Origin of the West Nile Virus Responsible for an Outbreak of Encephalitis in the Northeastern United States

See allHide authors and affiliations

Science  17 Dec 1999:
Vol. 286, Issue 5448, pp. 2333-2337
DOI: 10.1126/science.286.5448.2333


In late summer 1999, an outbreak of human encephalitis occurred in the northeastern United States that was concurrent with extensive mortality in crows (Corvusspecies) as well as the deaths of several exotic birds at a zoological park in the same area. Complete genome sequencing of a flavivirus isolated from the brain of a dead Chilean flamingo (Phoenicopterus chilensis), together with partial sequence analysis of envelope glycoprotein (E-glycoprotein) genes amplified from several other species including mosquitoes and two fatal human cases, revealed that West Nile (WN) virus circulated in natural transmission cycles and was responsible for the human disease. Antigenic mapping with E-glycoprotein–specific monoclonal antibodies and E-glycoprotein phylogenetic analysis confirmed these viruses as WN. This North American WN virus was most closely related to a WN virus isolated from a dead goose in Israel in 1998.

In late August and early September 1999, New York City and surrounding areas experienced an outbreak of human encephalitis consistent with an arboviral etiology. Serological evidence from this outbreak implicated a flavivirus as the etiologic agent. Concurrent with this human encephalitis outbreak, a viral encephalitis of unknown etiology was discovered in American crows (Corvus brachyrhynchos) and fish crows (Corvus ossifragus) dying in the same geographic area. Deaths were also observed among several exotic avian species, including a Chilean flamingo (Phoenicopterus chilensis) at the Bronx Zoo. Necropsy samples from these birds were submitted to the National Veterinary Services Laboratories, U.S. Department of Agriculture, and were inoculated into embryonated chicken eggs for virus isolation. Flavivirus-like particles (diameter 40 nm) were observed by electron microscopy in the allantoic fluid 4 days after inoculation. The isolates were forwarded to the Centers for Disease Control and Prevention (CDC) for identification.

The complete nucleotide sequence of one of these viral isolates (WN-NY99, from the dead Chilean flamingo) has now been determined. The viral genomic RNA was amplified and copied into overlapping DNA fragments of ∼2 to 3 kb by means of the reverse transcription polymerase chain reaction (RT-PCR) (1). Both strands of the purified DNAs were sequenced with the use of primers spaced about 400 bases apart along the entire genome. The complete 11,029-nucleotide genomic sequence of WN-NY99 has been submitted to GenBank (accession number AF196835). The deduced amino acid sequence of the coding region of WN-NY99 (genomic positions 97 to 10,395) is shown in Fig. 1. The WN-NY99 virus genome exhibited standard flavivirus genomic organization, the same overall genomic organization as was described for the WN-Nigeria and Kunjin (KUN) viruses (2). A short 5′ noncoding region of 96 nucleotides is followed by an ATG initiation codon at position 97 and a single open reading frame of 10,302 nucleotides coding for three structural proteins—capsid, premembrane (prM), and envelope (E)—and five nonstructural proteins (NS1, NS2a/NS2b, NS3, NS4a/NS4b, and NS5). The coding region of WN-NY99 is followed by a 3′ noncoding region of 631 nucleotides.

Figure 1

Deduced amino acid sequence of the polyprotein of West Nile virus WN-NY99. The start of each protein is marked by an arrow. Abbreviations for protein names: cap, nucleocapsid; prM, premembrane protein; M, viral membrane protein; E, viral envelope glycoprotein; NS1 to NS5, viral nonstructural proteins. The E-glycoprotein glycosylation motif (NYS) is underlined. Single-letter abbreviations for amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

To identify the New York virus antigenically, we performed indirect immunofluorescence antibody tests using a panel of well-defined monoclonal antibodies (mAbs) to map various isolates from birds and mosquitoes. The mAb end-point titers with the North American isolates were compared to titers derived with other representatives of the Japanese encephalitis (JE) virus serocomplex of flaviviruses (Table 1). These mAbs, which are specific for the E-glycoprotein, can distinguish WN virus from KUN virus and can also distinguish either of these viruses from other members of the JE virus serocomplex. Viruses were grown in Vero cells, spotted onto 12-well slides, air-dried, and fixed with acetone before staining. All viruses reacted similarly with the broad flavivirus-reactive, positive-control mAb 4G2 (3, 4). None of these viruses reacted with the negative-control antibody, which is specific for the E1 glycoprotein of eastern equine encephalitis (EEE), an unrelated alphavirus (5). All WN isolates, including those from North America, reacted specifically with the WN virus–specific mAb H5.46, but not with the KUN virus–specific mAb 10A1 (four- to eightfold titer differences) (6–8). Similarly, only KUN virus reacted with mAb 10A1, but not with mAb H5.46. No KUN or WN viruses reacted with either the St. Louis encephalitis (SLE) virus–specific mAb 6B5A-2 or the Murray Valley encephalitis (MVE) virus–specific mAb 4B6C-2 (9, 10). These results type the North American isolates as WN virus and not as KUN, SLE, or MVE viruses.

