Emergence and predominance of an H5N1 influenza …that H5N1-positive chicken were detected in 11 of the last 12 months, a marked increase from only 4 positive months in 2004 2005

Emergence and predominance of an H5N1 influenzavariant in ChinaG. J. D. Smith†, X. H. Fan†, J. Wang†, K. S. Li†, K. Qin†, J. X. Zhang†, D. Vijaykrishna†, C. L. Cheung†, K. Huang†,J. M. Rayner†, J. S. M. Peiris†, H. Chen†, R. G. Webster†‡§, and Y. Guan†§

†State Key Laboratory of Emerging Infectious Diseases, Department of Microbiology, Li Ka Shing Faculty of Medicine, University of Hong Kong,21 Sassoon Road, Pokfulam, Hong Kong Special Administration Region, China; and ‡Virology Division, Department of Infectious Diseases, St. Jude Children’sResearch Hospital, Memphis, TN 38105

Contributed by R. G. Webster, September 20, 2006

The development of highly pathogenic avian H5N1 influenzaviruses in poultry in Eurasia accompanied with the increase inhuman infection in 2006 suggests that the virus has not beeneffectively contained and that the pandemic threat persists. Up-dated virological and epidemiological findings from our marketsurveillance in southern China demonstrate that H5N1 influenzaviruses continued to be panzootic in different types of poultry.Genetic and antigenic analyses revealed the emergence and pre-dominance of a previously uncharacterized H5N1 virus sublineage(Fujian-like) in poultry since late 2005. Viruses from this sublineagegradually replaced those multiple regional distinct sublineages andcaused recent human infection in China. These viruses have alreadytransmitted to Hong Kong, Laos, Malaysia, and Thailand, resultingin a new transmission and outbreak wave in Southeast Asia.Serological studies suggest that H5N1 seroconversion in marketpoultry is low and that vaccination may have facilitated theselection of the Fujian-like sublineage. The predominance of thisvirus over a large geographical region within a short period directlychallenges current disease control measures.

influenza A � molecular epidemiology � virus evolution

Extensive surveillance and genetic studies have revealed thathighly pathogenic avian influenza H5N1 viruses have be-

come first predominant and then endemic in poultry in southernChina and Southeast Asia since 2003 (1). This endemicityresulted in the establishment of multiple distinct regional sub-lineages (2). The recognition of multiple different H5N1 sub-lineages makes it possible to identify the source and to under-stand the evolutionary and transmission pathways of H5N1viruses that have become widespread in Southeast Asia, Europe,and Africa.

Since the H5N1 influenza virus caused the first outbreak inmigratory waterfowls at Qinghai Lake in May 2005 (3), a newtransmission and outbreak wave was initiated. The virus expandedits geographical distribution and caused outbreaks in poultry in over30 countries from Central Asia, the Middle East, Europe, andAfrica (4). This expansion led directly to a marked increase inhuman infection cases and escalated the pandemic threat. In thefirst 8 months of 2006, the World Health Organization confirmed96 cases from 9 countries; whereas for the whole of 2005 there wasa total of 95 cases from 5 countries (5). In addition, in Indonesiarecently there were suspected cases of human-to-human transmis-sion involving members of an extended family, and the infectionsources of other human cases have not been identified (6).

In China, despite a compulsory program for the vaccination of allpoultry commencing in September 2005 (7), H5N1 influenza virushas caused outbreaks in poultry in 12 provinces from October 2005to August 2006 (4). At the same time, 22 human infection cases havebeen confirmed from 14 provinces since November 2005 (4, 5).Some of those cases were residents of metropolitan areas remotefrom poultry farms, such as Guangzhou, Wuhan, and Shanghai (4).Furthermore, there were no obvious poultry outbreaks reported inneighboring markets or farms before or after the onset of those

human infections. Therefore, whether those people were infectedlocally and directly from affected poultry or other sources, includinghumans, is still unknown. This situation directly challenges currentpandemic preparedness plans, raising concern that a pandemiccould emerge not only from the countryside but also from an urbanarea, just as severe acute respiratory syndrome emerged from thelive-animal markets of Guangzhou and the Pearl River delta (8, 9).

