2',3'-Didehydro-2',3'-dideoxythymidine (D4T) is a potent inhibitor of human immunodeficiency virus (HIV), with low hematological toxicity. In the present study, the cellular pharmacology of D4T was investigated in human bone marrow cells (BMC), in an attempt to understand the mechanism of the observed low bone marrow toxicity. After exposure of human BMC to 10 microM [3H]D4T for 24 hr, D4T-5'-triphosphate (D4T-TP) was the predominant metabolite, reaching a concentration of 0.3 pmol/10(6) cells. The D4T-5'-monophosphate levels were slightly lower, whereas the D4T-5'-diphosphate levels were about 6-fold lower than those of D4T-TP at 24 hr. Nucleic acids of human BMC exposed to 10 microM [3H]D4T for 24 hr were purified and analyzed by cesium sulfate density gradient centrifugation. No radioactivity was detected in the RNA region, whereas a limited amount was associated with the DNA region. The amount of label incorporated into DNA correlated with the extracellular D4T concentration and the length of incubation time. Enzymatic hydrolysis of radiolabeled DNA and subsequent analysis by high performance liquid chromatography demonstrated incorporation of both D4T and thymidine (dThd) into DNA. Degradation of D4T to thymine and subsequent formation of labeled dThd was also detected in human BMC. Pulse (24 hr)-chase (48 hr) experiments with 10 microM [3H]D4T demonstrated that the amount of radiolabel from D4T in DNA decreased over time during the chase. Under similar conditions, [3H]3'-azido-3'-deoxythymidine (AZT) incorporated into DNA of human BMC did not decrease during the chase. Although D4T-TP standard was demonstrated to be unstable at 37 degrees and neutral pH, D4T was much more stable in solution when incorporated into newly synthesized DNA isolated from human BMC, suggesting that enzymatic excision may be the mechanism for D4T removal from DNA. In summary, although higher concentrations of D4T-TP, compared with AZT-5'-triphosphate, are observed in human BMC, after exposure of cells to similar extracellular concentrations of parent drug, steady state levels of D4T incorporated into DNA are 10-50-fold lower, compared with AZT. Competition with dTTP formed by D4T metabolism and excision of D4T from DNA may be responsible, in part, for these effects. This study further demonstrates that incorporation of 2',3'-dideoxynucleosides into nuclear DNA of human BMC may be related to the ability of these anti-HIV agents to induce hematological side effects.