Table 1

Antigenic characterization of North American WN viruses. Ig, immunoglobulin; nd, not done.

View this table:

WN virus belongs to the family Flaviviridae, genusFlavivirus, and is a member of the JE virus serocomplex, which also includes JE, SLE, MVE, and KUN viruses, among others (11). Flaviviruses are plus-sense, single-stranded RNA viruses with a genome of ∼11,000 nucleotides (2). Recently published sequence and phylogenetic data suggest that, within this serocomplex, KUN viruses appear to be a subtype of WN virus rather than a separate viral species (12). Although flaviviruses are closely related to each other antigenically and cross-react in serological tests with polyclonal antisera, most have a rather distinctive geographic distribution. Those of the JE serocomplex are maintained in a natural transmission cycle involving mosquito vectors and bird reservoir hosts. Humans and horses are usually incidental hosts.

To determine more precisely the relationships between the WN-NY99 virus and other related virus strains, we performed a phylogenetic analysis on an informative region of the E-glycoprotein gene (genome positions 1402 to 1656) (12, 13). Aligned nucleic acid sequence data from 33 WN viruses, seven KUN viruses, and one JE virus were analyzed with the use of algorithms for parsimony (PAUP), distance (MEGA; Fig. 2 ), and maximum likelihood (fastDNAml) (14–17). The phylogenetic trees generated by these analyses had the same overall topology as that previously observed, insofar as all WN and KUN viruses are separated into two major lineages (12, 13). Viruses in lineage 1 are primarily of West African, Middle Eastern, Eastern European, and Australian origin. Lineage 2 consists exclusively of viruses from the African continent that have apparently not been involved in human or equine outbreaks, but rather are maintained in enzootic cycles.

Figure 2

Phylogenetic tree based on E-glycoprotein nucleic acid sequence data (255 base pairs). The tree was constructed with the program MEGA by neighbor-joining with Kimura two-parameter distance (scale bar). Bootstrap confidence level (500 replicates) and a confidence probability value based on the standard error test (22, 23) were calculated using MEGA and are included on the tree (top and bottom values, respectively), illustrating support for the division between the lineage 1 WN virus group (not including the India isolates) and the KUN virus group. The best estimated length of the segment (bold line) separating these groups, in units of expected nucleotide substitutions per site, is 0.06928 and is statistically significantly positive (P< 0.01) by the likelihood ratio test (fastDNAml maximum likelihood program). An approximate 95% confidence interval for the true length of this segment is 0.03347, 0.10737. The isolate history of strains used in this tree and the alignment used for analysis are available upon request from the authors. GenBank accession numbers for the sequences included in the tree are as follows: WN-Romania 1996 H,AF130363; WN-Romania 1996, AF205879; WN–South Africa, AF205880; WN-Israel 1952, AF205881; WN-Egypt 1951, AF001568; WN-France 1965,AF001560; WN-Senegal 1979, AF001569; WN-Algeria 1968, AF001567; WN–New York 1999, AF196835; WN-Israel 1998, AF205882; WN-C.Afr.Rep. 1989,AF001558; WN-Italy 1998, AF205883; WN-Morocco 1996, AF205884; WN-Romania 1996 M, AF130362; WN-Kenya 1998, AF146082; WN-Senegal 1993,AF001570; WN-C.Afr.Rep. 1967, AF001566; WN–Ivory Coast 1981, AF001561; Kunjin 1994, AF196495; Kunjin 1966, AF196509; Kunjin 1973, AF196515; Kunjin 1960, D00246; Kunjin 1984b, AF196498; Kunjin 1991, AF196491; Kunjin 1984a, AF196519; WN-India 1955a, AF205885; WN-India 1955b,AF196525; WN-India 1980, AF196526; WN-India 1958, AF196524; WN-Madagascar 1978, AF001559; WN-Madagascar 1988, AF001574; WN-Kenya,AF001571; WN-Madagascar 1986, AF001564; WN-Uganda 1959, AF001562; WN-C.Afr.Rep. 1972a, AF001563; WN-C.Afr.Rep. 1983, AF001557; WN-C.Afr.Rep. 1972b, AF001565; WN-Nigeria, M12294; WN-Uganda, AF001573; WN-Senegal 1990, AF001556; JE SA 14, U04522.