Here we report updated virological and epidemiological find-ings from our market surveillance in southern China. Epidemi-ological analysis showed that H5N1 influenza viruses werecontinued to be perpetuated in poultry in each of the provincestested, mainly in domestic duck and geese. Genetic analysisrevealed that an H5N1 influenza variant had emerged andbecome predominant in each of the provinces, replacing thosepreviously established multiple sublineages in different regionsof southern China. This virus had also transmitted to HongKong, Laos, Malaysia, and Thailand. Serological studies sug-gested that the seroconversion rate in poultry in China is low andthat the emergence and predominance of this H5N1 strain maybe associated with vaccination in poultry. Genetic findings alsorevealed that these viruses also were responsible for all recentlyreported human infection cases in China. The predominance ofthis virus over a large geographical region within a short periodquestions the efficacy of current disease control measures inpoultry and revealed that a new transmission and outbreak wavehas been initiated from China to Southeast Asia since early 2006.

ResultsSurveillance. From July 2005 to June 2006 our influenza surveil-lance in live-poultry markets in six provinces of southern Chinashowed that 1,294 of 53,220 (overall isolation rate 2.4%) poultrywere H5N1-positive (Fig. 1; see also Fig. 5 and Table 3, which arepublished as supporting information on the PNAS web site). Themain body of H5N1 isolates was from duck and goose, with onlya small number isolated from chicken (chicken 0.5%, duck 3.3%,goose 3.5%). The prevalence of H5N1 viruses in southern Chinahas increased when compared with the period July 2004 to June2005 (overall, 0.9%; chicken, 0.2%; duck, 1.3%; and goose,2.0%) (Fig. 1 and Table 3). A winter-seasonal peak was observedfrom October 2005 to March 2006 as in previous years (1, 2),during which H5N1 influenza viruses were isolated in each

Author contributions: G.J.D.S., X.H.F., K.S.L., J.S.M.P., H.C., R.G.W., and Y.G. designedresearch; G.J.D.S., J.W., K.Q., J.X.Z., D.V., C.L.C., K.H., and J.M.R. performed research;G.J.D.S., J.M.R., and H.C. contributed new reagents�analytical tools; G.J.D.S., X.H.F., K.S.L.,D.V., J.S.M.P., H.C., R.G.W., and Y.G. analyzed data; G.J.D.S. and Y.G. wrote the paper; andY.G. coordinated the study.

The authors declare no conflict of interest.

Abbreviation: HI, HA inhibition.

Data deposition: The sequences reported in this paper have been deposited in the GenBankdatabase (accession nos. DQ992562–DQ993117).

§To whom correspondence may be addressed. E-mail: [email protected] [email protected].

© 2006 by The National Academy of Sciences of the USA

16936–16941 � PNAS � November 7, 2006 � vol. 103 � no. 45 www.pnas.org�cgi�doi�10.1073�pnas.0608157103

Dow

nloa

ded

by g

uest

on

May

20,

202

0

province tested (Fig. 1). However, an extension of the peakseason was observed in April to June 2006 because isolation ratesremained high in these warmer months (Fig. 1) (1, 2).

Comparison between different types of poultry shows that H5N1viruses were mainly isolated from domestic duck and goose whereinthe viruses were prevalent year-round, whereas chicken testedpositive mostly during the winter (Fig. 1 and Table 3). It is notablethat H5N1-positive chicken were detected in 11 of the last 12months, a marked increase from only 4 positive months in 2004�2005 (Fig. 1A). These findings indicate an escalation of H5N1activity in poultry in 2005�2006 compared with previous years andsuggest that H5N1 influenza viruses have not been effectivelycontained in this region and have maintained endemicity broadly inpoultry, especially domestic duck and goose.

Regarding virus names and sublineages, the following nomen-clature applies: BH goose, bar-headed goose; Ck, chicken; FJ,Fujian; Dk, duck; GD, Guangdong; Gf, Guinea fowl; Gs, goose;GX, Guangxi; GY, Guiyang; HK, Hong Kong; HN, Hunan; IDN,Indonesia; Mixed, southern China isolates; QH, Qinghai; ST,Shantou; YN, Yunnan; VNM, Vietnam; VNM2, second Viet-

nam introduction in March 2005;. The year the virus wasidentified is represented by the last two digits of the year: e.g.,02, 2002.

