West Nile virus (WNV) is a member of the Japanese encephalitis antigenic complex (JEV) within the Flavivirus genus and is transmitted by Culex spp. mosquitoes among birds. Other vertebrates such as mammals and reptiles also become infected [1]. Human infection with WNV causes mild to severe illness, sometimes affecting the nervous systems and provoking encephalitis, meningitis and death [2].

Since the first reports of WNV isolation from Africa, Europe, India, Russia, Israel, France, and its 1999 introduction into North America, WNV has extended its geographic distribution throughout the United States [3]. Subsequently, in the following years this arbovirus was detected in Canada, Mexico, Guatemala, Caribbean islands, and South America [46]. Serologic evidence for the natural circulation of WNV in Colombia has been observed in equids sampled from the department of Córdoba [7, 8] and other regions from the Caribbean [911]. Despite WNV being isolated for the first time from captive flamingoes in Santa Fé Zoo (Medellín, Colombia) [12], it is not clear why WNV has not been isolated from or been the cause of detectable disease in horses or humans in Colombia. Possible explanations include: circulation of WNV in remote enzootic cycles away from human settlements, limited vector competence of mosquito species, ornitophilic blood-feeding preferences, cross-protective immunity in humans from other flaviviruses (dengue, Saint Louis encephalitis viruses), or the circulation of WNV-strains with low or attenuated virulence [1012].

WNV occurs in four major lineages, but lineage 1 is epidemiologically relevant. Lineage 1 is subdivided into three clades (1a, 1b, 1c); clade 1a contains isolates from Africa, Europe, the Middle East, Russia and Americas [1, 3]. WNV-American strains have close relationships with three Old World isolates: IS98-ST1 (Israel - 1998), PaH001 (Tunez-1997) and goose-03 (Hungary-2003), and extensive studies have allowed detailed investigations of WNV microevolution in different areas over time and also the emergence of new genotypes [3]. In this sense, phylogenetic analysis has enabled the understanding of epidemiological patterns of emergence, dispersal routes, adaptation to new hosts/mosquitoes species, and spatio-temporal patterns of evolution [1, 3, 5, 12]. Evolutionary studies are necessary to identify “drivers” of emergence, molecular evolution of virulence and associations to ecological factors that allow the establishment of this arbovirus pathogen in human populations.

Between 2011 and 2013, a surveillance study was performed at one locality that is characteristic of a large migratory bird population in northern Colombia. We detected WNV in pools of mosquitoes and amplified two viral regions that were used to establish the phylogenetic relationships with strains isolated in the U.S., Mexico and Argentina, with the goal of defining the evolutionary relationships with genotypes previously isolated from outbreaks in the geographic areas mentioned above.

Material and methods


During virological surveillance for detection of emerging and re-emerging arboviruses between 2011–2013 in San Bernardo del Viento (Córdoba, Colombia) (9° 21´ 30.97” N, 75° 58´ 37.28” W) (Fig. 1), mosquitoes were collected using CDC-light/EVS traps that were baited with dry ice (CO2). All the insects sampled were separated into pools through morphological identification and triturated using minimum essential medium (MEM) supplemented with 10 % fetal bovine serum, 1 % penicillin, then clarified by centrifugation at 13,000 rpm for 30 min. Supernatant was used for RNA extraction and generic nested reverse transcriptase polymerase chain reaction (RT-PCR) for detection of flaviviruses (Table 1) [13]. Two mosquito pools, each containing 40 and 50 specimens (whose mosquitoes had a high percentage −99.57 % of similarity in their DNA Barcode sequences with Culex (Melanoconion) erraticus) were positive for a member of the Flavivirus genus. Sequencing of PCR-products, BLASTN and Neighbor-Joining phylogenetic analysis allowed identification of these sequences as West Nile virus.

Fig. 1
figure 1

Map of the study area showing San Bernardo del Viento (Córdoba, Colombia)

Table 1 Primers and RT-PCR (generic/nested) conditions for arboviral detection in target groups

Molecular protocols

Total RNA was extracted from the PCR-positive homogenates with the RNeasy kit (Qiagen, Valencia) and used for RT-PCR using the One-Step RT-PCR Kit (Qiagen, Valencia, CA) as described previously [14, 15]. Two viral genes were amplified: NS5 (primers: FU1PMF-TACAACATGATGGGVAARAGWGARAA/cFD3PMR-ARCATGTCTTCYGTBGTCATCCA) and Envelope (primers: WN1101-GATGAATATGGAGGCGGTCA/WN1816A-CCGACGTCAACTTGACAGTG and WN1751-TGCATCAAGCTTTGGCTGGA/WN2504A TCTTGCCGGCTGATGTCTAT).

