All work involving infectious YFV was performed in biosafety level 3 facilities at the University of Texas Medical Branch. Three human isolates of YFV were used: CENETROP-322 (La Paz Department, Bolivia, 1999), Jimenez (Panama, 1974), and Asibi (Ghana, 1927). To facilitate transmission from a viremic vertebrate, viruses were adapted by serial passage through Syrian golden hamsters (Mesocricetus auratus), following the model of Tesh et al. (4,5); CENETROP-322 and Jimenez were passaged 11 times, and the Asibi strain was passaged 10 times.

The SC strain of Ae. aegypti was started with mosquitoes collected from Santa Cruz, Bolivia, in 2001. Mosquitoes used in this experiment were from laboratory-reared F2-F3 generation. The REX-D strain, an old laboratory colony originally started with mosquitoes collected in Rexville, Puerto Rico and of previously defined susceptibility to YFV infection (6) was used as a control. Mosquitoes were maintained as previously described (7).

Three hamsters were injected intraperitoneally (IP) with 100 µL of clarified liver homogenate, which contained approximately 10⁶ log₁₀ tissue culture infectious dose 50% (TCID₅₀/mL) of each YFV strain. Three days after infection, when viremia levels have been shown to peak (4), hamsters were anesthetized (50 mg Pentobarb/kg IP) and simultaneously exposed to 10-day-old Ae. aegypti SC or REX-D mosquitoes for 1 h. Fully engorged mosquitoes in each group were placed in separate cages and incubated for 15 days at 28°C and 80% relative humidity on a diet of 10% sucrose. Hamster blood samples were collected immediately afterward and stored at –80°C for viral assay.

Susceptibility to YFV infection is highly variable in mosquitoes from different locations (10–12) and may be influenced by selection (6) and colonization (13). Although the use of Bolivian mosquitoes with few laboratory-reared generations compromised our ability to use large numbers, obtaining competence data as possibly representative of wild, noncolonized, mosquitoes was important. Dissemination and transmission by Santa Cruz Ae. aegypti indicate their ability to serve as vectors for a Bolivian strain of YFV. A critical component of this study was the use of a hamster model for YFV (4). The high viremia levels in hamsters (4,5) facilitate oral infection of mosquitoes and more closely resemble natural infection than feeding the insects on artificial blood meals. Suckling mice remain useful because of their sensitivity to YFV infection.

We could argue that by passing the virus in hamsters, the virus phenotype may be altered with respect to vector infectivity. However, after equivalent passages, the infectivity of the Bolivian, Panamanian, and African strains differed. The Jimenez strain was highly infectious for Bolivian Ae. aegypti (93.5%), with a relatively high dissemination rate (34.5%). In contrast, the Asibi was relatively noninfectious for the Bolivian mosquitoes. Considering the numbers of mosquitoes and virus strains involved, we cannot conclude that this finding reflects a general trend of incompatibility between South American Ae. aegypti and YFV of African origin. However, the results obtained are in close agreement with the findings by Tabachnick et al. (12). Johnson et al. (14), using Brazilian strains of Ae. aegypti and YFV, reported similar results of 35% infection rates and 25% dissemination rates. Lourenço-de-Oliveira et al. (11,15) observed infection rates from 0% to 48.6% in Brazilian Ae. aegypti infected with Brazilian YFV. In comparison, we found higher infection rates for Panamian and Bolivian YF viruses (63.3%–100%), but this finding may reflect our use of a viremic animal to infect the mosquitoes, whereas Tabachnick et al. (12) and Johnson et al. (14) used artificially prepared blood meals. Our results also demonstrate that passaging YFV in hamsters does not compromise the ability of the virus to infect mosquitoes and that the hamster model is useful to study mosquito competence for YFV.

In conclusion, our results do not support the hypothesis that Bolivian strains of YFV cannot infect Bolivian Ae. aegypti and demonstrate that the recolonizing (after 1980) South American strains of Ae. aegypti are potential YFV vectors. The reason urban YF epidemics have not yet occurred in South America, including in the city of Santa Cruz, Bolivia, where some cases were recently reported within the city limits (3), is still unknown. The mosquito infection rates observed in our study were higher that those reported by Lourenço-de-Oliviera et al. (11), and we also demonstrated YFV transmission (albeit at a low level). Thus, if YFV were to be reintroduced into urban areas of South America, a transmission cycle could possibly be established. The absence of epidemic YF may be the result of other factors, including widespread deforestation and less opportunity for YFV to move out of the sylvatic cycle, better mosquito control, the local population's YFV vaccine status, and, possibly, heterologous antibodies to other flaviviruses such as dengue (5).