai, they fell within two separate subclades, which implies that the genetic diversity of H9N2 avian inuenza viruses in chickens in this area is high (appendix). We believe that, similar to the other three internal genes, PB2, PB2 and PA might have arisen from H9N2

appendix). 11 aminoacid mutations and a ve aminoacid deletion in the NA protein distinguished the novel H7N9 viruses from those noted in wild birds; two of these mutations were noted in Shanghai/1 only (gure 4). One glutamine (Gln) to leucine (Leu) mutation (position 226 in H3 numbering and 217 in H7 numbering) at the receptor-binding site might cause the H7N9 virus to bind with high anity to the human receptor.15 This nding could be explained by the hydrophilic characteristics of the Gln that interacts with the avian receptor and the hydrophobic characteristics of the Leu that favours the human receptor (gure 4).16 The substitution Gln226Leu, which was noted in Anhui/1 and Shanghai/2, suggests that the novel H7N9 virus might have changed receptor-binding properties. Additionally, the substitution Gln226Ile (isoleucine) was noted in Hangzhou/1, which probably increases the potential for high-anity binding to human receptors because isoleucine is hydrophobic. This Gln226Ile substitution has not been noted previously for any HA subtypes (gure 4). Although we integrated data from various sources and proposed the potential origins and reassortment routes of the novel avian origin H7N9 inuenza virus, we emphasise that our analysis did not have extensive surveillance dataa bottleneck for all researchers partly because, compared with highly pathogenic H5N1 avian inuenza viruses, viruses of the H7 subtype have low pathogenicity for avian hosts. Hence, avian hosts could harbour viruses for a long time without showing symptoms. However, we have analysed most publicly available, genetically related sequences.
6

www.thelancet.com Published online May 1, 2013 http://dx.doi.org/10.1016/S0140-6736(13)60938-1

Articles

virus within chicken populations, although wild birds could have brought earlier lineages of viruses to southern China. Increased surveillance data are needed in this region to further elucidate the origin of this virus and trace the evolution of the polymerase genes. We hypothesise that the H7N9 viruses that infected human beings resulted from a reassortment of avian inuenza viruses of at least four originsduck origin for HA, duck (probably also wild bird) origin for NA, and at least two H9N2 chicken viruses for the internal genes (gure 1). Regarding the potential intermediate hosts and the place where the reassortment events could have taken place, we extrapolated, on the basis of available evidence, that the HA genes were circulating in the east Asian yway in both wild birds and ducks, the NA genes were introduced from European lineages at an early time and transferred to ducks in China by wild birds through migration along the east Asian yway, and H9N2 avian inuenza viruses were circulating in chicken and duck populations in eastern China and possibly reassorted with the H7 and N9 avian inuenza viruses in ducks, which led to the emergence of the new H7N9 lineage. After these reassortment events, the new viruses started to circulate in chickens, with low pathogenicity. Furthermore, we propose that these reassortment events most probably took place in Shanghai or the adjacent provinces, such as Zhejiang or Anhui. Position 292 of the NA gene (position 294 in N9) has been previously reported to confer resistance to oseltamivir.18,19 The strain Shanghai/1 had a lysine at this position and was thus deemed resistant to oseltamivir, whereas all the other strains had an arginine at this site, and were thus supposed to be sensitive to oseltamivir.19 These genotypic and postulated phenotypic dierences imply that the H7N9 lineage has diversied since its emergence and emphasise the necessity of extensive surveillance of this virus in human beings, poultry, and wild birds. Several long branches existed between the novel H7N9 clade and closely related sequences, which could be explained as unknown intermediate hosts or an absence of genetic diversity because few sequence data are available. Because few surveillance data are available, to establish whether the noted aminoacid substitutions that increased receptor binding in people resulted from xation of genetic drifts or under host-specic selective pressures is dicult. So far, no mammals other than human beings have been reported to be infected with H7 or N9 inuenza viruses in China, but H9N2 viruses have been isolated from pigs. In conclusion, on the basis of available evidence, we believe that the novel avian inuenza A H7N9 virus was a multiple reassortant. The HA and NA genes might originate from duck avian inuenza viruses, which might have obtained the viral genes from migratory birds a year previously, whereas the internal genes might come from chicken avian inuenza viruses. We believe that the estimated times to most recent common