Antigenic Analysis. Antigenic analysis with World Health Organi-zation H5N1 reference antisera with representative viruses fromdifferent sublineages (see below) demonstrated a diversity ofreaction patterns that generally corresponded to their phylogeneticrelationships (Fig. 2; see also Table 4, which is published assupporting information on the PNAS web site). All tested FJ-likeH5N1 viruses showed high HA inhibition (HI) titers to antiserumof Anhui�1�06, a virus from the same sublineage. Those viruses hadmoderate to low reactivity to antisera of IDN�357�06, Iraq�1�06,Dk�HN�101�04, and BH goose�QH�1A�05 but no reactivity withantisera of IDN�5�05, VNM�1203�04, Turkey�65596�06, andWhooping swan�Mongolia�244�05 (Table 4).

Numerical analysis of HI titers conducted to visualize simi-larity between the antigenic reactivity of different virusesshowed that those FJ-like H5N1 viruses had a distinguishableantigenic reaction pattern. However, three viruses from thissublineage (Ck�GX�463�06, Dk�HN�856�06, and Gs�ST�18442�05) differed slightly (Fig. 2). This analysis also revealedtwo other major reactivity groups, one of Indonesia isolates andanother that contained viruses from three different sublineages(GD�06, QH-like, and Mixed�VNM2) (Fig. 2). Four viruses(VNM�1203�06, Ck�GY�3570�05, Gs�YN�5539�05, and Gs�GY�337�06) representing four other distinct H5N1 sublineageshad different reactivity patterns and did not group with otherviruses (Fig. 2 and Table 4).

Phylogenetic Analysis. To better understand the increased preva-lence of H5N1 in poultry and the emergence of human infection inChina, 390 (30% of total new isolates) of those avian H5N1influenza viruses isolated from July 2005 to June 2006 plus 16viruses isolated from smuggled poultry and dead wild birds in HongKong in early 2006 were sequenced and analyzed together withsequences available from public databases. Phylogenetic analysis ofthe HA gene revealed that 266 of 390 (68%) of those recent H5N1viruses from southern China formed a previously uncharacterizedand distinct H5N1 sublineage (FJ-like) (Fig. 3 and 4A). Twenty-eight viruses isolated in Guiyang from November 2005 to January2006 formed a sublineage (GY2) that is the sister group to theFJ-like sublineage (Figs. 3 and 4A). Another three sublineages fromGuangdong (GD�06, n � 6), Guiyang (GY1, n � 14), and Yunnan(YN2, n � 13) also were identified. A further 59 viruses groupedin the Mixed�VNM2 sublineage, and only a single virus (Gf�ST�1341�06) belonged to the QH-like sublineage currently circulatingin Africa and Europe (Fig. 3 and 4A). The remaining virusesanalyzed in this study belonged to previously reported sublineagesfrom China and Southeast Asia, except two isolates from Hunan(Ck�HN�2246�06 and Ck�HN�2292�06), isolated in May 2006 thatdo not fall with any of these sublineages (Fig. 4A).

The prototype virus of the FJ-like sublineage (Dk�FJ�1734�05)was detected in March 2005. From July to September 2005, only asingle strain of 33 sequenced viruses was FJ-like (Table 1). Re-markably, from October 2005 onwards the percentage of FJ-likeviruses detected increased dramatically, until from April to June2006, 103 of the 108 H5N1 poultry isolates tested (95%) wereFJ-like (Table 1). Viruses from other sublineages (YN2, GY2, GY1,GD�06, and Mixed�VNM2) were not detected in our surveillancesince October 2005, November 2005, March 2006, April 2006, andMay 2006, respectively. These findings reflect the process of FJ-likeviruses gradually becoming predominant in this region.

Phylogenetic analysis also revealed that the HA gene of fiverecent human H5N1 viruses from different provinces of China (4,10, 11) belong to this FJ-like sublineage and were most closelyrelated to poultry isolates (Fig. 4A). This finding suggests thatH5N1 human infection from China since November 2005 was

Fig. 1. Comparison of H5N1 influenza virus isolation rate (%) in chicken (A),duck (B) and goose (C) from southern China, July 2004 to June 2006. Surveil-lance was conducted in live-poultry markets in Fujian, Guangdong, Guangxi,Guiyang, Hunan, and Yunnan.

