The relationships between West Nile and Kunjin viruses.

Until recently, West Nile (WN) and Kunjin (KUN) viruses were classified as distinct types in the Flavivirus genus. However, genetic and antigenic studies on isolates of these two viruses indicate that the relationship between them is more complex. To better define this relationship, we performed sequence analyses on 32 isolates of KUN virus and 28 isolates of WN virus from different geographic areas, including a WN isolate from the recent outbreak in New York. Sequence comparisons showed that the KUN virus isolates from Australia were tightly grouped but that the WN virus isolates exhibited substantial divergence and could be differentiated into four distinct groups. KUN virus isolates from Australia were antigenically homologous and distinct from the WN isolates and a Malaysian KUN virus. Our results suggest that KUN and WN viruses comprise a group of closely related viruses that can be differentiated into subgroups on the basis of genetic and antigenic analyses.

Until recently, West Nile (WN) and Kunjin (KUN) viruses were classified as distinct types in the Flavivirus genus. However, genetic and antigenic studies on isolates of these two viruses indicate that the relationship between them is more complex. To better define this relationship, we performed sequence analyses on 32 isolates of KUN virus and 28 isolates of WN virus from different geographic areas, including a WN isolate from the recent outbreak in New York. Sequence comparisons showed that the KUN virus isolates from Australia were tightly grouped but that the WN virus isolates exhibited substantial divergence and could be differentiated into four distinct groups. KUN virus isolates from Australia were antigenically homologous and distinct from the WN isolates and a Malaysian KUN virus. Our results suggest that KUN and WN viruses comprise a group of closely related viruses that can be differentiated into subgroups on the basis of genetic and antigenic analyses.  (23). Both strands of the PCR product were then sequenced on a 377 automated sequencer (Applied Biosystems International [ABI], Foster City, CA, USA) by using the same primer pair. The two sequences derived from each PCR product were initially aligned by using the program SeqEd (ABI) and a consensus sequence determined. The consensus sequences were then aligned by using the program Clustal W (24), and results were further analyzed by using phylogenetic programs in Bionavigator (http://www.bionavigator.com). Percentage nucleotide similarity was calculated by the Old Distance (GCG) program, and bootstrap confidence levels were calculated with 1,000 replicates by using the Consense program (25). Sequences determined in this study have been deposited in GenBank (National Institutes of Health, Bethesda, MD, USA) ( Table 1). Additional sequences included in this analysis are listed in Table 2.

Enzyme-Linked Immunosorbent Assay (ELISA)
Antigenic profiles of each isolate were compared by using a panel of anti-KUN monoclonal antibodies (MAbs) (26,27) and anti-WN MAbs (28,29) in ELISA as described (26). All MAbs were produced to the E protein except for 3.1112G, which was specific for the NS1 protein.

