ABSTRACT. A potential in vitvo model of varicella-zoster virus (VZV) latency was developed. Dissociated human dorsal root ganglion cultures were infected with VZV and maintained for 1 wk in the presence of bromovinyl arabinosy1 uracil, a potent inhibitor of VZV. Seven to 21 d after removing the inhibitor (214 d after infection), the cells were trypsinized, passed to monolayers of human embryonic lung fibroblasts, and observed for VZV reactivation as indicated by typical cytopathic effects and the appearance of VZV antigens. VZV reactivated from 56% of the cultures containing both neurons and satellite cells but not from cultures specifically enriched for either neurons, satellite cells, or ganglion-derived fibroblasts. The failure to isolate VZV from cell suspensions that were sonicated before cocultivation with fibroblasts indicated that infectious VZV was not present before reactivation. Moreover, immunohistochemical and immunoprecipitation studies revealed no VZV-specific antigens in any cultures before the reactivation stimulus. VZV antigens were detected after trypsinization and cocultivation. These findings suggest that cultures containing both neurons and satellite cells provide a model system for VZV persistence that possesses many properties of a latent infection. (Pediatr Res 32: 699-703,1992) Abbreviations VZV, varicella-zoster virus HSV, herpes simplex virus NGF, nerve growth factor HELF, human embryonic lung fibroblast CPE, cytopathic effect DRG, dorsal root ganglion DMEM, Dulbecco's modified essential medium EMEM, Eagle's minimal essential medium FUDR, fluorodeoxyuridine BVaraU, bromovinyl arabinosyl uracil pfu, plaque forming units VZV, which produces latent infection of human DRG during primary infection (varicella), can reactivate many years later to produce herpes zoster. The sites of latency were initially assumed to be the DRG and trigeminal ganglia, on the basis of the characteristic dermatomal distribution of herpes zoster (1-3) and the detection of VZV within the ganglia during the acute phase of herpes zoster (4, 5). However, numerous attempts to recover VZV from DRG and trigeminal ganglia of adult cadavers have all failed, in contrast to the frequent recovery of HSV from trigeminal and sacral ganglia (6, 7).These sites of VZV latency were established, finally, by detecting VZV nucleic acids in human ganglia harvested at autopsy from individuals who died without active infection (8-1 1). However, in situ hybridization studies using VZV-specific probes have yielded conflicting data concerning the cellular location of the latent VZV and the prevalence of infected cells within these ganglia (8,10,12,13).A model of VZV latency was approached in adult rats by S.C. paraspinous injection of virus and its subsequent detection in neurons by in situ hybridization. VZV was also reported to be reactivated from ganglion cells from these rats upon trypsinization and cocultivation with permissive fibroblasts (14). These findings seem surprising because the rat is not readily infected by VZV. Moreover...