D. L. Horton, L. M. McElhinney, D. A. Marston, J. L. N. Wood, C. A. Russell, N. Lewis, I. V. Kuzmin, R. A. M. Fouchier, A. D. M. E. Osterhaus, A. R. Fooks, D. J. Smith

All lyssaviruses cause fatal encephalitis in mammals. There is sufficient antigenic variation within the genus to cause variable vaccine efficacy, but this variation is difficult to characterize quantitatively: sequence analysis cannot yet provide detailed antigenic information, and antigenic neutralization data have been refractory to high-resolution robust interpretation. Here, we address these issues by using state-of-the-art antigenic analyses to generate a high-resolution antigenic map of a global panel of 25 lyssaviruses. We compared the calculated antigenic distances with viral glycoprotein ectodomain sequence data. Although 67% of antigenic variation was predictable from the glycoprotein amino acid sequence, there are in some cases substantial differences between genetic and antigenic distances, thus highlighting the risk of inferring antigenic relationships solely from sequence data at this time. These differences included epidemiologically important antigenic differences between vaccine strains and wild-type rabies viruses. Further, we quantitatively assessed the antigenic relationships measured by using rabbit, mouse, and human sera, validating the use of nonhuman experimental animals as a model for determining antigenic variation in humans. The use of passive immune globulin is a crucial component of rabies postexposure prophylaxis, and here we also show that it is possible to predict the reactivity of immune globulin against divergent lyssaviruses.