The results of inoculation of birds with WNV strains of Mexican origin are difficult to compare with results of published WNV infection studies because of variation in bird species, virus strains, and inoculation method (e.g., needle vs. mosquito bite). House sparrows from Mexico that were needle inoculated with the Tabasco strain were almost as infectious as house sparrows originating in the United States that were needle inoculated with the WNV NY99 strain but 3-fold less infectious than mosquito-inoculated sparrows from the United States (9
). Responses of adult chickens to needle inoculation with strains from Mexico and the NY99 strain were similar (10
). Chickens and pigeons consistently failed to become infectious regardless of infection method or virus strain (11,12
). Common grackles (Quiscalus quiscula
) inoculated by mosquito bite with NY99 have been found to be more competent amplifying hosts than American robins (Turdus migratorius
) in concordance with the findings of our studies that used the 2 strains from Mexico in species of these same genera. However, the American robin was a moderately competent host for NY99, whereas the clay-colored thrush was minimally competent for the Tabasco strain. The 20-fold greater competence of the clay-colored thrush for the Tecate strain was still relatively less than competence for its counterpart from the United States for the NY99 strain and is somewhat contrived because the clay-colored thrush is absent from northwest Mexico where the Tecate strain was isolated. The observed competence of the great-tailed grackle for the Tabasco strain is about as high as or higher than that observed for any species regardless of strain or inoculation method (8
). This species is likely a major reservoir throughout its range from the central plains of the United States south throughout Mexico and Central America and into Colombia.
Observed differences in quantitative viremia measurements and calculated competence index values between the 2 virus strains from Mexico could not be substantiated by using statistical tests on direct comparisons of like measurements. However, binomial statistics applied across measurement categories did indicate a significant difference. The Tecate strain generally seemed to be more virulent than the Tabasco strain for the avian species examined, except for grackles. One basis for these differing outcomes could be the presence of virulence factors in the viral genome of the Tecate strain that may be absent from the Tabasco strain.
Genetic determinants associated with high titers of viremia in the American crow (Corvus brachyrhynchos
) include a glycosylation site on the E protein (5
) and an amino acid substitution in the NS3 helicase (14
). The original, unpassaged Tabasco strain was shown to contain at least 2 genotypes, 1 of which was determined to have reduced virulence and lacks the glycosylation site on the E protein (5
). However, the passaged Tabasco strain we used to inoculate birds from Mexico retained the E protein glycosylation site. Previous data have indicated the association of glycosylation of the E protein for virulence in mice and hatchling chickens (13
). The absence of this reported virulence factor, as well as the presence of additional potential attenuating mutations in the original unpassaged Tabasco strain, could explain the lack of observed bird deaths in Mexico and Latin America, assuming a widespread circulation of this genotype throughout the region. This attenuated phenotype could be explained by selection for lower virulence in migrating birds (4
). Migrating birds with WNV infections have been shown to maintain their migratory behavior during the viremic phase of infection (17
). Therefore, viremic migrating birds that survive long-distance (i.e., trans-Gulf) migration to the Tabasco region may have contributed to selection for mutant, low-virulence genomes. However, after serial passage in vertebrate cells, the Tabasco strain used in this study could have reverted to a more virulent phenotype. As evidence of this potential, previous studies have demonstrated that repeated passage of WNV (Kunjin) has resulted in attainment of a glycosylated phenotype after as few as 2 passages in Vero cells (18
). The 3 additional amino acid differences (prM-T141I, NS4BV245I, or NS5-I898T) in this Tabasco stock that were identical at these positions to the NY99 avian virulent strain could individually or in combination also impart the enhanced avian virulent phenotype observed in these studies.
Both WNV strains from Mexico were pathogenic, leading to death in birds of 2 species (house sparrows and great-tailed grackles), although the stress of captivity and handling may have exacerbated illness among these birds (as indicated by the death of the 1 sham-inoculated grackle). This result signifies that birds in the tropics are probably dying of WNV infection. Therefore, surveillance of bird deaths from WNV may be useful for early warnings of outbreaks in Mexico, as it has been in the United States (19
), although challenges include lack of public involvement and rapid disappearance of carcasses. The link between surveillance of bird illness and deaths and emerging zoonotic pathogens such as highly pathogenic avian influenza virus (Asian strain subtype H5N1) (20
) suggests that a large effort should be made to investigate bird deaths.
From our data, thrushes do not appear to be amplifying hosts of the WNV Tabasco strain. However, house sparrows and great-tailed grackles are highly competent hosts and susceptible to infection and some associated deaths, suggesting that high rates of WNV transmission in the American tropics is being overlooked. Alternatively, infection rates are not high among sensitive species such as house sparrows and great-tailed grackles, or these species are fed upon by vectors at lower rates than expected. Blood meal identification studies of Culex
spp. mosquitoes have demonstrated that these mosquitoes feed on house sparrows and common grackles (Q. quiscula
) at a frequency lower than expected from the relative abundance of these avian species (21,22
). The same studies report a strong preference of Culex
spp. mosquitoes for blood meals from American robins. Because clay-colored thrushes were not highly competent hosts for strains of WNV from Mexico, preference of infected Culex
spp. mosquito vectors for blood meals from this congener could lead to zooprophylaxis. Relatively high viral loads in tissues of birds infected with Tabasco or Tecate WNV strains (i.e., loads greater than or equal to those of birds infected with WNV NY99) (12
) could result in higher rates of oral transmission to predatory or carrion-eating vertebrates, even if mosquito-borne transmission is less supported in tropical than in temperate regions. For example, feral dogs and cats, raptors, corvids, and other animals may become orally infected by eating WNV-infected birds and carcasses (23–25
). In addition, birds from Mexico that are inoculated with Tabasco and Tecate strains shed infectious WNV from oral and cloacal cavities, as did birds infected with the WNV NY99 strain (12
). Shedding of WNV could serve as an additional source of non–vector-borne transmission.
This study indicates that WNV is probably contributing to deaths of some species of birds in the tropics, where numerous unique bird populations are often geographically isolated because of islands of fragmented habitats. In contrast, high levels of biodiversity, such as those found in the Neotropics, lend themselves to reduced WNV transmission (26,27
). Urban locations are less biodiverse yet colonized with numerous species of birds competent for amplifying WNV, like house sparrows and, even more so, great-tailed grackles. The lack of peridomestic corvid populations in the Neotropics would seemingly contribute to reduced human risk for WNV infection in the region (28
). However, there could be an alternative, less susceptible, super spreader, as with the American robin in the United States.
Aside from ecological explanations of reduced WNV transmission in the tropics, human factors may ultimately explain the lack of an obvious public health problem. Three major human factors stand out in this regard: 1) the high incidence of secondary flavivirus infections, mainly caused by dengue virus holoendemicity, which may cause high levels of cross-reactive flavivirus-reactive antibodies; 2) the low investment in surveillance and diagnostic services because of the lack of confirmed human cases of West Nile neurologic disease, and finally, 3) the inability of arbovirus reference laboratories to use serologic methods to diagnose WNV-induced neurologic illness in persons with circulating heterologous antibodies against flavivirus. This last possibility raises concerns that WNV might indeed cause a substantial amount of disease in Mexico, as it does in the United States, but it might be difficult to detect.