Sequence analysis

A total of 46 genome sequences of envelope/NS5 from WNV isolates, representative of American clade 1a-Cluster 4 geographic locations [16], were downloaded from GenBank (Table 2). Nucleotide sequences were aligned using the MAFFTv7.0 ( and then transferred in FASTA format to BioEdit ( for manual editing, keeping gaps consistent within the reading frame. The sequences of NS5 and envelope of the WNV-strains were aligned and evaluated in jModelTest v2.1.4 [17] using the Akaike criterion informative for identify appropriate substitution model nucleotide. One test was also performed for concatenated sequences. The XML file for Bayesian analysis was created in BEAUti v1.5.4 (, describing model of sequences, invariants, gamma distribution, size of the chain run (20 million of generations), coalescent constant population and for accommodate the variation in substitution rate among branches, a random local clock model was chosen for this analysis [18].

Table 2 Sequences downloaded of Genbank and background information of WNV strain/isolates used in this study

Bayesian phylogenetic analysis was performed using the BEAST software package v2.1.3 [19], and estimation of the maximum clade credibility (MCC) phylogenetic tree was achieved using TreeAnnotator-v2.0.2. BEAST output was viewed with TRACERv-1.5 and evolutionary trees were generated in the FigTree-v1.3.1.

DNAsp-v5.0 [20] was used to establish polymorphic sites between NS5/envelope sequences characterized in our study and reference sequences of representative WNV strains.

Results and discussion

The two pools infected with WNV, corresponding probably to the mosquito species Culex (Melanoconion) erraticus (codes: CDCCA4-12 – CDCCZ2-21), were collected in September, 2012 from coastal mangroves in a migratory bird zone located in San Bernardo del Viento (Córdoba Department). Both isolates were RT-PCR amplified, including one sequence of 1504 nt in length from the envelope gene (Genbank accession number: KM212943 – KM212944) and another 1186 nt long from the NS5 gene (Genbank accession number: KM212941 – KM212942). jModelTest-v2.1.4 estimated the same model of substitution nucleotide, General-Time-Reversible (GTR) + gamma distribution (−lnL = 4792.34, AICc = 9798.9686) for both viral regions. A concatenated file with two NS5/envelope sequences (2690 nt) estimated the same model and was used for phylogenetic inference.

The phylogenetic tree inferred with Bayesian methods indicated that our two WNV-samples were closely related to strains Mosquito-v4369, Bird1519 and v4095, all belonging to southeastern coastal Texas genotype (Fig. 2). Our sequences differed from that of this WNV-genotype in three positions in the envelope gene: only two positions are unique (shaded positions: 547,641) and position 555 shared the same nucleotide with strains HNY1999, COL524 and COL9835. These substitutions were all synonymous. No unique mutations were found in NS5, but our sequences shared similar differences to the WNV-Texas genotype and other WNV genotypes (Table 3). The southeastern coastal Texas genotype includes several isolates collected in 2002 from Texas, and is considered to have an attenuated phenotype with a small plaques (sp) size, temperature sensitivity (ts), reduced replication in cell culture and reduced neuroinvasiveness that is dose-dependent [21]. The southeastern coastal Texas genotype has not been detected since 2002, suggesting its possible extinction [3]. In Colombia, previous work characterizing WNV-strains isolated from flamingoes (COL524/COL9835) showed a close genetic relationship with WNV strains isolated in Louisiana in 2001 and the NY99 strain (Fig. 2), but in vitro phenotypic characterization showed differences with the attenuated Texas-genotype. In fact, COL524/COL9835 has high virulence in chicken eggs and newborn/4-week-old Balb/c mice [12].

Fig. 2
figure 2

Phylogenetic tree estimated by Bayesian analysis of 46 sequence strains (NS5-Envelope) of West Nile virus data under the GTR + G model of nucleotide substitution. The number accession of Genbank is followed by abbreviated name of WNV-strain. The branches in red belong to characterized previously by Osorio et al. (2012) and characterized by ours

Table 3 Polymorphic sites between envelope and NS5 sequences belonging to southeastern coastal Texas genotype and our samples from Northern of Colombia (Córdoba, San Bernardo del Viento)

Our results demonstrate genetic diversity of WNV strains circulating in Colombia. The presence of the attenuated coastal Texas genotype could explain, in part, the lack of human and equine cases detected. Previous work has shown WNV-seropositivity in horses from the Caribbean region with no disease reported [711]. Different WNV genotypes may converge along migratory bird flyways, which pass from the U.S. to Mexico, Colombia and Venezuela, Caribbean Islands, and other parts of South America south to Argentina. The high diversity of migratory birds in certain areas of Caribbean Colombia [3, 4, 12, 22] may be important for WNV maintenance.

Finally, our results indicate a close evolutionary relationship with the attenuated coastal Texas genotype requires further studies in cell culture and animal models to confirm the attenuated phenotype. Additional surveillance focused on avian and mosquito fauna is also needed to obtain more isolates of WNV in conserved Colombian ecosystems to further examine the genetic diversity of WNV and possible strain dissemination to another geographic areas.