Smith et al. PNAS � November 7, 2006 � vol. 103 � no. 45 � 16937

MIC

ROBI

OLO

GY

Dow

nloa

ded

by g

uest

on

May

20,

202

0

directly from affected poultry. Furthermore, H5N1 viruses isolatedin early 2006 from neighboring regions of mainland China, includ-ing the 16 Hong Kong viruses and two poultry isolates from Laosand Malaysia, all joined the FJ-like sublineage (Fig. 4A).

Of those genotyped viruses, 99 of 137 (72%) were from thispredominant FJ-like sublineage, all of which belonged to H5N1genotype Z, except for a single genotype G virus (Gs�GY�1794�06)(Fig. 4B) (1, 2). Thirty-two viruses from other sublineages belongedto genotypes G and Z, whereas six GD�06 viruses belonged togenotype X, which has only previously detected in Hong Kong andGuangdong (Fig. 4B) (1). Of the 16 H5N1 virus isolates from Hong

Kong in 2006, there were 7 genotype Z viruses, whereas 9 genotypeG viruses were isolated from dead wild birds, suggesting a differentecology. Genotype G has seven gene segments in common withgenotype Z but has a Gs�GD-like PB2 gene (Fig. 4B) (1, 2). Thesefindings show that FJ-like H5N1 influenza viruses have becomepredominant and panzootic in southern China since mid-October2005, have gradually replaced other H5N1 sublineages, and havebeen transmitted to Laos and Malaysia.

Molecular Characterization. All viruses characterized in this studymaintained the motif of multiple basic amino acids at the HAcleavage site characteristic of highly pathogenic avian influenza.Furthermore, these viruses also kill embryonated eggs within24 h of inoculation, which also is a characteristic of highlypathogenic avian influenza. However, those viruses from theFJ-like sublineage have a Gln3Leu substitution at position �9from the cleavage site (LRERRR-KR�G). The receptor-bindingpocket of HA1 retains amino acid residues Gln-222 and Gly-224(H5 numbering used throughout) that preferentially bind to�2,3-NeuAcGal linkages of avian cell-surface receptors (12, 13).Other amino acid residues relevant to receptor binding wereidentical to those of Gs�GD-like viruses (14), with the exceptionof Dk�HN�5128�05, which had Val-131-Met substitution.

Only six viruses from the FJ-like lineage (Ck�ST�3840�06,Ck�ST�3923�06, Ck�FJ�11933�05, Ck�FJ�12239�05, Ck�FJ�584�06, and Dk�GX�1550�06) had the Ser-31-Asn mutation in

Table 1. Number of viruses from the Fujian-like sublineage inChina

Collection date No. of FJ-like viruses

2005July–September 1�33 (3)October–December 72�136 (53)

2006January–March 90�113 (80)April–June 103�108 (95)

Total 266�390 (68)

Shown are the number of FJ-like sequences�total number (percentage) ofH5N1-positive samples sequenced.

Fig. 2. Numerical analysis of HI titers (see Table 4) by using hierarchical agglomerative clustering (A) and nonmetric multidimensional ordination in twodimensions (B). Colors indicate viruses from the FJ-like (red), GY1 (blue), GY2 (green), YN2 (purple), and GD�06 (orange) sublineages. MYS, Malaysia; Ph, pheasant;VTM, Vietnam�Thailand�Malaysia.

16938 � www.pnas.org�cgi�doi�10.1073�pnas.0608157103 Smith et al.

Dow

nloa

ded

by g

uest

on

May

20,

202

0

the matrix protein 2, which confers resistance to amantadine(15). All other viruses have residues indicating sensitivity toamantadine. All viruses characterized from the FJ-like lineagehad His at position 274 of the neuraminidase protein and arepredicted to be sensitive to oseltamivir (16).

