2',3'-Dideoxyuridine (ddU) is ineffective at controlling human immunodeficiency virus type 1 (HIV-1) infection in human T cells, because it is not biotransformed to the active 5'-triphosphate. The metabolic block resides in the poor substrate affinity of ddU for cellular nucleoside kinases. This problem cannot be overcome by supplying the preformed nucleotides, because such compounds are unable to penetrate cells. To circumvent the requirement of ddU for enzymic phosphorylation, we have prepared bis(pivaloyloxymethyl) 2',3'-dideoxyuridine 5'-monophosphate (piv2 ddUMP), as a potential membrane-permeable prodrug of ddUMP, and investigated its metabolism and anti-HIV activity in two human T cell lines, one with wild-type thymidine kinase activity (MT-4) and the other deficient in thymidine kinase activity (CEM-tk-). The 5'-mono-, di-, and triphosphates of ddU were formed in both cell lines after exposure to piv2-ddUMP. In contrast, phosphorylated metabolites were not observed in cells treated with ddU or ddUMP alone. piv2-ddUMP also reduced the cytopathic effects of HIV-1 in MT-4 cells (ED50, 4.75 microM) and inhibited virus production in culture fluid (ED50, 20 microM). In addition, piv2-ddUMP protected CEM-tk- cells from HIV-1 infection, as demonstrated by inhibition of intracellular p24 antigen levels (ED50, 3 microM) and reverse transcriptase activity in culture medium (Ed50, 2.5 microM). Based on these findings, we propose that the "masked nucleotide" strategy may make available for development nucleoside analogues hitherto considered inactive because of failure to undergo biotransformation to the corresponding 5'-monophosphates. Moreover, by circumventing metabolic dependency on nucleoside kinases, the strategy may overcome acquired resistance to nucleoside analogues caused by the loss or depletion of nucleoside kinases.