West Nile Virus Prevalence in Wild Birds from Mexico Open Access

West Nile virus (WNV, Flavivirus) is maintained in a natural transmission cycle primarily between Culex spp. mosquitoes and birds, with other vertebrates as incidental hosts (Reed et al. 2003; van der Meulen et al. 2005; Chancey et al. 2015). It is recognized as a major etiology of bird mortality, with the order Passeriformes being the most susceptible, and cloacal and oral shedding of virions can be detected in most bird species when infected naturally or experimentally (Komar et al. 2002, 2003). Migratory birds are the main geographic dispersers of WNV (Reed et al. 2003; Dusek et al. 2009).

West Nile virus was first detected in the Americas in New York in 1999, and within 3 yr, the virus had spread throughout the US (where it is now considered an endemic disease) moving through Canada, Mexico, Central America, the Caribbean, and down to Argentina by 2006 (Chancey et al. 2015). The clade introduced in 1999, genotype NY99, was partially substituted by genotype WN02 in 2002 (Ebel et al. 2004), and new genotypes (SW/WN03 and WN06) emerged in 2005 and 2006, respectively (McMullen et al. 2011; Añez et al. 2013).

In Mexico, WNV was first detected in 2002 in resident birds and horses in states bordering the US and along the coast of the Gulf of Mexico (Estrada-Franco et al. 2003). Virus or antibody detection has been reported in several domestic and wild animals and in humans across the country (Ulloa et al. 2009; Rios-Ibarra et al. 2010; Chaves et al. 2016); however, it is widely considered to be an underreported disease. The objective of this study was to assess the prevalence of WNV in a sample of birds from Mexico.

Samples were selected for a retrospective analysis of WNV prevalence from a collection originally used for an avian influenza study. The selected samples were from 200 birds (98 resident, 102 migratory) collected from September 2007 to March 2008 in 11 Mexican states (Table 1 and Fig. 1). Samples were oral and cloacal swabs pooled per individual in viral transport media supplemented with antibiotics and stored at –80 C until their analysis in the laboratory following a previously described technique (Komar et al. 2002).

We extracted RNA from samples (RNeasy Kit, Qiagen, Germantown, Maryland, USA) and reverse transcription (RT-)PCR was performed to detect WNV (One-Step RTPCR Kit, Qiagen, Valencia, California, USA) following protocols previously published to amplify a 408-base pair fragment and posterior nested PCR amplification of 104-base pair fragment spanning portions of the C and prM genes (Shi et al. 2001). A Student's t-test was used to determine the statistical differences in WNV prevalence between migratory and resident birds at α=0.05.

The RNA from one of the positive samples was used to amplify a fragment of the prM-E genes of WNV using previously described primers (Beasley et al. 2003), at an alignment temperature of 54 C, and sequenced (termination chemistry, BigDye Terminator, Applied Biosystems, Foster City, California, USA). For genotype classification analysis, five sequences representative of North American WNV genotypes—WN99, WN02, SW/WN03, and MW/WN06 (Añez et al. 2013)— were retrieved from GenBank. A similarity search was performed with BLAST (National Center for Biotechnology Information 2017) using a filtering query for sequences from Mexico; sequences classified as primary isolation and that spanned most of the sequence obtained in this study (total 17) were used for phylogenetic comparison. Sequences were aligned by ClustalW (Kumar et al. 2016) and manually edited to the same length. The phylogenetic analysis was conducted with Bayesian inference (MrBayes version 3.2.5 software; Huelsenbeck and Ronquist 2001) by the TIM2ef evolutionary model (Fig. 2A) or TrNef (Fig. 2B) selected by best fit with jModelTest version 0.1.1 (Posada 2008) and Kunjin virus as outgroup. The phylogenetic analysis was run for 10⁶ generations, sampling every 1,000 generations. Bayesian posterior probability values were calculated. Nodes were significantly supported if posterior probabilities were ≥0.95. Trees were visualized in FigTree version 1.4.3 (Rambaut and Drummond 2010).

Sixteen of 200 samples (8%) were positive for WNV (Table 1 and Fig. 1). The positive birds were from the states of Oaxaca, Tamaulipas, and Chiapas, with prevalences of 23, 17, and 10%, respectively. The prevalence by bird species was 33% for Ruby-throated Hummingbirds (Archilochus colubris), 33% for Double-crested Cormorants (Phalacrocorax auritus), 28% for Ring-billed Gulls (Larus delawarensis), and 27% for Mourning Doves (Zenaida macroura). The Ruby-throated Hummingbird and Ring-billed Gull are considered migratory species, whereas the Mourning Dove and Double-crested Cormorant are resident species (Pollet et al. 2012; Weidensaul et al. 2013). There was no significant difference in WNV prevalence between the migratory and resident birds (t=0.3956, df=198, P=0.69).