Serological Analysis. To estimate the seroconversion rate ofmarket poultry after vaccination, 1,113 chicken sera fromGuangdong and Guiyang Provinces were collected from Novem-ber 2005 to April 2006. HI assay revealed that 180 sera (16%)were positive (HI titer � 20) against Ck�HK�YU22�02 (H5N1).

Of 180 positive sera, 76 were randomly selected for a neu-tralization test against three representative viruses from theFJ-like, GY2, and Mixed�VNM2 sublineages. Most HI-positivesera tested showed neutralizing titers to Ck�GY�3570�05 andDk�YN�4400�05, whereas 55 had low or no neutralization to

Dk�FJ�1734�05 (Table 2; see also Table 5, which is published assupporting information on the PNAS web site). Almost all serahad 2- to 4-fold higher titers to Ck�GY�3570�05 and Dk�YN�4400�05, which are from the GY2 and Mixed�VNM2 sublin-eages, in comparison with Dk�FJ�1734�05 (Tables 4 and 5).Only four sera were neutralizing-negative against all threerepresentative strains, although they were positive in the HIassay (Table 5). These findings suggest that chicken in southernChina are poorly immunized against FJ-like viruses in compar-ison with other sublineages.

DiscussionThe highly pathogenic H5N1 influenza virus currently panzooticin Eurasian and African poultry populations is considered themost likely candidate for a new pandemic influenza. The devel-opment of more and more avian-to-human interspecies trans-mission events in the last 12 months seems to favor such ahypothesis (4, 5). Our results demonstrate that the emergenceand predominance of a H5N1 influenza virus sublineage inChina has initiated a new transmission wave in Southeast Asia.

The emergence of this FJ-like sublineage has had similar con-sequences to the first wave of virus transmission throughout South-east Asia in early 2004 (1) and the second wave to Europe andAfrica that followed the Qinghai Lake H5N1 outbreak (3, 4). Thefindings of our study show that this virus has replaced most of thosepreviously established regional sublineages across a large geograph-ical area in China (2). The predominance of this FJ-like virusappears to be responsible for the increased prevalence of H5N1 inpoultry since October 2005 and recent human infection cases inChina (4, 5). Furthermore, it has already caused poultry outbreaksin Laos, Malaysia, and Thailand and human disease in Thailand (4,5). As such, it is likely that this variant has already initiated a thirdwave of transmission throughout Southeast Asia and may spreadfurther in Eurasia. It is also probable that this virus will continue toevolve to form other regionally distinct sublineages, as witnessedwith the H5N1 genotype viruses in the first and second transmissionwaves (1, 2, 17).

The mechanism for the emergence and prevalence of FJ-likeH5N1 variant is still unknown. The compulsory vaccination of allpoultry was ordered in China beginning September 2005 (7), butour data indicate that seroconversion rates are still low and thatpoultry are poorly immunized against FJ-like viruses, which sug-gests that the poultry vaccine currently used in China may onlygenerate very low neutralizing antibodies to FJ-like viruses incomparison to other previously cocirculating H5N1 sublineages.This situation could have helped to select for the FJ-like sublineagein poultry, because our results also show that these viruses hadreplaced the GY2 and YN2 virus sublineages, both of which hadhigh titers in the serological tests. As such, this information suggeststhat the predominance of FJ-like viruses may be associated withimmune escape from the current vaccine strain in poultry.

Previously, we described the establishment of multiple sublin-eages of H5N1 virus in southern China and Southeast Asia (2). Theemergence and replacement of these sublineages by FJ-like viruses

Table 2. Summary of serological analysis of chicken serumcollected in southern China from April to November 2006against H5N1 influenza viruses from different sublineages

Virus Sublineage

NI titer range

� 20–40 80–160 �320

Dk�FJ�1734�05 FJ-like 30 25 18 3Ck�GY�3570�05 GY2 9 16 35 16Dk�YN�4400�05 YN2 4 14 41 17

Virus sublineages were defined by phylogenetic analysis of the HA gene(Fig. 3). �, lowest dilution tested (1:20); NI, neutralizing inhibition.