ancestor for the eight genomic fragments and the frequent poultry transportation in China account for the increased number of conrmed sporadic cases of human infection. In particular, this novel H7N9 virus has diversied into dierent lineages since its emergence several months ago.
Contributors GFG, FL, DL, WS, and YShu designed the study. DL, WS, YShi, HX, DW, WLi, YB, YWu, XL, JY, WLiu, and FL analysed the data. DL, WS, YShi, GZ, WY, YWa, JM, YShu, FL, and GFG interpreted the data. DL, WS, YShi, FL, and GFG prepared the gures. DL, WS, YShi, YShu, FL, and GFG wrote the paper. Conicts of interest We declare that we have no conicts of interest. Acknowledgments The study was supported by grants from China Ministry of Science and Technology Project 973 (numbers 2010CB530303, 2011CB504703, and 2012CB955501), National Natural Science Foundation of China (numbers 30925008 and 81290342), Grand S&T project of China Health and Family Planning Commission (number 2012ZX10004201-009), and intramural special grant for inuenza virus research from the Chinese Academy of Sciences (KSZD-EW-Z-002). GFG is a leading principal investigator of the Innovative Research Group of the National Natural Science Foundation of China (grant number 81021003). We thank Chris Vavricka and Joel Heywood for critical reading and English editing, and Yan Wu for helpful discussion. References 1 Gao R, Cao B, Hu Y, et al. Human infection with a novel avian-origin inuenza A (H7N9) virus. N Engl J Med 2013; published online April 11. DOI:10.1056/NEJMoa1304459. 2 Nguyen-Van-Tam JS, Nair P, Acheson P, et al. Outbreak of low pathogenicity H7N3 avian inuenza in UK, including associated case of human conjunctivitis. Euro Surveill 2006; 11: E060504 2. 3 Stegeman A, Bouma A, Elbers AR, et al. Avian inuenza A virus (H7N7) epidemic in the Netherlands in 2003: course of the epidemic and eectiveness of control measures. J Infect Dis 2004; 190: 208895. 4 Fouchier RA, Schneeberger PM, Rozendaal FW, et al. Avian inuenza A virus (H7N7) associated with human conjunctivitis and a fatal case of acute respiratory distress syndrome. Proc Natl Acad Sci USA 2004; 101: 135661. 5 Belser JA, Bridges CB, Katz JM, Tumpey TM. Past, present, and possible future human infection with inuenza virus A subtype H7. Emerg Infect Dis 2009; 15: 85965. 6 Krauss S, Obert CA, Franks J, et al. Inuenza in migratory birds and evidence of limited intercontinental virus exchange. PLoS Pathog 2007; 3: e167. 7 Kageyama T, Fujisaki S, Takashita E, et al. Genetic analysis of novel avian A(H7N9) inuenza viruses isolated from patients in China, February to April 2013. Euro Surveill 2013; 18: pii=20453. 8 Bao Y, Bolotov P, Dernovoy D, et al. The inuenza virus resource at the National Center for Biotechnology Information. J Virol 2008; 82: 596601. 9 Rodriguez F, Oliver JL, Marin A, Medina JR. The general stochastic model of nucleotide substitution. J Theor Biol 1990; 142: 485501. 10 Stamatakis A. RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 2006; 22: 268890. 11 Drummond AJ, Rambaut A. BEAST: bayesian evolutionary analysis by sampling trees. BMC Evol Biol 2007; 7: 214. 12 Drummond AJ, Ho SY, Phillips MJ, Rambaut A. Relaxed phylogenetics and dating with condence. PLoS Biol 2006; 4: e88. 13 Drummond AJ, Nicholls GK, Rodrigo AG, Solomon W. Estimating mutation parameters, population history and genealogy simultaneously from temporally spaced sequence data. Genetics 2002; 161: 130720. 14 Kim HR, Park CK, Lee YJ, et al. Low pathogenic H7 subtype avian inuenza viruses isolated from domestic ducks in South Korea and the close association with isolates of wild birds. J Gen Virol 2012; 93: 127887.

www.thelancet.com Published online May 1, 2013 http://dx.doi.org/10.1016/S0140-6736(13)60938-1

Articles

15 16

17

Herfst S, Schrauwen EJ, Linster M, et al. Airborne transmission of inuenza A/H5N1 virus between ferrets. Science 2012; 336: 153441. Ha Y, Stevens DJ, Skehel JJ, Wiley DC. X-ray structures of H5 avian and H9 swine inuenza virus hemagglutinins bound to avian and human receptor analogs. Proc Natl Acad Sci USA 2001; 98: 1118186. Cheng TH. Fauna sinica: aves, volume 2: anseriformes. Beijing: Science Press, 1979.

18 19

Colman PM. Structural basis of antigenic variation: studies of inuenza virus neuraminidase. Immunol Cell Biol 1992; 70: 20914. Kiso M, Ozawa M, Le MT, et al. Eect of an asparagine-to-serine mutation at position 294 in neuraminidase on the pathogenicity of highly pathogenic H5N1 inuenza A virus. J Virol 2011; 85: 466772.

www.thelancet.com Published online May 1, 2013 http://dx.doi.org/10.1016/S0140-6736(13)60938-1