Genetic Analysis
In accordance with previous reports (16,18,21), the phylogenetic trees generated from both E gene and NS5/ 3'UTR sequences grouped most of the isolates into two major lineages (Figures 1 and 2). Australian KUN isolates and WN isolates from North, West, and central Africa; southern and eastern Europe; India; the Middle East; and New York constituted lineage I. Lineage II comprised WN isolates from West, central, and East Africa and Madagascar. Genetic lineage was not significantly associated with date or source of isolation, with most isolates of both lineages coming from human, mosquito, and avian sources between 1950 and 1990. However, as noted, all viruses isolated during outbreaks of human or avian disease in the last decade belonged to lineage I. Lineage I viruses grouped together with an average sequence identity of 80% (E gene) and 77% (NS5/3'UTR), while the viruses of lineage II contained a single cluster with an average identity of 82% and 83%, respectively. The lineage I viruses were further separated into three clusters: the Australian KUN isolates; the Indian WN viruses; and WN isolates from Africa, the Middle East, Europe, and North America. The divergence observed between lineage I and lineage II viruses was in the range of 16.5% to 30.8% and 19% to 36.5% for sequences of the E gene and NS5/3'UTR, respectively. High bootstrap confidence levels (100%) for the sequences of the NS5/3'UTR also support the separation of the  (Table 3) shows that these two isolates display similar divergence from all other isolates in this study, supporting their grouping outside the two main lineages.
The viruses of lineage I group together in three tight clusters. The first of these includes the Australian KUN viruses, which were 94% identical when sequences of the E gene were compared and 90% when the sequences of the NS5/ 3'UTR were compared. High bootstrap confidence levels (100% for sequences from the E gene and 99% for sequences from the NS5/3'UTR) separated the Australian KUN viruses from the other isolates. However, extremely low bootstrap confidence levels were observed for most of the branches between the Australian KUN viruses in both dendrograms, which also suggests that these viruses are closely related and cannot be definitively separated from each other. The Indian viruses also cluster together, with a sequence identity of 97% and 98% for sequences of the E gene and NS5/3'UTR, respectively. The WN isolates in the remaining cluster of lineage I are 90% and 97% identical, respectively, for the regions sequenced. When compared with the Australian KUN isolates, this cluster, which includes the 1999 New York isolate, shared a sequence identity of 89% for the E gene and   The high degree of nucleotide sequence homology within clusters is consistent with the observed similarity of the amino acid sequences. The most notable variation in amino acid sequence in this study appears around the potential glycosylation site at amino acid 154 of the E protein ( Figure  3). The Australian KUN viruses generally contain either the glycosylation motif NYS at this position or the sequence NYF, which abolishes glycosylation of the E protein. In contrast, the KUN virus SH183 has a 154N→K substitution, which also ablates the potential for glycosylation at this site. In comparison with the KUN prototype, the amino acids 159 (T→I, T→V, or T→Q) and 162 (A→T) of all the WN isolates in this study contain an amino acid substitution. The KUN isolate P1553 also differs from the KUN prototype at amino acid 159 (T→I). Two aberrant isolates, 63134Ent280 and WNFCG, incur a deletion of four amino acids (154 through 157), which also abolishes the glycosylation site.
Our results concur with those of Berthet et al. (16), who suggested the presence of signature motifs within the E gene that support the segregation of WN viruses into two lineages. These signature residues include the amino acid substitutions from lineage I→II as follows: 172A→S, 205T→S, and 210T→S. The amino acid substitution 208T→A holds true in general; however, two of the Indian isolates (lineage I) have K at this position and WNFCG (lineage II) has E. Of particular note is the substitution at amino acid 199. The Australian KUN isolates (199S) share the same amino acid as the lineage II WN viruses, while the lineage I viruses contain an N residue at this position. We have also identified an additional three signature motifs (I→II) at amino acids 128R→W, 129T→I, and 131L→Q. When we attempted to place the Malaysian KUN isolate within either lineage by using these signature motifs, the residues at 128, 129, 131, 172, and 208 were similar to those of lineage I viruses, but the residues at  205 and 210 were consistent with those of lineage II viruses. Residue 199 (D) was unlike any of the other viruses. The KOU isolate displayed more similarities with the lineage II WN viruses (residues 131, 172, 199, and 210) when signature motifs were compared. Residues 129 and 208 differed from viruses of both lineages.
We have identified signature motifs within the NS5 protein that correlate with the separation of the two lineages. Substitutions between lineages I→II include 860A→T, 869Q→H, 878I→V (except for the isolate MgAn798, which has 878I→L), and 899L→V (except for the isolate ArNa1047, which has 899L→I) (Figure 4) Nucleotide sequences in the 3'UTR of the viruses included in this study had a highly variable region in both length and nucleotide sequence immediately downstream of the open reading frame stop codon ( Figure 5). Deletions as well as point mutations were observed in this region, which varied from 38 (MgAn798) to 129 (ArNa1047) nt in length. The Australian KUN viruses displayed only point mutations when compared with the KUN prototype, except for the isolate P1553, which contained a 7-nt insertion, consistent with the WN viruses of lineage I. The long deletion in the nucleotide sequence immediately downstream of the stop codon of the WN prototype virus, WNFCG (53 nt), has been described (31); it is also present in the sequences of another two lineage II WN viruses analyzed in this study, Sarafend (53 nt) and MgAn798 (65 nt). The rest of the 3'UTR for these viruses was found to be highly conserved.

Antigenic Analysis
The MAb 10A1, produced to the KUN isolate OR393 (26), reacted specifically with the Australian KUN isolates in ELISA and did not react with the KUN isolate from Malaysia (MP502-66) nor with KOU virus or any of the lineage I or lineage II WN viruses ( Table 4). The MAb 546 (29), produced to the WN strain Eg101, reacted with all the lineage I and lineage II WN isolates except WN-Sarafend; it did not react with the KOU, KUN, or Malaysian viruses. The MAbs 2B2, produced to the KUN isolate MRM 16 (27), and 2B4, produced to the WN isolate H442 (28), reacted with all the isolates in the study, while the MAbs 3.67G and 3.91D, again produced to the KUN isolate OR393 (26), reacted with all the isolates except WN-Sarafend. The MAb 3.1112G, produced to the NS1 protein of KUN isolate OR393 (26), reacted with all isolates except KOU. The Mab binding patterns (