Fig. 3. Phylogenetic relationships of the HA genes of representative influ-enza A viruses isolated in Asia. Trees were generated by the neighbor-joiningmethod in the PAUP* program (27). Numbers above or below branchesindicate neighbor-joining bootstrap values. Not all supports are shown be-cause of space constraints. Analysis was based on nucleotides 1–1011, and thetree was rooted to A�tern�South Africa�61. Colors indicate viruses from theFJ-like (red), GY1 (blue), GY2 (green), YN2 (purple), and GD�06 (orange)sublineages. VTM, Vietnam�Thailand�Malaysia. (Scale bar, 0.01 substitutionsper site.)

Smith et al. PNAS � November 7, 2006 � vol. 103 � no. 45 � 16939

MIC

ROBI

OLO

GY

Dow

nloa

ded

by g

uest

on

May

20,

202

0

within a short period highlights the difficulties faced in controllingH5N1 virus in China. A complex ecology and highly diverse viruspopulations make it almost impossible to capture each circulatingvirus sublineage, even with the application of mass vaccination. Thiscomplexity has resulted in recurrent H5N1 outbreaks in poultry indifferent regions and has led to occasional human infection.

Since November 2005, 22 H5N1 human infection cases from 14provinces of China have been reported (4, 5). It is noteworthy thatfour of the provinces (Fujian, Guangdong, Shanghai, and Zhejiang)with human cases have not recorded any outbreaks in poultry (4).

However, phylogenetic and antigenic analyses in this study clearlyshow that those recent human H5N1 isolates from different prov-inces are FJ-like viruses, which suggests that this virus may beprevalent in an area much larger than we have identified. Given thelack of systematic influenza surveillance in poultry at a nationallevel, the timely identification of the source of human infection isalmost impossible. Therefore, to understand and identify possibleinfection sources and to avert a potential pandemic, comprehensiveinfluenza surveillance in both human and animal populations isurgently required in H5N1-affected regions.

Fig. 4. Phylogenetic relationships of the HA (A) and PB2 (B) genes of representative influenza A viruses isolated in Asia. Trees were generated by theneighbor-joining method in the PAUP* program (27) (Bayesian analysis revealed the same relationships.) Numbers above and below branches indicateneighbor-joining bootstrap values and Bayesian posterior probabilities, respectively. Not all supports are shown because of space constraints. Analysis was basedon nucleotides 1–1011 of the HA gene and 985-2233 of the PB2 gene. The HA and PB2 trees were rooted to A�tern�South Africa�61 and A�equine�Prague�1�56,respectively. Colors indicate viruses from the FJ-like (red), GY1 (blue), GY2 (green), YN2 (purple), and GD�06 (orange) sublineages. Recent human isolates fromChina are underlined. *, Viruses included in antigenic analysis (Fig. 2 and Table 4). JX, Jiangxi; MDK, migratory duck; MYS, Malaysia; Ph, pheasant; Qa, quail; SCK,silky chicken; VTM, Vietnam�Thailand�Malaysia. (Scale bar, 0.01 substitutions per site.)

16940 � www.pnas.org�cgi�doi�10.1073�pnas.0608157103 Smith et al.

Dow

nloa

ded

by g

uest

on

May

20,

202

0

The repeated emergence of H5N1 variants from southernChina and their subsequent spread to other parts of the world(1–3) makes it increasingly apparent that implementation ofeffective control measures in this region is of paramount im-portance. Such a system of control measures could be achievedby integrated real-time virological and genetic information withrapid diagnostic approaches and vaccine production accompa-nied by strict quality control. Our surveillance network insouthern China is the longest running, and the data this networkhas generated have provided the most comprehensive insightinto the ecology and evolution of H5N1 virus in its natural host(1, 2, 18–20). Despite these efforts, there remains a lack ofinformation in the broader region, and it is critical that similarsurveillance programs begin in other areas, including Indonesia,Vietnam, Thailand, and India. Perhaps most importantly, infor-mation from northern China is required, because it could answerkey questions regarding the movement of H5N1 in and out ofsouthern China, the hypothetical influenza epicenter (21).