A partial sequence of the prM-E gene was obtained from one of the Double-crested Cormorant samples from Oaxaca state (sample OAX193, GenBank no. JX444751.1, 779 nucleotides). Analysis of the partial sequence of prM-E genes strongly suggested that sample OAX193 belonged to genotype NY99 (Fig. 2A). The sequence from sample OAX193 was found to be 99% identical to all selected sequences of Mexican origin, and no sequence on the GenBank database showed 100% nucleotide identity. Phylogenetic reconstruction demonstrated a clade organization based on geographic localization, separating sequences of north or south origin (Fig. 2B). Reanalysis of sequence alignments by the maximum likelihood method for phylogeny rendered both trees with similar topology.

Because oral and cloacal swabs were the only samples available in the bank used for this retrospective study, we performed molecular methods to obtain prevalence and genetic information. The use of antibodybased methods can increase epidemiologic sensitivity because it detects past infections; however, it can cross-react with related viral infections, compared with the method used in this study, which can detect only current infections with high specificity. Detection of WNV RNA from cloacal swabs can be equally effective as organ samples of infected birds (Komar et al. 2002).

The prevalence of WNV in birds from Mexico reported in this study was low (8%) compared with reports from countries where WNV is regularly detected, such as Nigeria (26%; van der Meulen et al. 2005) and the US (60%; Dusek et al. 2009). Most published epidemiologic studies for WNV report prevalence with antibody-based methods (Ulloa et al. 2009; Barros et al. 2011; Machain-Williams et al. 2013), typically higher than the PCRbased prevalence studies (including the one reported in our study) because of the lower sensitivity but higher specificity of PCR. Before this report, WNV has been reported in animals in the states of Oaxaca (Loza-Rubio et al. 2016), Chiapas (Ulloa et al. 2009), and Tamaulipas (Fernández-Salas et al. 2003). A study performed in wild birds from south Mexico, also with the use of molecular tools for the detection of the virus, reported a prevalence of 13.5% (Chaves et al. 2016), similar to that described in this article. In the present study, we found that, of the birds with WNV-positive samples, 50% are considered resident and 50% are considered migratory, unlike the results reported in studies of Cuba (66.66% and 33.33%, respectively) and Puerto Rico (10% and 90%, respectively; Dupuis et al. 2005). West Nile virus has been previously detected in more than 300 wild and domestic bird species (Nasci et al. 2013), including the species reported in this study.

The differentiation of WNV genotypes is based on conserved nucleotide substitutions along the genome sequence (McMullen et al. 2011). The sequence obtained from this study, partially covering prM-E genes, shared the highest sequence similarity with sequences that are firmly classified as NY99 genotype (Fig. 2A). Strain OAX193 does not share total identity with any other sequence in the GenBank database; two nucleotide changes distinguished it from its closest relatives. A query of WNV sequences of Mexican origin deposited in GenBank rendered a database of only 43 sequences, of which only 17 are primary isolate sequences that span most of the OAX193 sequence. Phylogenetic analyses of these 17 sequences and the OAX193 strongly suggest a geographic separation of north and south sequences (Fig. 2B); however, sequences from other states and spanning a greater range of years are needed for a more confident geographic classification.

This study provides evidence that WNV circulated in wild birds in Chiapas, Oaxaca, and Tamaulipas in 2008; however, our analysis cannot confirm that these birds were shedding infectious virions. Extensive surveillance of wild and domestic animals is needed to increase the epidemiologic database of WNV in Mexico and understand the natural history of this virus.

A.B.-G. was supported by fellowships from the Consejo Nacional de Ciencia y Tecnología, and Instituto Politécnico Nacional, Mexico. A.A.V.-A. was supported partially by grant PINV11-15 from the Instituto de Ciencia y Tecnología of Mexico. We acknowledge Dolores Hernández-Rodríguez, Universidad Autónoma Metropolitana, Mexico, for her support on statistical analysis. Also, we are extremely grateful to Claudia E. Chacón Zendejas for collection and preparation of field samples with hunters and other collaborators. Specimens were collected under the Mexican Federal Government permit SGPA/DGVS/00422/09 license FAUT-0016.