Within lineage 1, the KUN viruses and the Indian WN viruses both appear as monophyletic sister clades to the European and African WN viruses. The WN-NY99 virus is found within lineage 1 and is most closely related to WN viruses that have recently been isolated from North Africa, Romania, Kenya, Italy, and the Middle East. Of particular note is the close relationship between the WN-NY99 virus and a WN virus isolated from the brain of a dead goose in Israel in 1998. Phylogenetic analysis of a portion of the gene encoding the NS5 protein and of the 3′ noncoding region (830 bases) of 12 WN and KUN viruses also generates trees with nearly identical topology, with WN-NY99 demonstrating the closest relationship with lineage 1 WN viruses (18). Flavivirus sequences amplified from brain specimens from fatal human cases occurring in the New York City outbreak have been analyzed by comparing genomic sequences from the nonstructural protein genes to single strains of KUN and WN viruses (19). Brieseet al. concluded that the agent responsible for the New York City area outbreak was most closely related to KUN virus, and accordingly called the outbreak virus Kunjin/West Nile– like. The phylogenetic tree in Fig. 2 compares 40 WN and KUN viruses and is more representative of the phylogeny among WN viruses.

To investigate these relationships further, we derived additional sequence data from the structural gene region of the genome (positions 549 to 1826 in the genes encoding the prM and E proteins) from selected isolates obtained during the 1999 epidemic and compared them to other WN strains within lineage 1 for which sequence data from this region were available. Table 2 displays the percent identity among these viruses. The high degree of sequence similarity (>99.8%) among the various strains circulating throughout New York City and surrounding counties and states indicates that a single WN strain was introduced and circulated during the U.S. WN virus outbreak. The identical genomic sequences identified from human brain specimens also confirm the association of this WN-NY99 virus with human disease. The small number of nucleotide substitutions observed among the strains analyzed is indicative of viral microevolution occurring during the outbreak.

Table 2

Percent identity among West Nile virus strains over a 1278-nucleotide base region of prM and E proteins.

View this table:

A high degree of similarity between all of the U.S. WN viruses and the WN virus isolated in Israel in 1998 (>99.8%) was observed. Within these 1278 nucleotides (genome positions 549 to 1826), only two nucleotide differences occurred between WN-NY99 and WN-Israel 1998. Although this high degree of homology was unexpected, it could not have resulted from cross-contamination of U.S. viruses with the Israeli virus; the sequencing of the WN-Israel 1998 virus was performed independently at the Pasteur Institute, whereas the isolation and sequencing of New York isolates was carried out independently at CDC. For comparison, analysis of this same region of WN-NY99 with another virus within the same lineage (Romania 1996, mosquito isolate) revealed 37 nucleotide differences (96.9% identity). The cumulative data support the hypothesis that the epidemic and epizootic observed in the late summer of 1999 in the northeastern United States (primarily New York, New Jersey, and Connecticut) are attributable to a WN virus that has been circulating in the Mediterranean region since 1998. It is noteworthy that the WN-Israel 1998 virus was associated with increased pathogenicity for birds, a property also observed in the U.S. outbreak and previously observed only experimentally (20). The absence of reported human cases during this Israeli epizootic may be due to background human immunity to the WN virus in Israel.

The northeastern U.S. outbreak is the first documented incidence of the WN virus in the Western Hemisphere. This virus has a widespread distribution in Africa, West Asia, and the Middle East, occasionally causing epidemics in Europe that are thought to be initiated by viruses introduced by migrant birds (21). The current epidemic of WN virus in New York City is unprecedented and underscores the ease with which pathogens can move among the population centers of the world. It is not yet known how the virus was introduced, nor how long it has been in the United States. The extent of its geographic distribution remains a mystery, as does the long-term impact it may have on human and animal health. The WN virus could have entered the Western Hemisphere through a number of mechanisms, including travel by infected humans, importation of illegal birds or other domestic pets, or unintentional introduction of virus-infected ticks or mosquitoes. Additional surveillance as well as field and laboratory studies are in progress to help address these questions. Because it cannot be predicted whether the WN virus will reappear in the year 2000 transmission season, all components of the public health system must be prepared with rapid surveillance and clinical detection systems in place.

  • * To whom correspondence should be addressed. E-mail: rsl2{at}


View Abstract

Stay Connected to Science

Navigate This Article