MethodsVirological Surveillance, Isolation, and Characterization. Cloacal,tracheal, and fecal samples were collected once every 7–10 daysfrom apparently healthy poultry in live-poultry markets inFujian, Guangdong, Guangxi, Guiyang, Hunan, and Yunnan(Fig. 5). Specimens were first screened by RT-PCR for H5subtype influenza virus. All PCR-positive swabs were shipped tothe State Key Laboratory of Emerging Infectious Diseases atThe University of Hong Kong and grown in embryonated eggs.All isolates were identified and subtyped by using a panel ofreference antisera as previously described (22).

Antigenic Analysis. The antigenic characteristics of the H5N1influenza viruses from different sublineages were compared byHI assay with ferret antisera to the World Health Organizationreference H5 subtype viruses as previously described (20). TheDepartment of Infectious Diseases at St. Jude Children’s Re-search Hospital produced the ferret antisera against Dk�HN�101�04, IDN�5�05, BH goose�QH�1A�05, and VNM�1203�04.Antisera to Anhui�1�06, IDN�357�06, Iraq�1�06, Whoopingswan�Mongolia�244�05, and Turkey�65596�06 were kindly pro-vided by Nancy Cox (Centers for Disease Control and Preven-tion, Atlanta, GA). The HI assay started at 1:20 dilution.

To visualize similarity between the antigenic reaction patterns

of different viruses, numerical analysis of HI titers was con-ducted by using PRIMER version 5.2.9 (PRIMER-E, Plymouth,United Kingdom). The data were standardized and square-root-transformed, and the Bray–Curtis coefficient (23) was used toconstruct a similarity matrix. Hierarchical agglomerative clus-tering with group-average linking (24) was conducted, and adendrogram was produced. Nonmetric multidimensional scaling(25) also was used to produce two- and three-dimensionalordinations over 100 iterations. The two-dimensional configu-ration with lowest overall stress was presented.

Phylogenetic Analysis and Molecular Characterization. We se-quenced the HA gene of 390 of the 1,294 (30%) H5N1 influenzaviruses isolated from poultry market surveillance in southernChina from July 2005 to June 2006 along with 16 viruses isolatedfrom smuggled poultry and dead wild birds in Hong Kong inJanuary and February 2006. In addition, 137 of those 1,294(11%) viruses, plus the 16 Hong Kong isolates, were partiallysequenced for each of the 8 gene segments and genotyped aspreviously described (1, 2, 18). Sequence assembly, editing,alignment, and residue analysis were performed as previouslydescribed (2). Phylogenetic analysis using MrModelTest 2.2 (26),PAUP* 4.0 (27), and MrBayes 3.1 (28) also was carried out aspreviously described (2).

Serological Analysis. Chicken sera (n � 1,113) were collected fromlive-poultry markets in different provinces from November 2005to April 2006. The antibodies for H5 subtype influenza viruswere detected by HI assay, with Ck�HK�YU22�02 as theantigen. A serum was considered positive for H5N1 virus if itsHI titer was �20. Some of those HI-positive sera were randomlyselected and further tested by neutralization assay for antibodiesagainst the representative isolates: Ck�FJ�1734�05, Ck�GY�3570�05, and Dk�YN�4400�05, which belong to the FJ-like,GY2, and Mixed�VNM2 sublineages, respectively. Sera werescreened at a 1:10 series dilution against 100 TCID50 (50% tissueculture infective dose) of those three representative viruses toexclude negative samples, as previously described (2). Titers of�20 were regarded as positive.

We thank L. J. Zhang, J. Wong, L. Duan, and W. S. Hong for excellenttechnical support. Sequence data from human cases in Indonesia werekindly provided by the Indonesian Department of Health. This work wassupported by the Li Ka Shing Foundation and by National Institute ofAllergy and Infectious Disease Contract AI95357.

1. Li KS, Guan Y, Wang J, Smith GJD, Xu KM, Duan L, Rahardjo AP,Puthavathana P, Buranathai C, Nguyen TD, et al. (2004) Nature 430:209–213.

2. Chen H, Smith GJD, Li KS, Wang J, Fan XH, Rayner JM, Vijaykrishna D,Zhang JX, Zhang LJ, Guo CT, et al. (2006) Proc Natl Acad Sci USA103:2845–2850.