Conclusion
The results of the phylogenic analysis in this report clearly illustrate that the KUN, WN, and KOU viruses make up a closely related group of viruses, which can be further subdivided into several subgroups on the basis of genetic and antigenic data. Previous phylogenic studies have also shown that KUN and WN viruses share a close relationship (16)(17)(18)21). This report however, further defines this relationship by using a comprehensive panel of both viruses. Also included in this study were several anomalous isolates, including an isolate from Southeast Asia (MP502-66), a laboratory-adapted WN strain of uncertain passage history and origin (Sarafend), and a flavivirus from West Africa (KOU), which has been shown to be closely related to the KUN/WN group of viruses.
The region sequenced in the E gene spans a glycosylation site that, although highly conserved among viruses of the JE antigenic subgroup, is absent from many KUN and WN isolates (16,26; Scherret JH, Khromykh AA, Mackenzie JS, Hall RA, unpub. data). While glycosylation at this site has been associated with neuroinvasiveness of WN isolates in mice (32,33), the biological significance of E protein glycosylation is still unclear. Indeed, sequence analysis of the E gene of WN viruses responsible for fatal outbreaks of encephalitis in Romania (Rom 96) and New York (NY99) showed that only the latter contained a potential glycosylation site, casting doubt on the importance of E protein glycosylation in viral pathogenesis. However, our studies and those of others have shown that limited passage of WN and KUN viruses in some cell types can alter the glycosylation status of the E protein and that analysis of passaged viral isolates should be interpreted with caution The 3'UTR of flaviviruses ranges in length from 400 nt to 600 nt and is thought to play a crucial role in the initiation and regulation of viral translation, replication, and assembly. It includes a potential stable secondary RNA structure at its terminus (2,(34)(35)(36)(37)(38), and upstream it contains several domains that appear to be conserved among mosquito-borne flaviviruses (2,39,40). Men et al. (41) have suggested that deletions in the distal 80 nt to 90 nt would most likely lead to disruption of the stem-loop and loss of viability. In contrast, the region sequenced in this study contains highly variable regions suitable for genetic classification and analysis of the relationships among viruses, which had been subjected to deletions or insertions or both during evolution (17).
Phylogenetic trees constructed from sequence data from both regions identified two major lineages, consistent with previous reports (16,18,21). These two lineages did not separate the KUN isolates from the WN isolates; rather, they emphasized the close link between KUN and WN viruses of lineage I. Nevertheless, within lineage I, the Australian KUN isolates formed a tight cluster with an average nucleotide divergence of 6% for the E gene and 10% for the NS5/3'UTR. In contrast, the WN isolates were spread between the two lineages in three clusters, with a divergence of up to 30.6% for sequences of the E gene and 28.3% for sequences of the NS5/3'UTR. Signature motifs in the deduced amino acid sequences of the E and NS5 proteins also support the separation of the viruses into two lineages.
The virus from Malaysia, KUN MP502-66, and the African virus, KOU, pose a conundrum as to their relationship with the WN and KUN group of viruses. Statistical support for clustering with either of the WN lineages was poor, suggesting that they represent two single-isolate lineages. Although our previous findings suggested that the Malaysian KUN isolate may represent an evolutionary link between the KUN and WN viruses (17), the lack of sequence identity between KUN MP502-66 and the KUN/WN group of viruses in our study suggests that these viruses have evolved separately from a common ancestor.

West Nile Virus
The binding patterns of MAbs to KUN and WN isolates did not differentiate these viruses into the same phylogenetic lineages observed in the dendrograms, although they did support the sequencing results by identifying the Australian KUN viruses, the Malaysian KUN virus, and KOU virus as distinct antigenic groups. The WN-specific MAb used in this study, 546, could not distinguish subgroups within the WN group of viruses; however, Besselaar and Blackburn (28) and Damle et al. (42) have differentiated Indian WN isolates from lineage I South African strains by using MAbs, consistent with the earlier studies of Hammam et al. (43,44). These findings support our sequence data, which show tight clustering of the Indian isolates on a separate branch from other WN isolates in the phylogenetic trees (Figures 1 and 2). Additional MAbs to the E protein of WN viruses may be required to differentiate between lineage I and lineage II viruses.
The unique binding pattern of anti-E MAbs to the Sarafend WN isolate is difficult to explain in light of the E gene sequencing results and amino acid alignments, which show that this virus is similar to other lineage II viruses. However, Sarafend also differs from other WN viruses in the way that it buds from the cell membrane of infected cells (45). Sequencing of the entire prM and E genes of this virus may identify the basis for structural differences in the envelope heterodimer that account for the loss of MAb binding sites and unusual virion maturation.
Phylogenetic analyses enable more precise determination of the relationships among similar viruses and consequently aid in identifying the origin of unknown viruses in subsequent outbreaks. The importance of defining the relationship between the KUN and WN viruses was emphasized during the 1999 outbreak of viral encephalitis in New York City (46,47). Until recently, WN and KUN had been classified as distinct virus types in the Flavivirus genus. However, the latest report by the International Committee on Taxonomy of Viruses (25) recognized that KUN and WN should not be classified as two separate species and designated KUN as a subtype of WN. Our results suggest that this definition requires further consideration. The species should perhaps be further subdivided into at least six subtypes on the basis of the clusters of viruses displayed in the phylogenetic trees. Subtypes would then include lineage II WN group, Indian WN group, Australian KUN group, lineage I WN group, Malaysian group, and KOU group.
Indeed, the assessment of viruses from each subgroup for transmissibility by the major mosquito vectors of each geographic region and relative virulence and amplification in primate, equine, and avian species will provide valuable information on the likelihood and possible consequences of the spread of these viruses to new geographic regions. Additional studies of cross-protection between subgroups by natural infection or immunization with vaccines derived from these viruses and the specificity and sensitivity of serologic and molecular assays for each subgroup in monitoring and diagnostic applications will be useful in defining control strategies.