3. Chen H, Smith GJD, Zhang SY, Qin K, Wang J, Li KS, Webster RG, PeirisJSM, Guan Y (2005) Nature 436:191–192.

4. Centre for Health Protection (2006) Global Statistics of Avian Influenza (As of30 August 2006) (Centre for Health Protection, Department of Health, HongKong SAR, China), available at www.info.gov.hk�info�f lu�eng�global.htm.

5. World Health Organization (2006) Cumulative Number of Confirmed HumanCases of Avian Influenza A (H5N1) (WHO, Geneva), available at www.who.int�csr�disease�avian�influenza�country�en.

6. World Health Organization (2006) Avian Influenza: Situation in Indonesia(WHO, Geneva), available at www.who.int�csr�don�archive�country�idn�en.

7. Cyranowski D (2005) Nature 438:406.8. Guan Y, Zheng BJ, He YQ, Liu XL, Zhuang ZX, Cheung CL, Luo SW, Li PH,

Zhang LJ, et al. (2003) Science 302:276–278.9. Guan Y, Zhong NS, Chen H, Smith GJD, Zheng BJ (2006) in Challenges of

Severe Acute Respiratory Syndrome, eds Chan JCK, Taam Wong VCW (Elsevier,Singapore), pp 93–100.

10. Yu H, Shu Y, Hu S, Zhang H, Gao Z, Chen H, Dong J, Xu C, Zhang, YxiangN, et al. (2006) Lancet 367:84.

11. Shu Y, Yu H, Li D (2006) N Engl J Med 354:1421–1422.12. Ha Y, Stevens DJ, Skehel JJ, Wiley DC (2001) Proc Natl Acad Sci USA

98:11181–11186.

13. Stevens J, Blixt O, Tumpey TM, Taubenberger JK, Paulson JC, Wilson IA(2006) Science 312:404–410.

14. Claas EC, Osterhaus AD, van Beek R, De Jong JC, Rimmelzwaan GF, SenneDA, Krauss S, Shortridge KF, Webster RG (1998) Lancet 351:472–477.

15. Scholtissek C, Quack G, Klenk HD, Webster RG (1998) Antiviral Res 37:83–95.16. Treanor JJ, Hayden FG, Vrooman PS, Barbarash R, Bettis R, Riff D, Singh S,

Kinnersley N, Ward P, Mills RG (2000) J Am Med Assoc 283:1016–1024.17. The World Health Organization Global Influenza Program Surveillance

Network (2005) Emerg Inf Dis 11:1515–1521.18. Guan Y, Peiris JS, Lipatov AS, Ellis TM, Dyrting KC, Krauss S, Zhang LJ,

Webster RG, Shortridge KF (2002) Proc Natl Acad Sci USA 99:8950–8955.19. Guan Y, Peiris M, Kong KF, Dyrting KC, Ellis TM, Sit T, Zhang LJ, Shortridge

KF (2002) Virology 292:16–23.20. Guan Y, Poon LLM, Cheung CY, Ellis TM, Lim W, Lipatov AS, Chan KH,

Sturm-Ramirez KM, Cheung CL, Leung YHC, et al. (2004) Proc Natl Acad SciUSA 101:8156–8161.

21. Shortridge KF, Stuart-Harris CH (1982) Lancet 2:812–813.22. Guan Y, Shortridge KF, Krauss S, Chin PS, Dyrting KC, Ellis TM, Webster RG,

Peiris M (2000) J Virol 74:9372–9380.23. Bray RJ, Curtis JT (1957) Ecol Monogr 27:325–349.24. Sokal RR, Michener CD (1958) Univ Kans Sci Bull 38:1409–1438.25. Kruskal JB (1964) Psychometrika 29:115–129.26. Nylander JAA (2004) MrModeltest (Evolutionary Biology Centre, Uppsala

University, Uppsala, Sweden), Version 2.27. Swofford DL (2001) PAUP*: Phylogenetic Analysis Using Parsimony (and

Other Methods) (Sinauer, Sunderland, MA), Version 4.0 beta.28. Ronquist F, Huelsenbeck JP (2003) Bioinformatics 19:1572–1574.

Smith et al. PNAS � November 7, 2006 � vol. 103 � no. 45 � 16941

MIC

ROBI

OLO

GY

Dow

nloa

ded

by g

uest

on

May

20,

202

0